notebook.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 103,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 397
    },
    "colab_type": "code",
    "id": "HEP6rjTRGuWC",
    "outputId": "53c53f88-264e-4135-93f1-28d51a923335"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Requirement already satisfied: pescador in /usr/local/lib/python3.6/dist-packages (2.0.2)\n",
      "Requirement already satisfied: joblib>=0.9 in /usr/local/lib/python3.6/dist-packages (from pescador) (0.13.2)\n",
      "Requirement already satisfied: numpy>=1.9 in /usr/local/lib/python3.6/dist-packages (from pescador) (1.16.4)\n",
      "Requirement already satisfied: pyzmq>=15.0 in /usr/local/lib/python3.6/dist-packages (from pescador) (17.0.0)\n",
      "Requirement already satisfied: decorator>=4.0 in /usr/local/lib/python3.6/dist-packages (from pescador) (4.4.0)\n",
      "Requirement already satisfied: six>=1.8 in /usr/local/lib/python3.6/dist-packages (from pescador) (1.12.0)\n",
      "Requirement already satisfied: museval in /usr/local/lib/python3.6/dist-packages (0.2.1)\n",
      "Requirement already satisfied: scipy in /usr/local/lib/python3.6/dist-packages (from museval) (1.3.0)\n",
      "Requirement already satisfied: soundfile in /usr/local/lib/python3.6/dist-packages (from museval) (0.10.2)\n",
      "Requirement already satisfied: simplejson in /usr/local/lib/python3.6/dist-packages (from museval) (3.16.0)\n",
      "Requirement already satisfied: numpy in /usr/local/lib/python3.6/dist-packages (from museval) (1.16.4)\n",
      "Requirement already satisfied: musdb==0.2.3 in /usr/local/lib/python3.6/dist-packages (from museval) (0.2.3)\n",
      "Requirement already satisfied: jsonschema in /usr/local/lib/python3.6/dist-packages (from museval) (2.6.0)\n",
      "Requirement already satisfied: six in /usr/local/lib/python3.6/dist-packages (from museval) (1.12.0)\n",
      "Requirement already satisfied: cffi>=1.0 in /usr/local/lib/python3.6/dist-packages (from soundfile->museval) (1.12.3)\n",
      "Requirement already satisfied: pyaml in /usr/local/lib/python3.6/dist-packages (from musdb==0.2.3->museval) (19.4.1)\n",
      "Requirement already satisfied: tqdm in /usr/local/lib/python3.6/dist-packages (from musdb==0.2.3->museval) (4.28.1)\n",
      "Requirement already satisfied: stempeg>=0.1.1 in /usr/local/lib/python3.6/dist-packages (from musdb==0.2.3->museval) (0.1.8)\n",
      "Requirement already satisfied: pycparser in /usr/local/lib/python3.6/dist-packages (from cffi>=1.0->soundfile->museval) (2.19)\n",
      "Requirement already satisfied: PyYAML in /usr/local/lib/python3.6/dist-packages (from pyaml->musdb==0.2.3->museval) (3.13)\n"
     ]
    }
   ],
   "source": [
    "!pip install pescador\n",
    "!pip install museval"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 104,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 35
    },
    "colab_type": "code",
    "id": "h40ZAyQD9sXb",
    "outputId": "7bc558ec-e9a6-44bd-c7c4-08554c5fc4bd"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Requirement already satisfied: lmdb in /usr/local/lib/python3.6/dist-packages (0.96)\n"
     ]
    }
   ],
   "source": [
    "!pip install lmdb"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "slOot9JrG1FK"
   },
   "outputs": [],
   "source": [
    "import os\n",
    "import time\n",
    "import math \n",
    "import random\n",
    "import librosa\n",
    "import librosa.display\n",
    "import lmdb\n",
    "\n",
    "import numpy as np\n",
    "from torch.utils import data\n",
    "\n",
    "import glob\n",
    "from scipy.io.wavfile import read as wavread #TODO needs to check how to make the reader faster\n",
    "import pescador\n",
    "import torch\n",
    "import torch.nn as nn\n",
    "import torch.nn.functional as F  \n",
    "import torch.optim as optim\n",
    "from torch.autograd import grad, Variable\n",
    "from torchsummary import summary  \n",
    "import logging \n",
    "\n",
    "import matplotlib.pyplot as plt\n",
    "import matplotlib.animation as animation\n",
    "import matplotlib\n",
    "import copy\n",
    "from IPython.display import HTML\n",
    "\n",
    "from tqdm import tqdm_notebook as tqdm"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "DyA4N5jdIhh-"
   },
   "source": [
    "## Config"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "yrCdycz2HFBK"
   },
   "outputs": [],
   "source": [
    "#############################\n",
    "# Model Params\n",
    "#############################\n",
    "model_prefix = 'exp_musdb_1_wide_unpaired_ralsgan_4'#'exp_large_1' # name of the model to be saved\n",
    "n_iterations = 100000\n",
    "# for the cyclic gan use these param with ncritic 1\n",
    "# but for training wavegan model it is better to use ttur lr_g = 1e-4 lr_d=3e-4 and n_critic=1\n",
    "lr_g = 2e-4\n",
    "lr_d = 2e-4\n",
    "beta1 = 0.5\n",
    "beta2 = 0.9\n",
    "decay_lr = False\n",
    "generator_batch_size_factor = 1 # in some cases we might try to update the generator with double batch size used in the discriminator\n",
    "n_critic = 5 # update generator every n_critic steps \n",
    "# gradient penalty regularization factor.\n",
    "p_coeff = 10\n",
    "batch_size = 10\n",
    "noise_latent_dim = 100 \n",
    "model_capacity_size = 64 # reduce the capacity to 32 for faster training also in case of generating larger window size"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "FEeBC1QvQh0h"
   },
   "outputs": [],
   "source": [
    "# rate of storing validation and costs params\n",
    "store_cost_every = 300\n",
    "progress_bar_step_iter_size = 400"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "iZF94hUxIqjO"
   },
   "outputs": [],
   "source": [
    "# backup params\n",
    "take_backup = True\n",
    "backup_every_n_iters = 1000\n",
    "save_samples_every = 1000 \n",
    "# folder including data under each folder train, valid and test subfolders\n",
    "# code can be updated to use mancini dataset piano and noise\n",
    "target_signals_dir = 'mancini_piano/piano'\n",
    "other_signals_dir  = 'mancini_piano/noise'\n",
    "\n",
    "output_dir = 'output'"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "qO0VUGnJI_mC"
   },
   "outputs": [],
   "source": [
    "#############################\n",
    "# Audio Reading Params\n",
    "#############################\n",
    "window_length = 16384 #[16384, 32768, 65536] change model_capacity size to 32 in case of slice>16384\n",
    "sampling_rate = 16000\n",
    "normalize_audio = True "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 110,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 35
    },
    "colab_type": "code",
    "id": "tI5ZYc_1JCxH",
    "outputId": "579bf630-0c5b-4d78-c8a4-cc25c0cfc5f0"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Random Seed:  2014\n"
     ]
    }
   ],
   "source": [
    "#############################\n",
    "# Torch Init and seed setting\n",
    "#############################\n",
    "cuda = torch.cuda.is_available()\n",
    "device = torch.device(\"cuda:0\" if (torch.cuda.is_available()) else \"cpu\")\n",
    "# update the seed\n",
    "manual_seed = 2014 #@param {type: \"integer\"}\n",
    "#manualSeed = random.randint(1, 10000) # use if you want new results\n",
    "print(\"Random Seed: \", manual_seed)\n",
    "random.seed(manual_seed)\n",
    "torch.manual_seed(manual_seed)\n",
    "np.random.seed(manual_seed)\n",
    "if cuda:\n",
    "    torch.cuda.manual_seed(manual_seed)\n",
    "    torch.cuda.empty_cache()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "QPKhpO3rJCOr"
   },
   "outputs": [],
   "source": [
    "#############################\n",
    "# Logger init\n",
    "#############################\n",
    "LOGGER = logging.getLogger('wavegan')\n",
    "LOGGER.setLevel(logging.DEBUG)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "BLMhCmpbHeJm"
   },
   "source": [
    "## Utils"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "s9HuhlS9ozMA"
   },
   "source": [
    "### File Utils"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "vm2BniJ6HQsS"
   },
   "outputs": [],
   "source": [
    "def get_recursive_files(folderPath,ext):\n",
    "\t\tresults  = os.listdir(folderPath)\n",
    "\t\toutFiles = [] \n",
    "\t\tfor file in results:\n",
    "\t\t\tif os.path.isdir(os.path.join(folderPath,file)):\n",
    "\t\t\t\toutFiles+=get_recursive_files(os.path.join(folderPath,file),ext)\n",
    "\t\t\telif file.endswith(ext) :\n",
    "\t\t\t\toutFiles.append(os.path.join(folderPath,file))\n",
    "\t\t\t\n",
    "\t\treturn outFiles\n",
    "\n",
    "def make_path(output_path):\n",
    "    if not os.path.isdir(output_path):\n",
    "        os.makedirs(output_path)\n",
    "    return output_path"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "K1tRItDco3bK"
   },
   "source": [
    "###  Wav files utils"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "RJG0ti9RG1sp"
   },
   "outputs": [],
   "source": [
    "#Fast loading used with wav files only of 8 bits\n",
    "def load_wav(wav_file_path, fast_loading=False ):\n",
    "    try:\n",
    "        if fast_loading:\n",
    "            file_sampling_rate, audio_data = wavread(wav_file_path)\n",
    "            if file_sampling_rate is not None and sampling_rate != file_sampling_rate:\n",
    "                raise NotImplementedError('Scipy cannot resample audio.')\n",
    "            if audio_data.dtype == np.int16:\n",
    "                audio_data = audio_data.astype(np.float32)\n",
    "                audio_data /= 32768.\n",
    "            elif audio_data.dtype == np.float32:\n",
    "                audio_data = np.copy(audio_data)\n",
    "            else:\n",
    "                raise NotImplementedError('Scipy cannot process atypical WAV files.')\n",
    "            \n",
    "        else:\n",
    "            audio_data, _ = librosa.load(wav_file_path, sr=sampling_rate)\n",
    "\n",
    "        if normalize_audio:\n",
    "            # Clip magnitude\n",
    "            max_mag = np.max(np.abs(audio_data))\n",
    "            if max_mag > 1:\n",
    "                audio_data /= max_mag\n",
    "    except Exception as e:\n",
    "        print(wav_file_path)\n",
    "        print(str(e))\n",
    "        LOGGER.error(\"Could not load {}: {}\".format(wav_file_path, str(e)))\n",
    "        raise e\n",
    "    audio_len = len(audio_data)\n",
    "    if audio_len < window_length:\n",
    "        pad_length = window_length - audio_len\n",
    "        left_pad = pad_length // 2\n",
    "        right_pad = pad_length - left_pad \n",
    "        audio_data = np.pad(audio_data, (left_pad, right_pad), mode='constant')\n",
    "        \n",
    "    \n",
    "    return audio_data.astype('float32')\n",
    "\n",
    "\n",
    "def save_samples(epoch_samples, epoch ,prefix=''):\n",
    "    \"\"\"\n",
    "    Save output samples.\n",
    "    \"\"\"\n",
    "    sample_dir = make_path(os.path.join(output_dir, str(epoch)))\n",
    "\n",
    "    for idx, sample in enumerate(epoch_samples):\n",
    "        output_path = os.path.join(sample_dir, \"{}_{}.wav\".format(prefix,idx+1))\n",
    "        sample = sample[0]\n",
    "        librosa.output.write_wav(output_path, sample, sampling_rate)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "offZXw-rHWzw"
   },
   "outputs": [],
   "source": [
    "def sample_audio(audio_data, start_idx=None, end_idx=None):\n",
    "    audio_len = len(audio_data)\n",
    "    if audio_len == window_length:\n",
    "        # If we only have a single 1*window_length audio, just yield.\n",
    "        sample = audio_data\n",
    "    else:\n",
    "        # Sample a random window from the audio\n",
    "        if start_idx is None or end_idx is None:\n",
    "            start_idx = np.random.randint(0, (audio_len - window_length) // 2)\n",
    "            end_idx = start_idx + window_length\n",
    "        sample = audio_data[start_idx:end_idx]\n",
    "    sample =  sample.astype('float32')\n",
    "    assert not np.any(np.isnan(sample))\n",
    "    return sample, start_idx, end_idx\n",
    "\n",
    "def audio_generator(audio_data):\n",
    "    audio_len = len(audio_data)\n",
    "    n_iters = audio_len // window_length\n",
    "    for i in range(n_iters+1):\n",
    "        start_idx = i * window_length\n",
    "        end_idx = start_idx  + window_length\n",
    "        result = np.zeros(window_length)\n",
    "        audio_size = audio_data[start_idx:end_idx].shape[0]\n",
    "        result[:audio_size] = audio_data[start_idx:end_idx]\n",
    "        yield result\n",
    "    \n",
    "    \n",
    "def sample_buffer(buffer_data, start_idx=None, end_idx=None):\n",
    "    audio_len = len(buffer_data) // 4\n",
    "    if audio_len == window_length:\n",
    "        # If we only have a single 1*window_length audio, just yield.\n",
    "        sample = buffer_data\n",
    "    else:\n",
    "        # Sample a random window from the audio\n",
    "        if start_idx is None or end_idx is None:\n",
    "            start_idx = np.random.randint(0, (audio_len - window_length) // 2)\n",
    "            end_idx = start_idx + window_length\n",
    "        sample = buffer_data[start_idx * 4:end_idx*4]\n",
    "    return sample, start_idx, end_idx\n",
    "\n",
    "\n",
    "def wav_generator(file_path, mixing_signal_path):\n",
    "    audio_data = load_wav(file_path)\n",
    "    mixing_data = load_wav(mixing_signal_path)\n",
    "    while True:\n",
    "        sample, start_idx, end_idx = sample_audio(audio_data)\n",
    "        mixing_sample, _, _ = sample_audio(mixing_data,start_idx, end_idx)\n",
    "\n",
    "        mixing_ratio = np.random.uniform(0, 1)\n",
    "        mixed_signal = mixing_ratio * mixing_sample + (1-mixing_ratio) * sample\n",
    "\n",
    "        yield {'single':sample, 'mixed':mixed_signal}\n",
    "\n",
    "def create_stream_reader(single_signal_file_list, other_signal_file_list):\n",
    "    data_streams = []\n",
    "    other_signal_len  = len(other_signal_file_list)\n",
    "    for audio_path in single_signal_file_list:\n",
    "        other_signal_indx = np.random.randint(0,other_signal_len)\n",
    "        stream = pescador.Streamer(wav_generator, audio_path, other_signal_file_list[other_signal_indx])\n",
    "        data_streams.append(stream)\n",
    "    mux = pescador.ShuffledMux(data_streams)\n",
    "    batch_gen = pescador.buffer_stream(mux, batch_size)\n",
    "    return batch_gen"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "NKeUFiF0o-Jh"
   },
   "source": [
    "### Model Utils"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "5-YmpyOZHNk8"
   },
   "outputs": [],
   "source": [
    "\n",
    "def sample_noise(size):\n",
    "    z = torch.FloatTensor(size,noise_latent_dim).to(device)\n",
    "    z.data.normal_() # generating latent space based on normal distribution\n",
    "    return z\n",
    "\n",
    "def weights_init(m): \n",
    "    if isinstance(m, nn.Conv1d):\n",
    "        m.weight.data.normal_(0.0, 0.02)\n",
    "        if m.bias is not None:\n",
    "            m.bias.data.fill_(0)\n",
    "        m.bias.data.fill_(0)\n",
    "    elif isinstance(m, nn.Linear):\n",
    "        m.bias.data.fill_(0)\n",
    "\n",
    "def update_optimizer_lr(optimizer,lr,decay):\n",
    "  for param_group in optimizer.param_groups:\n",
    "      param_group['lr'] = lr* decay   \n",
    "\n",
    "\n",
    "def gradients_status(model,flag):\n",
    "    for p in model.parameters():\n",
    "        p.requires_grad = flag"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "1Qcmv23crtN_"
   },
   "outputs": [],
   "source": [
    "\n",
    "# used to replay some history generate to the discriminator to avoid greedy disc.\n",
    "class ReplayBuffer():\n",
    "    def __init__(self, max_size=50):\n",
    "        assert (max_size > 0), 'Empty buffer or trying to create a black hole. Be careful.'\n",
    "        self.max_size = max_size\n",
    "        self.data = []\n",
    "\n",
    "    def push_and_pop(self, data):\n",
    "        to_return = []\n",
    "        for element in data.data:\n",
    "            element = torch.unsqueeze(element, 0)\n",
    "            if len(self.data) < self.max_size:\n",
    "                self.data.append(element)\n",
    "                to_return.append(element)\n",
    "            else:\n",
    "                if random.uniform(0,1) > 0.5:\n",
    "                    i = random.randint(0, self.max_size-1)\n",
    "                    to_return.append(self.data[i].clone())\n",
    "                    self.data[i] = element\n",
    "                else:\n",
    "                    to_return.append(element)\n",
    "        return Variable(torch.cat(to_return))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "lgCxXMPApNo-"
   },
   "source": [
    "### Data Loader and Sampler"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "QKOlZPMOHK6N"
   },
   "outputs": [],
   "source": [
    "def numpy_to_tensor(numpy_array):\n",
    "    numpy_array = numpy_array[:, np.newaxis, :]\n",
    "    return torch.Tensor(numpy_array).to(device)\n",
    "\n",
    "#############################\n",
    "# Creating Data Loader and Sampler\n",
    "#############################\n",
    "class WavDataLoader():\n",
    "    def __init__(self, folder_path, other_signals_folder, audio_extension='wav'):\n",
    "        self.signal_paths = get_recursive_files(folder_path, audio_extension)\n",
    "        self.mixed_wav_files = get_recursive_files(other_signals_folder, audio_extension)\n",
    "        self.data_iter = None\n",
    "        self.initialize_iterator()\n",
    "\n",
    "    def initialize_iterator(self):\n",
    "        data_iter = create_stream_reader(self.signal_paths, self.mixed_wav_files)\n",
    "        self.data_iter = iter(data_iter)\n",
    "\n",
    "    def __len__(self):\n",
    "        return len(self.signal_paths)\n",
    "        \n",
    "\n",
    "\n",
    "    def __iter__(self):\n",
    "        return self\n",
    "\n",
    "    def __next__(self):\n",
    "        x =  next(self.data_iter)\n",
    "        return (numpy_to_tensor(x['single']), numpy_to_tensor(x['mixed']), numpy_to_tensor(x['foreground']))\n",
    "\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "uN8qDopu9sYm"
   },
   "outputs": [],
   "source": [
    "class LMDBWavLoader(data.Dataset):\n",
    "    def __init__(self, lmdb_file_path, is_test=False):\n",
    "        self.env = lmdb.open(lmdb_file_path, max_readers=1, readonly=True, lock=False,\n",
    "                             readahead=False, meminit=False)\n",
    "        self.datum = datanum_pb2.DataNum()\n",
    "        self.is_test = is_test\n",
    "        \n",
    "\n",
    "    def __len__(self):\n",
    "        n_entries =  int(self.env.stat()['entries'])\n",
    "        return n_entries\n",
    "\n",
    "\n",
    "    def __getitem__(self, data_indx):\n",
    "        index = None\n",
    "        audio_indx = None\n",
    "        index = data_indx\n",
    "        with self.env.begin(write=False) as cursor:\n",
    "            raw_datum = cursor.get('{:08}'.format(index).encode('ascii'))\n",
    "        self.datum.ParseFromString(raw_datum)\n",
    "        # float is represented by 4 bytes\n",
    "        start_idx = None\n",
    "        end_idx = None\n",
    "        if self.is_test:\n",
    "            return np.array(np.frombuffer(self.datum.vocals, dtype=np.float32)).reshape(-1),np.array(np.frombuffer(self.datum.mixture, dtype=np.float32)).reshape(-1)\n",
    "                \n",
    "        mixture, start_idx, end_idx = sample_buffer(self.datum.mixture, start_idx, end_idx)\n",
    "        mixture = np.array(np.frombuffer(mixture, dtype=np.float32)).reshape(1,-1)\n",
    "        vocals, _, _ = sample_buffer(self.datum.vocals, start_idx, end_idx)\n",
    "        vocals = np.frombuffer(vocals, dtype=np.float32).reshape(1,-1)\n",
    "        return vocals, mixture"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "vtrJ-eP0pQjS"
   },
   "source": [
    "### Visualize utils"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "ukzdK3k6MTnP"
   },
   "outputs": [],
   "source": [
    "def visualize_audio(audio_tensor, is_monphonic=False):\n",
    "    # takes a batch ,n channels , window length and plots the spectogram\n",
    "    input_audios = audio_tensor.detach().cpu().numpy()\n",
    "    plt.figure(figsize=(18, 50))\n",
    "    for i , audio in enumerate(input_audios):\n",
    "        plt.subplot(10, 2, i+1)\n",
    "        if is_monphonic:\n",
    "            plt.title('Monophonic %i' % (i+1))\n",
    "            librosa.display.waveplot(audio[0], sr=sampling_rate)\n",
    "        else:\n",
    "            D = librosa.amplitude_to_db(np.abs(librosa.stft(audio[0])), ref=np.max)\n",
    "            librosa.display.specshow(D, y_axis='linear')\n",
    "            plt.colorbar(format='%+2.0f dB')\n",
    "            plt.title('Linear-frequency power spectrogram %i' % (i+1))\n",
    "    plt.show()\n",
    "\n",
    "def visualize_loss(loss_1, loss_2, first_legend, second_legend, y_label):\n",
    "    plt.figure(figsize=(10,5))\n",
    "    plt.title(\"{} and {} Loss During Training\".format(first_legend, second_legend))\n",
    "    plt.plot(loss_1,label=first_legend)\n",
    "    plt.plot(loss_2,label=second_legend)\n",
    "    plt.xlabel(\"iterations\")\n",
    "    plt.ylabel(y_label)\n",
    "    plt.grid(True)\n",
    "    plt.tight_layout()\n",
    "    plt.legend()\n",
    "    plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "PxHweiv2pW-5"
   },
   "outputs": [],
   "source": [
    "def latent_space_interpolation(model, n_samples=10 ):\n",
    "    z_test = sample_noise(2)\n",
    "    with torch.no_grad():  \n",
    "        interpolates = []\n",
    "        for alpha in np.linspace(0, 1, n_samples):\n",
    "            interpolate_vec = alpha * z_test[0] + ((1 - alpha) * z_test[1])\n",
    "            interpolates.append(interpolate_vec)     \n",
    "            \n",
    "        interpolates = torch.stack(interpolates)\n",
    "        generated_audio = model(interpolates)\n",
    "    visualize_audio(generated_audio, True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 123,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 35
    },
    "colab_type": "code",
    "id": "7GGWmjTDIyIv",
    "outputId": "3e23f00e-b246-4f35-d9f9-46a27158950f"
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "'output'"
      ]
     },
     "execution_count": 123,
     "metadata": {
      "tags": []
     },
     "output_type": "execute_result"
    }
   ],
   "source": [
    "if os.path.isdir(output_dir):\n",
    "  os.system('rm -r {}'.format(output_dir))\n",
    "make_path(output_dir)\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "3PgUiN6uJYtz"
   },
   "source": [
    "## Data Download and Loading"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 124,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 161
    },
    "colab_type": "code",
    "id": "vuTx8G429sY6",
    "outputId": "c8333871-0e99-4e83-ee10-94021d9d60e8"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Reading package lists... Done\n",
      "Building dependency tree       \n",
      "Reading state information... Done\n",
      "p7zip-full is already the newest version (16.02+dfsg-6).\n",
      "The following package was automatically installed and is no longer required:\n",
      "  libnvidia-common-410\n",
      "Use 'apt autoremove' to remove it.\n",
      "0 upgraded, 0 newly installed, 0 to remove and 4 not upgraded.\n"
     ]
    }
   ],
   "source": [
    "!apt-get install p7zip-full"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "MUSDB download from MUSDB18 corpus for music separation https://zenodo.org/record/1117372\n",
    "\n",
    "```\n",
    "@misc{MUSDB18,\n",
    "  author       = {Rafii, Zafar and\n",
    "                  Liutkus, Antoine and\n",
    "                  Fabian-Robert St{\\\"o}ter and\n",
    "                  Mimilakis, Stylianos Ioannis and\n",
    "                  Bittner, Rachel},\n",
    "  title        = {The {MUSDB18} corpus for music separation},\n",
    "  month        = dec,\n",
    "  year         = 2017,\n",
    "  doi          = {10.5281/zenodo.1117372},\n",
    "  url          = {https://doi.org/10.5281/zenodo.1117372}\n",
    "}\n",
    "```"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "RCG_OJ0wJxr2"
   },
   "outputs": [],
   "source": [
    "\n",
    "if not(os.path.isfile('musdb_.7z')):\n",
    "  os.system('7z x  musdb_.7z')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "Q3E6Oa0o9sZD"
   },
   "outputs": [],
   "source": [
    "from protocol_buffer import  datanum_pb2"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "oEO_FDuOKfId"
   },
   "outputs": [],
   "source": [
    "# loading data using data loader\n",
    "train_dataset = LMDBWavLoader('musdb_train')\n",
    "train_loader = torch.utils.data.DataLoader(train_dataset, batch_size=batch_size,\n",
    "                                        shuffle=True, num_workers=2,drop_last=True,pin_memory=True)\n",
    "val_dataset = LMDBWavLoader('musdb_valid')\n",
    "val_loader = torch.utils.data.DataLoader(val_dataset, batch_size=batch_size,\n",
    "                                        shuffle=True, num_workers=2,drop_last=True,pin_memory=True)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "mvsDhwKeHf4r"
   },
   "source": [
    "## Model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "uu2hKN9kHgsl"
   },
   "outputs": [],
   "source": [
    "class Transpose1dLayer(nn.Module):\n",
    "    def __init__(self, in_channels, out_channels, kernel_size, stride, padding=11, upsample=None, output_padding=1, use_batch_norm = False):\n",
    "        super(Transpose1dLayer, self).__init__()\n",
    "        self.upsample = upsample\n",
    "        reflection_pad = nn.ConstantPad1d(kernel_size // 2, value=0)\n",
    "        conv1d = nn.Conv1d(in_channels, out_channels, kernel_size, stride)\n",
    "        conv1d.weight.data.normal_(0.0, 0.02)\n",
    "        Conv1dTrans = nn.ConvTranspose1d(in_channels, out_channels, kernel_size, stride, padding, output_padding)\n",
    "        batch_norm = nn.BatchNorm1d(out_channels)\n",
    "        if self.upsample:\n",
    "            operation_list = [\n",
    "                reflection_pad,\n",
    "                conv1d  \n",
    "            ]\n",
    "        else:\n",
    "            operation_list = [\n",
    "                Conv1dTrans\n",
    "            ]\n",
    "\n",
    "        if use_batch_norm:\n",
    "            operation_list.append(batch_norm)\n",
    "        self.transpose_ops = nn.Sequential(*operation_list )\n",
    "\n",
    "    def forward(self, x):\n",
    "        if self.upsample:\n",
    "            # recommended by wavgan paper to use nearest upsampling\n",
    "            x = nn.functional.interpolate(x,scale_factor=self.upsample, mode='nearest')\n",
    "        return self.transpose_ops(x)\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "o5U4rCRHHiJp"
   },
   "outputs": [],
   "source": [
    "class Conv1D(nn.Module):\n",
    "    def __init__(self,input_channels, output_channels, kernel_size,alpha = 0.2,shift_factor=2, stride=4, padding=11, use_batch_norm=False, drop_prob = 0):\n",
    "        super(Conv1D, self).__init__()\n",
    "        self.conv1d = nn.Conv1d(input_channels, output_channels, kernel_size, stride=stride, padding=padding)\n",
    "        self.batch_norm = nn.BatchNorm1d(output_channels)\n",
    "        self.phase_shuffle =  PhaseShuffle(shift_factor)\n",
    "        self.alpha = alpha\n",
    "        self.use_batch_norm = use_batch_norm\n",
    "        self.use_phase_shuffle = shift_factor==0\n",
    "        self.use_drop = drop_prob>0\n",
    "        self.dropout = nn.Dropout2d(drop_prob)\n",
    "    \n",
    "    def forward(self, x):\n",
    "        x = self.conv1d(x)\n",
    "        if self.use_batch_norm:\n",
    "            x = self.batch_norm(x)\n",
    "        x = F.leaky_relu(x, negative_slope=self.alpha)\n",
    "        if self.use_phase_shuffle: \n",
    "            x = self.phase_shuffle(x)\n",
    "        if self.use_drop:\n",
    "            x = self.dropout(x)\n",
    "        return x"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "tCZoaeDOHjaP"
   },
   "outputs": [],
   "source": [
    "class PhaseShuffle(nn.Module):\n",
    "    \"\"\"\n",
    "    Performs phase shuffling, i.e. shifting feature axis of a 3D tensor\n",
    "    by a random integer in {-n, n} and performing reflection padding where\n",
    "    necessary.\n",
    "    \"\"\"\n",
    "    # Copied from https://github.com/jtcramer/wavegan/blob/master/wavegan.py#L8\n",
    "    def __init__(self, shift_factor):\n",
    "        super(PhaseShuffle, self).__init__()\n",
    "        self.shift_factor = shift_factor\n",
    "\n",
    "    def forward(self, x):\n",
    "        if self.shift_factor == 0:\n",
    "            return x\n",
    "        # uniform in (L, R)\n",
    "        k_list = torch.Tensor(x.shape[0]).random_(0, 2 * self.shift_factor + 1) - self.shift_factor\n",
    "        k_list = k_list.numpy().astype(int)\n",
    "\n",
    "        # Combine sample indices into lists so that less shuffle operations\n",
    "        # need to be performed\n",
    "        k_map = {}\n",
    "        for idx, k in enumerate(k_list):\n",
    "            k = int(k)\n",
    "            if k not in k_map:\n",
    "                k_map[k] = []\n",
    "            k_map[k].append(idx)\n",
    "\n",
    "        # Make a copy of x for our output\n",
    "        x_shuffle = x.clone()\n",
    "\n",
    "        # Apply shuffle to each sample\n",
    "        for k, idxs in k_map.items():\n",
    "            if k > 0:\n",
    "                x_shuffle[idxs] = F.pad(x[idxs][..., :-k], (k, 0), mode='reflect')\n",
    "            else:\n",
    "                x_shuffle[idxs] = F.pad(x[idxs][..., -k:], (0, -k), mode='reflect')\n",
    "\n",
    "        assert x_shuffle.shape == x.shape, \"{}, {}\".format(x_shuffle.shape,\n",
    "                                                       x.shape)\n",
    "        return x_shuffle"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "UI5KHANTHnq7"
   },
   "source": [
    "### Generator Model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "b_pE1roiHlMJ"
   },
   "outputs": [],
   "source": [
    "class WaveGANGenerator(nn.Module):\n",
    "    def __init__(self, model_size=64, num_channels=1,\n",
    "                  verbose=False,\n",
    "                  upsample=True, slice_len=16384, use_batch_norm = False):\n",
    "        super(WaveGANGenerator, self).__init__()\n",
    "        assert slice_len in [16384, 32768, 65536] # used to predict longer utterances\n",
    "\n",
    "        self.model_size = model_size  # d \n",
    "        self.verbose = verbose\n",
    "        self.use_batch_norm = use_batch_norm\n",
    "\n",
    "        self.dim_mul = 16 if slice_len == 16384 else 32\n",
    "\n",
    "        self.fc1 = nn.Linear(noise_latent_dim, 4*4* model_size * self.dim_mul)\n",
    "        self.bn1 = nn.BatchNorm1d(num_features= model_size * self.dim_mul)\n",
    "\n",
    "        stride = 4\n",
    "        if upsample:\n",
    "            stride = 1\n",
    "            upsample = 4\n",
    "        \n",
    "        deconv_layers = [\n",
    "            Transpose1dLayer( self.dim_mul* model_size, (self.dim_mul* model_size) //2, 25, stride, upsample=upsample,use_batch_norm=use_batch_norm),\n",
    "            Transpose1dLayer((self.dim_mul* model_size) //2, (self.dim_mul* model_size) //4, 25, stride, upsample=upsample,use_batch_norm=use_batch_norm),\n",
    "            Transpose1dLayer( (self.dim_mul* model_size) //4,  (self.dim_mul* model_size) //8, 25, stride, upsample=upsample,use_batch_norm=use_batch_norm),\n",
    "            Transpose1dLayer( (self.dim_mul* model_size) //8,  (self.dim_mul* model_size) //16, 25, stride, upsample=upsample,use_batch_norm=use_batch_norm),\n",
    "        ]\n",
    "        \n",
    "\n",
    "        if slice_len== 16384:\n",
    "            deconv_layers.append( Transpose1dLayer((self.dim_mul* model_size) //16, num_channels, 25, stride, upsample=upsample))\n",
    "        elif slice_len == 32768 :\n",
    "            deconv_layers +=[ \n",
    "                Transpose1dLayer((self.dim_mul* model_size) //16, model_size, 25, stride, upsample=upsample,use_batch_norm=use_batch_norm)\n",
    "                ,Transpose1dLayer(model_size, num_channels, 25, 2, upsample=upsample)\n",
    "            ]\n",
    "        elif slice_len == 65536:\n",
    "            deconv_layers +=[\n",
    "                Transpose1dLayer((self.dim_mul* model_size) //16, model_size, 25, stride, upsample=upsample,use_batch_norm=use_batch_norm)\n",
    "                ,Transpose1dLayer(model_size, num_channels, 25, stride, upsample=upsample)\n",
    "            ]\n",
    "        else:\n",
    "            raise ValueError('slice_len {} value is not supported'.format(slice_len))\n",
    "        \n",
    "\n",
    "        self.deconv_list = nn.ModuleList(deconv_layers)\n",
    "        for m in self.modules():\n",
    "            if isinstance(m, nn.ConvTranspose1d) or isinstance(m, nn.Linear):\n",
    "                nn.init.kaiming_normal_(m.weight.data)\n",
    "\n",
    "    def forward(self, x):\n",
    "        x = self.fc1(x).view(-1, self.dim_mul* self.model_size, 16)\n",
    "        if self.use_batch_norm:\n",
    "            x = self.bn1(x)\n",
    "        x = F.relu(x)\n",
    "        if self.verbose:\n",
    "            print(x.shape)\n",
    "\n",
    "        for deconv in self.deconv_list[:-1]:\n",
    "            x = F.relu(deconv(x))\n",
    "            if self.verbose:\n",
    "                print(x.shape)\n",
    "        output = torch.tanh(self.deconv_list[-1](x))\n",
    "        return output"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "VeGi98DxHqH1"
   },
   "source": [
    "### Discriminator Model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "WL4yeoYXHsNS"
   },
   "outputs": [],
   "source": [
    "class WaveGANDiscriminator(nn.Module):\n",
    "    def __init__(self, model_size=64, ngpus=1, num_channels=1, shift_factor=2,\n",
    "                 alpha=0.2, verbose=False, slice_len=16384, use_batch_norm = False):\n",
    "        super(WaveGANDiscriminator, self).__init__()\n",
    "        assert slice_len in [16384, 32768, 65536] # used to predict longer utterances\n",
    "        \n",
    "        self.model_size = model_size  # d\n",
    "        self.ngpus = ngpus\n",
    "        self.use_batch_norm = use_batch_norm\n",
    "        self.num_channels = num_channels  # c\n",
    "        self.shift_factor = shift_factor  # n\n",
    "        self.alpha = alpha\n",
    "        self.verbose = verbose\n",
    "\n",
    "\n",
    "        conv_layers = [\n",
    "            Conv1D(num_channels, model_size, 25, stride=4, padding=11, use_batch_norm=use_batch_norm, alpha=alpha,shift_factor=shift_factor),\n",
    "            Conv1D(model_size, 2 * model_size, 25, stride=4, padding=11, use_batch_norm=use_batch_norm, alpha=alpha,shift_factor=shift_factor),\n",
    "            Conv1D(2 * model_size, 4 * model_size, 25, stride=4, padding=11, use_batch_norm=use_batch_norm, alpha=alpha,shift_factor=shift_factor),\n",
    "            Conv1D(4 * model_size, 8 * model_size, 25, stride=4, padding=11, use_batch_norm=use_batch_norm, alpha=alpha,shift_factor=shift_factor),\n",
    "            Conv1D(8 * model_size, 16 * model_size, 25, stride=4, padding=11, use_batch_norm=use_batch_norm, alpha=alpha,shift_factor=0 if slice_len==16384 else shift_factor)\n",
    "        ]\n",
    "        self.fc_input_size = 256 * model_size\n",
    "        if slice_len == 32768 :\n",
    "            conv_layers.append(\n",
    "                 Conv1D(16 * model_size, 32 * model_size, 25, stride=2, padding=11,use_batch_norm=use_batch_norm, alpha=alpha,shift_factor=0)\n",
    "            ) \n",
    "            self.fc_input_size = 480 * model_size\n",
    "        elif slice_len == 65536:\n",
    "            conv_layers.append(\n",
    "                 Conv1D(16 * model_size, 32 * model_size, 25, stride=4, padding=11,use_batch_norm=use_batch_norm, alpha=alpha,shift_factor=0)\n",
    "            )\n",
    "            self.fc_input_size = 512 * model_size\n",
    "        \n",
    "        self.conv_layers = nn.ModuleList(conv_layers)\n",
    "        \n",
    "        self.fc1 = nn.Linear(self.fc_input_size, 1)\n",
    "\n",
    "        for m in self.modules():\n",
    "            if isinstance(m, nn.Conv1d) or isinstance(m, nn.Linear):\n",
    "                nn.init.kaiming_normal_(m.weight.data)\n",
    "\n",
    "    def forward(self, x):\n",
    "        for conv in self.conv_layers:\n",
    "            x= conv(x)\n",
    "            if self.verbose:\n",
    "                print(x.shape)\n",
    "        x = x.view(-1, self.fc_input_size)\n",
    "        if self.verbose:\n",
    "            print(x.shape)\n",
    "\n",
    "        return self.fc1(x)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "naAboFqLq9-m"
   },
   "source": [
    "### Cyclic Generator with resblocks"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "cBkTYn7ErBHB"
   },
   "outputs": [],
   "source": [
    "class ResidualBlock(nn.Module):\n",
    "    def __init__(self, in_features, use_batch_norm=True, alpha=0.2, shift_factor=2):\n",
    "        super(ResidualBlock, self).__init__()\n",
    "        conv_blocks = [\n",
    "            Conv1D(in_features, in_features, 21, stride=1, padding=10, use_batch_norm=use_batch_norm, alpha=alpha,shift_factor=shift_factor),\n",
    "            Conv1D(in_features, in_features, 21, stride=1, padding=10, use_batch_norm=use_batch_norm, alpha=alpha,shift_factor=shift_factor)\n",
    "        ]\n",
    "        self.conv_blocks = nn.ModuleList(conv_blocks)\n",
    "    \n",
    "    def forward(self, x):\n",
    "        #down_sampled_x = nn.functional.interpolate(x,scale_factor=0.25, mode='nearest')\n",
    "        output = x\n",
    "        for conv in self.conv_blocks:\n",
    "            output = conv(output)\n",
    "        return x + output"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "f4TD3PoHrDtn"
   },
   "outputs": [],
   "source": [
    "class CyclicWaveGanGenerator(nn.Module):\n",
    "    def __init__(self,model_size=64,  num_channels=1, shift_factor=2,\n",
    "                 alpha=0.2, verbose=False, slice_len=16384, use_batch_norm = False, upsample=True):\n",
    "        super(CyclicWaveGanGenerator, self).__init__()\n",
    "        assert slice_len in [16384, 32768, 65536] # used to predict longer utterances\n",
    "        self.dim_mul = 16 if slice_len == 16384 else 32\n",
    "        self.verbose = verbose\n",
    "        encoder_conv = [ \n",
    "            Conv1D(num_channels, model_size, 25, stride=4, padding=11, use_batch_norm=use_batch_norm, alpha=alpha,shift_factor=shift_factor),\n",
    "            Conv1D(model_size, 2 * model_size, 25, stride=4, padding=13 if slice_len==32768 else 11, use_batch_norm=use_batch_norm, alpha=alpha,shift_factor=shift_factor),\n",
    "            #Conv1D(2 * model_size, 4*model_size , 25, stride=4, padding=13 if slice_len==32768 else 11, use_batch_norm=use_batch_norm, alpha=alpha,shift_factor=shift_factor)\n",
    "        ]\n",
    "        n_resblocks = 8\n",
    "        if slice_len == 32768:\n",
    "            encoder_conv.append(\n",
    "                Conv1D(2*model_size , (self.dim_mul* model_size) //8, 25, stride=2, padding=11, use_batch_norm=use_batch_norm, alpha=alpha,shift_factor=shift_factor)\n",
    "            )\n",
    "            n_resblocks = 9\n",
    "        elif slice_len == 65536:\n",
    "            encoder_conv.append(\n",
    "                Conv1D(2*model_size , (self.dim_mul* model_size) //8, 25, stride=4, padding=11, use_batch_norm=use_batch_norm, alpha=alpha,shift_factor=shift_factor)\n",
    "            )\n",
    "            n_resblocks = 9\n",
    "        \n",
    "        self.encoder = nn.Sequential(*encoder_conv)\n",
    "        # encoder output 1 256 256\n",
    "        \n",
    "        transformation = []\n",
    "        for _ in range(n_resblocks):\n",
    "            transformation.append(ResidualBlock((self.dim_mul* model_size) //8))\n",
    "        self.transformation = nn.Sequential(*transformation)\n",
    "\n",
    "        # Upsampling\n",
    "        stride = 4\n",
    "        if upsample:\n",
    "            stride = 1\n",
    "            upsample = 4\n",
    "        \n",
    "        deconv_layers = [\n",
    "            Transpose1dLayer( (self.dim_mul* model_size) //8,  (self.dim_mul* model_size) //16, 25, stride, upsample=upsample,use_batch_norm=use_batch_norm),\n",
    "            #Transpose1dLayer( (self.dim_mul* model_size) //8,  (self.dim_mul* model_size) //16, 25, stride, upsample=upsample,use_batch_norm=use_batch_norm),\n",
    "        ]\n",
    "        \n",
    "\n",
    "        if slice_len== 16384:\n",
    "            deconv_layers.append( Transpose1dLayer((self.dim_mul* model_size) //16, num_channels, 25, stride, upsample=upsample))\n",
    "        elif slice_len == 32768 :\n",
    "            deconv_layers +=[ \n",
    "                Transpose1dLayer((self.dim_mul* model_size) //16, model_size, 25, stride, upsample=upsample,use_batch_norm=use_batch_norm)\n",
    "                ,Transpose1dLayer(model_size, num_channels, 25, 2, upsample=upsample)\n",
    "            ]\n",
    "        elif slice_len == 65536:\n",
    "            deconv_layers +=[\n",
    "                Transpose1dLayer((self.dim_mul* model_size) //16, model_size, 25, stride, upsample=upsample,use_batch_norm=use_batch_norm)\n",
    "                ,Transpose1dLayer(model_size, num_channels, 25, stride, upsample=upsample)\n",
    "            ]\n",
    "        else:\n",
    "            raise ValueError('slice_len {} value is not supported'.format(slice_len))\n",
    "        self.upsampling = nn.Sequential( *deconv_layers)\n",
    "\n",
    "    def forward(self, x):\n",
    "        x = self.encoder(x)\n",
    "        if self.verbose:\n",
    "            print(x.shape)\n",
    "        x = self.transformation(x)\n",
    "        if self.verbose:\n",
    "            print(x.shape)\n",
    "        x = self.upsampling(x)\n",
    "        if self.verbose:\n",
    "            print(x.shape)\n",
    "        return x"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "G0BEGQGtQwCC"
   },
   "source": [
    "### Discriminator Model With Latent Code Eval"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "6ip7uE5jQ0Af"
   },
   "outputs": [],
   "source": [
    "class WaveGANDiscriminatorSep(nn.Module):\n",
    "    def __init__(self, model_size=64,  num_channels=1, shift_factor=2,\n",
    "                 alpha=0.2, verbose=False, slice_len=16384, use_batch_norm = False, drop_prob=0.2):\n",
    "        super(WaveGANDiscriminatorSep, self).__init__()\n",
    "        assert slice_len in [16384, 32768, 65536] # used to predict longer utterances\n",
    "        \n",
    "        self.verbose = verbose\n",
    "        self.model_size = model_size\n",
    " \n",
    "        conv_layers = [\n",
    "            Conv1D(num_channels, model_size, 25, stride=4, padding=11, use_batch_norm=use_batch_norm, alpha=alpha,shift_factor=shift_factor, drop_prob = drop_prob),\n",
    "            Conv1D(model_size, 2 * model_size, 25, stride=4, padding=11, use_batch_norm=use_batch_norm, alpha=alpha,shift_factor=shift_factor, drop_prob = drop_prob),\n",
    "            Conv1D(2 * model_size, 4 * model_size, 25, stride=4, padding=11, use_batch_norm=use_batch_norm, alpha=alpha,shift_factor=shift_factor, drop_prob = drop_prob),\n",
    "        ]\n",
    "\n",
    "        if slice_len == 32768 :\n",
    "            conv_layers.append(\n",
    "                 Conv1D(4 * model_size, 4 * model_size, 25, stride=2, padding=12,use_batch_norm=use_batch_norm, alpha=alpha,shift_factor=shift_factor)\n",
    "            ) \n",
    "        elif slice_len == 65536:\n",
    "            conv_layers.append(\n",
    "                 Conv1D(4 * model_size, 4 * model_size, 25, stride=4, padding=12,use_batch_norm=use_batch_norm, alpha=alpha,shift_factor=shift_factor)\n",
    "            )\n",
    "        self.infer_input_audio = nn.ModuleList(conv_layers)\n",
    "\n",
    "        self.move_z_space = nn.Linear(noise_latent_dim, 64 * model_size)\n",
    "        self.infer_z = nn.ModuleList([\n",
    "            Conv1D( 1, model_size//8, 25, stride=4, padding=11, use_batch_norm=use_batch_norm, drop_prob = drop_prob),\n",
    "            Conv1D(model_size//8, model_size//4, 25, stride=4, padding=11,use_batch_norm=use_batch_norm, drop_prob = drop_prob),\n",
    "        ])\n",
    "        \n",
    "        conv_joint_layers = [\n",
    "            Conv1D((4 * model_size)+(model_size//4), 8 * model_size, 25, stride=4, padding=11, use_batch_norm=use_batch_norm, alpha=alpha,shift_factor=shift_factor, drop_prob = drop_prob),\n",
    "            Conv1D(8 * model_size, 16 * model_size, 25, stride=4, padding=11, use_batch_norm=use_batch_norm, alpha=alpha,shift_factor=0, drop_prob = 0 if slice_len==16384 else drop_prob)\n",
    "        ]\n",
    "        \n",
    "        self.fc_input_size = 8 * 32 * model_size\n",
    "        if slice_len == 32768 :\n",
    "            conv_joint_layers.append(\n",
    "                 Conv1D(16 * model_size, 32 * model_size, 25, stride=2, padding=11,use_batch_norm=use_batch_norm, alpha=alpha,shift_factor=shift_factor)\n",
    "            ) \n",
    "            self.fc_input_size = 7 * 32 * model_size\n",
    "        elif slice_len == 65536:\n",
    "            conv_joint_layers.append(\n",
    "                 Conv1D(16 * model_size, 32 * model_size, 25, stride=4, padding=11,use_batch_norm=use_batch_norm, alpha=alpha,shift_factor=shift_factor)\n",
    "            )\n",
    "            self.fc_input_size = 4 * 32 * model_size\n",
    "        self.infer_join = nn.ModuleList(conv_joint_layers)\n",
    "        self.fc1 = nn.Linear(self.fc_input_size, 1)\n",
    "        \n",
    "\n",
    "        for m in self.modules():\n",
    "            if isinstance(m, nn.Conv1d) or isinstance(m, nn.Linear):\n",
    "                nn.init.kaiming_normal_(m.weight.data)\n",
    "\n",
    "    def forward(self, x, z):\n",
    "        for conv in self.infer_input_audio:\n",
    "            x = conv(x)\n",
    "            if self.verbose:\n",
    "                print(x.shape)\n",
    "        if self.verbose:\n",
    "            print(x.shape)\n",
    "        z = self.move_z_space(z).view(-1, 1, self.model_size * 64)\n",
    "        if self.verbose:\n",
    "            print(z.shape)\n",
    "        for conv in self.infer_z:\n",
    "            z = conv(z)\n",
    "            if self.verbose:\n",
    "                print(z.shape)\n",
    "        \n",
    "        # join z and x\n",
    "        output = torch.cat([x, z], dim=1)\n",
    "        if self.verbose:\n",
    "            print(output.shape)\n",
    "        for conv in self.infer_join:\n",
    "            output = conv(output)\n",
    "            if self.verbose:\n",
    "                print(output.shape)\n",
    "        output = output.view(-1, self.fc_input_size)\n",
    "        if self.verbose:\n",
    "            print(output.shape)\n",
    "        return self.fc1(output)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "anx6LvECQ41-"
   },
   "source": [
    "### Latent Space Encoder"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "ehfs42VsQ7l8"
   },
   "outputs": [],
   "source": [
    "class WaveGanEncoder(nn.Module):\n",
    "    def __init__(self,model_size=64,  num_channels=1, shift_factor=2,\n",
    "                 alpha=0.2, verbose=False, slice_len=16384, use_batch_norm = False):\n",
    "        super(WaveGanEncoder, self).__init__()\n",
    "        self.verbose = verbose\n",
    "        conv_layers = [\n",
    "            Conv1D(num_channels, model_size, 25, stride=4, padding=11, use_batch_norm=use_batch_norm, alpha=alpha,shift_factor=shift_factor),\n",
    "            Conv1D(model_size, 2 * model_size, 25, stride=4, padding=11, use_batch_norm=use_batch_norm, alpha=alpha,shift_factor=shift_factor),\n",
    "            Conv1D(2 * model_size, 4 * model_size, 25, stride=4, padding=11, use_batch_norm=use_batch_norm, alpha=alpha,shift_factor=shift_factor),\n",
    "            Conv1D(4 * model_size, 8 * model_size, 25, stride=4, padding=11, use_batch_norm=use_batch_norm, alpha=alpha,shift_factor=shift_factor),\n",
    "        ]\n",
    "        self.fc_input_size = 8 * model_size * 16\n",
    "        if slice_len == 32768 :\n",
    "            conv_layers.append(\n",
    "                 Conv1D(8 * model_size, 16 * model_size, 25, stride=2, padding=11,use_batch_norm=use_batch_norm, alpha=alpha,shift_factor=shift_factor)\n",
    "            ) \n",
    "            self.fc_input_size = 16 * model_size * 16\n",
    "        elif slice_len == 65536:\n",
    "            conv_layers.append(\n",
    "                 Conv1D(8 * model_size, 16 * model_size, 25, stride=4, padding=11,use_batch_norm=use_batch_norm, alpha=alpha,shift_factor=shift_factor)\n",
    "            )\n",
    "            self.fc_input_size = 16 * model_size * 16\n",
    "        self.conv_layers = nn.ModuleList(conv_layers)\n",
    "\n",
    "        self.adaptive_pooling = nn.AdaptiveAvgPool1d(16) # needs to be checked later\n",
    "\n",
    "        self.mu_linear = nn.Linear(self.fc_input_size, noise_latent_dim)\n",
    "        self.logvar_linear = nn.Linear(self.fc_input_size, noise_latent_dim)\n",
    "        for m in self.modules():\n",
    "            if isinstance(m, nn.Conv1d) or isinstance(m, nn.Linear):\n",
    "                nn.init.kaiming_normal_(m.weight.data)\n",
    "    \n",
    "    def reparametrize(self, mu, logvar):\n",
    "        std = logvar.mul(0.5).exp_()\n",
    "        z = mu + std * sample_noise(std.size(0))\n",
    "        return z\n",
    "\n",
    "    def bottleneck(self, h):\n",
    "        mu, logvar = self.mu_linear(h), self.logvar_linear(h)\n",
    "        z = self.reparametrize(mu, logvar)\n",
    "        return z, mu, logvar\n",
    "\n",
    "    def forward(self, x, inference=True):\n",
    "        for conv in self.conv_layers:\n",
    "            x= conv(x)\n",
    "            if self.verbose:\n",
    "                print(x.shape)\n",
    "        x = self.adaptive_pooling(x)\n",
    "        if self.verbose:\n",
    "            print(x.shape)\n",
    "        x = x.view(-1, self.fc_input_size)\n",
    "        if self.verbose:\n",
    "            print(x.shape)\n",
    "        z, _, _ = self.bottleneck(x)\n",
    "        return z"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "JXDsLG4xHuIi"
   },
   "source": [
    "### Model Init and Testing"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 138,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 125
    },
    "colab_type": "code",
    "id": "6IkIOCh_Hv09",
    "outputId": "d6cce5a4-cfae-4cdf-9aee-1916fd8b065c"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "torch.Size([10, 1024, 16])\n",
      "torch.Size([10, 512, 64])\n",
      "torch.Size([10, 256, 256])\n",
      "torch.Size([10, 128, 1024])\n",
      "torch.Size([10, 64, 4096])\n",
      "torch.Size([10, 1, 16384])\n"
     ]
    }
   ],
   "source": [
    "z = sample_noise(10)\n",
    "generator_model = WaveGANGenerator(verbose=True, upsample=True,  use_batch_norm=False ,slice_len = window_length).to(device)\n",
    "generated_audio_test = generator_model(z)\n",
    "print(generated_audio_test.shape)\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 139,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 539
    },
    "colab_type": "code",
    "id": "M_KXlWMWIEDQ",
    "outputId": "88745063-6360-424b-b378-e6b4bec41c17"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "----------------------------------------------------------------\n",
      "        Layer (type)               Output Shape         Param #\n",
      "================================================================\n",
      "            Linear-1             [-1, 1, 16384]       1,654,784\n",
      "     ConstantPad1d-2             [-1, 1024, 88]               0\n",
      "            Conv1d-3              [-1, 512, 64]      13,107,712\n",
      "  Transpose1dLayer-4              [-1, 512, 64]               0\n",
      "     ConstantPad1d-5             [-1, 512, 280]               0\n",
      "            Conv1d-6             [-1, 256, 256]       3,277,056\n",
      "  Transpose1dLayer-7             [-1, 256, 256]               0\n",
      "     ConstantPad1d-8            [-1, 256, 1048]               0\n",
      "            Conv1d-9            [-1, 128, 1024]         819,328\n",
      " Transpose1dLayer-10            [-1, 128, 1024]               0\n",
      "    ConstantPad1d-11            [-1, 128, 4120]               0\n",
      "           Conv1d-12             [-1, 64, 4096]         204,864\n",
      " Transpose1dLayer-13             [-1, 64, 4096]               0\n",
      "    ConstantPad1d-14            [-1, 64, 16408]               0\n",
      "           Conv1d-15             [-1, 1, 16384]           1,601\n",
      " Transpose1dLayer-16             [-1, 1, 16384]               0\n",
      "================================================================\n",
      "Total params: 19,065,345\n",
      "Trainable params: 19,065,345\n",
      "Non-trainable params: 0\n",
      "----------------------------------------------------------------\n",
      "Input size (MB): 0.00\n",
      "Forward/backward pass size (MB): 23.74\n",
      "Params size (MB): 72.73\n",
      "Estimated Total Size (MB): 96.47\n",
      "----------------------------------------------------------------\n"
     ]
    }
   ],
   "source": [
    "generator_model.verbose=False\n",
    "summary(generator_model, input_size=(1,noise_latent_dim))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 140,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 143
    },
    "colab_type": "code",
    "id": "3uWoLAztIIR5",
    "outputId": "51d033a0-f6d3-40cc-a51a-6d9f8a3dea64"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "torch.Size([10, 64, 4096])\n",
      "torch.Size([10, 128, 1024])\n",
      "torch.Size([10, 256, 256])\n",
      "torch.Size([10, 512, 64])\n",
      "torch.Size([10, 1024, 16])\n",
      "torch.Size([10, 16384])\n",
      "torch.Size([10, 1])\n"
     ]
    }
   ],
   "source": [
    "discriminator_model = WaveGANDiscriminator(verbose=True, use_batch_norm=True  ,slice_len = window_length).to(device)\n",
    "discriminating_output = discriminator_model(generated_audio_test)\n",
    "print(discriminating_output.shape)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 141,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 557
    },
    "colab_type": "code",
    "id": "NQ67upDZIZqq",
    "outputId": "8e0a2296-0568-4c06-8423-77e7ecc7030b"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "----------------------------------------------------------------\n",
      "        Layer (type)               Output Shape         Param #\n",
      "================================================================\n",
      "            Conv1d-1             [-1, 64, 4096]           1,664\n",
      "       BatchNorm1d-2             [-1, 64, 4096]             128\n",
      "            Conv1D-3             [-1, 64, 4096]               0\n",
      "            Conv1d-4            [-1, 128, 1024]         204,928\n",
      "       BatchNorm1d-5            [-1, 128, 1024]             256\n",
      "            Conv1D-6            [-1, 128, 1024]               0\n",
      "            Conv1d-7             [-1, 256, 256]         819,456\n",
      "       BatchNorm1d-8             [-1, 256, 256]             512\n",
      "            Conv1D-9             [-1, 256, 256]               0\n",
      "           Conv1d-10              [-1, 512, 64]       3,277,312\n",
      "      BatchNorm1d-11              [-1, 512, 64]           1,024\n",
      "           Conv1D-12              [-1, 512, 64]               0\n",
      "           Conv1d-13             [-1, 1024, 16]      13,108,224\n",
      "      BatchNorm1d-14             [-1, 1024, 16]           2,048\n",
      "     PhaseShuffle-15             [-1, 1024, 16]               0\n",
      "           Conv1D-16             [-1, 1024, 16]               0\n",
      "           Linear-17                    [-1, 1]          16,385\n",
      "================================================================\n",
      "Total params: 17,431,937\n",
      "Trainable params: 17,431,937\n",
      "Non-trainable params: 0\n",
      "----------------------------------------------------------------\n",
      "Input size (MB): 0.06\n",
      "Forward/backward pass size (MB): 11.75\n",
      "Params size (MB): 66.50\n",
      "Estimated Total Size (MB): 78.31\n",
      "----------------------------------------------------------------\n"
     ]
    }
   ],
   "source": [
    "discriminator_model.verbose=False\n",
    "summary(discriminator_model, input_size=(1,window_length))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 142,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 233
    },
    "colab_type": "code",
    "id": "1qq1MUFeRKI-",
    "outputId": "91960b91-873f-4197-9523-3c909d1ad39c"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "torch.Size([10, 64, 4096])\n",
      "torch.Size([10, 128, 1024])\n",
      "torch.Size([10, 256, 256])\n",
      "torch.Size([10, 256, 256])\n",
      "torch.Size([10, 1, 4096])\n",
      "torch.Size([10, 8, 1024])\n",
      "torch.Size([10, 16, 256])\n",
      "torch.Size([10, 272, 256])\n",
      "torch.Size([10, 512, 64])\n",
      "torch.Size([10, 1024, 16])\n",
      "torch.Size([10, 16384])\n",
      "torch.Size([10, 1])\n"
     ]
    }
   ],
   "source": [
    "discriminator_model = WaveGANDiscriminatorSep(verbose=True  ,slice_len = window_length).to(device)\n",
    "discriminating_output = discriminator_model(generated_audio_test, sample_noise(10))\n",
    "print(discriminating_output.shape)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 143,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 665
    },
    "colab_type": "code",
    "id": "FnL_bKItRjUr",
    "outputId": "dfc0bc0e-6800-4c34-cd87-b07dc6787f72"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "----------------------------------------------------------------\n",
      "        Layer (type)               Output Shape         Param #\n",
      "================================================================\n",
      "            Conv1d-1             [-1, 64, 4096]           1,664\n",
      "         Dropout2d-2             [-1, 64, 4096]               0\n",
      "            Conv1D-3             [-1, 64, 4096]               0\n",
      "            Conv1d-4            [-1, 128, 1024]         204,928\n",
      "         Dropout2d-5            [-1, 128, 1024]               0\n",
      "            Conv1D-6            [-1, 128, 1024]               0\n",
      "            Conv1d-7             [-1, 256, 256]         819,456\n",
      "         Dropout2d-8             [-1, 256, 256]               0\n",
      "            Conv1D-9             [-1, 256, 256]               0\n",
      "           Linear-10              [-1, 1, 4096]         413,696\n",
      "           Conv1d-11              [-1, 8, 1024]             208\n",
      "        Dropout2d-12              [-1, 8, 1024]               0\n",
      "           Conv1D-13              [-1, 8, 1024]               0\n",
      "           Conv1d-14              [-1, 16, 256]           3,216\n",
      "        Dropout2d-15              [-1, 16, 256]               0\n",
      "           Conv1D-16              [-1, 16, 256]               0\n",
      "           Conv1d-17              [-1, 512, 64]       3,482,112\n",
      "        Dropout2d-18              [-1, 512, 64]               0\n",
      "           Conv1D-19              [-1, 512, 64]               0\n",
      "           Conv1d-20             [-1, 1024, 16]      13,108,224\n",
      "     PhaseShuffle-21             [-1, 1024, 16]               0\n",
      "           Conv1D-22             [-1, 1024, 16]               0\n",
      "           Linear-23                    [-1, 1]          16,385\n",
      "================================================================\n",
      "Total params: 18,049,889\n",
      "Trainable params: 18,049,889\n",
      "Non-trainable params: 0\n",
      "----------------------------------------------------------------\n",
      "Input size (MB): 6.25\n",
      "Forward/backward pass size (MB): 11.94\n",
      "Params size (MB): 68.85\n",
      "Estimated Total Size (MB): 87.04\n",
      "----------------------------------------------------------------\n"
     ]
    }
   ],
   "source": [
    "discriminator_model.verbose=False\n",
    "summary(discriminator_model, input_size=[(1,window_length),(1,noise_latent_dim)])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 144,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 143
    },
    "colab_type": "code",
    "id": "jkSypp8RRKUZ",
    "outputId": "c7a6b99d-f493-45cc-9429-111cfa14d29f"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "torch.Size([10, 64, 4096])\n",
      "torch.Size([10, 128, 1024])\n",
      "torch.Size([10, 256, 256])\n",
      "torch.Size([10, 512, 64])\n",
      "torch.Size([10, 512, 16])\n",
      "torch.Size([10, 8192])\n",
      "torch.Size([10, 100])\n"
     ]
    }
   ],
   "source": [
    "encoder_model = WaveGanEncoder(verbose=True  ,slice_len = window_length).to(device)\n",
    "latent_encoded = encoder_model(generated_audio_test)\n",
    "print(latent_encoded.shape)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 145,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 449
    },
    "colab_type": "code",
    "id": "c9DIl5VxRqgk",
    "outputId": "72ca9a6e-5627-44e8-8302-97a611cbbaac"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "----------------------------------------------------------------\n",
      "        Layer (type)               Output Shape         Param #\n",
      "================================================================\n",
      "            Conv1d-1             [-1, 64, 4096]           1,664\n",
      "            Conv1D-2             [-1, 64, 4096]               0\n",
      "            Conv1d-3            [-1, 128, 1024]         204,928\n",
      "            Conv1D-4            [-1, 128, 1024]               0\n",
      "            Conv1d-5             [-1, 256, 256]         819,456\n",
      "            Conv1D-6             [-1, 256, 256]               0\n",
      "            Conv1d-7              [-1, 512, 64]       3,277,312\n",
      "            Conv1D-8              [-1, 512, 64]               0\n",
      " AdaptiveAvgPool1d-9              [-1, 512, 16]               0\n",
      "           Linear-10                  [-1, 100]         819,300\n",
      "           Linear-11                  [-1, 100]         819,300\n",
      "================================================================\n",
      "Total params: 5,941,960\n",
      "Trainable params: 5,941,960\n",
      "Non-trainable params: 0\n",
      "----------------------------------------------------------------\n",
      "Input size (MB): 0.06\n",
      "Forward/backward pass size (MB): 7.56\n",
      "Params size (MB): 22.67\n",
      "Estimated Total Size (MB): 30.29\n",
      "----------------------------------------------------------------\n"
     ]
    }
   ],
   "source": [
    "encoder_model.verbose=False\n",
    "summary(encoder_model, input_size=(1,window_length))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 146,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 89
    },
    "colab_type": "code",
    "id": "XUPONVxSrOF1",
    "outputId": "8dd20e3c-393e-44d4-aef4-c0e7848f79b0"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "torch.Size([10, 128, 1024])\n",
      "torch.Size([10, 128, 1024])\n",
      "torch.Size([10, 1, 16384])\n",
      "torch.Size([10, 1, 16384])\n"
     ]
    }
   ],
   "source": [
    "cyclic_generator = CyclicWaveGanGenerator(slice_len=window_length, verbose = True).to(device)\n",
    "generated = cyclic_generator(Variable(torch.rand(10,1,window_length)).to(device))\n",
    "print(generated.shape)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 147,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 1000
    },
    "colab_type": "code",
    "id": "VL0-X0FAraW7",
    "outputId": "84f5a380-0cb9-4c8c-c27e-527bb3abab63"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "----------------------------------------------------------------\n",
      "        Layer (type)               Output Shape         Param #\n",
      "================================================================\n",
      "            Conv1d-1             [-1, 64, 4096]           1,664\n",
      "            Conv1D-2             [-1, 64, 4096]               0\n",
      "            Conv1d-3            [-1, 128, 1024]         204,928\n",
      "            Conv1D-4            [-1, 128, 1024]               0\n",
      "            Conv1d-5            [-1, 128, 1024]         344,192\n",
      "       BatchNorm1d-6            [-1, 128, 1024]             256\n",
      "            Conv1D-7            [-1, 128, 1024]               0\n",
      "            Conv1d-8            [-1, 128, 1024]         344,192\n",
      "       BatchNorm1d-9            [-1, 128, 1024]             256\n",
      "           Conv1D-10            [-1, 128, 1024]               0\n",
      "    ResidualBlock-11            [-1, 128, 1024]               0\n",
      "           Conv1d-12            [-1, 128, 1024]         344,192\n",
      "      BatchNorm1d-13            [-1, 128, 1024]             256\n",
      "           Conv1D-14            [-1, 128, 1024]               0\n",
      "           Conv1d-15            [-1, 128, 1024]         344,192\n",
      "      BatchNorm1d-16            [-1, 128, 1024]             256\n",
      "           Conv1D-17            [-1, 128, 1024]               0\n",
      "    ResidualBlock-18            [-1, 128, 1024]               0\n",
      "           Conv1d-19            [-1, 128, 1024]         344,192\n",
      "      BatchNorm1d-20            [-1, 128, 1024]             256\n",
      "           Conv1D-21            [-1, 128, 1024]               0\n",
      "           Conv1d-22            [-1, 128, 1024]         344,192\n",
      "      BatchNorm1d-23            [-1, 128, 1024]             256\n",
      "           Conv1D-24            [-1, 128, 1024]               0\n",
      "    ResidualBlock-25            [-1, 128, 1024]               0\n",
      "           Conv1d-26            [-1, 128, 1024]         344,192\n",
      "      BatchNorm1d-27            [-1, 128, 1024]             256\n",
      "           Conv1D-28            [-1, 128, 1024]               0\n",
      "           Conv1d-29            [-1, 128, 1024]         344,192\n",
      "      BatchNorm1d-30            [-1, 128, 1024]             256\n",
      "           Conv1D-31            [-1, 128, 1024]               0\n",
      "    ResidualBlock-32            [-1, 128, 1024]               0\n",
      "           Conv1d-33            [-1, 128, 1024]         344,192\n",
      "      BatchNorm1d-34            [-1, 128, 1024]             256\n",
      "           Conv1D-35            [-1, 128, 1024]               0\n",
      "           Conv1d-36            [-1, 128, 1024]         344,192\n",
      "      BatchNorm1d-37            [-1, 128, 1024]             256\n",
      "           Conv1D-38            [-1, 128, 1024]               0\n",
      "    ResidualBlock-39            [-1, 128, 1024]               0\n",
      "           Conv1d-40            [-1, 128, 1024]         344,192\n",
      "      BatchNorm1d-41            [-1, 128, 1024]             256\n",
      "           Conv1D-42            [-1, 128, 1024]               0\n",
      "           Conv1d-43            [-1, 128, 1024]         344,192\n",
      "      BatchNorm1d-44            [-1, 128, 1024]             256\n",
      "           Conv1D-45            [-1, 128, 1024]               0\n",
      "    ResidualBlock-46            [-1, 128, 1024]               0\n",
      "           Conv1d-47            [-1, 128, 1024]         344,192\n",
      "      BatchNorm1d-48            [-1, 128, 1024]             256\n",
      "           Conv1D-49            [-1, 128, 1024]               0\n",
      "           Conv1d-50            [-1, 128, 1024]         344,192\n",
      "      BatchNorm1d-51            [-1, 128, 1024]             256\n",
      "           Conv1D-52            [-1, 128, 1024]               0\n",
      "    ResidualBlock-53            [-1, 128, 1024]               0\n",
      "           Conv1d-54            [-1, 128, 1024]         344,192\n",
      "      BatchNorm1d-55            [-1, 128, 1024]             256\n",
      "           Conv1D-56            [-1, 128, 1024]               0\n",
      "           Conv1d-57            [-1, 128, 1024]         344,192\n",
      "      BatchNorm1d-58            [-1, 128, 1024]             256\n",
      "           Conv1D-59            [-1, 128, 1024]               0\n",
      "    ResidualBlock-60            [-1, 128, 1024]               0\n",
      "    ConstantPad1d-61            [-1, 128, 4120]               0\n",
      "           Conv1d-62             [-1, 64, 4096]         204,864\n",
      " Transpose1dLayer-63             [-1, 64, 4096]               0\n",
      "    ConstantPad1d-64            [-1, 64, 16408]               0\n",
      "           Conv1d-65             [-1, 1, 16384]           1,601\n",
      " Transpose1dLayer-66             [-1, 1, 16384]               0\n",
      "================================================================\n",
      "Total params: 5,924,225\n",
      "Trainable params: 5,924,225\n",
      "Non-trainable params: 0\n",
      "----------------------------------------------------------------\n",
      "Input size (MB): 0.06\n",
      "Forward/backward pass size (MB): 78.29\n",
      "Params size (MB): 22.60\n",
      "Estimated Total Size (MB): 100.95\n",
      "----------------------------------------------------------------\n"
     ]
    }
   ],
   "source": [
    "cyclic_generator.verbose=False\n",
    "summary(cyclic_generator, input_size=(1,window_length))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "NqLXp66Qq2mI"
   },
   "source": [
    "### Cyclic Gan"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "qvy6nHfUq4KA"
   },
   "outputs": [],
   "source": [
    "class CycleGan(object):\n",
    "    def __init__(self, train_loader, val_loader, validate=True):\n",
    "        # train cyclic to just disable the model from single source to mixed as an experiment\n",
    "\n",
    "        self.validate = validate\n",
    "\n",
    "        self.val_g_cost = []\n",
    "        self.train_g_cost = []\n",
    "        self.cyclic_loss = []\n",
    "        self.valid_reconstruction = []\n",
    "\n",
    "        self.discriminator_loss = []\n",
    "\n",
    "        self.generator = CyclicWaveGanGenerator(slice_len = window_length, model_size = model_capacity_size).to(device)\n",
    "        self.generator.apply(weights_init)\n",
    "        \n",
    "        self.discriminator_1 = WaveGANDiscriminator(slice_len = window_length, model_size = model_capacity_size).to(device)\n",
    "        self.discriminator_1.apply(weights_init)\n",
    "    \n",
    "        self.discriminator_2 = WaveGANDiscriminator(slice_len = window_length, model_size = model_capacity_size).to(device)\n",
    "        self.discriminator_2.apply(weights_init)\n",
    "\n",
    "\n",
    "        self.optimizer_d_1 = optim.Adam(self.discriminator_1.parameters(), lr=lr_d, betas=(beta1,  beta2))\n",
    "        self.optimizer_d_2 = optim.Adam(self.discriminator_2.parameters(), lr=lr_d, betas=(beta1,  beta2))\n",
    "        \n",
    "\n",
    "        \n",
    "\n",
    "        self.optimizer_g = optim.Adam(self.generator.parameters(), lr=lr_g, betas=(beta1,  beta2))\n",
    "\n",
    "        self.train_loader = train_loader\n",
    "        self.val_loader = val_loader          \n",
    "    \n",
    "    def apply_zero_grad(self):\n",
    "        self.discriminator_1.zero_grad()\n",
    "        self.discriminator_2.zero_grad()\n",
    "        self.generator.zero_grad()\n",
    "\n",
    "    def enable_gen_disable_disc(self):\n",
    "        gradients_status(self.generator, True)\n",
    "        gradients_status(self.discriminator_1, False)\n",
    "        gradients_status(self.discriminator_2, False)\n",
    "    \n",
    "    def disable_all(self):\n",
    "        gradients_status(self.generator, False)\n",
    "        gradients_status(self.discriminator_1, False)\n",
    "        gradients_status(self.discriminator_2, False)\n",
    "    \n",
    "  \n",
    "    def train(self):\n",
    "        real_label = 0.9\n",
    "        progress_bar =  tqdm(total=n_iterations//progress_bar_step_iter_size)\n",
    "        val_set = iter(self.val_loader)\n",
    "        val_data = next(val_set)\n",
    "        fixed_mixed_signal = val_data[1].to(device)\n",
    "        fixed_single_signal = val_data[0].to(device)\n",
    "        save_samples(fixed_mixed_signal.detach().cpu().numpy(), 'fixed_mixed' ) \n",
    "        save_samples(fixed_single_signal.detach().cpu().numpy(), 'fixed_single' ) \n",
    "        \n",
    "        gan_model_name = 'gan_cyclic_single_2disc_{}.tar'.format(model_prefix)\n",
    "        \n",
    "        first_iter = 0\n",
    "        if take_backup and os.path.isfile(gan_model_name):\n",
    "            if cuda:\n",
    "                checkpoint = torch.load(gan_model_name)\n",
    "            else:\n",
    "                checkpoint = torch.load(gan_model_name, map_location='cpu')\n",
    "            self.generator.load_state_dict(checkpoint['generator'])\n",
    "            self.discriminator_1.load_state_dict(checkpoint['discriminator_1'])\n",
    "            self.discriminator_2.load_state_dict(checkpoint['discriminator_2'])\n",
    "            self.optimizer_d_1.load_state_dict(checkpoint['optimizer_d_1'])\n",
    "            self.optimizer_d_2.load_state_dict(checkpoint['optimizer_d_2'])\n",
    "            self.optimizer_g.load_state_dict(checkpoint['optimizer_g'])\n",
    "            self.val_g_cost = checkpoint['val_g_cost']\n",
    "            self.train_g_cost = checkpoint['train_g_cost']\n",
    "            self.cyclic_loss = checkpoint['cyclic_loss']\n",
    "            self.discriminator_loss = checkpoint['discriminator_loss']\n",
    "            first_iter = checkpoint['n_iterations'] + 1\n",
    "            for _ in range(0, first_iter, progress_bar_step_iter_size):\n",
    "                progress_bar.update()\n",
    "            \n",
    "            \n",
    "            self.generator.eval()\n",
    "        criterion_GAN = nn.MSELoss()\n",
    "        criterion_cycle = nn.L1Loss()\n",
    "        criterion_identity = nn.L1Loss()\n",
    "\n",
    "        target_real = Variable(torch.Tensor(batch_size,1).fill_(1.0), requires_grad=False).to(device)\n",
    "        target_fake = Variable(torch.Tensor(batch_size,1).fill_(0.0), requires_grad=False).to(device)\n",
    "        \n",
    "        generated_single_buffer = ReplayBuffer()\n",
    "        train_set = iter(self.train_loader)\n",
    "        for iter_indx in range(first_iter, n_iterations): \n",
    "            self.generator.train()\n",
    "            self.discriminator_1.train()\n",
    "            self.discriminator_2.train()\n",
    "            try:\n",
    "                data = next(train_set)\n",
    "            except StopIteration:\n",
    "                train_set = iter(self.train_loader)\n",
    "                data = next(train_set)\n",
    "            \n",
    "            # in case of unpaired data\n",
    "            single_signal = data[0].to(device)\n",
    "            mixed_signal = data[1].to(device)\n",
    "\n",
    "\n",
    "            #############################\n",
    "            # Training First Discriminator\n",
    "            #############################\n",
    "            self.apply_zero_grad()\n",
    "            self.disable_all()\n",
    "            gradients_status(self.discriminator_1, True)\n",
    "            generated_single_signal = self.generator(single_signal)\n",
    "            rest_of_signal = mixed_signal - generated_single_signal\n",
    "            new_mixed_signal = rest_of_signal + single_signal\n",
    "\n",
    "            # Real loss\n",
    "            is_single_signal_r = self.discriminator_1(single_signal)\n",
    "            #d_loss_real_1 = criterion_GAN(is_single_signal_r,target_real)\n",
    "\n",
    "            # generated loss\n",
    "            #generated_single_signal = generated_single_buffer.push_and_pop(generated_single_signal)\n",
    "            is_single_signal_f = self.discriminator_1(generated_single_signal.detach())\n",
    "            #d_loss_generated_1 = criterion_GAN(is_single_signal_f, target_fake)\n",
    "\n",
    "            d_loss_1 = (torch.mean((is_single_signal_r - torch.mean(is_single_signal_f) - target_real) ** 2) +\n",
    "                torch.mean((is_single_signal_f - torch.mean(is_single_signal_r) + target_real) ** 2))/2#(d_loss_real_1 + d_loss_generated_1)/2\n",
    "            d_loss_1.backward()\n",
    "            self.optimizer_d_1.step()\n",
    "            #############################\n",
    "            # Training Second Discriminator\n",
    "            #############################\n",
    "            self.apply_zero_grad()\n",
    "            self.disable_all()\n",
    "            gradients_status(self.discriminator_2, True)\n",
    "            # Real loss\n",
    "            is_mixed_signal_r = self.discriminator_2(mixed_signal)\n",
    "            #d_loss_real_1 = criterion_GAN(is_mixed_signal_r,target_real)\n",
    "\n",
    "            # generated loss\n",
    "            #generated_single_signal = generated_single_buffer.push_and_pop(generated_single_signal)\n",
    "            is_mixed_signal_f = self.discriminator_2(new_mixed_signal.detach())\n",
    "            #d_loss_generated_1 = criterion_GAN(is_mixed_signal_f, target_fake)\n",
    "\n",
    "            d_loss_2 = (torch.mean((is_mixed_signal_r - torch.mean(is_mixed_signal_f) - target_real) ** 2) +\n",
    "                torch.mean((is_mixed_signal_f - torch.mean(is_mixed_signal_r) + target_real) ** 2))/2#(d_loss_real_1 + d_loss_generated_1)/2\n",
    "            d_loss_2.backward()\n",
    "            self.optimizer_d_2.step()\n",
    "\n",
    "            #############################\n",
    "            # Training  generator\n",
    "            #############################\n",
    "            self.apply_zero_grad()\n",
    "            self.enable_gen_disable_disc()\n",
    "            \n",
    "\n",
    "            # Identity loss without it the model would make changes to input even without any need\n",
    "\n",
    "            \n",
    "            identity_loss_1 = criterion_identity(generated_single_signal, single_signal)\n",
    "        \n",
    "            # Gan Loss\n",
    "            #generated_single_signal = self.generator(mixed_signal)\n",
    "            is_single_signal_r = self.discriminator_1(single_signal)\n",
    "            is_single_signal_f = self.discriminator_1(generated_single_signal)\n",
    "            gan_loss_1 = (torch.mean((is_single_signal_r - torch.mean(is_single_signal_f) + target_real) ** 2) +\n",
    "                torch.mean((is_single_signal_f - torch.mean(is_single_signal_r) - target_real) ** 2))/2\n",
    "\n",
    "            #gan_loss_1 =criterion_GAN(is_single_signal, target_real )\n",
    "\n",
    "\n",
    "            is_mixed_signal_r = self.discriminator_2(mixed_signal)\n",
    "            is_mixed_signal_f = self.discriminator_2(new_mixed_signal)\n",
    "            gan_loss_2 = (torch.mean((is_mixed_signal_r - torch.mean(is_mixed_signal_f) + target_real) ** 2) +\n",
    "                torch.mean((is_mixed_signal_f - torch.mean(is_mixed_signal_r) - target_real) ** 2))/2\n",
    "\n",
    "            #gan_loss_2 = criterion_GAN(is_mixed_signal, target_real )\n",
    "\n",
    "            reconstructed_single_sinal = self.generator(new_mixed_signal)\n",
    "\n",
    "            cycle_loss_1 = criterion_cycle(reconstructed_single_sinal, single_signal)\n",
    "            # Total Loss\n",
    "            g_cost  =  identity_loss_1*0.5 + (gan_loss_1 + gan_loss_2)  + 10*cycle_loss_1 \n",
    "            g_cost.backward()\n",
    "            self.optimizer_g.step()\n",
    "\n",
    "\n",
    "            if self.validate and iter_indx%store_cost_every==0:\n",
    "                self.discriminator_loss.append(d_loss_1.item())\n",
    "                self.train_g_cost.append(g_cost.item())\n",
    "                # validating\n",
    "                self.disable_all()\n",
    "                with torch.no_grad():\n",
    "                    try:\n",
    "                        val_data = next(val_set)\n",
    "                    except StopIteration:\n",
    "                        val_set = iter(self.val_loader)\n",
    "                        val_data = next(val_set)\n",
    "                    val_single = val_data[0].to(device)\n",
    "                    val_mixed = val_data[1].to(device)\n",
    "                    val_cost= criterion_GAN(self.discriminator_1(val_single), target_real) + criterion_GAN(self.discriminator_2(val_single), target_real)\n",
    "                    self.val_g_cost.append(val_cost.item())\n",
    "                    reconstructed_music = self.generator(val_mixed)\n",
    "                    self.valid_reconstruction.append(F.mse_loss(reconstructed_music, val_single, reduction='sum').item())\n",
    "                    \n",
    "\n",
    "            if iter_indx%store_cost_every==0 :\n",
    "                progress_updates = {'Reconstruction': str(self.valid_reconstruction[-1]),'Loss_D1': str(d_loss_1.item()), 'Loss_g':str(g_cost.item())}\n",
    "                progress_bar.set_postfix(progress_updates)\n",
    "\n",
    "\n",
    "            if iter_indx%progress_bar_step_iter_size==0:\n",
    "                progress_bar.update()\n",
    "            # lr decay \n",
    "            if decay_lr:\n",
    "                decay = max(0.0,1.0-(iter_indx*1.0/n_iterations))\n",
    "                # update the learning rate\n",
    "                update_optimizer_lr(self.optimizer_d, lr_d, decay)\n",
    "                update_optimizer_lr(self.optimizer_g, lr_g, decay)\n",
    "\n",
    "            if (iter_indx%save_samples_every==0):\n",
    "                with torch.no_grad():\n",
    "                    fake = self.generator(fixed_mixed_signal).detach().cpu().numpy()\n",
    "                save_samples(fake, iter_indx , prefix='predictions')\n",
    "            \n",
    "            if take_backup and iter_indx%backup_every_n_iters==0:\n",
    "                saving_dict = {\n",
    "                    'generator': self.generator.state_dict(),\n",
    "                    'discriminator_1': self.discriminator_1.state_dict(),\n",
    "                    'optimizer_d_1': self.optimizer_d_1.state_dict(), \n",
    "                    'discriminator_2': self.discriminator_2.state_dict(),\n",
    "                    'optimizer_d_2': self.optimizer_d_2.state_dict(), \n",
    "                    'optimizer_g': self.optimizer_g.state_dict(),\n",
    "                    'val_g_cost': self.val_g_cost,\n",
    "                    'train_g_cost': self.train_g_cost,\n",
    "                    'cyclic_loss': self.cyclic_loss,\n",
    "                    'discriminator_loss': self.discriminator_loss,\n",
    "                    'n_iterations': iter_indx\n",
    "                }\n",
    "                torch.save(saving_dict,gan_model_name)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "Cz3XfTqIJRBx"
   },
   "source": [
    "## Train"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 77,
     "resources": {
      "http://localhost:8080/nbextensions/google.colab/colabwidgets/controls.css": {
       "data": "LyogQ29weXJpZ2h0IChjKSBKdXB5dGVyIERldmVsb3BtZW50IFRlYW0uCiAqIERpc3RyaWJ1dGVkIHVuZGVyIHRoZSB0ZXJtcyBvZiB0aGUgTW9kaWZpZWQgQlNEIExpY2Vuc2UuCiAqLwoKIC8qIFdlIGltcG9ydCBhbGwgb2YgdGhlc2UgdG9nZXRoZXIgaW4gYSBzaW5nbGUgY3NzIGZpbGUgYmVjYXVzZSB0aGUgV2VicGFjawpsb2FkZXIgc2VlcyBvbmx5IG9uZSBmaWxlIGF0IGEgdGltZS4gVGhpcyBhbGxvd3MgcG9zdGNzcyB0byBzZWUgdGhlIHZhcmlhYmxlCmRlZmluaXRpb25zIHdoZW4gdGhleSBhcmUgdXNlZC4gKi8KCiAvKi0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tCnwgQ29weXJpZ2h0IChjKSBKdXB5dGVyIERldmVsb3BtZW50IFRlYW0uCnwgRGlzdHJpYnV0ZWQgdW5kZXIgdGhlIHRlcm1zIG9mIHRoZSBNb2RpZmllZCBCU0QgTGljZW5zZS4KfC0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0tLS0qLwoKIC8qClRoaXMgZmlsZSBpcyBjb3BpZWQgZnJvbSB0aGUgSnVweXRlckxhYiBwcm9qZWN0IHRvIGRlZmluZSBkZWZhdWx0IHN0eWxpbmcgZm9yCndoZW4gdGhlIHdpZGdldCBzdHlsaW5nIGlzIGNvbXBpbGVkIGRvd24gdG8gZWxpbWluYXRlIENTUyB2YXJpYWJsZXMuIFdlIG1ha2Ugb25lCmNoYW5nZSAtIHdlIGNvbW1lbnQgb3V0IHRoZSBmb250IGltcG9ydCBiZWxvdy4KKi8KCiAvKioKICogVGhlIG1hdGVyaWFsIGRlc2lnbiBjb2xvcnMgYXJlIGFkYXB0ZWQgZnJvbSBnb29nbGUtbWF0ZXJpYWwtY29sb3IgdjEuMi42CiAqIGh0dHBzOi8vZ2l0aHViLmNvbS9kYW5sZXZhbi9nb29nbGUtbWF0ZXJpYWwtY29sb3IKICogaHR0cHM6Ly9naXRodWIuY29tL2RhbmxldmFuL2dvb2dsZS1tYXRlcmlhbC1jb2xvci9ibG9iL2Y2N2NhNWY0MDI4YjJmMWIzNDg2MmY2NGIwY2E2NzMyM2Y5MWIwODgvZGlzdC9wYWxldHRlLnZhci5jc3MKICoKICogVGhlIGxpY2Vuc2UgZm9yIHRoZSBtYXRlcmlhbCBkZXNpZ24gY29sb3IgQ1NTIHZhcmlhYmxlcyBpcyBhcyBmb2xsb3dzIChzZWUKICogaHR0cHM6Ly9naXRodWIuY29tL2RhbmxldmFuL2dvb2dsZS1tYXRlcmlhbC1jb2xvci9ibG9iL2Y2N2NhNWY0MDI4YjJmMWIzNDg2MmY2NGIwY2E2NzMyM2Y5MWIwODgvTElDRU5TRSkKICoKICogVGhlIE1JVCBMaWNlbnNlIChNSVQpCiAqCiAqIENvcHlyaWdodCAoYykgMjAxNCBEYW4gTGUgVmFuCiAqCiAqIFBlcm1pc3Npb24gaXMgaGVyZWJ5IGdyYW50ZWQsIGZyZWUgb2YgY2hhcmdlLCB0byBhbnkgcGVyc29uIG9idGFpbmluZyBhIGNvcHkKICogb2YgdGhpcyBzb2Z0d2FyZSBhbmQgYXNzb2NpYXRlZCBkb2N1bWVudGF0aW9uIGZpbGVzICh0aGUgIlNvZnR3YXJlIiksIHRvIGRlYWwKICogaW4gdGhlIFNvZnR3YXJlIHdpdGhvdXQgcmVzdHJpY3Rpb24sIGluY2x1ZGluZyB3aXRob3V0IGxpbWl0YXRpb24gdGhlIHJpZ2h0cwogKiB0byB1c2UsIGNvcHksIG1vZGlmeSwgbWVyZ2UsIHB1Ymxpc2gsIGRpc3RyaWJ1dGUsIHN1YmxpY2Vuc2UsIGFuZC9vciBzZWxsCiAqIGNvcGllcyBvZiB0aGUgU29mdHdhcmUsIGFuZCB0byBwZXJtaXQgcGVyc29ucyB0byB3aG9tIHRoZSBTb2Z0d2FyZSBpcwogKiBmdXJuaXNoZWQgdG8gZG8gc28sIHN1YmplY3QgdG8gdGhlIGZvbGxvd2luZyBjb25kaXRpb25zOgogKgogKiBUaGUgYWJvdmUgY29weXJpZ2h0IG5vdGljZSBhbmQgdGhpcyBwZXJtaXNzaW9uIG5vdGljZSBzaGFsbCBiZSBpbmNsdWRlZCBpbgogKiBhbGwgY29waWVzIG9yIHN1YnN0YW50aWFsIHBvcnRpb25zIG9mIHRoZSBTb2Z0d2FyZS4KICoKICogVEhFIFNPRlRXQVJFIElTIFBST1ZJREVEICJBUyBJUyIsIFdJVEhPVVQgV0FSUkFOVFkgT0YgQU5ZIEtJTkQsIEVYUFJFU1MgT1IKICogSU1QTElFRCwgSU5DTFVESU5HIEJVVCBOT1QgTElNSVRFRCBUTyBUSEUgV0FSUkFOVElFUyBPRiBNRVJDSEFOVEFCSUxJVFksCiAqIEZJVE5FU1MgRk9SIEEgUEFSVElDVUxBUiBQVVJQT1NFIEFORCBOT05JTkZSSU5HRU1FTlQuIElOIE5PIEVWRU5UIFNIQUxMIFRIRQogKiBBVVRIT1JTIE9SIENPUFlSSUdIVCBIT0xERVJTIEJFIExJQUJMRSBGT1IgQU5ZIENMQUlNLCBEQU1BR0VTIE9SIE9USEVSCiAqIExJQUJJTElUWSwgV0hFVEhFUiBJTiBBTiBBQ1RJT04gT0YgQ09OVFJBQ1QsIFRPUlQgT1IgT1RIRVJXSVNFLCBBUklTSU5HIEZST00sCiAqIE9VVCBPRiBPUiBJTiBDT05ORUNUSU9OIFdJVEggVEhFIFNPRlRXQVJFIE9SIFRIRSBVU0UgT1IgT1RIRVIgREVBTElOR1MgSU4gVEhFCiAqIFNPRlRXQVJFLgogKi8KCiAvKgpUaGUgZm9sbG93aW5nIENTUyB2YXJpYWJsZXMgZGVmaW5lIHRoZSBtYWluLCBwdWJsaWMgQVBJIGZvciBzdHlsaW5nIEp1cHl0ZXJMYWIuClRoZXNlIHZhcmlhYmxlcyBzaG91bGQgYmUgdXNlZCBieSBhbGwgcGx1Z2lucyB3aGVyZXZlciBwb3NzaWJsZS4gSW4gb3RoZXIKd29yZHMsIHBsdWdpbnMgc2hvdWxkIG5vdCBkZWZpbmUgY3VzdG9tIGNvbG9ycywgc2l6ZXMsIGV0YyB1bmxlc3MgYWJzb2x1dGVseQpuZWNlc3NhcnkuIFRoaXMgZW5hYmxlcyB1c2VycyB0byBjaGFuZ2UgdGhlIHZpc3VhbCB0aGVtZSBvZiBKdXB5dGVyTGFiCmJ5IGNoYW5naW5nIHRoZXNlIHZhcmlhYmxlcy4KCk1hbnkgdmFyaWFibGVzIGFwcGVhciBpbiBhbiBvcmRlcmVkIHNlcXVlbmNlICgwLDEsMiwzKS4gVGhlc2Ugc2VxdWVuY2VzCmFyZSBkZXNpZ25lZCB0byB3b3JrIHdlbGwgdG9nZXRoZXIsIHNvIGZvciBleGFtcGxlLCBgLS1qcC1ib3JkZXItY29sb3IxYCBzaG91bGQKYmUgdXNlZCB3aXRoIGAtLWpwLWxheW91dC1jb2xvcjFgLiBUaGUgbnVtYmVycyBoYXZlIHRoZSBmb2xsb3dpbmcgbWVhbmluZ3M6CgoqIDA6IHN1cGVyLXByaW1hcnksIHJlc2VydmVkIGZvciBzcGVjaWFsIGVtcGhhc2lzCiogMTogcHJpbWFyeSwgbW9zdCBpbXBvcnRhbnQgdW5kZXIgbm9ybWFsIHNpdHVhdGlvbnMKKiAyOiBzZWNvbmRhcnksIG5leHQgbW9zdCBpbXBvcnRhbnQgdW5kZXIgbm9ybWFsIHNpdHVhdGlvbnMKKiAzOiB0ZXJ0aWFyeSwgbmV4dCBtb3N0IGltcG9ydGFudCB1bmRlciBub3JtYWwgc2l0dWF0aW9ucwoKVGhyb3VnaG91dCBKdXB5dGVyTGFiLCB3ZSBhcmUgbW9zdGx5IGZvbGxvd2luZyBwcmluY2lwbGVzIGZyb20gR29vZ2xlJ3MKTWF0ZXJpYWwgRGVzaWduIHdoZW4gc2VsZWN0aW5nIGNvbG9ycy4gV2UgYXJlIG5vdCwgaG93ZXZlciwgZm9sbG93aW5nCmFsbCBvZiBNRCBhcyBpdCBpcyBub3Qgb3B0aW1pemVkIGZvciBkZW5zZSwgaW5mb3JtYXRpb24gcmljaCBVSXMuCiovCgogLyoKICogT3B0aW9uYWwgbW9ub3NwYWNlIGZvbnQgZm9yIGlucHV0L291dHB1dCBwcm9tcHQuCiAqLwoKIC8qIENvbW1lbnRlZCBvdXQgaW4gaXB5d2lkZ2V0cyBzaW5jZSB3ZSBkb24ndCBuZWVkIGl0LiAqLwoKIC8qIEBpbXBvcnQgdXJsKCdodHRwczovL2ZvbnRzLmdvb2dsZWFwaXMuY29tL2Nzcz9mYW1pbHk9Um9ib3RvK01vbm8nKTsgKi8KCiAvKgogKiBBZGRlZCBmb3IgY29tcGFiaXRpbGl0eSB3aXRoIG91dHB1dCBhcmVhCiAqLwoKIDpyb290IHsKCiAgLyogQm9yZGVycwoKICBUaGUgZm9sbG93aW5nIHZhcmlhYmxlcywgc3BlY2lmeSB0aGUgdmlzdWFsIHN0eWxpbmcgb2YgYm9yZGVycyBpbiBKdXB5dGVyTGFiLgogICAqLwoKICAvKiBVSSBGb250cwoKICBUaGUgVUkgZm9udCBDU1MgdmFyaWFibGVzIGFyZSB1c2VkIGZvciB0aGUgdHlwb2dyYXBoeSBhbGwgb2YgdGhlIEp1cHl0ZXJMYWIKICB1c2VyIGludGVyZmFjZSBlbGVtZW50cyB0aGF0IGFyZSBub3QgZGlyZWN0bHkgdXNlciBnZW5lcmF0ZWQgY29udGVudC4KICAqLyAvKiBCYXNlIGZvbnQgc2l6ZSAqLyAvKiBFbnN1cmVzIHB4IHBlcmZlY3QgRm9udEF3ZXNvbWUgaWNvbnMgKi8KCiAgLyogVXNlIHRoZXNlIGZvbnQgY29sb3JzIGFnYWluc3QgdGhlIGNvcnJlc3BvbmRpbmcgbWFpbiBsYXlvdXQgY29sb3JzLgogICAgIEluIGEgbGlnaHQgdGhlbWUsIHRoZXNlIGdvIGZyb20gZGFyayB0byBsaWdodC4KICAqLwoKICAvKiBVc2UgdGhlc2UgYWdhaW5zdCB0aGUgYnJhbmQvYWNjZW50L3dhcm4vZXJyb3IgY29sb3JzLgogICAgIFRoZXNlIHdpbGwgdHlwaWNhbGx5IGdvIGZyb20gbGlnaHQgdG8gZGFya2VyLCBpbiBib3RoIGEgZGFyayBhbmQgbGlnaHQgdGhlbWUKICAgKi8KCiAgLyogQ29udGVudCBGb250cwoKICBDb250ZW50IGZvbnQgdmFyaWFibGVzIGFyZSB1c2VkIGZvciB0eXBvZ3JhcGh5IG9mIHVzZXIgZ2VuZXJhdGVkIGNvbnRlbnQuCiAgKi8gLyogQmFzZSBmb250IHNpemUgKi8KCgogIC8qIExheW91dAoKICBUaGUgZm9sbG93aW5nIGFyZSB0aGUgbWFpbiBsYXlvdXQgY29sb3JzIHVzZSBpbiBKdXB5dGVyTGFiLiBJbiBhIGxpZ2h0CiAgdGhlbWUgdGhlc2Ugd291bGQgZ28gZnJvbSBsaWdodCB0byBkYXJrLgogICovCgogIC8qIEJyYW5kL2FjY2VudCAqLwoKICAvKiBTdGF0ZSBjb2xvcnMgKHdhcm4sIGVycm9yLCBzdWNjZXNzLCBpbmZvKSAqLwoKICAvKiBDZWxsIHNwZWNpZmljIHN0eWxlcyAqLwogIC8qIEEgY3VzdG9tIGJsZW5kIG9mIE1EIGdyZXkgYW5kIGJsdWUgNjAwCiAgICogU2VlIGh0dHBzOi8vbWV5ZXJ3ZWIuY29tL2VyaWMvdG9vbHMvY29sb3ItYmxlbmQvIzU0NkU3QToxRTg4RTU6NTpoZXggKi8KICAvKiBBIGN1c3RvbSBibGVuZCBvZiBNRCBncmV5IGFuZCBvcmFuZ2UgNjAwCiAgICogaHR0cHM6Ly9tZXllcndlYi5jb20vZXJpYy90b29scy9jb2xvci1ibGVuZC8jNTQ2RTdBOkY0NTExRTo1OmhleCAqLwoKICAvKiBOb3RlYm9vayBzcGVjaWZpYyBzdHlsZXMgKi8KCiAgLyogQ29uc29sZSBzcGVjaWZpYyBzdHlsZXMgKi8KCiAgLyogVG9vbGJhciBzcGVjaWZpYyBzdHlsZXMgKi8KfQoKIC8qIENvcHlyaWdodCAoYykgSnVweXRlciBEZXZlbG9wbWVudCBUZWFtLgogKiBEaXN0cmlidXRlZCB1bmRlciB0aGUgdGVybXMgb2YgdGhlIE1vZGlmaWVkIEJTRCBMaWNlbnNlLgogKi8KCiAvKgogKiBXZSBhc3N1bWUgdGhhdCB0aGUgQ1NTIHZhcmlhYmxlcyBpbgogKiBodHRwczovL2dpdGh1Yi5jb20vanVweXRlcmxhYi9qdXB5dGVybGFiL2Jsb2IvbWFzdGVyL3NyYy9kZWZhdWx0LXRoZW1lL3ZhcmlhYmxlcy5jc3MKICogaGF2ZSBiZWVuIGRlZmluZWQuCiAqLwoKIC8qIFRoaXMgZmlsZSBoYXMgY29kZSBkZXJpdmVkIGZyb20gUGhvc3Bob3JKUyBDU1MgZmlsZXMsIGFzIG5vdGVkIGJlbG93LiBUaGUgbGljZW5zZSBmb3IgdGhpcyBQaG9zcGhvckpTIGNvZGUgaXM6CgpDb3B5cmlnaHQgKGMpIDIwMTQtMjAxNywgUGhvc3Bob3JKUyBDb250cmlidXRvcnMKQWxsIHJpZ2h0cyByZXNlcnZlZC4KClJlZGlzdHJpYnV0aW9uIGFuZCB1c2UgaW4gc291cmNlIGFuZCBiaW5hcnkgZm9ybXMsIHdpdGggb3Igd2l0aG91dAptb2RpZmljYXRpb24sIGFyZSBwZXJtaXR0ZWQgcHJvdmlkZWQgdGhhdCB0aGUgZm9sbG93aW5nIGNvbmRpdGlvbnMgYXJlIG1ldDoKCiogUmVkaXN0cmlidXRpb25zIG9mIHNvdXJjZSBjb2RlIG11c3QgcmV0YWluIHRoZSBhYm92ZSBjb3B5cmlnaHQgbm90aWNlLCB0aGlzCiAgbGlzdCBvZiBjb25kaXRpb25zIGFuZCB0aGUgZm9sbG93aW5nIGRpc2NsYWltZXIuCgoqIFJlZGlzdHJpYnV0aW9ucyBpbiBiaW5hcnkgZm9ybSBtdXN0IHJlcHJvZHVjZSB0aGUgYWJvdmUgY29weXJpZ2h0IG5vdGljZSwKICB0aGlzIGxpc3Qgb2YgY29uZGl0aW9ucyBhbmQgdGhlIGZvbGxvd2luZyBkaXNjbGFpbWVyIGluIHRoZSBkb2N1bWVudGF0aW9uCiAgYW5kL29yIG90aGVyIG1hdGVyaWFscyBwcm92aWRlZCB3aXRoIHRoZSBkaXN0cmlidXRpb24uCgoqIE5laXRoZXIgdGhlIG5hbWUgb2YgdGhlIGNvcHlyaWdodCBob2xkZXIgbm9yIHRoZSBuYW1lcyBvZiBpdHMKICBjb250cmlidXRvcnMgbWF5IGJlIHVzZWQgdG8gZW5kb3JzZSBvciBwcm9tb3RlIHByb2R1Y3RzIGRlcml2ZWQgZnJvbQogIHRoaXMgc29mdHdhcmUgd2l0aG91dCBzcGVjaWZpYyBwcmlvciB3cml0dGVuIHBlcm1pc3Npb24uCgpUSElTIFNPRlRXQVJFIElTIFBST1ZJREVEIEJZIFRIRSBDT1BZUklHSFQgSE9MREVSUyBBTkQgQ09OVFJJQlVUT1JTICJBUyBJUyIKQU5EIEFOWSBFWFBSRVNTIE9SIElNUExJRUQgV0FSUkFOVElFUywgSU5DTFVESU5HLCBCVVQgTk9UIExJTUlURUQgVE8sIFRIRQpJTVBMSUVEIFdBUlJBTlRJRVMgT0YgTUVSQ0hBTlRBQklMSVRZIEFORCBGSVRORVNTIEZPUiBBIFBBUlRJQ1VMQVIgUFVSUE9TRSBBUkUKRElTQ0xBSU1FRC4gSU4gTk8gRVZFTlQgU0hBTEwgVEhFIENPUFlSSUdIVCBIT0xERVIgT1IgQ09OVFJJQlVUT1JTIEJFIExJQUJMRQpGT1IgQU5ZIERJUkVDVCwgSU5ESVJFQ1QsIElOQ0lERU5UQUwsIFNQRUNJQUwsIEVYRU1QTEFSWSwgT1IgQ09OU0VRVUVOVElBTApEQU1BR0VTIChJTkNMVURJTkcsIEJVVCBOT1QgTElNSVRFRCBUTywgUFJPQ1VSRU1FTlQgT0YgU1VCU1RJVFVURSBHT09EUyBPUgpTRVJWSUNFUzsgTE9TUyBPRiBVU0UsIERBVEEsIE9SIFBST0ZJVFM7IE9SIEJVU0lORVNTIElOVEVSUlVQVElPTikgSE9XRVZFUgpDQVVTRUQgQU5EIE9OIEFOWSBUSEVPUlkgT0YgTElBQklMSVRZLCBXSEVUSEVSIElOIENPTlRSQUNULCBTVFJJQ1QgTElBQklMSVRZLApPUiBUT1JUIChJTkNMVURJTkcgTkVHTElHRU5DRSBPUiBPVEhFUldJU0UpIEFSSVNJTkcgSU4gQU5ZIFdBWSBPVVQgT0YgVEhFIFVTRQpPRiBUSElTIFNPRlRXQVJFLCBFVkVOIElGIEFEVklTRUQgT0YgVEhFIFBPU1NJQklMSVRZIE9GIFNVQ0ggREFNQUdFLgoKKi8KCiAvKgogKiBUaGUgZm9sbG93aW5nIHNlY3Rpb24gaXMgZGVyaXZlZCBmcm9tIGh0dHBzOi8vZ2l0aHViLmNvbS9waG9zcGhvcmpzL3Bob3NwaG9yL2Jsb2IvMjNiOWQwNzVlYmM1YjczYWIxNDhiNmViZmMyMGFmOTdmODU3MTRjNC9wYWNrYWdlcy93aWRnZXRzL3N0eWxlL3RhYmJhci5jc3MgCiAqIFdlJ3ZlIHNjb3BlZCB0aGUgcnVsZXMgc28gdGhhdCB0aGV5IGFyZSBjb25zaXN0ZW50IHdpdGggZXhhY3RseSBvdXIgY29kZS4KICovCgogLmp1cHl0ZXItd2lkZ2V0cy53aWRnZXQtdGFiID4gLnAtVGFiQmFyIHsKICBkaXNwbGF5OiAtd2Via2l0LWJveDsKICBkaXNwbGF5OiAtbXMtZmxleGJveDsKICBkaXNwbGF5OiBmbGV4OwogIC13ZWJraXQtdXNlci1zZWxlY3Q6IG5vbmU7CiAgLW1vei11c2VyLXNlbGVjdDogbm9uZTsKICAtbXMtdXNlci1zZWxlY3Q6IG5vbmU7CiAgdXNlci1zZWxlY3Q6IG5vbmU7Cn0KCiAuanVweXRlci13aWRnZXRzLndpZGdldC10YWIgPiAucC1UYWJCYXJbZGF0YS1vcmllbnRhdGlvbj0naG9yaXpvbnRhbCddIHsKICAtd2Via2l0LWJveC1vcmllbnQ6IGhvcml6b250YWw7CiAgLXdlYmtpdC1ib3gtZGlyZWN0aW9uOiBub3JtYWw7CiAgICAgIC1tcy1mbGV4LWRpcmVjdGlvbjogcm93OwogICAgICAgICAgZmxleC1kaXJlY3Rpb246IHJvdzsKfQoKIC5qdXB5dGVyLXdpZGdldHMud2lkZ2V0LXRhYiA+IC5wLVRhYkJhcltkYXRhLW9yaWVudGF0aW9uPSd2ZXJ0aWNhbCddIHsKICAtd2Via2l0LWJveC1vcmllbnQ6IHZlcnRpY2FsOwogIC13ZWJraXQtYm94LWRpcmVjdGlvbjogbm9ybWFsOwogICAgICAtbXMtZmxleC1kaXJlY3Rpb246IGNvbHVtbjsKICAgICAgICAgIGZsZXgtZGlyZWN0aW9uOiBjb2x1bW47Cn0KCiAuanVweXRlci13aWRnZXRzLndpZGdldC10YWIgPiAucC1UYWJCYXIgPiAucC1UYWJCYXItY29udGVudCB7CiAgbWFyZ2luOiAwOwogIHBhZGRpbmc6IDA7CiAgZGlzcGxheTogLXdlYmtpdC1ib3g7CiAgZGlzcGxheTogLW1zLWZsZXhib3g7CiAgZGlzcGxheTogZmxleDsKICAtd2Via2l0LWJveC1mbGV4OiAxOwogICAgICAtbXMtZmxleDogMSAxIGF1dG87CiAgICAgICAgICBmbGV4OiAxIDEgYXV0bzsKICBsaXN0LXN0eWxlLXR5cGU6IG5vbmU7Cn0KCiAuanVweXRlci13aWRnZXRzLndpZGdldC10YWIgPiAucC1UYWJCYXJbZGF0YS1vcmllbnRhdGlvbj0naG9yaXpvbnRhbCddID4gLnAtVGFiQmFyLWNvbnRlbnQgewogIC13ZWJraXQtYm94LW9yaWVudDogaG9yaXpvbnRhbDsKICAtd2Via2l0LWJveC1kaXJlY3Rpb246IG5vcm1hbDsKICAgICAgLW1zLWZsZXgtZGlyZWN0aW9uOiByb3c7CiAgICAgICAgICBmbGV4LWRpcmVjdGlvbjogcm93Owp9CgogLmp1cHl0ZXItd2lkZ2V0cy53aWRnZXQtdGFiID4gLnAtVGFiQmFyW2RhdGEtb3JpZW50YXRpb249J3ZlcnRpY2FsJ10gPiAucC1UYWJCYXItY29udGVudCB7CiAgLXdlYmtpdC1ib3gtb3JpZW50OiB2ZXJ0aWNhbDsKICAtd2Via2l0LWJveC1kaXJlY3Rpb246IG5vcm1hbDsKICAgICAgLW1zLWZsZXgtZGlyZWN0aW9uOiBjb2x1bW47CiAgICAgICAgICBmbGV4LWRpcmVjdGlvbjogY29sdW1uOwp9CgogLmp1cHl0ZXItd2lkZ2V0cy53aWRnZXQtdGFiID4gLnAtVGFiQmFyIC5wLVRhYkJhci10YWIgewogIGRpc3BsYXk6IC13ZWJraXQtYm94OwogIGRpc3BsYXk6IC1tcy1mbGV4Ym94OwogIGRpc3BsYXk6IGZsZXg7CiAgLXdlYmtpdC1ib3gtb3JpZW50OiBob3Jpem9udGFsOwogIC13ZWJraXQtYm94LWRpcmVjdGlvbjogbm9ybWFsOwogICAgICAtbXMtZmxleC1kaXJlY3Rpb246IHJvdzsKICAgICAgICAgIGZsZXgtZGlyZWN0aW9uOiByb3c7CiAgLXdlYmtpdC1ib3gtc2l6aW5nOiBib3JkZXItYm94OwogICAgICAgICAgYm94LXNpemluZzogYm9yZGVyLWJveDsKICBvdmVyZmxvdzogaGlkZGVuOwp9CgogLmp1cHl0ZXItd2lkZ2V0cy53aWRnZXQtdGFiID4gLnAtVGFiQmFyIC5wLVRhYkJhci10YWJJY29uLAouanVweXRlci13aWRnZXRzLndpZGdldC10YWIgPiAucC1UYWJCYXIgLnAtVGFiQmFyLXRhYkNsb3NlSWNvbiB7CiAgLXdlYmtpdC1ib3gtZmxleDogMDsKICAgICAgLW1zLWZsZXg6IDAgMCBhdXRvOwogICAgICAgICAgZmxleDogMCAwIGF1dG87Cn0KCiAuanVweXRlci13aWRnZXRzLndpZGdldC10YWIgPiAucC1UYWJCYXIgLnAtVGFiQmFyLXRhYkxhYmVsIHsKICAtd2Via2l0LWJveC1mbGV4OiAxOwogICAgICAtbXMtZmxleDogMSAxIGF1dG87CiAgICAgICAgICBmbGV4OiAxIDEgYXV0bzsKICBvdmVyZmxvdzogaGlkZGVuOwogIHdoaXRlLXNwYWNlOiBub3dyYXA7Cn0KCiAuanVweXRlci13aWRnZXRzLndpZGdldC10YWIgPiAucC1UYWJCYXIgLnAtVGFiQmFyLXRhYi5wLW1vZC1oaWRkZW4gewogIGRpc3BsYXk6IG5vbmUgIWltcG9ydGFudDsKfQoKIC5qdXB5dGVyLXdpZGdldHMud2lkZ2V0LXRhYiA+IC5wLVRhYkJhci5wLW1vZC1kcmFnZ2luZyAucC1UYWJCYXItdGFiIHsKICBwb3NpdGlvbjogcmVsYXRpdmU7Cn0KCiAuanVweXRlci13aWRnZXRzLndpZGdldC10YWIgPiAucC1UYWJCYXIucC1tb2QtZHJhZ2dpbmdbZGF0YS1vcmllbnRhdGlvbj0naG9yaXpvbnRhbCddIC5wLVRhYkJhci10YWIgewogIGxlZnQ6IDA7CiAgLXdlYmtpdC10cmFuc2l0aW9uOiBsZWZ0IDE1MG1zIGVhc2U7CiAgdHJhbnNpdGlvbjogbGVmdCAxNTBtcyBlYXNlOwp9CgogLmp1cHl0ZXItd2lkZ2V0cy53aWRnZXQtdGFiID4gLnAtVGFiQmFyLnAtbW9kLWRyYWdnaW5nW2RhdGEtb3JpZW50YXRpb249J3ZlcnRpY2FsJ10gLnAtVGFiQmFyLXRhYiB7CiAgdG9wOiAwOwogIC13ZWJraXQtdHJhbnNpdGlvbjogdG9wIDE1MG1zIGVhc2U7CiAgdHJhbnNpdGlvbjogdG9wIDE1MG1zIGVhc2U7Cn0KCiAuanVweXRlci13aWRnZXRzLndpZGdldC10YWIgPiAucC1UYWJCYXIucC1tb2QtZHJhZ2dpbmcgLnAtVGFiQmFyLXRhYi5wLW1vZC1kcmFnZ2luZyB7CiAgLXdlYmtpdC10cmFuc2l0aW9uOiBub25lOwogIHRyYW5zaXRpb246IG5vbmU7Cn0KCiAvKiBFbmQgdGFiYmFyLmNzcyAqLwoKIDpyb290IHsgLyogbWFyZ2luIGJldHdlZW4gaW5saW5lIGVsZW1lbnRzICovCgogICAgLyogRnJvbSBNYXRlcmlhbCBEZXNpZ24gTGl0ZSAqLwp9CgogLmp1cHl0ZXItd2lkZ2V0cyB7CiAgICBtYXJnaW46IDJweDsKICAgIC13ZWJraXQtYm94LXNpemluZzogYm9yZGVyLWJveDsKICAgICAgICAgICAgYm94LXNpemluZzogYm9yZGVyLWJveDsKICAgIGNvbG9yOiBibGFjazsKICAgIG92ZXJmbG93OiB2aXNpYmxlOwp9CgogLmp1cHl0ZXItd2lkZ2V0cy5qdXB5dGVyLXdpZGdldHMtZGlzY29ubmVjdGVkOjpiZWZvcmUgewogICAgbGluZS1oZWlnaHQ6IDI4cHg7CiAgICBoZWlnaHQ6IDI4cHg7Cn0KCiAuanAtT3V0cHV0LXJlc3VsdCA+IC5qdXB5dGVyLXdpZGdldHMgewogICAgbWFyZ2luLWxlZnQ6IDA7CiAgICBtYXJnaW4tcmlnaHQ6IDA7Cn0KCiAvKiB2Ym94IGFuZCBoYm94ICovCgogLndpZGdldC1pbmxpbmUtaGJveCB7CiAgICAvKiBIb3Jpem9udGFsIHdpZGdldHMgKi8KICAgIC13ZWJraXQtYm94LXNpemluZzogYm9yZGVyLWJveDsKICAgICAgICAgICAgYm94LXNpemluZzogYm9yZGVyLWJveDsKICAgIGRpc3BsYXk6IC13ZWJraXQtYm94OwogICAgZGlzcGxheTogLW1zLWZsZXhib3g7CiAgICBkaXNwbGF5OiBmbGV4OwogICAgLXdlYmtpdC1ib3gtb3JpZW50OiBob3Jpem9udGFsOwogICAgLXdlYmtpdC1ib3gtZGlyZWN0aW9uOiBub3JtYWw7CiAgICAgICAgLW1zLWZsZXgtZGlyZWN0aW9uOiByb3c7CiAgICAgICAgICAgIGZsZXgtZGlyZWN0aW9uOiByb3c7CiAgICAtd2Via2l0LWJveC1hbGlnbjogYmFzZWxpbmU7CiAgICAgICAgLW1zLWZsZXgtYWxpZ246IGJhc2VsaW5lOwogICAgICAgICAgICBhbGlnbi1pdGVtczogYmFzZWxpbmU7Cn0KCiAud2lkZ2V0LWlubGluZS12Ym94IHsKICAgIC8qIFZlcnRpY2FsIFdpZGdldHMgKi8KICAgIC13ZWJraXQtYm94LXNpemluZzogYm9yZGVyLWJveDsKICAgICAgICAgICAgYm94LXNpemluZzogYm9yZGVyLWJveDsKICAgIGRpc3BsYXk6IC13ZWJraXQtYm94OwogICAgZGlzcGxheTogLW1zLWZsZXhib3g7CiAgICBkaXNwbGF5OiBmbGV4OwogICAgLXdlYmtpdC1ib3gtb3JpZW50OiB2ZXJ0aWNhbDsKICAgIC13ZWJraXQtYm94LWRpcmVjdGlvbjogbm9ybWFsOwogICAgICAgIC1tcy1mbGV4LWRpcmVjdGlvbjogY29sdW1uOwogICAgICAgICAgICBmbGV4LWRpcmVjdGlvbjogY29sdW1uOwogICAgLXdlYmtpdC1ib3gtYWxpZ246IGNlbnRlcjsKICAgICAgICAtbXMtZmxleC1hbGlnbjogY2VudGVyOwogICAgICAgICAgICBhbGlnbi1pdGVtczogY2VudGVyOwp9CgogLndpZGdldC1ib3ggewogICAgLXdlYmtpdC1ib3gtc2l6aW5nOiBib3JkZXItYm94OwogICAgICAgICAgICBib3gtc2l6aW5nOiBib3JkZXItYm94OwogICAgZGlzcGxheTogLXdlYmtpdC1ib3g7CiAgICBkaXNwbGF5OiAtbXMtZmxleGJveDsKICAgIGRpc3BsYXk6IGZsZXg7CiAgICBtYXJnaW46IDA7CiAgICBvdmVyZmxvdzogYXV0bzsKfQoKIC53aWRnZXQtZ3JpZGJveCB7CiAgICAtd2Via2l0LWJveC1zaXppbmc6IGJvcmRlci1ib3g7CiAgICAgICAgICAgIGJveC1zaXppbmc6IGJvcmRlci1ib3g7CiAgICBkaXNwbGF5OiBncmlkOwogICAgbWFyZ2luOiAwOwogICAgb3ZlcmZsb3c6IGF1dG87Cn0KCiAud2lkZ2V0LWhib3ggewogICAgLXdlYmtpdC1ib3gtb3JpZW50OiBob3Jpem9udGFsOwogICAgLXdlYmtpdC1ib3gtZGlyZWN0aW9uOiBub3JtYWw7CiAgICAgICAgLW1zLWZsZXgtZGlyZWN0aW9uOiByb3c7CiAgICAgICAgICAgIGZsZXgtZGlyZWN0aW9uOiByb3c7Cn0KCiAud2lkZ2V0LXZib3ggewogICAgLXdlYmtpdC1ib3gtb3JpZW50OiB2ZXJ0aWNhbDsKICAgIC13ZWJraXQtYm94LWRpcmVjdGlvbjogbm9ybWFsOwogICAgICAgIC1tcy1mbGV4LWRpcmVjdGlvbjogY29sdW1uOwogICAgICAgICAgICBmbGV4LWRpcmVjdGlvbjogY29sdW1uOwp9CgogLyogR2VuZXJhbCBCdXR0b24gU3R5bGluZyAqLwoKIC5qdXB5dGVyLWJ1dHRvbiB7CiAgICBwYWRkaW5nLWxlZnQ6IDEwcHg7CiAgICBwYWRkaW5nLXJpZ2h0OiAxMHB4OwogICAgcGFkZGluZy10b3A6IDBweDsKICAgIHBhZGRpbmctYm90dG9tOiAwcHg7CiAgICBkaXNwbGF5OiBpbmxpbmUtYmxvY2s7CiAgICB3aGl0ZS1zcGFjZTogbm93cmFwOwogICAgb3ZlcmZsb3c6IGhpZGRlbjsKICAgIHRleHQtb3ZlcmZsb3c6IGVsbGlwc2lzOwogICAgdGV4dC1hbGlnbjogY2VudGVyOwogICAgZm9udC1zaXplOiAxM3B4OwogICAgY3Vyc29yOiBwb2ludGVyOwoKICAgIGhlaWdodDogMjhweDsKICAgIGJvcmRlcjogMHB4IHNvbGlkOwogICAgbGluZS1oZWlnaHQ6IDI4cHg7CiAgICAtd2Via2l0LWJveC1zaGFkb3c6IG5vbmU7CiAgICAgICAgICAgIGJveC1zaGFkb3c6IG5vbmU7CgogICAgY29sb3I6IHJnYmEoMCwgMCwgMCwgLjgpOwogICAgYmFja2dyb3VuZC1jb2xvcjogI0VFRUVFRTsKICAgIGJvcmRlci1jb2xvcjogI0UwRTBFMDsKICAgIGJvcmRlcjogbm9uZTsKfQoKIC5qdXB5dGVyLWJ1dHRvbiBpLmZhIHsKICAgIG1hcmdpbi1yaWdodDogNHB4OwogICAgcG9pbnRlci1ldmVudHM6IG5vbmU7Cn0KCiAuanVweXRlci1idXR0b246ZW1wdHk6YmVmb3JlIHsKICAgIGNvbnRlbnQ6ICJcMjAwYiI7IC8qIHplcm8td2lkdGggc3BhY2UgKi8KfQoKIC5qdXB5dGVyLXdpZGdldHMuanVweXRlci1idXR0b246ZGlzYWJsZWQgewogICAgb3BhY2l0eTogMC42Owp9CgogLmp1cHl0ZXItYnV0dG9uIGkuZmEuY2VudGVyIHsKICAgIG1hcmdpbi1yaWdodDogMDsKfQoKIC5qdXB5dGVyLWJ1dHRvbjpob3ZlcjplbmFibGVkLCAuanVweXRlci1idXR0b246Zm9jdXM6ZW5hYmxlZCB7CiAgICAvKiBNRCBMaXRlIDJkcCBzaGFkb3cgKi8KICAgIC13ZWJraXQtYm94LXNoYWRvdzogMCAycHggMnB4IDAgcmdiYSgwLCAwLCAwLCAuMTQpLAogICAgICAgICAgICAgICAgMCAzcHggMXB4IC0ycHggcmdiYSgwLCAwLCAwLCAuMiksCiAgICAgICAgICAgICAgICAwIDFweCA1cHggMCByZ2JhKDAsIDAsIDAsIC4xMik7CiAgICAgICAgICAgIGJveC1zaGFkb3c6IDAgMnB4IDJweCAwIHJnYmEoMCwgMCwgMCwgLjE0KSwKICAgICAgICAgICAgICAgIDAgM3B4IDFweCAtMnB4IHJnYmEoMCwgMCwgMCwgLjIpLAogICAgICAgICAgICAgICAgMCAxcHggNXB4IDAgcmdiYSgwLCAwLCAwLCAuMTIpOwp9CgogLmp1cHl0ZXItYnV0dG9uOmFjdGl2ZSwgLmp1cHl0ZXItYnV0dG9uLm1vZC1hY3RpdmUgewogICAgLyogTUQgTGl0ZSA0ZHAgc2hhZG93ICovCiAgICAtd2Via2l0LWJveC1zaGFkb3c6IDAgNHB4IDVweCAwIHJnYmEoMCwgMCwgMCwgLjE0KSwKICAgICAgICAgICAgICAgIDAgMXB4IDEwcHggMCByZ2JhKDAsIDAsIDAsIC4xMiksCiAgICAgICAgICAgICAgICAwIDJweCA0cHggLTFweCByZ2JhKDAsIDAsIDAsIC4yKTsKICAgICAgICAgICAgYm94LXNoYWRvdzogMCA0cHggNXB4IDAgcmdiYSgwLCAwLCAwLCAuMTQpLAogICAgICAgICAgICAgICAgMCAxcHggMTBweCAwIHJnYmEoMCwgMCwgMCwgLjEyKSwKICAgICAgICAgICAgICAgIDAgMnB4IDRweCAtMXB4IHJnYmEoMCwgMCwgMCwgLjIpOwogICAgY29sb3I6IHJnYmEoMCwgMCwgMCwgLjgpOwogICAgYmFja2dyb3VuZC1jb2xvcjogI0JEQkRCRDsKfQoKIC5qdXB5dGVyLWJ1dHRvbjpmb2N1czplbmFibGVkIHsKICAgIG91dGxpbmU6IDFweCBzb2xpZCAjNjRCNUY2Owp9CgogLyogQnV0dG9uICJQcmltYXJ5IiBTdHlsaW5nICovCgogLmp1cHl0ZXItYnV0dG9uLm1vZC1wcmltYXJ5IHsKICAgIGNvbG9yOiByZ2JhKDI1NSwgMjU1LCAyNTUsIDEuMCk7CiAgICBiYWNrZ3JvdW5kLWNvbG9yOiAjMjE5NkYzOwp9CgogLmp1cHl0ZXItYnV0dG9uLm1vZC1wcmltYXJ5Lm1vZC1hY3RpdmUgewogICAgY29sb3I6IHJnYmEoMjU1LCAyNTUsIDI1NSwgMSk7CiAgICBiYWNrZ3JvdW5kLWNvbG9yOiAjMTk3NkQyOwp9CgogLmp1cHl0ZXItYnV0dG9uLm1vZC1wcmltYXJ5OmFjdGl2ZSB7CiAgICBjb2xvcjogcmdiYSgyNTUsIDI1NSwgMjU1LCAxKTsKICAgIGJhY2tncm91bmQtY29sb3I6ICMxOTc2RDI7Cn0KCiAvKiBCdXR0b24gIlN1Y2Nlc3MiIFN0eWxpbmcgKi8KCiAuanVweXRlci1idXR0b24ubW9kLXN1Y2Nlc3MgewogICAgY29sb3I6IHJnYmEoMjU1LCAyNTUsIDI1NSwgMS4wKTsKICAgIGJhY2tncm91bmQtY29sb3I6ICM0Q0FGNTA7Cn0KCiAuanVweXRlci1idXR0b24ubW9kLXN1Y2Nlc3MubW9kLWFjdGl2ZSB7CiAgICBjb2xvcjogcmdiYSgyNTUsIDI1NSwgMjU1LCAxKTsKICAgIGJhY2tncm91bmQtY29sb3I6ICMzODhFM0M7CiB9CgogLmp1cHl0ZXItYnV0dG9uLm1vZC1zdWNjZXNzOmFjdGl2ZSB7CiAgICBjb2xvcjogcmdiYSgyNTUsIDI1NSwgMjU1LCAxKTsKICAgIGJhY2tncm91bmQtY29sb3I6ICMzODhFM0M7CiB9CgogLyogQnV0dG9uICJJbmZvIiBTdHlsaW5nICovCgogLmp1cHl0ZXItYnV0dG9uLm1vZC1pbmZvIHsKICAgIGNvbG9yOiByZ2JhKDI1NSwgMjU1LCAyNTUsIDEuMCk7CiAgICBiYWNrZ3JvdW5kLWNvbG9yOiAjMDBCQ0Q0Owp9CgogLmp1cHl0ZXItYnV0dG9uLm1vZC1pbmZvLm1vZC1hY3RpdmUgewogICAgY29sb3I6IHJnYmEoMjU1LCAyNTUsIDI1NSwgMSk7CiAgICBiYWNrZ3JvdW5kLWNvbG9yOiAjMDA5N0E3Owp9CgogLmp1cHl0ZXItYnV0dG9uLm1vZC1pbmZvOmFjdGl2ZSB7CiAgICBjb2xvcjogcmdiYSgyNTUsIDI1NSwgMjU1LCAxKTsKICAgIGJhY2tncm91bmQtY29sb3I6ICMwMDk3QTc7Cn0KCiAvKiBCdXR0b24gIldhcm5pbmciIFN0eWxpbmcgKi8KCiAuanVweXRlci1idXR0b24ubW9kLXdhcm5pbmcgewogICAgY29sb3I6IHJnYmEoMjU1LCAyNTUsIDI1NSwgMS4wKTsKICAgIGJhY2tncm91bmQtY29sb3I6ICNGRjk4MDA7Cn0KCiAuanVweXRlci1idXR0b24ubW9kLXdhcm5pbmcubW9kLWFjdGl2ZSB7CiAgICBjb2xvcjogcmdiYSgyNTUsIDI1NSwgMjU1LCAxKTsKICAgIGJhY2tncm91bmQtY29sb3I6ICNGNTdDMDA7Cn0KCiAuanVweXRlci1idXR0b24ubW9kLXdhcm5pbmc6YWN0aXZlIHsKICAgIGNvbG9yOiByZ2JhKDI1NSwgMjU1LCAyNTUsIDEpOwogICAgYmFja2dyb3VuZC1jb2xvcjogI0Y1N0MwMDsKfQoKIC8qIEJ1dHRvbiAiRGFuZ2VyIiBTdHlsaW5nICovCgogLmp1cHl0ZXItYnV0dG9uLm1vZC1kYW5nZXIgewogICAgY29sb3I6IHJnYmEoMjU1LCAyNTUsIDI1NSwgMS4wKTsKICAgIGJhY2tncm91bmQtY29sb3I6ICNGNDQzMzY7Cn0KCiAuanVweXRlci1idXR0b24ubW9kLWRhbmdlci5tb2QtYWN0aXZlIHsKICAgIGNvbG9yOiByZ2JhKDI1NSwgMjU1LCAyNTUsIDEpOwogICAgYmFja2dyb3VuZC1jb2xvcjogI0QzMkYyRjsKfQoKIC5qdXB5dGVyLWJ1dHRvbi5tb2QtZGFuZ2VyOmFjdGl2ZSB7CiAgICBjb2xvcjogcmdiYSgyNTUsIDI1NSwgMjU1LCAxKTsKICAgIGJhY2tncm91bmQtY29sb3I6ICNEMzJGMkY7Cn0KCiAvKiBXaWRnZXQgQnV0dG9uKi8KCiAud2lkZ2V0LWJ1dHRvbiwgLndpZGdldC10b2dnbGUtYnV0dG9uIHsKICAgIHdpZHRoOiAxNDhweDsKfQoKIC8qIFdpZGdldCBMYWJlbCBTdHlsaW5nICovCgogLyogT3ZlcnJpZGUgQm9vdHN0cmFwIGxhYmVsIGNzcyAqLwoKIC5qdXB5dGVyLXdpZGdldHMgbGFiZWwgewogICAgbWFyZ2luLWJvdHRvbTogMDsKICAgIG1hcmdpbi1ib3R0b206IGluaXRpYWw7Cn0KCiAud2lkZ2V0LWxhYmVsLWJhc2ljIHsKICAgIC8qIEJhc2ljIExhYmVsICovCiAgICBjb2xvcjogYmxhY2s7CiAgICBmb250LXNpemU6IDEzcHg7CiAgICBvdmVyZmxvdzogaGlkZGVuOwogICAgdGV4dC1vdmVyZmxvdzogZWxsaXBzaXM7CiAgICB3aGl0ZS1zcGFjZTogbm93cmFwOwogICAgbGluZS1oZWlnaHQ6IDI4cHg7Cn0KCiAud2lkZ2V0LWxhYmVsIHsKICAgIC8qIExhYmVsICovCiAgICBjb2xvcjogYmxhY2s7CiAgICBmb250LXNpemU6IDEzcHg7CiAgICBvdmVyZmxvdzogaGlkZGVuOwogICAgdGV4dC1vdmVyZmxvdzogZWxsaXBzaXM7CiAgICB3aGl0ZS1zcGFjZTogbm93cmFwOwogICAgbGluZS1oZWlnaHQ6IDI4cHg7Cn0KCiAud2lkZ2V0LWlubGluZS1oYm94IC53aWRnZXQtbGFiZWwgewogICAgLyogSG9yaXpvbnRhbCBXaWRnZXQgTGFiZWwgKi8KICAgIGNvbG9yOiBibGFjazsKICAgIHRleHQtYWxpZ246IHJpZ2h0OwogICAgbWFyZ2luLXJpZ2h0OiA4cHg7CiAgICB3aWR0aDogODBweDsKICAgIC1tcy1mbGV4LW5lZ2F0aXZlOiAwOwogICAgICAgIGZsZXgtc2hyaW5rOiAwOwp9CgogLndpZGdldC1pbmxpbmUtdmJveCAud2lkZ2V0LWxhYmVsIHsKICAgIC8qIFZlcnRpY2FsIFdpZGdldCBMYWJlbCAqLwogICAgY29sb3I6IGJsYWNrOwogICAgdGV4dC1hbGlnbjogY2VudGVyOwogICAgbGluZS1oZWlnaHQ6IDI4cHg7Cn0KCiAvKiBXaWRnZXQgUmVhZG91dCBTdHlsaW5nICovCgogLndpZGdldC1yZWFkb3V0IHsKICAgIGNvbG9yOiBibGFjazsKICAgIGZvbnQtc2l6ZTogMTNweDsKICAgIGhlaWdodDogMjhweDsKICAgIGxpbmUtaGVpZ2h0OiAyOHB4OwogICAgb3ZlcmZsb3c6IGhpZGRlbjsKICAgIHdoaXRlLXNwYWNlOiBub3dyYXA7CiAgICB0ZXh0LWFsaWduOiBjZW50ZXI7Cn0KCiAud2lkZ2V0LXJlYWRvdXQub3ZlcmZsb3cgewogICAgLyogT3ZlcmZsb3dpbmcgUmVhZG91dCAqLwoKICAgIC8qIEZyb20gTWF0ZXJpYWwgRGVzaWduIExpdGUKICAgICAgICBzaGFkb3cta2V5LXVtYnJhLW9wYWNpdHk6IDAuMjsKICAgICAgICBzaGFkb3cta2V5LXBlbnVtYnJhLW9wYWNpdHk6IDAuMTQ7CiAgICAgICAgc2hhZG93LWFtYmllbnQtc2hhZG93LW9wYWNpdHk6IDAuMTI7CiAgICAgKi8KICAgIC13ZWJraXQtYm94LXNoYWRvdzogMCAycHggMnB4IDAgcmdiYSgwLCAwLCAwLCAuMiksCiAgICAgICAgICAgICAgICAgICAgICAgIDAgM3B4IDFweCAtMnB4IHJnYmEoMCwgMCwgMCwgLjE0KSwKICAgICAgICAgICAgICAgICAgICAgICAgMCAxcHggNXB4IDAgcmdiYSgwLCAwLCAwLCAuMTIpOwoKICAgIGJveC1zaGFkb3c6IDAgMnB4IDJweCAwIHJnYmEoMCwgMCwgMCwgLjIpLAogICAgICAgICAgICAgICAgMCAzcHggMXB4IC0ycHggcmdiYSgwLCAwLCAwLCAuMTQpLAogICAgICAgICAgICAgICAgMCAxcHggNXB4IDAgcmdiYSgwLCAwLCAwLCAuMTIpOwp9CgogLndpZGdldC1pbmxpbmUtaGJveCAud2lkZ2V0LXJlYWRvdXQgewogICAgLyogSG9yaXpvbnRhbCBSZWFkb3V0ICovCiAgICB0ZXh0LWFsaWduOiBjZW50ZXI7CiAgICBtYXgtd2lkdGg6IDE0OHB4OwogICAgbWluLXdpZHRoOiA3MnB4OwogICAgbWFyZ2luLWxlZnQ6IDRweDsKfQoKIC53aWRnZXQtaW5saW5lLXZib3ggLndpZGdldC1yZWFkb3V0IHsKICAgIC8qIFZlcnRpY2FsIFJlYWRvdXQgKi8KICAgIG1hcmdpbi10b3A6IDRweDsKICAgIC8qIGFzIHdpZGUgYXMgdGhlIHdpZGdldCAqLwogICAgd2lkdGg6IGluaGVyaXQ7Cn0KCiAvKiBXaWRnZXQgQ2hlY2tib3ggU3R5bGluZyAqLwoKIC53aWRnZXQtY2hlY2tib3ggewogICAgd2lkdGg6IDMwMHB4OwogICAgaGVpZ2h0OiAyOHB4OwogICAgbGluZS1oZWlnaHQ6IDI4cHg7Cn0KCiAud2lkZ2V0LWNoZWNrYm94IGlucHV0W3R5cGU9ImNoZWNrYm94Il0gewogICAgbWFyZ2luOiAwcHggOHB4IDBweCAwcHg7CiAgICBsaW5lLWhlaWdodDogMjhweDsKICAgIGZvbnQtc2l6ZTogbGFyZ2U7CiAgICAtd2Via2l0LWJveC1mbGV4OiAxOwogICAgICAgIC1tcy1mbGV4LXBvc2l0aXZlOiAxOwogICAgICAgICAgICBmbGV4LWdyb3c6IDE7CiAgICAtbXMtZmxleC1uZWdhdGl2ZTogMDsKICAgICAgICBmbGV4LXNocmluazogMDsKICAgIC1tcy1mbGV4LWl0ZW0tYWxpZ246IGNlbnRlcjsKICAgICAgICBhbGlnbi1zZWxmOiBjZW50ZXI7Cn0KCiAvKiBXaWRnZXQgVmFsaWQgU3R5bGluZyAqLwoKIC53aWRnZXQtdmFsaWQgewogICAgaGVpZ2h0OiAyOHB4OwogICAgbGluZS1oZWlnaHQ6IDI4cHg7CiAgICB3aWR0aDogMTQ4cHg7CiAgICBmb250LXNpemU6IDEzcHg7Cn0KCiAud2lkZ2V0LXZhbGlkIGk6YmVmb3JlIHsKICAgIGxpbmUtaGVpZ2h0OiAyOHB4OwogICAgbWFyZ2luLXJpZ2h0OiA0cHg7CiAgICBtYXJnaW4tbGVmdDogNHB4OwoKICAgIC8qIGZyb20gdGhlIGZhIGNsYXNzIGluIEZvbnRBd2Vzb21lOiBodHRwczovL2dpdGh1Yi5jb20vRm9ydEF3ZXNvbWUvRm9udC1Bd2Vzb21lL2Jsb2IvNDkxMDBjN2MzYTdiNThkNTBiYWE3MWVmZWYxMWFmNDFhNjZiMDNkMy9jc3MvZm9udC1hd2Vzb21lLmNzcyNMMTQgKi8KICAgIGRpc3BsYXk6IGlubGluZS1ibG9jazsKICAgIGZvbnQ6IG5vcm1hbCBub3JtYWwgbm9ybWFsIDE0cHgvMSBGb250QXdlc29tZTsKICAgIGZvbnQtc2l6ZTogaW5oZXJpdDsKICAgIHRleHQtcmVuZGVyaW5nOiBhdXRvOwogICAgLXdlYmtpdC1mb250LXNtb290aGluZzogYW50aWFsaWFzZWQ7CiAgICAtbW96LW9zeC1mb250LXNtb290aGluZzogZ3JheXNjYWxlOwp9CgogLndpZGdldC12YWxpZC5tb2QtdmFsaWQgaTpiZWZvcmUgewogICAgY29udGVudDogIlxmMDBjIjsKICAgIGNvbG9yOiBncmVlbjsKfQoKIC53aWRnZXQtdmFsaWQubW9kLWludmFsaWQgaTpiZWZvcmUgewogICAgY29udGVudDogIlxmMDBkIjsKICAgIGNvbG9yOiByZWQ7Cn0KCiAud2lkZ2V0LXZhbGlkLm1vZC12YWxpZCAud2lkZ2V0LXZhbGlkLXJlYWRvdXQgewogICAgZGlzcGxheTogbm9uZTsKfQoKIC8qIFdpZGdldCBUZXh0IGFuZCBUZXh0QXJlYSBTdHlpbmcgKi8KCiAud2lkZ2V0LXRleHRhcmVhLCAud2lkZ2V0LXRleHQgewogICAgd2lkdGg6IDMwMHB4Owp9CgogLndpZGdldC10ZXh0IGlucHV0W3R5cGU9InRleHQiXSwgLndpZGdldC10ZXh0IGlucHV0W3R5cGU9Im51bWJlciJdewogICAgaGVpZ2h0OiAyOHB4OwogICAgbGluZS1oZWlnaHQ6IDI4cHg7Cn0KCiAud2lkZ2V0LXRleHQgaW5wdXRbdHlwZT0idGV4dCJdOmRpc2FibGVkLCAud2lkZ2V0LXRleHQgaW5wdXRbdHlwZT0ibnVtYmVyIl06ZGlzYWJsZWQsIC53aWRnZXQtdGV4dGFyZWEgdGV4dGFyZWE6ZGlzYWJsZWQgewogICAgb3BhY2l0eTogMC42Owp9CgogLndpZGdldC10ZXh0IGlucHV0W3R5cGU9InRleHQiXSwgLndpZGdldC10ZXh0IGlucHV0W3R5cGU9Im51bWJlciJdLCAud2lkZ2V0LXRleHRhcmVhIHRleHRhcmVhIHsKICAgIC13ZWJraXQtYm94LXNpemluZzogYm9yZGVyLWJveDsKICAgICAgICAgICAgYm94LXNpemluZzogYm9yZGVyLWJveDsKICAgIGJvcmRlcjogMXB4IHNvbGlkICM5RTlFOUU7CiAgICBiYWNrZ3JvdW5kLWNvbG9yOiB3aGl0ZTsKICAgIGNvbG9yOiByZ2JhKDAsIDAsIDAsIC44KTsKICAgIGZvbnQtc2l6ZTogMTNweDsKICAgIHBhZGRpbmc6IDRweCA4cHg7CiAgICAtd2Via2l0LWJveC1mbGV4OiAxOwogICAgICAgIC1tcy1mbGV4LXBvc2l0aXZlOiAxOwogICAgICAgICAgICBmbGV4LWdyb3c6IDE7CiAgICBtaW4td2lkdGg6IDA7IC8qIFRoaXMgbWFrZXMgaXQgcG9zc2libGUgZm9yIHRoZSBmbGV4Ym94IHRvIHNocmluayB0aGlzIGlucHV0ICovCiAgICAtbXMtZmxleC1uZWdhdGl2ZTogMTsKICAgICAgICBmbGV4LXNocmluazogMTsKICAgIG91dGxpbmU6IG5vbmUgIWltcG9ydGFudDsKfQoKIC53aWRnZXQtdGV4dGFyZWEgdGV4dGFyZWEgewogICAgaGVpZ2h0OiBpbmhlcml0OwogICAgd2lkdGg6IGluaGVyaXQ7Cn0KCiAud2lkZ2V0LXRleHQgaW5wdXQ6Zm9jdXMsIC53aWRnZXQtdGV4dGFyZWEgdGV4dGFyZWE6Zm9jdXMgewogICAgYm9yZGVyLWNvbG9yOiAjNjRCNUY2Owp9CgogLyogV2lkZ2V0IFNsaWRlciAqLwoKIC53aWRnZXQtc2xpZGVyIC51aS1zbGlkZXIgewogICAgLyogU2xpZGVyIFRyYWNrICovCiAgICBib3JkZXI6IDFweCBzb2xpZCAjQkRCREJEOwogICAgYmFja2dyb3VuZDogI0JEQkRCRDsKICAgIC13ZWJraXQtYm94LXNpemluZzogYm9yZGVyLWJveDsKICAgICAgICAgICAgYm94LXNpemluZzogYm9yZGVyLWJveDsKICAgIHBvc2l0aW9uOiByZWxhdGl2ZTsKICAgIGJvcmRlci1yYWRpdXM6IDBweDsKfQoKIC53aWRnZXQtc2xpZGVyIC51aS1zbGlkZXIgLnVpLXNsaWRlci1oYW5kbGUgewogICAgLyogU2xpZGVyIEhhbmRsZSAqLwogICAgb3V0bGluZTogbm9uZSAhaW1wb3J0YW50OyAvKiBmb2N1c2VkIHNsaWRlciBoYW5kbGVzIGFyZSBjb2xvcmVkIC0gc2VlIGJlbG93ICovCiAgICBwb3NpdGlvbjogYWJzb2x1dGU7CiAgICBiYWNrZ3JvdW5kLWNvbG9yOiB3aGl0ZTsKICAgIGJvcmRlcjogMXB4IHNvbGlkICM5RTlFOUU7CiAgICAtd2Via2l0LWJveC1zaXppbmc6IGJvcmRlci1ib3g7CiAgICAgICAgICAgIGJveC1zaXppbmc6IGJvcmRlci1ib3g7CiAgICB6LWluZGV4OiAxOwogICAgYmFja2dyb3VuZC1pbWFnZTogbm9uZTsgLyogT3ZlcnJpZGUganF1ZXJ5LXVpICovCn0KCiAvKiBPdmVycmlkZSBqcXVlcnktdWkgKi8KCiAud2lkZ2V0LXNsaWRlciAudWktc2xpZGVyIC51aS1zbGlkZXItaGFuZGxlOmhvdmVyLCAud2lkZ2V0LXNsaWRlciAudWktc2xpZGVyIC51aS1zbGlkZXItaGFuZGxlOmZvY3VzIHsKICAgIGJhY2tncm91bmQtY29sb3I6ICMyMTk2RjM7CiAgICBib3JkZXI6IDFweCBzb2xpZCAjMjE5NkYzOwp9CgogLndpZGdldC1zbGlkZXIgLnVpLXNsaWRlciAudWktc2xpZGVyLWhhbmRsZTphY3RpdmUgewogICAgYmFja2dyb3VuZC1jb2xvcjogIzIxOTZGMzsKICAgIGJvcmRlci1jb2xvcjogIzIxOTZGMzsKICAgIHotaW5kZXg6IDI7CiAgICAtd2Via2l0LXRyYW5zZm9ybTogc2NhbGUoMS4yKTsKICAgICAgICAgICAgdHJhbnNmb3JtOiBzY2FsZSgxLjIpOwp9CgogLndpZGdldC1zbGlkZXIgIC51aS1zbGlkZXIgLnVpLXNsaWRlci1yYW5nZSB7CiAgICAvKiBJbnRlcnZhbCBiZXR3ZWVuIHRoZSB0d28gc3BlY2lmaWVkIHZhbHVlIG9mIGEgZG91YmxlIHNsaWRlciAqLwogICAgcG9zaXRpb246IGFic29sdXRlOwogICAgYmFja2dyb3VuZDogIzIxOTZGMzsKICAgIHotaW5kZXg6IDA7Cn0KCiAvKiBTaGFwZXMgb2YgU2xpZGVyIEhhbmRsZXMgKi8KCiAud2lkZ2V0LWhzbGlkZXIgLnVpLXNsaWRlciAudWktc2xpZGVyLWhhbmRsZSB7CiAgICB3aWR0aDogMTZweDsKICAgIGhlaWdodDogMTZweDsKICAgIG1hcmdpbi10b3A6IC03cHg7CiAgICBtYXJnaW4tbGVmdDogLTdweDsKICAgIGJvcmRlci1yYWRpdXM6IDUwJTsKICAgIHRvcDogMDsKfQoKIC53aWRnZXQtdnNsaWRlciAudWktc2xpZGVyIC51aS1zbGlkZXItaGFuZGxlIHsKICAgIHdpZHRoOiAxNnB4OwogICAgaGVpZ2h0OiAxNnB4OwogICAgbWFyZ2luLWJvdHRvbTogLTdweDsKICAgIG1hcmdpbi1sZWZ0OiAtN3B4OwogICAgYm9yZGVyLXJhZGl1czogNTAlOwogICAgbGVmdDogMDsKfQoKIC53aWRnZXQtaHNsaWRlciAudWktc2xpZGVyIC51aS1zbGlkZXItcmFuZ2UgewogICAgaGVpZ2h0OiA4cHg7CiAgICBtYXJnaW4tdG9wOiAtM3B4Owp9CgogLndpZGdldC12c2xpZGVyIC51aS1zbGlkZXIgLnVpLXNsaWRlci1yYW5nZSB7CiAgICB3aWR0aDogOHB4OwogICAgbWFyZ2luLWxlZnQ6IC0zcHg7Cn0KCiAvKiBIb3Jpem9udGFsIFNsaWRlciAqLwoKIC53aWRnZXQtaHNsaWRlciB7CiAgICB3aWR0aDogMzAwcHg7CiAgICBoZWlnaHQ6IDI4cHg7CiAgICBsaW5lLWhlaWdodDogMjhweDsKCiAgICAvKiBPdmVycmlkZSB0aGUgYWxpZ24taXRlbXMgYmFzZWxpbmUuIFRoaXMgd2F5LCB0aGUgZGVzY3JpcHRpb24gYW5kIHJlYWRvdXQKICAgIHN0aWxsIHNlZW0gdG8gYWxpZ24gdGhlaXIgYmFzZWxpbmUgcHJvcGVybHksIGFuZCB3ZSBkb24ndCBoYXZlIHRvIGhhdmUKICAgIGFsaWduLXNlbGY6IHN0cmV0Y2ggaW4gdGhlIC5zbGlkZXItY29udGFpbmVyLiAqLwogICAgLXdlYmtpdC1ib3gtYWxpZ246IGNlbnRlcjsKICAgICAgICAtbXMtZmxleC1hbGlnbjogY2VudGVyOwogICAgICAgICAgICBhbGlnbi1pdGVtczogY2VudGVyOwp9CgogLndpZGdldHMtc2xpZGVyIC5zbGlkZXItY29udGFpbmVyIHsKICAgIG92ZXJmbG93OiB2aXNpYmxlOwp9CgogLndpZGdldC1oc2xpZGVyIC5zbGlkZXItY29udGFpbmVyIHsKICAgIGhlaWdodDogMjhweDsKICAgIG1hcmdpbi1sZWZ0OiA2cHg7CiAgICBtYXJnaW4tcmlnaHQ6IDZweDsKICAgIC13ZWJraXQtYm94LWZsZXg6IDE7CiAgICAgICAgLW1zLWZsZXg6IDEgMSAxNDhweDsKICAgICAgICAgICAgZmxleDogMSAxIDE0OHB4Owp9CgogLndpZGdldC1oc2xpZGVyIC51aS1zbGlkZXIgewogICAgLyogSW5uZXIsIGludmlzaWJsZSBzbGlkZSBkaXYgKi8KICAgIGhlaWdodDogNHB4OwogICAgbWFyZ2luLXRvcDogMTJweDsKICAgIHdpZHRoOiAxMDAlOwp9CgogLyogVmVydGljYWwgU2xpZGVyICovCgogLndpZGdldC12Ym94IC53aWRnZXQtbGFiZWwgewogICAgaGVpZ2h0OiAyOHB4OwogICAgbGluZS1oZWlnaHQ6IDI4cHg7Cn0KCiAud2lkZ2V0LXZzbGlkZXIgewogICAgLyogVmVydGljYWwgU2xpZGVyICovCiAgICBoZWlnaHQ6IDIwMHB4OwogICAgd2lkdGg6IDcycHg7Cn0KCiAud2lkZ2V0LXZzbGlkZXIgLnNsaWRlci1jb250YWluZXIgewogICAgLXdlYmtpdC1ib3gtZmxleDogMTsKICAgICAgICAtbXMtZmxleDogMSAxIDE0OHB4OwogICAgICAgICAgICBmbGV4OiAxIDEgMTQ4cHg7CiAgICBtYXJnaW4tbGVmdDogYXV0bzsKICAgIG1hcmdpbi1yaWdodDogYXV0bzsKICAgIG1hcmdpbi1ib3R0b206IDZweDsKICAgIG1hcmdpbi10b3A6IDZweDsKICAgIGRpc3BsYXk6IC13ZWJraXQtYm94OwogICAgZGlzcGxheTogLW1zLWZsZXhib3g7CiAgICBkaXNwbGF5OiBmbGV4OwogICAgLXdlYmtpdC1ib3gtb3JpZW50OiB2ZXJ0aWNhbDsKICAgIC13ZWJraXQtYm94LWRpcmVjdGlvbjogbm9ybWFsOwogICAgICAgIC1tcy1mbGV4LWRpcmVjdGlvbjogY29sdW1uOwogICAgICAgICAgICBmbGV4LWRpcmVjdGlvbjogY29sdW1uOwp9CgogLndpZGdldC12c2xpZGVyIC51aS1zbGlkZXItdmVydGljYWwgewogICAgLyogSW5uZXIsIGludmlzaWJsZSBzbGlkZSBkaXYgKi8KICAgIHdpZHRoOiA0cHg7CiAgICAtd2Via2l0LWJveC1mbGV4OiAxOwogICAgICAgIC1tcy1mbGV4LXBvc2l0aXZlOiAxOwogICAgICAgICAgICBmbGV4LWdyb3c6IDE7CiAgICBtYXJnaW4tbGVmdDogYXV0bzsKICAgIG1hcmdpbi1yaWdodDogYXV0bzsKfQoKIC8qIFdpZGdldCBQcm9ncmVzcyBTdHlsaW5nICovCgogLnByb2dyZXNzLWJhciB7CiAgICAtd2Via2l0LXRyYW5zaXRpb246IG5vbmU7CiAgICB0cmFuc2l0aW9uOiBub25lOwp9CgogLnByb2dyZXNzLWJhciB7CiAgICBoZWlnaHQ6IDI4cHg7Cn0KCiAucHJvZ3Jlc3MtYmFyIHsKICAgIGJhY2tncm91bmQtY29sb3I6ICMyMTk2RjM7Cn0KCiAucHJvZ3Jlc3MtYmFyLXN1Y2Nlc3MgewogICAgYmFja2dyb3VuZC1jb2xvcjogIzRDQUY1MDsKfQoKIC5wcm9ncmVzcy1iYXItaW5mbyB7CiAgICBiYWNrZ3JvdW5kLWNvbG9yOiAjMDBCQ0Q0Owp9CgogLnByb2dyZXNzLWJhci13YXJuaW5nIHsKICAgIGJhY2tncm91bmQtY29sb3I6ICNGRjk4MDA7Cn0KCiAucHJvZ3Jlc3MtYmFyLWRhbmdlciB7CiAgICBiYWNrZ3JvdW5kLWNvbG9yOiAjRjQ0MzM2Owp9CgogLnByb2dyZXNzIHsKICAgIGJhY2tncm91bmQtY29sb3I6ICNFRUVFRUU7CiAgICBib3JkZXI6IG5vbmU7CiAgICAtd2Via2l0LWJveC1zaGFkb3c6IG5vbmU7CiAgICAgICAgICAgIGJveC1zaGFkb3c6IG5vbmU7Cn0KCiAvKiBIb3Jpc29udGFsIFByb2dyZXNzICovCgogLndpZGdldC1ocHJvZ3Jlc3MgewogICAgLyogUHJvZ3Jlc3MgQmFyICovCiAgICBoZWlnaHQ6IDI4cHg7CiAgICBsaW5lLWhlaWdodDogMjhweDsKICAgIHdpZHRoOiAzMDBweDsKICAgIC13ZWJraXQtYm94LWFsaWduOiBjZW50ZXI7CiAgICAgICAgLW1zLWZsZXgtYWxpZ246IGNlbnRlcjsKICAgICAgICAgICAgYWxpZ24taXRlbXM6IGNlbnRlcjsKCn0KCiAud2lkZ2V0LWhwcm9ncmVzcyAucHJvZ3Jlc3MgewogICAgLXdlYmtpdC1ib3gtZmxleDogMTsKICAgICAgICAtbXMtZmxleC1wb3NpdGl2ZTogMTsKICAgICAgICAgICAgZmxleC1ncm93OiAxOwogICAgbWFyZ2luLXRvcDogNHB4OwogICAgbWFyZ2luLWJvdHRvbTogNHB4OwogICAgLW1zLWZsZXgtaXRlbS1hbGlnbjogc3RyZXRjaDsKICAgICAgICBhbGlnbi1zZWxmOiBzdHJldGNoOwogICAgLyogT3ZlcnJpZGUgYm9vdHN0cmFwIHN0eWxlICovCiAgICBoZWlnaHQ6IGF1dG87CiAgICBoZWlnaHQ6IGluaXRpYWw7Cn0KCiAvKiBWZXJ0aWNhbCBQcm9ncmVzcyAqLwoKIC53aWRnZXQtdnByb2dyZXNzIHsKICAgIGhlaWdodDogMjAwcHg7CiAgICB3aWR0aDogNzJweDsKfQoKIC53aWRnZXQtdnByb2dyZXNzIC5wcm9ncmVzcyB7CiAgICAtd2Via2l0LWJveC1mbGV4OiAxOwogICAgICAgIC1tcy1mbGV4LXBvc2l0aXZlOiAxOwogICAgICAgICAgICBmbGV4LWdyb3c6IDE7CiAgICB3aWR0aDogMjBweDsKICAgIG1hcmdpbi1sZWZ0OiBhdXRvOwogICAgbWFyZ2luLXJpZ2h0OiBhdXRvOwogICAgbWFyZ2luLWJvdHRvbTogMDsKfQoKIC8qIFNlbGVjdCBXaWRnZXQgU3R5bGluZyAqLwoKIC53aWRnZXQtZHJvcGRvd24gewogICAgaGVpZ2h0OiAyOHB4OwogICAgd2lkdGg6IDMwMHB4OwogICAgbGluZS1oZWlnaHQ6IDI4cHg7Cn0KCiAud2lkZ2V0LWRyb3Bkb3duID4gc2VsZWN0IHsKICAgIHBhZGRpbmctcmlnaHQ6IDIwcHg7CiAgICBib3JkZXI6IDFweCBzb2xpZCAjOUU5RTlFOwogICAgYm9yZGVyLXJhZGl1czogMDsKICAgIGhlaWdodDogaW5oZXJpdDsKICAgIC13ZWJraXQtYm94LWZsZXg6IDE7CiAgICAgICAgLW1zLWZsZXg6IDEgMSAxNDhweDsKICAgICAgICAgICAgZmxleDogMSAxIDE0OHB4OwogICAgbWluLXdpZHRoOiAwOyAvKiBUaGlzIG1ha2VzIGl0IHBvc3NpYmxlIGZvciB0aGUgZmxleGJveCB0byBzaHJpbmsgdGhpcyBpbnB1dCAqLwogICAgLXdlYmtpdC1ib3gtc2l6aW5nOiBib3JkZXItYm94OwogICAgICAgICAgICBib3gtc2l6aW5nOiBib3JkZXItYm94OwogICAgb3V0bGluZTogbm9uZSAhaW1wb3J0YW50OwogICAgLXdlYmtpdC1ib3gtc2hhZG93OiBub25lOwogICAgICAgICAgICBib3gtc2hhZG93OiBub25lOwogICAgYmFja2dyb3VuZC1jb2xvcjogd2hpdGU7CiAgICBjb2xvcjogcmdiYSgwLCAwLCAwLCAuOCk7CiAgICBmb250LXNpemU6IDEzcHg7CiAgICB2ZXJ0aWNhbC1hbGlnbjogdG9wOwogICAgcGFkZGluZy1sZWZ0OiA4cHg7CglhcHBlYXJhbmNlOiBub25lOwoJLXdlYmtpdC1hcHBlYXJhbmNlOiBub25lOwoJLW1vei1hcHBlYXJhbmNlOiBub25lOwogICAgYmFja2dyb3VuZC1yZXBlYXQ6IG5vLXJlcGVhdDsKCWJhY2tncm91bmQtc2l6ZTogMjBweDsKCWJhY2tncm91bmQtcG9zaXRpb246IHJpZ2h0IGNlbnRlcjsKICAgIGJhY2tncm91bmQtaW1hZ2U6IHVybCgiZGF0YTppbWFnZS9zdmcreG1sO2Jhc2U2NCxQRDk0Yld3Z2RtVnljMmx2YmowaU1TNHdJaUJsYm1OdlpHbHVaejBpZFhSbUxUZ2lQejRLUENFdExTQkhaVzVsY21GMGIzSTZJRUZrYjJKbElFbHNiSFZ6ZEhKaGRHOXlJREU1TGpJdU1Td2dVMVpISUVWNGNHOXlkQ0JRYkhWbkxVbHVJQzRnVTFaSElGWmxjbk5wYjI0NklEWXVNREFnUW5WcGJHUWdNQ2tnSUMwdFBnbzhjM1puSUhabGNuTnBiMjQ5SWpFdU1TSWdhV1E5SWt4aGVXVnlYekVpSUhodGJHNXpQU0pvZEhSd09pOHZkM2QzTG5jekxtOXlaeTh5TURBd0wzTjJaeUlnZUcxc2JuTTZlR3hwYm1zOUltaDBkSEE2THk5M2QzY3Vkek11YjNKbkx6RTVPVGt2ZUd4cGJtc2lJSGc5SWpCd2VDSWdlVDBpTUhCNElnb0pJSFpwWlhkQ2IzZzlJakFnTUNBeE9DQXhPQ0lnYzNSNWJHVTlJbVZ1WVdKc1pTMWlZV05yWjNKdmRXNWtPbTVsZHlBd0lEQWdNVGdnTVRnN0lpQjRiV3c2YzNCaFkyVTlJbkJ5WlhObGNuWmxJajRLUEhOMGVXeGxJSFI1Y0dVOUluUmxlSFF2WTNOeklqNEtDUzV6ZERCN1ptbHNiRHB1YjI1bE8zMEtQQzl6ZEhsc1pUNEtQSEJoZEdnZ1pEMGlUVFV1TWl3MUxqbE1PU3c1TGpkc015NDRMVE11T0d3eExqSXNNUzR5YkMwMExqa3NOV3d0TkM0NUxUVk1OUzR5TERVdU9Yb2lMejRLUEhCaGRHZ2dZMnhoYzNNOUluTjBNQ0lnWkQwaVRUQXRNQzQyYURFNGRqRTRTREJXTFRBdU5ub2lMejRLUEM5emRtYytDZyIpOwp9CgogLndpZGdldC1kcm9wZG93biA+IHNlbGVjdDpmb2N1cyB7CiAgICBib3JkZXItY29sb3I6ICM2NEI1RjY7Cn0KCiAud2lkZ2V0LWRyb3Bkb3duID4gc2VsZWN0OmRpc2FibGVkIHsKICAgIG9wYWNpdHk6IDAuNjsKfQoKIC8qIFRvIGRpc2FibGUgdGhlIGRvdHRlZCBib3JkZXIgaW4gRmlyZWZveCBhcm91bmQgc2VsZWN0IGNvbnRyb2xzLgogICBTZWUgaHR0cDovL3N0YWNrb3ZlcmZsb3cuY29tL2EvMTg4NTMwMDIgKi8KCiAud2lkZ2V0LWRyb3Bkb3duID4gc2VsZWN0Oi1tb3otZm9jdXNyaW5nIHsKICAgIGNvbG9yOiB0cmFuc3BhcmVudDsKICAgIHRleHQtc2hhZG93OiAwIDAgMCAjMDAwOwp9CgogLyogU2VsZWN0IGFuZCBTZWxlY3RNdWx0aXBsZSAqLwoKIC53aWRnZXQtc2VsZWN0IHsKICAgIHdpZHRoOiAzMDBweDsKICAgIGxpbmUtaGVpZ2h0OiAyOHB4OwoKICAgIC8qIEJlY2F1c2UgRmlyZWZveCBkZWZpbmVzIHRoZSBiYXNlbGluZSBvZiBhIHNlbGVjdCBhcyB0aGUgYm90dG9tIG9mIHRoZQogICAgY29udHJvbCwgd2UgYWxpZ24gdGhlIGVudGlyZSBjb250cm9sIHRvIHRoZSB0b3AgYW5kIGFkZCBwYWRkaW5nIHRvIHRoZQogICAgc2VsZWN0IHRvIGdldCBhbiBhcHByb3hpbWF0ZSBmaXJzdCBsaW5lIGJhc2VsaW5lIGFsaWdubWVudC4gKi8KICAgIC13ZWJraXQtYm94LWFsaWduOiBzdGFydDsKICAgICAgICAtbXMtZmxleC1hbGlnbjogc3RhcnQ7CiAgICAgICAgICAgIGFsaWduLWl0ZW1zOiBmbGV4LXN0YXJ0Owp9CgogLndpZGdldC1zZWxlY3QgPiBzZWxlY3QgewogICAgYm9yZGVyOiAxcHggc29saWQgIzlFOUU5RTsKICAgIGJhY2tncm91bmQtY29sb3I6IHdoaXRlOwogICAgY29sb3I6IHJnYmEoMCwgMCwgMCwgLjgpOwogICAgZm9udC1zaXplOiAxM3B4OwogICAgLXdlYmtpdC1ib3gtZmxleDogMTsKICAgICAgICAtbXMtZmxleDogMSAxIDE0OHB4OwogICAgICAgICAgICBmbGV4OiAxIDEgMTQ4cHg7CiAgICBvdXRsaW5lOiBub25lICFpbXBvcnRhbnQ7CiAgICBvdmVyZmxvdzogYXV0bzsKICAgIGhlaWdodDogaW5oZXJpdDsKCiAgICAvKiBCZWNhdXNlIEZpcmVmb3ggZGVmaW5lcyB0aGUgYmFzZWxpbmUgb2YgYSBzZWxlY3QgYXMgdGhlIGJvdHRvbSBvZiB0aGUKICAgIGNvbnRyb2wsIHdlIGFsaWduIHRoZSBlbnRpcmUgY29udHJvbCB0byB0aGUgdG9wIGFuZCBhZGQgcGFkZGluZyB0byB0aGUKICAgIHNlbGVjdCB0byBnZXQgYW4gYXBwcm94aW1hdGUgZmlyc3QgbGluZSBiYXNlbGluZSBhbGlnbm1lbnQuICovCiAgICBwYWRkaW5nLXRvcDogNXB4Owp9CgogLndpZGdldC1zZWxlY3QgPiBzZWxlY3Q6Zm9jdXMgewogICAgYm9yZGVyLWNvbG9yOiAjNjRCNUY2Owp9CgogLndpZ2V0LXNlbGVjdCA+IHNlbGVjdCA+IG9wdGlvbiB7CiAgICBwYWRkaW5nLWxlZnQ6IDRweDsKICAgIGxpbmUtaGVpZ2h0OiAyOHB4OwogICAgLyogbGluZS1oZWlnaHQgZG9lc24ndCB3b3JrIG9uIHNvbWUgYnJvd3NlcnMgZm9yIHNlbGVjdCBvcHRpb25zICovCiAgICBwYWRkaW5nLXRvcDogY2FsYygyOHB4IC0gdmFyKC0tanAtd2lkZ2V0cy1mb250LXNpemUpIC8gMik7CiAgICBwYWRkaW5nLWJvdHRvbTogY2FsYygyOHB4IC0gdmFyKC0tanAtd2lkZ2V0cy1mb250LXNpemUpIC8gMik7Cn0KCiAvKiBUb2dnbGUgQnV0dG9ucyBTdHlsaW5nICovCgogLndpZGdldC10b2dnbGUtYnV0dG9ucyB7CiAgICBsaW5lLWhlaWdodDogMjhweDsKfQoKIC53aWRnZXQtdG9nZ2xlLWJ1dHRvbnMgLndpZGdldC10b2dnbGUtYnV0dG9uIHsKICAgIG1hcmdpbi1sZWZ0OiAycHg7CiAgICBtYXJnaW4tcmlnaHQ6IDJweDsKfQoKIC53aWRnZXQtdG9nZ2xlLWJ1dHRvbnMgLmp1cHl0ZXItYnV0dG9uOmRpc2FibGVkIHsKICAgIG9wYWNpdHk6IDAuNjsKfQoKIC8qIFJhZGlvIEJ1dHRvbnMgU3R5bGluZyAqLwoKIC53aWRnZXQtcmFkaW8gewogICAgd2lkdGg6IDMwMHB4OwogICAgbGluZS1oZWlnaHQ6IDI4cHg7Cn0KCiAud2lkZ2V0LXJhZGlvLWJveCB7CiAgICBkaXNwbGF5OiAtd2Via2l0LWJveDsKICAgIGRpc3BsYXk6IC1tcy1mbGV4Ym94OwogICAgZGlzcGxheTogZmxleDsKICAgIC13ZWJraXQtYm94LW9yaWVudDogdmVydGljYWw7CiAgICAtd2Via2l0LWJveC1kaXJlY3Rpb246IG5vcm1hbDsKICAgICAgICAtbXMtZmxleC1kaXJlY3Rpb246IGNvbHVtbjsKICAgICAgICAgICAgZmxleC1kaXJlY3Rpb246IGNvbHVtbjsKICAgIC13ZWJraXQtYm94LWFsaWduOiBzdHJldGNoOwogICAgICAgIC1tcy1mbGV4LWFsaWduOiBzdHJldGNoOwogICAgICAgICAgICBhbGlnbi1pdGVtczogc3RyZXRjaDsKICAgIC13ZWJraXQtYm94LXNpemluZzogYm9yZGVyLWJveDsKICAgICAgICAgICAgYm94LXNpemluZzogYm9yZGVyLWJveDsKICAgIC13ZWJraXQtYm94LWZsZXg6IDE7CiAgICAgICAgLW1zLWZsZXgtcG9zaXRpdmU6IDE7CiAgICAgICAgICAgIGZsZXgtZ3JvdzogMTsKICAgIG1hcmdpbi1ib3R0b206IDhweDsKfQoKIC53aWRnZXQtcmFkaW8tYm94IGxhYmVsIHsKICAgIGhlaWdodDogMjBweDsKICAgIGxpbmUtaGVpZ2h0OiAyMHB4OwogICAgZm9udC1zaXplOiAxM3B4Owp9CgogLndpZGdldC1yYWRpby1ib3ggaW5wdXQgewogICAgaGVpZ2h0OiAyMHB4OwogICAgbGluZS1oZWlnaHQ6IDIwcHg7CiAgICBtYXJnaW46IDAgOHB4IDAgMXB4OwogICAgZmxvYXQ6IGxlZnQ7Cn0KCiAvKiBDb2xvciBQaWNrZXIgU3R5bGluZyAqLwoKIC53aWRnZXQtY29sb3JwaWNrZXIgewogICAgd2lkdGg6IDMwMHB4OwogICAgaGVpZ2h0OiAyOHB4OwogICAgbGluZS1oZWlnaHQ6IDI4cHg7Cn0KCiAud2lkZ2V0LWNvbG9ycGlja2VyID4gLndpZGdldC1jb2xvcnBpY2tlci1pbnB1dCB7CiAgICAtd2Via2l0LWJveC1mbGV4OiAxOwogICAgICAgIC1tcy1mbGV4LXBvc2l0aXZlOiAxOwogICAgICAgICAgICBmbGV4LWdyb3c6IDE7CiAgICAtbXMtZmxleC1uZWdhdGl2ZTogMTsKICAgICAgICBmbGV4LXNocmluazogMTsKICAgIG1pbi13aWR0aDogNzJweDsKfQoKIC53aWRnZXQtY29sb3JwaWNrZXIgaW5wdXRbdHlwZT0iY29sb3IiXSB7CiAgICB3aWR0aDogMjhweDsKICAgIGhlaWdodDogMjhweDsKICAgIHBhZGRpbmc6IDAgMnB4OyAvKiBtYWtlIHRoZSBjb2xvciBzcXVhcmUgYWN0dWFsbHkgc3F1YXJlIG9uIENocm9tZSBvbiBPUyBYICovCiAgICBiYWNrZ3JvdW5kOiB3aGl0ZTsKICAgIGNvbG9yOiByZ2JhKDAsIDAsIDAsIC44KTsKICAgIGJvcmRlcjogMXB4IHNvbGlkICM5RTlFOUU7CiAgICBib3JkZXItbGVmdDogbm9uZTsKICAgIC13ZWJraXQtYm94LWZsZXg6IDA7CiAgICAgICAgLW1zLWZsZXgtcG9zaXRpdmU6IDA7CiAgICAgICAgICAgIGZsZXgtZ3JvdzogMDsKICAgIC1tcy1mbGV4LW5lZ2F0aXZlOiAwOwogICAgICAgIGZsZXgtc2hyaW5rOiAwOwogICAgLXdlYmtpdC1ib3gtc2l6aW5nOiBib3JkZXItYm94OwogICAgICAgICAgICBib3gtc2l6aW5nOiBib3JkZXItYm94OwogICAgLW1zLWZsZXgtaXRlbS1hbGlnbjogc3RyZXRjaDsKICAgICAgICBhbGlnbi1zZWxmOiBzdHJldGNoOwogICAgb3V0bGluZTogbm9uZSAhaW1wb3J0YW50Owp9CgogLndpZGdldC1jb2xvcnBpY2tlci5jb25jaXNlIGlucHV0W3R5cGU9ImNvbG9yIl0gewogICAgYm9yZGVyLWxlZnQ6IDFweCBzb2xpZCAjOUU5RTlFOwp9CgogLndpZGdldC1jb2xvcnBpY2tlciBpbnB1dFt0eXBlPSJjb2xvciJdOmZvY3VzLCAud2lkZ2V0LWNvbG9ycGlja2VyIGlucHV0W3R5cGU9InRleHQiXTpmb2N1cyB7CiAgICBib3JkZXItY29sb3I6ICM2NEI1RjY7Cn0KCiAud2lkZ2V0LWNvbG9ycGlja2VyIGlucHV0W3R5cGU9InRleHQiXSB7CiAgICAtd2Via2l0LWJveC1mbGV4OiAxOwogICAgICAgIC1tcy1mbGV4LXBvc2l0aXZlOiAxOwogICAgICAgICAgICBmbGV4LWdyb3c6IDE7CiAgICBvdXRsaW5lOiBub25lICFpbXBvcnRhbnQ7CiAgICBoZWlnaHQ6IDI4cHg7CiAgICBsaW5lLWhlaWdodDogMjhweDsKICAgIGJhY2tncm91bmQ6IHdoaXRlOwogICAgY29sb3I6IHJnYmEoMCwgMCwgMCwgLjgpOwogICAgYm9yZGVyOiAxcHggc29saWQgIzlFOUU5RTsKICAgIGZvbnQtc2l6ZTogMTNweDsKICAgIHBhZGRpbmc6IDRweCA4cHg7CiAgICBtaW4td2lkdGg6IDA7IC8qIFRoaXMgbWFrZXMgaXQgcG9zc2libGUgZm9yIHRoZSBmbGV4Ym94IHRvIHNocmluayB0aGlzIGlucHV0ICovCiAgICAtbXMtZmxleC1uZWdhdGl2ZTogMTsKICAgICAgICBmbGV4LXNocmluazogMTsKICAgIC13ZWJraXQtYm94LXNpemluZzogYm9yZGVyLWJveDsKICAgICAgICAgICAgYm94LXNpemluZzogYm9yZGVyLWJveDsKfQoKIC53aWRnZXQtY29sb3JwaWNrZXIgaW5wdXRbdHlwZT0idGV4dCJdOmRpc2FibGVkIHsKICAgIG9wYWNpdHk6IDAuNjsKfQoKIC8qIERhdGUgUGlja2VyIFN0eWxpbmcgKi8KCiAud2lkZ2V0LWRhdGVwaWNrZXIgewogICAgd2lkdGg6IDMwMHB4OwogICAgaGVpZ2h0OiAyOHB4OwogICAgbGluZS1oZWlnaHQ6IDI4cHg7Cn0KCiAud2lkZ2V0LWRhdGVwaWNrZXIgaW5wdXRbdHlwZT0iZGF0ZSJdIHsKICAgIC13ZWJraXQtYm94LWZsZXg6IDE7CiAgICAgICAgLW1zLWZsZXgtcG9zaXRpdmU6IDE7CiAgICAgICAgICAgIGZsZXgtZ3JvdzogMTsKICAgIC1tcy1mbGV4LW5lZ2F0aXZlOiAxOwogICAgICAgIGZsZXgtc2hyaW5rOiAxOwogICAgbWluLXdpZHRoOiAwOyAvKiBUaGlzIG1ha2VzIGl0IHBvc3NpYmxlIGZvciB0aGUgZmxleGJveCB0byBzaHJpbmsgdGhpcyBpbnB1dCAqLwogICAgb3V0bGluZTogbm9uZSAhaW1wb3J0YW50OwogICAgaGVpZ2h0OiAyOHB4OwogICAgYm9yZGVyOiAxcHggc29saWQgIzlFOUU5RTsKICAgIGJhY2tncm91bmQtY29sb3I6IHdoaXRlOwogICAgY29sb3I6IHJnYmEoMCwgMCwgMCwgLjgpOwogICAgZm9udC1zaXplOiAxM3B4OwogICAgcGFkZGluZzogNHB4IDhweDsKICAgIC13ZWJraXQtYm94LXNpemluZzogYm9yZGVyLWJveDsKICAgICAgICAgICAgYm94LXNpemluZzogYm9yZGVyLWJveDsKfQoKIC53aWRnZXQtZGF0ZXBpY2tlciBpbnB1dFt0eXBlPSJkYXRlIl06Zm9jdXMgewogICAgYm9yZGVyLWNvbG9yOiAjNjRCNUY2Owp9CgogLndpZGdldC1kYXRlcGlja2VyIGlucHV0W3R5cGU9ImRhdGUiXTppbnZhbGlkIHsKICAgIGJvcmRlci1jb2xvcjogI0ZGOTgwMDsKfQoKIC53aWRnZXQtZGF0ZXBpY2tlciBpbnB1dFt0eXBlPSJkYXRlIl06ZGlzYWJsZWQgewogICAgb3BhY2l0eTogMC42Owp9CgogLyogUGxheSBXaWRnZXQgKi8KCiAud2lkZ2V0LXBsYXkgewogICAgd2lkdGg6IDE0OHB4OwogICAgZGlzcGxheTogLXdlYmtpdC1ib3g7CiAgICBkaXNwbGF5OiAtbXMtZmxleGJveDsKICAgIGRpc3BsYXk6IGZsZXg7CiAgICAtd2Via2l0LWJveC1hbGlnbjogc3RyZXRjaDsKICAgICAgICAtbXMtZmxleC1hbGlnbjogc3RyZXRjaDsKICAgICAgICAgICAgYWxpZ24taXRlbXM6IHN0cmV0Y2g7Cn0KCiAud2lkZ2V0LXBsYXkgLmp1cHl0ZXItYnV0dG9uIHsKICAgIC13ZWJraXQtYm94LWZsZXg6IDE7CiAgICAgICAgLW1zLWZsZXgtcG9zaXRpdmU6IDE7CiAgICAgICAgICAgIGZsZXgtZ3JvdzogMTsKICAgIGhlaWdodDogYXV0bzsKfQoKIC53aWRnZXQtcGxheSAuanVweXRlci1idXR0b246ZGlzYWJsZWQgewogICAgb3BhY2l0eTogMC42Owp9CgogLyogVGFiIFdpZGdldCAqLwoKIC5qdXB5dGVyLXdpZGdldHMud2lkZ2V0LXRhYiB7CiAgICBkaXNwbGF5OiAtd2Via2l0LWJveDsKICAgIGRpc3BsYXk6IC1tcy1mbGV4Ym94OwogICAgZGlzcGxheTogZmxleDsKICAgIC13ZWJraXQtYm94LW9yaWVudDogdmVydGljYWw7CiAgICAtd2Via2l0LWJveC1kaXJlY3Rpb246IG5vcm1hbDsKICAgICAgICAtbXMtZmxleC1kaXJlY3Rpb246IGNvbHVtbjsKICAgICAgICAgICAgZmxleC1kaXJlY3Rpb246IGNvbHVtbjsKfQoKIC5qdXB5dGVyLXdpZGdldHMud2lkZ2V0LXRhYiA+IC5wLVRhYkJhciB7CiAgICAvKiBOZWNlc3Nhcnkgc28gdGhhdCBhIHRhYiBjYW4gYmUgc2hpZnRlZCBkb3duIHRvIG92ZXJsYXkgdGhlIGJvcmRlciBvZiB0aGUgYm94IGJlbG93LiAqLwogICAgb3ZlcmZsb3cteDogdmlzaWJsZTsKICAgIG92ZXJmbG93LXk6IHZpc2libGU7Cn0KCiAuanVweXRlci13aWRnZXRzLndpZGdldC10YWIgPiAucC1UYWJCYXIgPiAucC1UYWJCYXItY29udGVudCB7CiAgICAvKiBNYWtlIHN1cmUgdGhhdCB0aGUgdGFiIGdyb3dzIGZyb20gYm90dG9tIHVwICovCiAgICAtd2Via2l0LWJveC1hbGlnbjogZW5kOwogICAgICAgIC1tcy1mbGV4LWFsaWduOiBlbmQ7CiAgICAgICAgICAgIGFsaWduLWl0ZW1zOiBmbGV4LWVuZDsKICAgIG1pbi13aWR0aDogMDsKICAgIG1pbi1oZWlnaHQ6IDA7Cn0KCiAuanVweXRlci13aWRnZXRzLndpZGdldC10YWIgPiAud2lkZ2V0LXRhYi1jb250ZW50cyB7CiAgICB3aWR0aDogMTAwJTsKICAgIC13ZWJraXQtYm94LXNpemluZzogYm9yZGVyLWJveDsKICAgICAgICAgICAgYm94LXNpemluZzogYm9yZGVyLWJveDsKICAgIG1hcmdpbjogMDsKICAgIGJhY2tncm91bmQ6IHdoaXRlOwogICAgY29sb3I6IHJnYmEoMCwgMCwgMCwgLjgpOwogICAgYm9yZGVyOiAxcHggc29saWQgIzlFOUU5RTsKICAgIHBhZGRpbmc6IDE1cHg7CiAgICAtd2Via2l0LWJveC1mbGV4OiAxOwogICAgICAgIC1tcy1mbGV4LXBvc2l0aXZlOiAxOwogICAgICAgICAgICBmbGV4LWdyb3c6IDE7CiAgICBvdmVyZmxvdzogYXV0bzsKfQoKIC5qdXB5dGVyLXdpZGdldHMud2lkZ2V0LXRhYiA+IC5wLVRhYkJhciB7CiAgICBmb250OiAxM3B4IEhlbHZldGljYSwgQXJpYWwsIHNhbnMtc2VyaWY7CiAgICBtaW4taGVpZ2h0OiAyNXB4Owp9CgogLmp1cHl0ZXItd2lkZ2V0cy53aWRnZXQtdGFiID4gLnAtVGFiQmFyIC5wLVRhYkJhci10YWIgewogICAgLXdlYmtpdC1ib3gtZmxleDogMDsKICAgICAgICAtbXMtZmxleDogMCAxIDE0NHB4OwogICAgICAgICAgICBmbGV4OiAwIDEgMTQ0cHg7CiAgICBtaW4td2lkdGg6IDM1cHg7CiAgICBtaW4taGVpZ2h0OiAyNXB4OwogICAgbGluZS1oZWlnaHQ6IDI0cHg7CiAgICBtYXJnaW4tbGVmdDogLTFweDsKICAgIHBhZGRpbmc6IDBweCAxMHB4OwogICAgYmFja2dyb3VuZDogI0VFRUVFRTsKICAgIGNvbG9yOiByZ2JhKDAsIDAsIDAsIC41KTsKICAgIGJvcmRlcjogMXB4IHNvbGlkICM5RTlFOUU7CiAgICBib3JkZXItYm90dG9tOiBub25lOwogICAgcG9zaXRpb246IHJlbGF0aXZlOwp9CgogLmp1cHl0ZXItd2lkZ2V0cy53aWRnZXQtdGFiID4gLnAtVGFiQmFyIC5wLVRhYkJhci10YWIucC1tb2QtY3VycmVudCB7CiAgICBjb2xvcjogcmdiYSgwLCAwLCAwLCAxLjApOwogICAgLyogV2Ugd2FudCB0aGUgYmFja2dyb3VuZCB0byBtYXRjaCB0aGUgdGFiIGNvbnRlbnQgYmFja2dyb3VuZCAqLwogICAgYmFja2dyb3VuZDogd2hpdGU7CiAgICBtaW4taGVpZ2h0OiAyNnB4OwogICAgLXdlYmtpdC10cmFuc2Zvcm06IHRyYW5zbGF0ZVkoMXB4KTsKICAgICAgICAgICAgdHJhbnNmb3JtOiB0cmFuc2xhdGVZKDFweCk7CiAgICBvdmVyZmxvdzogdmlzaWJsZTsKfQoKIC5qdXB5dGVyLXdpZGdldHMud2lkZ2V0LXRhYiA+IC5wLVRhYkJhciAucC1UYWJCYXItdGFiLnAtbW9kLWN1cnJlbnQ6YmVmb3JlIHsKICAgIHBvc2l0aW9uOiBhYnNvbHV0ZTsKICAgIHRvcDogLTFweDsKICAgIGxlZnQ6IC0xcHg7CiAgICBjb250ZW50OiAnJzsKICAgIGhlaWdodDogMnB4OwogICAgd2lkdGg6IGNhbGMoMTAwJSArIDJweCk7CiAgICBiYWNrZ3JvdW5kOiAjMjE5NkYzOwp9CgogLmp1cHl0ZXItd2lkZ2V0cy53aWRnZXQtdGFiID4gLnAtVGFiQmFyIC5wLVRhYkJhci10YWI6Zmlyc3QtY2hpbGQgewogICAgbWFyZ2luLWxlZnQ6IDA7Cn0KCiAuanVweXRlci13aWRnZXRzLndpZGdldC10YWIgPiAucC1UYWJCYXIgLnAtVGFiQmFyLXRhYjpob3Zlcjpub3QoLnAtbW9kLWN1cnJlbnQpIHsKICAgIGJhY2tncm91bmQ6IHdoaXRlOwogICAgY29sb3I6IHJnYmEoMCwgMCwgMCwgLjgpOwp9CgogLmp1cHl0ZXItd2lkZ2V0cy53aWRnZXQtdGFiID4gLnAtVGFiQmFyIC5wLW1vZC1jbG9zYWJsZSA+IC5wLVRhYkJhci10YWJDbG9zZUljb24gewogICAgbWFyZ2luLWxlZnQ6IDRweDsKfQoKIC5qdXB5dGVyLXdpZGdldHMud2lkZ2V0LXRhYiA+IC5wLVRhYkJhciAucC1tb2QtY2xvc2FibGUgPiAucC1UYWJCYXItdGFiQ2xvc2VJY29uOmJlZm9yZSB7CiAgICBmb250LWZhbWlseTogRm9udEF3ZXNvbWU7CiAgICBjb250ZW50OiAnXGYwMGQnOyAvKiBjbG9zZSAqLwp9CgogLmp1cHl0ZXItd2lkZ2V0cy53aWRnZXQtdGFiID4gLnAtVGFiQmFyIC5wLVRhYkJhci10YWJJY29uLAouanVweXRlci13aWRnZXRzLndpZGdldC10YWIgPiAucC1UYWJCYXIgLnAtVGFiQmFyLXRhYkxhYmVsLAouanVweXRlci13aWRnZXRzLndpZGdldC10YWIgPiAucC1UYWJCYXIgLnAtVGFiQmFyLXRhYkNsb3NlSWNvbiB7CiAgICBsaW5lLWhlaWdodDogMjRweDsKfQoKIC8qIEFjY29yZGlvbiBXaWRnZXQgKi8KCiAucC1Db2xsYXBzZSB7CiAgICBkaXNwbGF5OiAtd2Via2l0LWJveDsKICAgIGRpc3BsYXk6IC1tcy1mbGV4Ym94OwogICAgZGlzcGxheTogZmxleDsKICAgIC13ZWJraXQtYm94LW9yaWVudDogdmVydGljYWw7CiAgICAtd2Via2l0LWJveC1kaXJlY3Rpb246IG5vcm1hbDsKICAgICAgICAtbXMtZmxleC1kaXJlY3Rpb246IGNvbHVtbjsKICAgICAgICAgICAgZmxleC1kaXJlY3Rpb246IGNvbHVtbjsKICAgIC13ZWJraXQtYm94LWFsaWduOiBzdHJldGNoOwogICAgICAgIC1tcy1mbGV4LWFsaWduOiBzdHJldGNoOwogICAgICAgICAgICBhbGlnbi1pdGVtczogc3RyZXRjaDsKfQoKIC5wLUNvbGxhcHNlLWhlYWRlciB7CiAgICBwYWRkaW5nOiA0cHg7CiAgICBjdXJzb3I6IHBvaW50ZXI7CiAgICBjb2xvcjogcmdiYSgwLCAwLCAwLCAuNSk7CiAgICBiYWNrZ3JvdW5kLWNvbG9yOiAjRUVFRUVFOwogICAgYm9yZGVyOiAxcHggc29saWQgIzlFOUU5RTsKICAgIHBhZGRpbmc6IDEwcHggMTVweDsKICAgIGZvbnQtd2VpZ2h0OiBib2xkOwp9CgogLnAtQ29sbGFwc2UtaGVhZGVyOmhvdmVyIHsKICAgIGJhY2tncm91bmQtY29sb3I6IHdoaXRlOwogICAgY29sb3I6IHJnYmEoMCwgMCwgMCwgLjgpOwp9CgogLnAtQ29sbGFwc2Utb3BlbiA+IC5wLUNvbGxhcHNlLWhlYWRlciB7CiAgICBiYWNrZ3JvdW5kLWNvbG9yOiB3aGl0ZTsKICAgIGNvbG9yOiByZ2JhKDAsIDAsIDAsIDEuMCk7CiAgICBjdXJzb3I6IGRlZmF1bHQ7CiAgICBib3JkZXItYm90dG9tOiBub25lOwp9CgogLnAtQ29sbGFwc2UgLnAtQ29sbGFwc2UtaGVhZGVyOjpiZWZvcmUgewogICAgY29udGVudDogJ1xmMGRhXDAwQTAnOyAgLyogY2FyZXQtcmlnaHQsIG5vbi1icmVha2luZyBzcGFjZSAqLwogICAgZGlzcGxheTogaW5saW5lLWJsb2NrOwogICAgZm9udDogbm9ybWFsIG5vcm1hbCBub3JtYWwgMTRweC8xIEZvbnRBd2Vzb21lOwogICAgZm9udC1zaXplOiBpbmhlcml0OwogICAgdGV4dC1yZW5kZXJpbmc6IGF1dG87CiAgICAtd2Via2l0LWZvbnQtc21vb3RoaW5nOiBhbnRpYWxpYXNlZDsKICAgIC1tb3otb3N4LWZvbnQtc21vb3RoaW5nOiBncmF5c2NhbGU7Cn0KCiAucC1Db2xsYXBzZS1vcGVuID4gLnAtQ29sbGFwc2UtaGVhZGVyOjpiZWZvcmUgewogICAgY29udGVudDogJ1xmMGQ3XDAwQTAnOyAvKiBjYXJldC1kb3duLCBub24tYnJlYWtpbmcgc3BhY2UgKi8KfQoKIC5wLUNvbGxhcHNlLWNvbnRlbnRzIHsKICAgIHBhZGRpbmc6IDE1cHg7CiAgICBiYWNrZ3JvdW5kLWNvbG9yOiB3aGl0ZTsKICAgIGNvbG9yOiByZ2JhKDAsIDAsIDAsIC44KTsKICAgIGJvcmRlci1sZWZ0OiAxcHggc29saWQgIzlFOUU5RTsKICAgIGJvcmRlci1yaWdodDogMXB4IHNvbGlkICM5RTlFOUU7CiAgICBib3JkZXItYm90dG9tOiAxcHggc29saWQgIzlFOUU5RTsKICAgIG92ZXJmbG93OiBhdXRvOwp9CgogLnAtQWNjb3JkaW9uIHsKICAgIGRpc3BsYXk6IC13ZWJraXQtYm94OwogICAgZGlzcGxheTogLW1zLWZsZXhib3g7CiAgICBkaXNwbGF5OiBmbGV4OwogICAgLXdlYmtpdC1ib3gtb3JpZW50OiB2ZXJ0aWNhbDsKICAgIC13ZWJraXQtYm94LWRpcmVjdGlvbjogbm9ybWFsOwogICAgICAgIC1tcy1mbGV4LWRpcmVjdGlvbjogY29sdW1uOwogICAgICAgICAgICBmbGV4LWRpcmVjdGlvbjogY29sdW1uOwogICAgLXdlYmtpdC1ib3gtYWxpZ246IHN0cmV0Y2g7CiAgICAgICAgLW1zLWZsZXgtYWxpZ246IHN0cmV0Y2g7CiAgICAgICAgICAgIGFsaWduLWl0ZW1zOiBzdHJldGNoOwp9CgogLnAtQWNjb3JkaW9uIC5wLUNvbGxhcHNlIHsKICAgIG1hcmdpbi1ib3R0b206IDA7Cn0KCiAucC1BY2NvcmRpb24gLnAtQ29sbGFwc2UgKyAucC1Db2xsYXBzZSB7CiAgICBtYXJnaW4tdG9wOiA0cHg7Cn0KCiAvKiBIVE1MIHdpZGdldCAqLwoKIC53aWRnZXQtaHRtbCwgLndpZGdldC1odG1sbWF0aCB7CiAgICBmb250LXNpemU6IDEzcHg7Cn0KCiAud2lkZ2V0LWh0bWwgPiAud2lkZ2V0LWh0bWwtY29udGVudCwgLndpZGdldC1odG1sbWF0aCA+IC53aWRnZXQtaHRtbC1jb250ZW50IHsKICAgIC8qIEZpbGwgb3V0IHRoZSBhcmVhIGluIHRoZSBIVE1MIHdpZGdldCAqLwogICAgLW1zLWZsZXgtaXRlbS1hbGlnbjogc3RyZXRjaDsKICAgICAgICBhbGlnbi1zZWxmOiBzdHJldGNoOwogICAgLXdlYmtpdC1ib3gtZmxleDogMTsKICAgICAgICAtbXMtZmxleC1wb3NpdGl2ZTogMTsKICAgICAgICAgICAgZmxleC1ncm93OiAxOwogICAgLW1zLWZsZXgtbmVnYXRpdmU6IDE7CiAgICAgICAgZmxleC1zaHJpbms6IDE7CiAgICAvKiBNYWtlcyBzdXJlIHRoZSBiYXNlbGluZSBpcyBzdGlsbCBhbGlnbmVkIHdpdGggb3RoZXIgZWxlbWVudHMgKi8KICAgIGxpbmUtaGVpZ2h0OiAyOHB4OwogICAgLyogTWFrZSBpdCBwb3NzaWJsZSB0byBoYXZlIGFic29sdXRlbHktcG9zaXRpb25lZCBlbGVtZW50cyBpbiB0aGUgaHRtbCAqLwogICAgcG9zaXRpb246IHJlbGF0aXZlOwp9CgovKiMgc291cmNlTWFwcGluZ1VSTD1kYXRhOmFwcGxpY2F0aW9uL2pzb247YmFzZTY0LGV5SjJaWEp6YVc5dUlqb3pMQ0p6YjNWeVkyVnpJanBiSWk0dUwyNXZaR1ZmYlc5a2RXeGxjeTlBYW5Wd2VYUmxjaTEzYVdSblpYUnpMMk52Ym5SeWIyeHpMMk56Y3k5M2FXUm5aWFJ6TG1OemN5SXNJaTR1TDI1dlpHVmZiVzlrZFd4bGN5OUFhblZ3ZVhSbGNpMTNhV1JuWlhSekwyTnZiblJ5YjJ4ekwyTnpjeTlzWVdKMllYSnBZV0pzWlhNdVkzTnpJaXdpTGk0dmJtOWtaVjl0YjJSMWJHVnpMMEJxZFhCNWRHVnlMWGRwWkdkbGRITXZZMjl1ZEhKdmJITXZZM056TDIxaGRHVnlhV0ZzWTI5c2IzSnpMbU56Y3lJc0lpNHVMMjV2WkdWZmJXOWtkV3hsY3k5QWFuVndlWFJsY2kxM2FXUm5aWFJ6TDJOdmJuUnliMnh6TDJOemN5OTNhV1JuWlhSekxXSmhjMlV1WTNOeklpd2lMaTR2Ym05a1pWOXRiMlIxYkdWekwwQnFkWEI1ZEdWeUxYZHBaR2RsZEhNdlkyOXVkSEp2YkhNdlkzTnpMM0JvYjNOd2FHOXlMbU56Y3lKZExDSnVZVzFsY3lJNlcxMHNJbTFoY0hCcGJtZHpJam9pUVVGQlFUczdSMEZGUnpzN1EwRkZSanM3YTBOQlJXbERPenREUTA1c1F6czdPeXRGUVVjclJUczdRMEZGTDBVN096czdSVUZKUlRzN1EwTlVSanM3T3pzN096czdPenM3T3pzN096czdPenM3T3pzN096czdPenM3UjBFMlFrYzdPME5FYUVKSU96czdPenM3T3pzN096czdPenM3T3pzN08wVkJiVUpGT3p0RFFVZEdPenRIUVVWSE96dERRVU5HTEhsRVFVRjVSRHM3UTBGRE1VUXNlVVZCUVhsRk96dERRVVY2UlRzN1IwRkZSenM3UTBGUFNEczdSVUZGUlRzN08wdEJSMGM3TzBWQlVVZzdPenM3U1VGSlJTeERRVWwzUWl4dlFrRkJiMElzUTBGSGFFSXNNRU5CUVRCRE96dEZRVWQ0UlRzN1NVRkZSVHM3UlVGUFJqczdTMEZGUnpzN1JVRlBTRHM3TzBsQlIwVXNRMEZYZDBJc2IwSkJRVzlDT3pzN1JVRlZPVU03T3pzN1NVRkpSVHM3UlVGUFJpeHJRa0ZCYTBJN08wVkJXV3hDTEN0RFFVRXJRenM3UlVGelFpOURMREJDUVVFd1FqdEZRV0V4UWpzMFJVRkRNRVU3UlVGRk1VVTdkMFZCUTNORk96dEZRVWQwUlN3NFFrRkJPRUk3TzBWQlN6bENMRFpDUVVFMlFqczdSVUZKTjBJc05rSkJRVFpDTzBOQlVUbENPenREUlhwTlJEczdSMEZGUnpzN1EwRkZTRHM3T3p0SFFVbEhPenREUTFKSU96czdPenM3T3pzN096czdPenM3T3pzN096czdPenM3T3pzN096czdSVUU0UWtVN08wTkJSVVk3T3p0SFFVZEhPenREUVVWSU8wVkJRMFVzY1VKQlFXTTdSVUZCWkN4eFFrRkJZenRGUVVGa0xHTkJRV003UlVGRFpDd3dRa0ZCTUVJN1JVRkRNVUlzZFVKQlFYVkNPMFZCUTNaQ0xITkNRVUZ6UWp0RlFVTjBRaXhyUWtGQmEwSTdRMEZEYmtJN08wTkJSMFE3UlVGRFJTd3JRa0ZCYjBJN1JVRkJjRUlzT0VKQlFXOUNPMDFCUVhCQ0xIZENRVUZ2UWp0VlFVRndRaXh2UWtGQmIwSTdRMEZEY2tJN08wTkJSMFE3UlVGRFJTdzJRa0ZCZFVJN1JVRkJka0lzT0VKQlFYVkNPMDFCUVhaQ0xESkNRVUYxUWp0VlFVRjJRaXgxUWtGQmRVSTdRMEZEZUVJN08wTkJSMFE3UlVGRFJTeFZRVUZWTzBWQlExWXNWMEZCVnp0RlFVTllMSEZDUVVGak8wVkJRV1FzY1VKQlFXTTdSVUZCWkN4alFVRmpPMFZCUTJRc2IwSkJRV1U3VFVGQlppeHRRa0ZCWlR0VlFVRm1MR1ZCUVdVN1JVRkRaaXh6UWtGQmMwSTdRMEZEZGtJN08wTkJSMFE3UlVGRFJTd3JRa0ZCYjBJN1JVRkJjRUlzT0VKQlFXOUNPMDFCUVhCQ0xIZENRVUZ2UWp0VlFVRndRaXh2UWtGQmIwSTdRMEZEY2tJN08wTkJSMFE3UlVGRFJTdzJRa0ZCZFVJN1JVRkJka0lzT0VKQlFYVkNPMDFCUVhaQ0xESkNRVUYxUWp0VlFVRjJRaXgxUWtGQmRVSTdRMEZEZUVJN08wTkJSMFE3UlVGRFJTeHhRa0ZCWXp0RlFVRmtMSEZDUVVGak8wVkJRV1FzWTBGQll6dEZRVU5rTEN0Q1FVRnZRanRGUVVGd1FpdzRRa0ZCYjBJN1RVRkJjRUlzZDBKQlFXOUNPMVZCUVhCQ0xHOUNRVUZ2UWp0RlFVTndRaXdyUWtGQmRVSTdWVUZCZGtJc2RVSkJRWFZDTzBWQlEzWkNMR2xDUVVGcFFqdERRVU5zUWpzN1EwRkhSRHM3UlVGRlJTeHZRa0ZCWlR0TlFVRm1MRzFDUVVGbE8xVkJRV1lzWlVGQlpUdERRVU5vUWpzN1EwRkhSRHRGUVVORkxHOUNRVUZsTzAxQlFXWXNiVUpCUVdVN1ZVRkJaaXhsUVVGbE8wVkJRMllzYVVKQlFXbENPMFZCUTJwQ0xHOUNRVUZ2UWp0RFFVTnlRanM3UTBGSFJEdEZRVU5GTEhsQ1FVRjVRanREUVVNeFFqczdRMEZIUkR0RlFVTkZMRzFDUVVGdFFqdERRVU53UWpzN1EwRkhSRHRGUVVORkxGRkJRVkU3UlVGRFVpeHZRMEZCTkVJN1JVRkJOVUlzTkVKQlFUUkNPME5CUXpkQ096dERRVWRFTzBWQlEwVXNUMEZCVHp0RlFVTlFMRzFEUVVFeVFqdEZRVUV6UWl3eVFrRkJNa0k3UTBGRE5VSTdPME5CUjBRN1JVRkRSU3g1UWtGQmFVSTdSVUZCYWtJc2FVSkJRV2xDTzBOQlEyeENPenREUVVWRUxHOUNRVUZ2UWpzN1EwUTVSM0JDTEZGQlZYRkRMRzlEUVVGdlF6czdTVUV5UW5KRkxDdENRVUVyUWp0RFFVbHNRenM3UTBGRlJEdEpRVU5KTEZsQlFXbERPMGxCUTJwRExDdENRVUYxUWp0WlFVRjJRaXgxUWtGQmRVSTdTVUZEZGtJc1lVRkJLMEk3U1VGREwwSXNhMEpCUVd0Q08wTkJRM0pDT3p0RFFVVkVPMGxCUTBrc2EwSkJRVFpETzBsQlF6ZERMR0ZCUVhkRE8wTkJRek5ET3p0RFFVVkVPMGxCUTBrc1pVRkJaVHRKUVVObUxHZENRVUZuUWp0RFFVTnVRanM3UTBGRlJDeHRRa0ZCYlVJN08wTkJSVzVDTzBsQlEwa3NkMEpCUVhkQ08wbEJRM2hDTEN0Q1FVRjFRanRaUVVGMlFpeDFRa0ZCZFVJN1NVRkRka0lzY1VKQlFXTTdTVUZCWkN4eFFrRkJZenRKUVVGa0xHTkJRV003U1VGRFpDd3JRa0ZCYjBJN1NVRkJjRUlzT0VKQlFXOUNPMUZCUVhCQ0xIZENRVUZ2UWp0WlFVRndRaXh2UWtGQmIwSTdTVUZEY0VJc05FSkJRWE5DTzFGQlFYUkNMSGxDUVVGelFqdFpRVUYwUWl4elFrRkJjMEk3UTBGRGVrSTdPME5CUlVRN1NVRkRTU3h6UWtGQmMwSTdTVUZEZEVJc0swSkJRWFZDTzFsQlFYWkNMSFZDUVVGMVFqdEpRVU4yUWl4eFFrRkJZenRKUVVGa0xIRkNRVUZqTzBsQlFXUXNZMEZCWXp0SlFVTmtMRFpDUVVGMVFqdEpRVUYyUWl3NFFrRkJkVUk3VVVGQmRrSXNNa0pCUVhWQ08xbEJRWFpDTEhWQ1FVRjFRanRKUVVOMlFpd3dRa0ZCYjBJN1VVRkJjRUlzZFVKQlFXOUNPMWxCUVhCQ0xHOUNRVUZ2UWp0RFFVTjJRanM3UTBGRlJEdEpRVU5KTEN0Q1FVRjFRanRaUVVGMlFpeDFRa0ZCZFVJN1NVRkRka0lzY1VKQlFXTTdTVUZCWkN4eFFrRkJZenRKUVVGa0xHTkJRV003U1VGRFpDeFZRVUZWTzBsQlExWXNaVUZCWlR0RFFVTnNRanM3UTBGRlJEdEpRVU5KTEN0Q1FVRjFRanRaUVVGMlFpeDFRa0ZCZFVJN1NVRkRka0lzWTBGQll6dEpRVU5rTEZWQlFWVTdTVUZEVml4bFFVRmxPME5CUTJ4Q096dERRVVZFTzBsQlEwa3NLMEpCUVc5Q08wbEJRWEJDTERoQ1FVRnZRanRSUVVGd1FpeDNRa0ZCYjBJN1dVRkJjRUlzYjBKQlFXOUNPME5CUTNaQ096dERRVVZFTzBsQlEwa3NOa0pCUVhWQ08wbEJRWFpDTERoQ1FVRjFRanRSUVVGMlFpd3lRa0ZCZFVJN1dVRkJka0lzZFVKQlFYVkNPME5CUXpGQ096dERRVVZFTERSQ1FVRTBRanM3UTBGRk5VSTdTVUZEU1N4dFFrRkJiVUk3U1VGRGJrSXNiMEpCUVc5Q08wbEJRM0JDTEdsQ1FVRnBRanRKUVVOcVFpeHZRa0ZCYjBJN1NVRkRjRUlzYzBKQlFYTkNPMGxCUTNSQ0xHOUNRVUZ2UWp0SlFVTndRaXhwUWtGQmFVSTdTVUZEYWtJc2QwSkJRWGRDTzBsQlEzaENMRzFDUVVGdFFqdEpRVU51UWl4blFrRkJkVU03U1VGRGRrTXNaMEpCUVdkQ096dEpRVVZvUWl4aFFVRjNRenRKUVVONFF5eHJRa0ZCYTBJN1NVRkRiRUlzYTBKQlFUWkRPMGxCUXpkRExIbENRVUZwUWp0WlFVRnFRaXhwUWtGQmFVSTdPMGxCUldwQ0xIbENRVUZuUXp0SlFVTm9ReXd3UWtGQk1FTTdTVUZETVVNc2MwSkJRWE5ETzBsQlEzUkRMR0ZCUVdFN1EwRkRhRUk3TzBOQlJVUTdTVUZEU1N4clFrRkJPRU03U1VGRE9VTXNjVUpCUVhGQ08wTkJRM2hDT3p0RFFVVkVPMGxCUTBrc2FVSkJRV2xDTEVOQlFVTXNjMEpCUVhOQ08wTkJRek5ET3p0RFFVVkVPMGxCUTBrc1lVRkJORU03UTBGREwwTTdPME5CUlVRN1NVRkRTU3huUWtGQlowSTdRMEZEYmtJN08wTkJSVVE3U1VGRFNTeDNRa0ZCZDBJN1NVRkRlRUk3T3l0RFFVVXJSVHRaUVVZdlJUczdLME5CUlN0Rk8wTkJRMnhHT3p0RFFVVkVPMGxCUTBrc2QwSkJRWGRDTzBsQlEzaENPenRwUkVGRk5rVTdXVUZHTjBVN08ybEVRVVUyUlR0SlFVTTNSU3g1UWtGQlowTTdTVUZEYUVNc01FSkJRVEJETzBOQlF6ZERPenREUVVWRU8wbEJRMGtzTWtKQlFUaEVPME5CUTJwRk96dERRVVZFTERoQ1FVRTRRanM3UTBGRk9VSTdTVUZEU1N4blEwRkJkME03U1VGRGVFTXNNRUpCUVhsRE8wTkJRelZET3p0RFFVVkVPMGxCUTBrc09FSkJRWGRETzBsQlEzaERMREJDUVVGNVF6dERRVU0xUXpzN1EwRkZSRHRKUVVOSkxEaENRVUYzUXp0SlFVTjRReXd3UWtGQmVVTTdRMEZETlVNN08wTkJSVVFzT0VKQlFUaENPenREUVVVNVFqdEpRVU5KTEdkRFFVRjNRenRKUVVONFF5d3dRa0ZCTWtNN1EwRkRPVU03TzBOQlJVUTdTVUZEU1N3NFFrRkJkME03U1VGRGVFTXNNRUpCUVRKRE8wVkJRemRET3p0RFFVVkdPMGxCUTBrc09FSkJRWGRETzBsQlEzaERMREJDUVVFeVF6dEZRVU0zUXpzN1EwRkZSQ3d5UWtGQk1rSTdPME5CUlRWQ08wbEJRMGtzWjBOQlFYZERPMGxCUTNoRExEQkNRVUYzUXp0RFFVTXpRenM3UTBGRlJEdEpRVU5KTERoQ1FVRjNRenRKUVVONFF5d3dRa0ZCZDBNN1EwRkRNME03TzBOQlJVUTdTVUZEU1N3NFFrRkJkME03U1VGRGVFTXNNRUpCUVhkRE8wTkJRek5ET3p0RFFVVkVMRGhDUVVFNFFqczdRMEZGT1VJN1NVRkRTU3huUTBGQmQwTTdTVUZEZUVNc01FSkJRWGRETzBOQlF6TkRPenREUVVWRU8wbEJRMGtzT0VKQlFYZERPMGxCUTNoRExEQkNRVUYzUXp0RFFVTXpRenM3UTBGRlJEdEpRVU5KTERoQ1FVRjNRenRKUVVONFF5d3dRa0ZCZDBNN1EwRkRNME03TzBOQlJVUXNOa0pCUVRaQ096dERRVVUzUWp0SlFVTkpMR2REUVVGM1F6dEpRVU40UXl3d1FrRkJlVU03UTBGRE5VTTdPME5CUlVRN1NVRkRTU3c0UWtGQmQwTTdTVUZEZUVNc01FSkJRWGxETzBOQlF6VkRPenREUVVWRU8wbEJRMGtzT0VKQlFYZERPMGxCUTNoRExEQkNRVUY1UXp0RFFVTTFRenM3UTBGRlJDeHJRa0ZCYTBJN08wTkJSV3hDTzBsQlEwa3NZVUZCTkVNN1EwRkRMME03TzBOQlJVUXNNRUpCUVRCQ096dERRVVV4UWl4clEwRkJhME03TzBOQlEyeERPMGxCUTBrc2FVSkJRWFZDTzBsQlFYWkNMSFZDUVVGMVFqdERRVU14UWpzN1EwRkZSRHRKUVVOSkxHbENRVUZwUWp0SlFVTnFRaXhoUVVGeFF6dEpRVU55UXl4blFrRkJkVU03U1VGRGRrTXNhVUpCUVdsQ08wbEJRMnBDTEhkQ1FVRjNRanRKUVVONFFpeHZRa0ZCYjBJN1NVRkRjRUlzYTBKQlFUWkRPME5CUTJoRU96dERRVVZFTzBsQlEwa3NWMEZCVnp0SlFVTllMR0ZCUVhGRE8wbEJRM0pETEdkQ1FVRjFRenRKUVVOMlF5eHBRa0ZCYVVJN1NVRkRha0lzZDBKQlFYZENPMGxCUTNoQ0xHOUNRVUZ2UWp0SlFVTndRaXhyUWtGQk5rTTdRMEZEYUVRN08wTkJSVVE3U1VGRFNTdzJRa0ZCTmtJN1NVRkROMElzWVVGQmNVTTdTVUZEY2tNc2EwSkJRV3RDTzBsQlEyeENMR3RDUVVFd1JEdEpRVU14UkN4WlFVRTBRenRKUVVNMVF5eHhRa0ZCWlR0UlFVRm1MR1ZCUVdVN1EwRkRiRUk3TzBOQlJVUTdTVUZEU1N3eVFrRkJNa0k3U1VGRE0wSXNZVUZCY1VNN1NVRkRja01zYlVKQlFXMUNPMGxCUTI1Q0xHdENRVUUyUXp0RFFVTm9SRHM3UTBGRlJDdzBRa0ZCTkVJN08wTkJSVFZDTzBsQlEwa3NZVUZCZFVNN1NVRkRka01zWjBKQlFYVkRPMGxCUTNaRExHRkJRWGRETzBsQlEzaERMR3RDUVVFMlF6dEpRVU0zUXl4cFFrRkJhVUk3U1VGRGFrSXNiMEpCUVc5Q08wbEJRM0JDTEcxQ1FVRnRRanREUVVOMFFqczdRMEZGUkR0SlFVTkpMSGxDUVVGNVFqczdTVUZGZWtJN096czdUMEZKUnp0SlFVTklPenQxUkVGRmIwUTdPMGxCVFhCRU96c3JRMEZGTkVNN1EwRkRMME03TzBOQlJVUTdTVUZEU1N4M1FrRkJkMEk3U1VGRGVFSXNiVUpCUVcxQ08wbEJRMjVDTEdsQ1FVRm5SRHRKUVVOb1JDeG5Ra0ZCSzBNN1NVRkRMME1zYVVKQlFUWkRPME5CUTJoRU96dERRVVZFTzBsQlEwa3NjMEpCUVhOQ08wbEJRM1JDTEdkQ1FVRTBRenRKUVVNMVF5d3lRa0ZCTWtJN1NVRkRNMElzWlVGQlpUdERRVU5zUWpzN1EwRkZSQ3cyUWtGQk5rSTdPME5CUlRkQ08wbEJRMGtzWVVGQmMwTTdTVUZEZEVNc1lVRkJkME03U1VGRGVFTXNhMEpCUVRaRE8wTkJRMmhFT3p0RFFVVkVPMGxCUTBrc2QwSkJRV2RGTzBsQlEyaEZMR3RDUVVFMlF6dEpRVU0zUXl4cFFrRkJhVUk3U1VGRGFrSXNiMEpCUVdFN1VVRkJZaXh4UWtGQllUdFpRVUZpTEdGQlFXRTdTVUZEWWl4eFFrRkJaVHRSUVVGbUxHVkJRV1U3U1VGRFppdzBRa0ZCYlVJN1VVRkJia0lzYlVKQlFXMUNPME5CUTNSQ096dERRVVZFTERCQ1FVRXdRanM3UTBGRk1VSTdTVUZEU1N4aFFVRjNRenRKUVVONFF5eHJRa0ZCTmtNN1NVRkROME1zWVVGQk5FTTdTVUZETlVNc1owSkJRWFZETzBOQlF6RkRPenREUVVWRU8wbEJRMGtzYTBKQlFUWkRPMGxCUXpkRExHdENRVUU0UXp0SlFVTTVReXhwUWtGQk5rTTdPMGxCUlRkRExEQktRVUV3U2p0SlFVTXhTaXh6UWtGQmMwSTdTVUZEZEVJc09FTkJRVGhETzBsQlF6bERMRzFDUVVGdFFqdEpRVU51UWl4eFFrRkJjVUk3U1VGRGNrSXNiME5CUVc5RE8wbEJRM0JETEcxRFFVRnRRenREUVVOMFF6czdRMEZGUkR0SlFVTkpMR2xDUVVGcFFqdEpRVU5xUWl4aFFVRmhPME5CUTJoQ096dERRVVZFTzBsQlEwa3NhVUpCUVdsQ08wbEJRMnBDTEZkQlFWYzdRMEZEWkRzN1EwRkZSRHRKUVVOSkxHTkJRV003UTBGRGFrSTdPME5CUlVRc2NVTkJRWEZET3p0RFFVVnlRenRKUVVOSkxHRkJRWE5ETzBOQlEzcERPenREUVVWRU8wbEJRMGtzWVVGQmQwTTdTVUZEZUVNc2EwSkJRVFpETzBOQlEyaEVPenREUVVWRU8wbEJRMGtzWVVGQk5FTTdRMEZETDBNN08wTkJSVVE3U1VGRFNTd3JRa0ZCZFVJN1dVRkJka0lzZFVKQlFYVkNPMGxCUTNaQ0xEQkNRVUYzUmp0SlFVTjRSaXgzUWtGQk1rUTdTVUZETTBRc2VVSkJRWEZETzBsQlEzSkRMR2RDUVVGMVF6dEpRVU4yUXl4cFFrRkJjMFk3U1VGRGRFWXNiMEpCUVdFN1VVRkJZaXh4UWtGQllUdFpRVUZpTEdGQlFXRTdTVUZEWWl4aFFVRmhMRU5CUVVNc2FVVkJRV2xGTzBsQlF5OUZMSEZDUVVGbE8xRkJRV1lzWlVGQlpUdEpRVU5tTEhsQ1FVRjVRanREUVVNMVFqczdRMEZGUkR0SlFVTkpMR2RDUVVGblFqdEpRVU5vUWl4bFFVRmxPME5CUTJ4Q096dERRVVZFTzBsQlEwa3NjMEpCUVhsRU8wTkJRelZFT3p0RFFVVkVMRzFDUVVGdFFqczdRMEZGYmtJN1NVRkRTU3hyUWtGQmEwSTdTVUZEYkVJc01FSkJRVFJGTzBsQlF6VkZMRzlDUVVGdlF6dEpRVU53UXl3clFrRkJkVUk3V1VGQmRrSXNkVUpCUVhWQ08wbEJRM1pDTEcxQ1FVRnRRanRKUVVOdVFpeHRRa0ZCYlVJN1EwRkRkRUk3TzBOQlJVUTdTVUZEU1N4dFFrRkJiVUk3U1VGRGJrSXNlVUpCUVhsQ0xFTkJRVU1zYjBSQlFXOUVPMGxCUXpsRkxHMUNRVUZ0UWp0SlFVTnVRaXgzUWtGQmJVVTdTVUZEYmtVc01FSkJRV2xITzBsQlEycEhMQ3RDUVVGMVFqdFpRVUYyUWl4MVFrRkJkVUk3U1VGRGRrSXNWMEZCVnp0SlFVTllMSFZDUVVGMVFpeERRVUZETEhkQ1FVRjNRanREUVVOdVJEczdRMEZGUkN4M1FrRkJkMEk3TzBOQlEzaENPMGxCUTBrc01FSkJRU3RFTzBsQlF5OUVMREJDUVVGcFJ6dERRVU53UnpzN1EwRkZSRHRKUVVOSkxEQkNRVUVyUkR0SlFVTXZSQ3h6UWtGQk1rUTdTVUZETTBRc1YwRkJWenRKUVVOWUxEaENRVUZ6UWp0WlFVRjBRaXh6UWtGQmMwSTdRMEZEZWtJN08wTkJSVVE3U1VGRFNTeHBSVUZCYVVVN1NVRkRha1VzYlVKQlFXMUNPMGxCUTI1Q0xHOUNRVUY1UkR0SlFVTjZSQ3hYUVVGWE8wTkJRMlE3TzBOQlJVUXNPRUpCUVRoQ096dERRVVU1UWp0SlFVTkpMRmxCUVRSRE8wbEJRelZETEdGQlFUWkRPMGxCUXpkRExHbENRVUZuU2p0SlFVTm9TaXhyUWtGQmNVYzdTVUZEY2tjc2JVSkJRVzFDTzBsQlEyNUNMRTlCUVU4N1EwRkRWanM3UTBGRlJEdEpRVU5KTEZsQlFUUkRPMGxCUXpWRExHRkJRVFpETzBsQlF6ZERMRzlDUVVGMVJ6dEpRVU4yUnl4clFrRkJhVW83U1VGRGFrb3NiVUpCUVcxQ08wbEJRMjVDTEZGQlFWRTdRMEZEV0RzN1EwRkZSRHRKUVVOSkxGbEJRVFpFTzBsQlF6ZEVMR2xDUVVGNVNqdERRVU0xU2pzN1EwRkZSRHRKUVVOSkxGZEJRVFJFTzBsQlF6VkVMR3RDUVVFd1NqdERRVU0zU2pzN1EwRkZSQ3gxUWtGQmRVSTdPME5CUlhaQ08wbEJRMGtzWVVGQmMwTTdTVUZEZEVNc1lVRkJkME03U1VGRGVFTXNhMEpCUVRaRE96dEpRVVUzUXpzN2IwUkJSV2RFTzBsQlEyaEVMREJDUVVGdlFqdFJRVUZ3UWl4MVFrRkJiMEk3V1VGQmNFSXNiMEpCUVc5Q08wTkJRM1pDT3p0RFFVVkVPMGxCUTBrc2EwSkJRV3RDTzBOQlEzSkNPenREUVVWRU8wbEJRMGtzWVVGQmQwTTdTVUZEZUVNc2FVSkJRWGRITzBsQlEzaEhMR3RDUVVGNVJ6dEpRVU42Unl4dlFrRkJLME03VVVGQkwwTXNiMEpCUVN0RE8xbEJRUzlETEdkQ1FVRXJRenREUVVOc1JEczdRMEZGUkR0SlFVTkpMR2REUVVGblF6dEpRVU5vUXl4WlFVRnBSRHRKUVVOcVJDeHBRa0ZCYlVjN1NVRkRia2NzV1VGQldUdERRVU5tT3p0RFFVVkVMSEZDUVVGeFFqczdRMEZGY2tJN1NVRkRTU3hoUVVGM1F6dEpRVU40UXl4clFrRkJOa003UTBGRGFFUTdPME5CUlVRN1NVRkRTU3h4UWtGQmNVSTdTVUZEY2tJc1kwRkJNRU03U1VGRE1VTXNXVUZCTWtNN1EwRkRPVU03TzBOQlJVUTdTVUZEU1N4dlFrRkJLME03VVVGQkwwTXNiMEpCUVN0RE8xbEJRUzlETEdkQ1FVRXJRenRKUVVNdlF5eHJRa0ZCYTBJN1NVRkRiRUlzYlVKQlFXMUNPMGxCUTI1Q0xHMUNRVUV3Unp0SlFVTXhSeXhuUWtGQmRVYzdTVUZEZGtjc2NVSkJRV003U1VGQlpDeHhRa0ZCWXp0SlFVRmtMR05CUVdNN1NVRkRaQ3cyUWtGQmRVSTdTVUZCZGtJc09FSkJRWFZDTzFGQlFYWkNMREpDUVVGMVFqdFpRVUYyUWl4MVFrRkJkVUk3UTBGRE1VSTdPME5CUlVRN1NVRkRTU3huUTBGQlowTTdTVUZEYUVNc1YwRkJaMFE3U1VGRGFFUXNiMEpCUVdFN1VVRkJZaXh4UWtGQllUdFpRVUZpTEdGQlFXRTdTVUZEWWl4clFrRkJhMEk3U1VGRGJFSXNiVUpCUVcxQ08wTkJRM1JDT3p0RFFVVkVMRFpDUVVFMlFqczdRMEZGTjBJN1NVRkRTU3g1UWtGQmVVSTdTVUZKZWtJc2FVSkJRV2xDTzBOQlEzQkNPenREUVVWRU8wbEJRMGtzWVVGQmQwTTdRMEZETTBNN08wTkJSVVE3U1VGRFNTd3dRa0ZCZVVNN1EwRkROVU03TzBOQlJVUTdTVUZEU1N3d1FrRkJNa003UTBGRE9VTTdPME5CUlVRN1NVRkRTU3d3UWtGQmQwTTdRMEZETTBNN08wTkJSVVE3U1VGRFNTd3dRa0ZCZDBNN1EwRkRNME03TzBOQlJVUTdTVUZEU1N3d1FrRkJlVU03UTBGRE5VTTdPME5CUlVRN1NVRkRTU3d3UWtGQk1FTTdTVUZETVVNc1lVRkJZVHRKUVVOaUxIbENRVUZwUWp0WlFVRnFRaXhwUWtGQmFVSTdRMEZEY0VJN08wTkJSVVFzZVVKQlFYbENPenREUVVWNlFqdEpRVU5KTEd0Q1FVRnJRanRKUVVOc1FpeGhRVUYzUXp0SlFVTjRReXhyUWtGQk5rTTdTVUZETjBNc1lVRkJjME03U1VGRGRFTXNNRUpCUVc5Q08xRkJRWEJDTEhWQ1FVRnZRanRaUVVGd1FpeHZRa0ZCYjBJN08wTkJSWFpDT3p0RFFVVkVPMGxCUTBrc2IwSkJRV0U3VVVGQllpeHhRa0ZCWVR0WlFVRmlMR0ZCUVdFN1NVRkRZaXhuUWtGQk5FTTdTVUZETlVNc2JVSkJRU3RETzBsQlF5OURMRFpDUVVGdlFqdFJRVUZ3UWl4dlFrRkJiMEk3U1VGRGNFSXNPRUpCUVRoQ08wbEJRemxDTEdGQlFXZENPMGxCUVdoQ0xHZENRVUZuUWp0RFFVTnVRanM3UTBGRlJDeDFRa0ZCZFVJN08wTkJSWFpDTzBsQlEwa3NZMEZCTUVNN1NVRkRNVU1zV1VGQk1rTTdRMEZET1VNN08wTkJSVVE3U1VGRFNTeHZRa0ZCWVR0UlFVRmlMSEZDUVVGaE8xbEJRV0lzWVVGQllUdEpRVU5pTEZsQlFUUkRPMGxCUXpWRExHdENRVUZyUWp0SlFVTnNRaXh0UWtGQmJVSTdTVUZEYmtJc2FVSkJRV2xDTzBOQlEzQkNPenREUVVWRUxESkNRVUV5UWpzN1EwRkZNMEk3U1VGRFNTeGhRVUYzUXp0SlFVTjRReXhoUVVGelF6dEpRVU4wUXl4clFrRkJOa003UTBGRGFFUTdPME5CUlVRN1NVRkRTU3h2UWtGQmIwSTdTVUZEY0VJc01FSkJRWGRHTzBsQlEzaEdMR2xDUVVGcFFqdEpRVU5xUWl4blFrRkJaMEk3U1VGRGFFSXNiMEpCUVN0RE8xRkJRUzlETEc5Q1FVRXJRenRaUVVFdlF5eG5Ra0ZCSzBNN1NVRkRMME1zWVVGQllTeERRVUZETEdsRlFVRnBSVHRKUVVNdlJTd3JRa0ZCZFVJN1dVRkJka0lzZFVKQlFYVkNPMGxCUTNaQ0xIbENRVUY1UWp0SlFVTjZRaXg1UWtGQmFVSTdXVUZCYWtJc2FVSkJRV2xDTzBsQlEycENMSGRDUVVFeVJEdEpRVU16UkN4NVFrRkJjVU03U1VGRGNrTXNaMEpCUVhWRE8wbEJRM1pETEc5Q1FVRnZRanRKUVVOd1FpeHJRa0ZCZVVRN1EwRkROVVFzYVVKQlFXbENPME5CUTJwQ0xIbENRVUY1UWp0RFFVTjZRaXh6UWtGQmMwSTdTVUZEYmtJc05rSkJRVFpDTzBOQlEyaERMSE5DUVVGelFqdERRVU4wUWl4clEwRkJhME03U1VGREwwSXNhM1ZDUVVGdFJEdERRVU4wUkRzN1EwRkRSRHRKUVVOSkxITkNRVUY1UkR0RFFVTTFSRHM3UTBGRlJEdEpRVU5KTEdGQlFUUkRPME5CUXk5RE96dERRVVZFT3paRFFVTTJRenM3UTBGRE4wTTdTVUZEU1N4dFFrRkJiVUk3U1VGRGJrSXNkMEpCUVhkQ08wTkJRek5DT3p0RFFVVkVMQ3RDUVVFclFqczdRMEZGTDBJN1NVRkRTU3hoUVVGelF6dEpRVU4wUXl4clFrRkJOa003TzBsQlJUZERPenRyUlVGRk9FUTdTVUZET1VRc2VVSkJRWGRDTzFGQlFYaENMSE5DUVVGM1FqdFpRVUY0UWl4M1FrRkJkMEk3UTBGRE0wSTdPME5CUlVRN1NVRkRTU3d3UWtGQmQwWTdTVUZEZUVZc2QwSkJRVEpFTzBsQlF6TkVMSGxDUVVGeFF6dEpRVU55UXl4blFrRkJkVU03U1VGRGRrTXNiMEpCUVN0RE8xRkJRUzlETEc5Q1FVRXJRenRaUVVFdlF5eG5Ra0ZCSzBNN1NVRkRMME1zZVVKQlFYbENPMGxCUTNwQ0xHVkJRV1U3U1VGRFppeG5Ra0ZCWjBJN08wbEJSV2hDT3p0clJVRkZPRVE3U1VGRE9VUXNhVUpCUVdsQ08wTkJRM0JDT3p0RFFVVkVPMGxCUTBrc2MwSkJRWGxFTzBOQlF6VkVPenREUVVWRU8wbEJRMGtzYTBKQlFUaERPMGxCUXpsRExHdENRVUUyUXp0SlFVTTNReXhyUlVGQmEwVTdTVUZEYkVVc01FUkJRV2xHTzBsQlEycEdMRFpFUVVGdlJqdERRVU4yUmpzN1EwRkpSQ3cwUWtGQk5FSTdPME5CUlRWQ08wbEJRMGtzYTBKQlFUWkRPME5CUTJoRU96dERRVVZFTzBsQlEwa3NhVUpCUVhORE8wbEJRM1JETEd0Q1FVRjFRenREUVVNeFF6czdRMEZGUkR0SlFVTkpMR0ZCUVRSRE8wTkJReTlET3p0RFFVVkVMREpDUVVFeVFqczdRMEZGTTBJN1NVRkRTU3hoUVVGelF6dEpRVU4wUXl4clFrRkJOa003UTBGRGFFUTdPME5CUlVRN1NVRkRTU3h4UWtGQll6dEpRVUZrTEhGQ1FVRmpPMGxCUVdRc1kwRkJZenRKUVVOa0xEWkNRVUYxUWp0SlFVRjJRaXc0UWtGQmRVSTdVVUZCZGtJc01rSkJRWFZDTzFsQlFYWkNMSFZDUVVGMVFqdEpRVU4yUWl3eVFrRkJjVUk3VVVGQmNrSXNkMEpCUVhGQ08xbEJRWEpDTEhGQ1FVRnhRanRKUVVOeVFpd3JRa0ZCZFVJN1dVRkJka0lzZFVKQlFYVkNPMGxCUTNaQ0xHOUNRVUZoTzFGQlFXSXNjVUpCUVdFN1dVRkJZaXhoUVVGaE8wbEJRMklzYlVKQlFUaEVPME5CUTJwRk96dERRVVZFTzBsQlEwa3NZVUZCTkVNN1NVRkROVU1zYTBKQlFXbEVPMGxCUTJwRUxHZENRVUYxUXp0RFFVTXhRenM3UTBGRlJEdEpRVU5KTEdGQlFUUkRPMGxCUXpWRExHdENRVUZwUkR0SlFVTnFSQ3h2UWtGQk5FUTdTVUZETlVRc1dVRkJXVHREUVVObU96dERRVVZFTERCQ1FVRXdRanM3UTBGRk1VSTdTVUZEU1N4aFFVRnpRenRKUVVOMFF5eGhRVUYzUXp0SlFVTjRReXhyUWtGQk5rTTdRMEZEYUVRN08wTkJSVVE3U1VGRFNTeHZRa0ZCWVR0UlFVRmlMSEZDUVVGaE8xbEJRV0lzWVVGQllUdEpRVU5pTEhGQ1FVRmxPMUZCUVdZc1pVRkJaVHRKUVVObUxHZENRVUVyUXp0RFFVTnNSRHM3UTBGRlJEdEpRVU5KTEZsQlFYVkRPMGxCUTNaRExHRkJRWGRETzBsQlEzaERMR1ZCUVdVc1EwRkJReXcyUkVGQk5rUTdTVUZETjBVc2EwSkJRWEZFTzBsQlEzSkVMSGxDUVVGeFF6dEpRVU55UXl3d1FrRkJkMFk3U1VGRGVFWXNhMEpCUVd0Q08wbEJRMnhDTEc5Q1FVRmhPMUZCUVdJc2NVSkJRV0U3V1VGQllpeGhRVUZoTzBsQlEySXNjVUpCUVdVN1VVRkJaaXhsUVVGbE8wbEJRMllzSzBKQlFYVkNPMWxCUVhaQ0xIVkNRVUYxUWp0SlFVTjJRaXcyUWtGQmIwSTdVVUZCY0VJc2IwSkJRVzlDTzBsQlEzQkNMSGxDUVVGNVFqdERRVU0xUWpzN1EwRkZSRHRKUVVOSkxDdENRVUUyUmp0RFFVTm9SenM3UTBGRlJEdEpRVU5KTEhOQ1FVRjVSRHREUVVNMVJEczdRMEZGUkR0SlFVTkpMRzlDUVVGaE8xRkJRV0lzY1VKQlFXRTdXVUZCWWl4aFFVRmhPMGxCUTJJc2VVSkJRWGxDTzBsQlEzcENMR0ZCUVhkRE8wbEJRM2hETEd0Q1FVRTJRenRKUVVNM1F5eHJRa0ZCY1VRN1NVRkRja1FzZVVKQlFYRkRPMGxCUTNKRExEQkNRVUYzUmp0SlFVTjRSaXhuUWtGQmRVTTdTVUZEZGtNc2FVSkJRWE5HTzBsQlEzUkdMR0ZCUVdFc1EwRkJReXhwUlVGQmFVVTdTVUZETDBVc2NVSkJRV1U3VVVGQlppeGxRVUZsTzBsQlEyWXNLMEpCUVhWQ08xbEJRWFpDTEhWQ1FVRjFRanREUVVNeFFqczdRMEZGUkR0SlFVTkpMR0ZCUVRSRE8wTkJReTlET3p0RFFVVkVMSGxDUVVGNVFqczdRMEZGZWtJN1NVRkRTU3hoUVVGelF6dEpRVU4wUXl4aFFVRjNRenRKUVVONFF5eHJRa0ZCTmtNN1EwRkRhRVE3TzBOQlJVUTdTVUZEU1N4dlFrRkJZVHRSUVVGaUxIRkNRVUZoTzFsQlFXSXNZVUZCWVR0SlFVTmlMSEZDUVVGbE8xRkJRV1lzWlVGQlpUdEpRVU5tTEdGQlFXRXNRMEZCUXl4cFJVRkJhVVU3U1VGREwwVXNlVUpCUVhsQ08wbEJRM3BDTEdGQlFYZERPMGxCUTNoRExEQkNRVUYzUmp0SlFVTjRSaXgzUWtGQk1rUTdTVUZETTBRc2VVSkJRWEZETzBsQlEzSkRMR2RDUVVGMVF6dEpRVU4yUXl4cFFrRkJjMFk3U1VGRGRFWXNLMEpCUVhWQ08xbEJRWFpDTEhWQ1FVRjFRanREUVVNeFFqczdRMEZGUkR0SlFVTkpMSE5DUVVGNVJEdERRVU0xUkRzN1EwRkZSRHRKUVVOSkxITkNRVUZ2UXp0RFFVTjJRenM3UTBGRlJEdEpRVU5KTEdGQlFUUkRPME5CUXk5RE96dERRVVZFTEdsQ1FVRnBRanM3UTBGRmFrSTdTVUZEU1N4aFFVRTBRenRKUVVNMVF5eHhRa0ZCWXp0SlFVRmtMSEZDUVVGak8wbEJRV1FzWTBGQll6dEpRVU5rTERKQ1FVRnhRanRSUVVGeVFpeDNRa0ZCY1VJN1dVRkJja0lzY1VKQlFYRkNPME5CUTNoQ096dERRVVZFTzBsQlEwa3NiMEpCUVdFN1VVRkJZaXh4UWtGQllUdFpRVUZpTEdGQlFXRTdTVUZEWWl4aFFVRmhPME5CUTJoQ096dERRVVZFTzBsQlEwa3NZVUZCTkVNN1EwRkRMME03TzBOQlJVUXNaMEpCUVdkQ096dERRVVZvUWp0SlFVTkpMSEZDUVVGak8wbEJRV1FzY1VKQlFXTTdTVUZCWkN4alFVRmpPMGxCUTJRc05rSkJRWFZDTzBsQlFYWkNMRGhDUVVGMVFqdFJRVUYyUWl3eVFrRkJkVUk3V1VGQmRrSXNkVUpCUVhWQ08wTkJRekZDT3p0RFFVVkVPMGxCUTBrc2VVWkJRWGxHTzBsQlEzcEdMRzlDUVVGdlFqdEpRVU53UWl4dlFrRkJiMEk3UTBGRGRrSTdPME5CUlVRN1NVRkRTU3hwUkVGQmFVUTdTVUZEYWtRc2RVSkJRWE5DTzFGQlFYUkNMRzlDUVVGelFqdFpRVUYwUWl4elFrRkJjMEk3U1VGRGRFSXNZVUZCWVR0SlFVTmlMR05CUVdNN1EwRkRha0k3TzBOQlJVUTdTVUZEU1N4WlFVRlpPMGxCUTFvc0swSkJRWFZDTzFsQlFYWkNMSFZDUVVGMVFqdEpRVU4yUWl4VlFVRlZPMGxCUTFZc2EwSkJRVzlETzBsQlEzQkRMSGxDUVVGblF6dEpRVU5vUXl3d1FrRkJOa1E3U1VGRE4wUXNZMEZCTmtNN1NVRkROME1zYjBKQlFXRTdVVUZCWWl4eFFrRkJZVHRaUVVGaUxHRkJRV0U3U1VGRFlpeGxRVUZsTzBOQlEyeENPenREUVVWRU8wbEJRMGtzZDBOQlFTdEVPMGxCUXk5RUxHbENRVUZ0Ump0RFFVTjBSanM3UTBGRlJEdEpRVU5KTEc5Q1FVRnBSRHRSUVVGcVJDeHZRa0ZCYVVRN1dVRkJha1FzWjBKQlFXbEVPMGxCUTJwRUxHZENRVUZuUWp0SlFVTm9RaXhwUWtGQmJVWTdTVUZEYmtZc2EwSkJRWEZFTzBsQlEzSkVMR3RDUVVFclF6dEpRVU12UXl4clFrRkJhMEk3U1VGRGJFSXNiMEpCUVc5RE8wbEJRM0JETEhsQ1FVRm5RenRKUVVOb1F5d3dRa0ZCTmtRN1NVRkROMFFzYjBKQlFXOUNPMGxCUTNCQ0xHMUNRVUZ0UWp0RFFVTjBRanM3UTBGRlJEdEpRVU5KTERCQ1FVRm5RenRKUVVOb1F5eG5SVUZCWjBVN1NVRkRhRVVzYTBKQlFXOURPMGxCUTNCRExHbENRVUYxUmp0SlFVTjJSaXh0UTBGQk9FTTdXVUZCT1VNc01rSkJRVGhETzBsQlF6bERMR3RDUVVGclFqdERRVU55UWpzN1EwRkZSRHRKUVVOSkxHMUNRVUZ0UWp0SlFVTnVRaXhWUVVGMVF6dEpRVU4yUXl4WFFVRjNRenRKUVVONFF5eFpRVUZaTzBsQlExb3NXVUZCYjBRN1NVRkRjRVFzZDBKQlFTdERPMGxCUXk5RExHOUNRVUZ0UXp0RFFVTjBRenM3UTBGRlJEdEpRVU5KTEdWQlFXVTdRMEZEYkVJN08wTkJSVVE3U1VGRFNTeHJRa0ZCYjBNN1NVRkRjRU1zZVVKQlFXZERPME5CUTI1RE96dERRVVZFTzBsQlEwa3NhVUpCUVdsQ08wTkJRM0JDT3p0RFFVVkVPMGxCUTBrc2VVSkJRWGxDTzBsQlEzcENMR2xDUVVGcFFpeERRVUZETEZkQlFWYzdRMEZEYUVNN08wTkJSVVE3T3p0SlFVZEpMR3RDUVVGeFJEdERRVU40UkRzN1EwRkZSQ3h6UWtGQmMwSTdPME5CUlhSQ08wbEJRMGtzY1VKQlFXTTdTVUZCWkN4eFFrRkJZenRKUVVGa0xHTkJRV003U1VGRFpDdzJRa0ZCZFVJN1NVRkJka0lzT0VKQlFYVkNPMUZCUVhaQ0xESkNRVUYxUWp0WlFVRjJRaXgxUWtGQmRVSTdTVUZEZGtJc01rSkJRWEZDTzFGQlFYSkNMSGRDUVVGeFFqdFpRVUZ5UWl4eFFrRkJjVUk3UTBGRGVFSTdPME5CUlVRN1NVRkRTU3hoUVVGNVF6dEpRVU42UXl4blFrRkJaMEk3U1VGRGFFSXNlVUpCUVdkRE8wbEJRMmhETERCQ1FVRXdRenRKUVVNeFF5d3dRa0ZCY1VVN1NVRkRja1VzYlVKQlFTdEdPMGxCUXk5R0xHdENRVUZyUWp0RFFVTnlRanM3UTBGRlJEdEpRVU5KTEhkQ1FVRXdRenRKUVVNeFF5eDVRa0ZCWjBNN1EwRkRia003TzBOQlJVUTdTVUZEU1N4M1FrRkJNRU03U1VGRE1VTXNNRUpCUVdkRE8wbEJRMmhETEdkQ1FVRm5RanRKUVVOb1FpeHZRa0ZCYjBJN1EwRkRka0k3TzBOQlJVUTdTVUZEU1N4elFrRkJjMElzUlVGQlJTeHhRMEZCY1VNN1NVRkROMFFzYzBKQlFYTkNPMGxCUTNSQ0xEaERRVUU0UXp0SlFVTTVReXh0UWtGQmJVSTdTVUZEYmtJc2NVSkJRWEZDTzBsQlEzSkNMRzlEUVVGdlF6dEpRVU53UXl4dFEwRkJiVU03UTBGRGRFTTdPME5CUlVRN1NVRkRTU3h6UWtGQmMwSXNRMEZCUXl4dlEwRkJiME03UTBGRE9VUTdPME5CUlVRN1NVRkRTU3hqUVVFMlF6dEpRVU0zUXl4M1FrRkJNRU03U1VGRE1VTXNlVUpCUVdkRE8wbEJRMmhETEN0Q1FVRXdSVHRKUVVNeFJTeG5RMEZCTWtVN1NVRkRNMFVzYVVOQlFUUkZPMGxCUXpWRkxHVkJRV1U3UTBGRGJFSTdPME5CUlVRN1NVRkRTU3h4UWtGQll6dEpRVUZrTEhGQ1FVRmpPMGxCUVdRc1kwRkJZenRKUVVOa0xEWkNRVUYxUWp0SlFVRjJRaXc0UWtGQmRVSTdVVUZCZGtJc01rSkJRWFZDTzFsQlFYWkNMSFZDUVVGMVFqdEpRVU4yUWl3eVFrRkJjVUk3VVVGQmNrSXNkMEpCUVhGQ08xbEJRWEpDTEhGQ1FVRnhRanREUVVONFFqczdRMEZGUkR0SlFVTkpMR2xDUVVGcFFqdERRVU53UWpzN1EwRkZSRHRKUVVOSkxHZENRVUZuUWp0RFFVTnVRanM3UTBGSlJDeHBRa0ZCYVVJN08wTkJSV3BDTzBsQlEwa3NaMEpCUVhWRE8wTkJRekZET3p0RFFVVkVPMGxCUTBrc01FTkJRVEJETzBsQlF6RkRMRFpDUVVGdlFqdFJRVUZ3UWl4dlFrRkJiMEk3U1VGRGNFSXNiMEpCUVdFN1VVRkJZaXh4UWtGQllUdFpRVUZpTEdGQlFXRTdTVUZEWWl4eFFrRkJaVHRSUVVGbUxHVkJRV1U3U1VGRFppeHJSVUZCYTBVN1NVRkRiRVVzYTBKQlFUWkRPMGxCUXpkRExIbEZRVUY1UlR0SlFVTjZSU3h0UWtGQmJVSTdRMEZEZEVJaUxDSm1hV3hsSWpvaVkyOXVkSEp2YkhNdVkzTnpJaXdpYzI5MWNtTmxjME52Ym5SbGJuUWlPbHNpTHlvZ1EyOXdlWEpwWjJoMElDaGpLU0JLZFhCNWRHVnlJRVJsZG1Wc2IzQnRaVzUwSUZSbFlXMHVYRzRnS2lCRWFYTjBjbWxpZFhSbFpDQjFibVJsY2lCMGFHVWdkR1Z5YlhNZ2IyWWdkR2hsSUUxdlpHbG1hV1ZrSUVKVFJDQk1hV05sYm5ObExseHVJQ292WEc1Y2JpQXZLaUJYWlNCcGJYQnZjblFnWVd4c0lHOW1JSFJvWlhObElIUnZaMlYwYUdWeUlHbHVJR0VnYzJsdVoyeGxJR056Y3lCbWFXeGxJR0psWTJGMWMyVWdkR2hsSUZkbFluQmhZMnRjYm14dllXUmxjaUJ6WldWeklHOXViSGtnYjI1bElHWnBiR1VnWVhRZ1lTQjBhVzFsTGlCVWFHbHpJR0ZzYkc5M2N5QndiM04wWTNOeklIUnZJSE5sWlNCMGFHVWdkbUZ5YVdGaWJHVmNibVJsWm1sdWFYUnBiMjV6SUhkb1pXNGdkR2hsZVNCaGNtVWdkWE5sWkM0Z0tpOWNibHh1UUdsdGNHOXlkQ0JjSWk0dmJHRmlkbUZ5YVdGaWJHVnpMbU56YzF3aU8xeHVRR2x0Y0c5eWRDQmNJaTR2ZDJsa1oyVjBjeTFpWVhObExtTnpjMXdpTzF4dUlpd2lMeW90TFMwdExTMHRMUzB0TFMwdExTMHRMUzB0TFMwdExTMHRMUzB0TFMwdExTMHRMUzB0TFMwdExTMHRMUzB0TFMwdExTMHRMUzB0TFMwdExTMHRMUzB0TFMwdExTMHRMUzB0TFMwdExWeHVmQ0JEYjNCNWNtbG5hSFFnS0dNcElFcDFjSGwwWlhJZ1JHVjJaV3h2Y0cxbGJuUWdWR1ZoYlM1Y2Jud2dSR2x6ZEhKcFluVjBaV1FnZFc1a1pYSWdkR2hsSUhSbGNtMXpJRzltSUhSb1pTQk5iMlJwWm1sbFpDQkNVMFFnVEdsalpXNXpaUzVjYm53dExTMHRMUzB0TFMwdExTMHRMUzB0TFMwdExTMHRMUzB0TFMwdExTMHRMUzB0TFMwdExTMHRMUzB0TFMwdExTMHRMUzB0TFMwdExTMHRMUzB0TFMwdExTMHRMUzB0TFMwdExTMHRLaTljYmx4dUx5cGNibFJvYVhNZ1ptbHNaU0JwY3lCamIzQnBaV1FnWm5KdmJTQjBhR1VnU25Wd2VYUmxja3hoWWlCd2NtOXFaV04wSUhSdklHUmxabWx1WlNCa1pXWmhkV3gwSUhOMGVXeHBibWNnWm05eVhHNTNhR1Z1SUhSb1pTQjNhV1JuWlhRZ2MzUjViR2x1WnlCcGN5QmpiMjF3YVd4bFpDQmtiM2R1SUhSdklHVnNhVzFwYm1GMFpTQkRVMU1nZG1GeWFXRmliR1Z6TGlCWFpTQnRZV3RsSUc5dVpWeHVZMmhoYm1kbElDMGdkMlVnWTI5dGJXVnVkQ0J2ZFhRZ2RHaGxJR1p2Ym5RZ2FXMXdiM0owSUdKbGJHOTNMbHh1S2k5Y2JseHVRR2x0Y0c5eWRDQmNJaTR2YldGMFpYSnBZV3hqYjJ4dmNuTXVZM056WENJN1hHNWNiaThxWEc1VWFHVWdabTlzYkc5M2FXNW5JRU5UVXlCMllYSnBZV0pzWlhNZ1pHVm1hVzVsSUhSb1pTQnRZV2x1TENCd2RXSnNhV01nUVZCSklHWnZjaUJ6ZEhsc2FXNW5JRXAxY0hsMFpYSk1ZV0l1WEc1VWFHVnpaU0IyWVhKcFlXSnNaWE1nYzJodmRXeGtJR0psSUhWelpXUWdZbmtnWVd4c0lIQnNkV2RwYm5NZ2QyaGxjbVYyWlhJZ2NHOXpjMmxpYkdVdUlFbHVJRzkwYUdWeVhHNTNiM0prY3l3Z2NHeDFaMmx1Y3lCemFHOTFiR1FnYm05MElHUmxabWx1WlNCamRYTjBiMjBnWTI5c2IzSnpMQ0J6YVhwbGN5d2daWFJqSUhWdWJHVnpjeUJoWW5OdmJIVjBaV3g1WEc1dVpXTmxjM05oY25rdUlGUm9hWE1nWlc1aFlteGxjeUIxYzJWeWN5QjBieUJqYUdGdVoyVWdkR2hsSUhacGMzVmhiQ0IwYUdWdFpTQnZaaUJLZFhCNWRHVnlUR0ZpWEc1aWVTQmphR0Z1WjJsdVp5QjBhR1Z6WlNCMllYSnBZV0pzWlhNdVhHNWNiazFoYm5rZ2RtRnlhV0ZpYkdWeklHRndjR1ZoY2lCcGJpQmhiaUJ2Y21SbGNtVmtJSE5sY1hWbGJtTmxJQ2d3TERFc01pd3pLUzRnVkdobGMyVWdjMlZ4ZFdWdVkyVnpYRzVoY21VZ1pHVnphV2R1WldRZ2RHOGdkMjl5YXlCM1pXeHNJSFJ2WjJWMGFHVnlMQ0J6YnlCbWIzSWdaWGhoYlhCc1pTd2dZQzB0YW5BdFltOXlaR1Z5TFdOdmJHOXlNV0FnYzJodmRXeGtYRzVpWlNCMWMyVmtJSGRwZEdnZ1lDMHRhbkF0YkdGNWIzVjBMV052Ykc5eU1XQXVJRlJvWlNCdWRXMWlaWEp6SUdoaGRtVWdkR2hsSUdadmJHeHZkMmx1WnlCdFpXRnVhVzVuY3pwY2JseHVLaUF3T2lCemRYQmxjaTF3Y21sdFlYSjVMQ0J5WlhObGNuWmxaQ0JtYjNJZ2MzQmxZMmxoYkNCbGJYQm9ZWE5wYzF4dUtpQXhPaUJ3Y21sdFlYSjVMQ0J0YjNOMElHbHRjRzl5ZEdGdWRDQjFibVJsY2lCdWIzSnRZV3dnYzJsMGRXRjBhVzl1YzF4dUtpQXlPaUJ6WldOdmJtUmhjbmtzSUc1bGVIUWdiVzl6ZENCcGJYQnZjblJoYm5RZ2RXNWtaWElnYm05eWJXRnNJSE5wZEhWaGRHbHZibk5jYmlvZ016b2dkR1Z5ZEdsaGNua3NJRzVsZUhRZ2JXOXpkQ0JwYlhCdmNuUmhiblFnZFc1a1pYSWdibTl5YldGc0lITnBkSFZoZEdsdmJuTmNibHh1VkdoeWIzVm5hRzkxZENCS2RYQjVkR1Z5VEdGaUxDQjNaU0JoY21VZ2JXOXpkR3g1SUdadmJHeHZkMmx1WnlCd2NtbHVZMmx3YkdWeklHWnliMjBnUjI5dloyeGxKM05jYmsxaGRHVnlhV0ZzSUVSbGMybG5iaUIzYUdWdUlITmxiR1ZqZEdsdVp5QmpiMnh2Y25NdUlGZGxJR0Z5WlNCdWIzUXNJR2h2ZDJWMlpYSXNJR1p2Ykd4dmQybHVaMXh1WVd4c0lHOW1JRTFFSUdGeklHbDBJR2x6SUc1dmRDQnZjSFJwYldsNlpXUWdabTl5SUdSbGJuTmxMQ0JwYm1admNtMWhkR2x2YmlCeWFXTm9JRlZKY3k1Y2Jpb3ZYRzVjYmx4dUx5cGNiaUFxSUU5d2RHbHZibUZzSUcxdmJtOXpjR0ZqWlNCbWIyNTBJR1p2Y2lCcGJuQjFkQzl2ZFhSd2RYUWdjSEp2YlhCMExseHVJQ292WEc0Z0x5b2dRMjl0YldWdWRHVmtJRzkxZENCcGJpQnBjSGwzYVdSblpYUnpJSE5wYm1ObElIZGxJR1J2YmlkMElHNWxaV1FnYVhRdUlDb3ZYRzR2S2lCQWFXMXdiM0owSUhWeWJDZ25hSFIwY0hNNkx5OW1iMjUwY3k1bmIyOW5iR1ZoY0dsekxtTnZiUzlqYzNNL1ptRnRhV3g1UFZKdlltOTBieXROYjI1dkp5azdJQ292WEc1Y2JpOHFYRzRnS2lCQlpHUmxaQ0JtYjNJZ1kyOXRjR0ZpYVhScGJHbDBlU0IzYVhSb0lHOTFkSEIxZENCaGNtVmhYRzRnS2k5Y2JqcHliMjkwSUh0Y2JpQWdMUzFxY0MxcFkyOXVMWE5sWVhKamFEb2dibTl1WlR0Y2JpQWdMUzFxY0MxMWFTMXpaV3hsWTNRdFkyRnlaWFE2SUc1dmJtVTdYRzU5WEc1Y2JseHVPbkp2YjNRZ2UxeHVYRzRnSUM4cUlFSnZjbVJsY25OY2JseHVJQ0JVYUdVZ1ptOXNiRzkzYVc1bklIWmhjbWxoWW14bGN5d2djM0JsWTJsbWVTQjBhR1VnZG1semRXRnNJSE4wZVd4cGJtY2diMllnWW05eVpHVnljeUJwYmlCS2RYQjVkR1Z5VEdGaUxseHVJQ0FnS2k5Y2JseHVJQ0F0TFdwd0xXSnZjbVJsY2kxM2FXUjBhRG9nTVhCNE8xeHVJQ0F0TFdwd0xXSnZjbVJsY2kxamIyeHZjakE2SUhaaGNpZ3RMVzFrTFdkeVpYa3ROekF3S1R0Y2JpQWdMUzFxY0MxaWIzSmtaWEl0WTI5c2IzSXhPaUIyWVhJb0xTMXRaQzFuY21WNUxUVXdNQ2s3WEc0Z0lDMHRhbkF0WW05eVpHVnlMV052Ykc5eU1qb2dkbUZ5S0MwdGJXUXRaM0psZVMwek1EQXBPMXh1SUNBdExXcHdMV0p2Y21SbGNpMWpiMnh2Y2pNNklIWmhjaWd0TFcxa0xXZHlaWGt0TVRBd0tUdGNibHh1SUNBdktpQlZTU0JHYjI1MGMxeHVYRzRnSUZSb1pTQlZTU0JtYjI1MElFTlRVeUIyWVhKcFlXSnNaWE1nWVhKbElIVnpaV1FnWm05eUlIUm9aU0IwZVhCdlozSmhjR2g1SUdGc2JDQnZaaUIwYUdVZ1NuVndlWFJsY2t4aFlseHVJQ0IxYzJWeUlHbHVkR1Z5Wm1GalpTQmxiR1Z0Wlc1MGN5QjBhR0YwSUdGeVpTQnViM1FnWkdseVpXTjBiSGtnZFhObGNpQm5aVzVsY21GMFpXUWdZMjl1ZEdWdWRDNWNiaUFnS2k5Y2JseHVJQ0F0TFdwd0xYVnBMV1p2Ym5RdGMyTmhiR1V0Wm1GamRHOXlPaUF4TGpJN1hHNGdJQzB0YW5BdGRXa3RabTl1ZEMxemFYcGxNRG9nWTJGc1l5aDJZWElvTFMxcWNDMTFhUzFtYjI1MExYTnBlbVV4S1M5MllYSW9MUzFxY0MxMWFTMW1iMjUwTFhOallXeGxMV1poWTNSdmNpa3BPMXh1SUNBdExXcHdMWFZwTFdadmJuUXRjMmw2WlRFNklERXpjSGc3SUM4cUlFSmhjMlVnWm05dWRDQnphWHBsSUNvdlhHNGdJQzB0YW5BdGRXa3RabTl1ZEMxemFYcGxNam9nWTJGc1l5aDJZWElvTFMxcWNDMTFhUzFtYjI1MExYTnBlbVV4S1NwMllYSW9MUzFxY0MxMWFTMW1iMjUwTFhOallXeGxMV1poWTNSdmNpa3BPMXh1SUNBdExXcHdMWFZwTFdadmJuUXRjMmw2WlRNNklHTmhiR01vZG1GeUtDMHRhbkF0ZFdrdFptOXVkQzF6YVhwbE1pa3FkbUZ5S0MwdGFuQXRkV2t0Wm05dWRDMXpZMkZzWlMxbVlXTjBiM0lwS1R0Y2JpQWdMUzFxY0MxMWFTMXBZMjl1TFdadmJuUXRjMmw2WlRvZ01UUndlRHNnTHlvZ1JXNXpkWEpsY3lCd2VDQndaWEptWldOMElFWnZiblJCZDJWemIyMWxJR2xqYjI1eklDb3ZYRzRnSUMwdGFuQXRkV2t0Wm05dWRDMW1ZVzFwYkhrNklGd2lTR1ZzZG1WMGFXTmhJRTVsZFdWY0lpd2dTR1ZzZG1WMGFXTmhMQ0JCY21saGJDd2djMkZ1Y3kxelpYSnBaanRjYmx4dUlDQXZLaUJWYzJVZ2RHaGxjMlVnWm05dWRDQmpiMnh2Y25NZ1lXZGhhVzV6ZENCMGFHVWdZMjl5Y21WemNHOXVaR2x1WnlCdFlXbHVJR3hoZVc5MWRDQmpiMnh2Y25NdVhHNGdJQ0FnSUVsdUlHRWdiR2xuYUhRZ2RHaGxiV1VzSUhSb1pYTmxJR2R2SUdaeWIyMGdaR0Z5YXlCMGJ5QnNhV2RvZEM1Y2JpQWdLaTljYmx4dUlDQXRMV3B3TFhWcExXWnZiblF0WTI5c2IzSXdPaUJ5WjJKaEtEQXNNQ3d3TERFdU1DazdYRzRnSUMwdGFuQXRkV2t0Wm05dWRDMWpiMnh2Y2pFNklISm5ZbUVvTUN3d0xEQXNNQzQ0S1R0Y2JpQWdMUzFxY0MxMWFTMW1iMjUwTFdOdmJHOXlNam9nY21kaVlTZ3dMREFzTUN3d0xqVXBPMXh1SUNBdExXcHdMWFZwTFdadmJuUXRZMjlzYjNJek9pQnlaMkpoS0RBc01Dd3dMREF1TXlrN1hHNWNiaUFnTHlvZ1ZYTmxJSFJvWlhObElHRm5ZV2x1YzNRZ2RHaGxJR0p5WVc1a0wyRmpZMlZ1ZEM5M1lYSnVMMlZ5Y205eUlHTnZiRzl5Y3k1Y2JpQWdJQ0FnVkdobGMyVWdkMmxzYkNCMGVYQnBZMkZzYkhrZ1oyOGdabkp2YlNCc2FXZG9kQ0IwYnlCa1lYSnJaWElzSUdsdUlHSnZkR2dnWVNCa1lYSnJJR0Z1WkNCc2FXZG9kQ0IwYUdWdFpWeHVJQ0FnS2k5Y2JseHVJQ0F0TFdwd0xXbHVkbVZ5YzJVdGRXa3RabTl1ZEMxamIyeHZjakE2SUhKblltRW9NalUxTERJMU5Td3lOVFVzTVNrN1hHNGdJQzB0YW5BdGFXNTJaWEp6WlMxMWFTMW1iMjUwTFdOdmJHOXlNVG9nY21kaVlTZ3lOVFVzTWpVMUxESTFOU3d4TGpBcE8xeHVJQ0F0TFdwd0xXbHVkbVZ5YzJVdGRXa3RabTl1ZEMxamIyeHZjakk2SUhKblltRW9NalUxTERJMU5Td3lOVFVzTUM0M0tUdGNiaUFnTFMxcWNDMXBiblpsY25ObExYVnBMV1p2Ym5RdFkyOXNiM0l6T2lCeVoySmhLREkxTlN3eU5UVXNNalUxTERBdU5TazdYRzVjYmlBZ0x5b2dRMjl1ZEdWdWRDQkdiMjUwYzF4dVhHNGdJRU52Ym5SbGJuUWdabTl1ZENCMllYSnBZV0pzWlhNZ1lYSmxJSFZ6WldRZ1ptOXlJSFI1Y0c5bmNtRndhSGtnYjJZZ2RYTmxjaUJuWlc1bGNtRjBaV1FnWTI5dWRHVnVkQzVjYmlBZ0tpOWNibHh1SUNBdExXcHdMV052Ym5SbGJuUXRabTl1ZEMxemFYcGxPaUF4TTNCNE8xeHVJQ0F0TFdwd0xXTnZiblJsYm5RdGJHbHVaUzFvWldsbmFIUTZJREV1TlR0Y2JpQWdMUzFxY0MxamIyNTBaVzUwTFdadmJuUXRZMjlzYjNJd09pQmliR0ZqYXp0Y2JpQWdMUzFxY0MxamIyNTBaVzUwTFdadmJuUXRZMjlzYjNJeE9pQmliR0ZqYXp0Y2JpQWdMUzFxY0MxamIyNTBaVzUwTFdadmJuUXRZMjlzYjNJeU9pQjJZWElvTFMxdFpDMW5jbVY1TFRjd01DazdYRzRnSUMwdGFuQXRZMjl1ZEdWdWRDMW1iMjUwTFdOdmJHOXlNem9nZG1GeUtDMHRiV1F0WjNKbGVTMDFNREFwTzF4dVhHNGdJQzB0YW5BdGRXa3RabTl1ZEMxelkyRnNaUzFtWVdOMGIzSTZJREV1TWp0Y2JpQWdMUzFxY0MxMWFTMW1iMjUwTFhOcGVtVXdPaUJqWVd4aktIWmhjaWd0TFdwd0xYVnBMV1p2Ym5RdGMybDZaVEVwTDNaaGNpZ3RMV3B3TFhWcExXWnZiblF0YzJOaGJHVXRabUZqZEc5eUtTazdYRzRnSUMwdGFuQXRkV2t0Wm05dWRDMXphWHBsTVRvZ01UTndlRHNnTHlvZ1FtRnpaU0JtYjI1MElITnBlbVVnS2k5Y2JpQWdMUzFxY0MxMWFTMW1iMjUwTFhOcGVtVXlPaUJqWVd4aktIWmhjaWd0TFdwd0xYVnBMV1p2Ym5RdGMybDZaVEVwS25aaGNpZ3RMV3B3TFhWcExXWnZiblF0YzJOaGJHVXRabUZqZEc5eUtTazdYRzRnSUMwdGFuQXRkV2t0Wm05dWRDMXphWHBsTXpvZ1kyRnNZeWgyWVhJb0xTMXFjQzExYVMxbWIyNTBMWE5wZW1VeUtTcDJZWElvTFMxcWNDMTFhUzFtYjI1MExYTmpZV3hsTFdaaFkzUnZjaWtwTzF4dVhHNGdJQzB0YW5BdFkyOWtaUzFtYjI1MExYTnBlbVU2SURFemNIZzdYRzRnSUMwdGFuQXRZMjlrWlMxc2FXNWxMV2hsYVdkb2REb2dNUzR6TURjN1hHNGdJQzB0YW5BdFkyOWtaUzF3WVdSa2FXNW5PaUExY0hnN1hHNGdJQzB0YW5BdFkyOWtaUzFtYjI1MExXWmhiV2xzZVRvZ2JXOXViM053WVdObE8xeHVYRzVjYmlBZ0x5b2dUR0Y1YjNWMFhHNWNiaUFnVkdobElHWnZiR3h2ZDJsdVp5QmhjbVVnZEdobElHMWhhVzRnYkdGNWIzVjBJR052Ykc5eWN5QjFjMlVnYVc0Z1NuVndlWFJsY2t4aFlpNGdTVzRnWVNCc2FXZG9kRnh1SUNCMGFHVnRaU0IwYUdWelpTQjNiM1ZzWkNCbmJ5Qm1jbTl0SUd4cFoyaDBJSFJ2SUdSaGNtc3VYRzRnSUNvdlhHNWNiaUFnTFMxcWNDMXNZWGx2ZFhRdFkyOXNiM0l3T2lCM2FHbDBaVHRjYmlBZ0xTMXFjQzFzWVhsdmRYUXRZMjlzYjNJeE9pQjNhR2wwWlR0Y2JpQWdMUzFxY0Mxc1lYbHZkWFF0WTI5c2IzSXlPaUIyWVhJb0xTMXRaQzFuY21WNUxUSXdNQ2s3WEc0Z0lDMHRhbkF0YkdGNWIzVjBMV052Ykc5eU16b2dkbUZ5S0MwdGJXUXRaM0psZVMwME1EQXBPMXh1WEc0Z0lDOHFJRUp5WVc1a0wyRmpZMlZ1ZENBcUwxeHVYRzRnSUMwdGFuQXRZbkpoYm1RdFkyOXNiM0l3T2lCMllYSW9MUzF0WkMxaWJIVmxMVGN3TUNrN1hHNGdJQzB0YW5BdFluSmhibVF0WTI5c2IzSXhPaUIyWVhJb0xTMXRaQzFpYkhWbExUVXdNQ2s3WEc0Z0lDMHRhbkF0WW5KaGJtUXRZMjlzYjNJeU9pQjJZWElvTFMxdFpDMWliSFZsTFRNd01DazdYRzRnSUMwdGFuQXRZbkpoYm1RdFkyOXNiM0l6T2lCMllYSW9MUzF0WkMxaWJIVmxMVEV3TUNrN1hHNWNiaUFnTFMxcWNDMWhZMk5sYm5RdFkyOXNiM0l3T2lCMllYSW9MUzF0WkMxbmNtVmxiaTAzTURBcE8xeHVJQ0F0TFdwd0xXRmpZMlZ1ZEMxamIyeHZjakU2SUhaaGNpZ3RMVzFrTFdkeVpXVnVMVFV3TUNrN1hHNGdJQzB0YW5BdFlXTmpaVzUwTFdOdmJHOXlNam9nZG1GeUtDMHRiV1F0WjNKbFpXNHRNekF3S1R0Y2JpQWdMUzFxY0MxaFkyTmxiblF0WTI5c2IzSXpPaUIyWVhJb0xTMXRaQzFuY21WbGJpMHhNREFwTzF4dVhHNGdJQzhxSUZOMFlYUmxJR052Ykc5eWN5QW9kMkZ5Yml3Z1pYSnliM0lzSUhOMVkyTmxjM01zSUdsdVptOHBJQ292WEc1Y2JpQWdMUzFxY0MxM1lYSnVMV052Ykc5eU1Eb2dkbUZ5S0MwdGJXUXRiM0poYm1kbExUY3dNQ2s3WEc0Z0lDMHRhbkF0ZDJGeWJpMWpiMnh2Y2pFNklIWmhjaWd0TFcxa0xXOXlZVzVuWlMwMU1EQXBPMXh1SUNBdExXcHdMWGRoY200dFkyOXNiM0l5T2lCMllYSW9MUzF0WkMxdmNtRnVaMlV0TXpBd0tUdGNiaUFnTFMxcWNDMTNZWEp1TFdOdmJHOXlNem9nZG1GeUtDMHRiV1F0YjNKaGJtZGxMVEV3TUNrN1hHNWNiaUFnTFMxcWNDMWxjbkp2Y2kxamIyeHZjakE2SUhaaGNpZ3RMVzFrTFhKbFpDMDNNREFwTzF4dUlDQXRMV3B3TFdWeWNtOXlMV052Ykc5eU1Ub2dkbUZ5S0MwdGJXUXRjbVZrTFRVd01DazdYRzRnSUMwdGFuQXRaWEp5YjNJdFkyOXNiM0l5T2lCMllYSW9MUzF0WkMxeVpXUXRNekF3S1R0Y2JpQWdMUzFxY0MxbGNuSnZjaTFqYjJ4dmNqTTZJSFpoY2lndExXMWtMWEpsWkMweE1EQXBPMXh1WEc0Z0lDMHRhbkF0YzNWalkyVnpjeTFqYjJ4dmNqQTZJSFpoY2lndExXMWtMV2R5WldWdUxUY3dNQ2s3WEc0Z0lDMHRhbkF0YzNWalkyVnpjeTFqYjJ4dmNqRTZJSFpoY2lndExXMWtMV2R5WldWdUxUVXdNQ2s3WEc0Z0lDMHRhbkF0YzNWalkyVnpjeTFqYjJ4dmNqSTZJSFpoY2lndExXMWtMV2R5WldWdUxUTXdNQ2s3WEc0Z0lDMHRhbkF0YzNWalkyVnpjeTFqYjJ4dmNqTTZJSFpoY2lndExXMWtMV2R5WldWdUxURXdNQ2s3WEc1Y2JpQWdMUzFxY0MxcGJtWnZMV052Ykc5eU1Eb2dkbUZ5S0MwdGJXUXRZM2xoYmkwM01EQXBPMXh1SUNBdExXcHdMV2x1Wm04dFkyOXNiM0l4T2lCMllYSW9MUzF0WkMxamVXRnVMVFV3TUNrN1hHNGdJQzB0YW5BdGFXNW1ieTFqYjJ4dmNqSTZJSFpoY2lndExXMWtMV041WVc0dE16QXdLVHRjYmlBZ0xTMXFjQzFwYm1adkxXTnZiRzl5TXpvZ2RtRnlLQzB0YldRdFkzbGhiaTB4TURBcE8xeHVYRzRnSUM4cUlFTmxiR3dnYzNCbFkybG1hV01nYzNSNWJHVnpJQ292WEc1Y2JpQWdMUzFxY0MxalpXeHNMWEJoWkdScGJtYzZJRFZ3ZUR0Y2JpQWdMUzFxY0MxalpXeHNMV1ZrYVhSdmNpMWlZV05yWjNKdmRXNWtPaUFqWmpkbU4yWTNPMXh1SUNBdExXcHdMV05sYkd3dFpXUnBkRzl5TFdKdmNtUmxjaTFqYjJ4dmNqb2dJMk5tWTJaalpqdGNiaUFnTFMxcWNDMWpaV3hzTFdWa2FYUnZjaTFpWVdOclozSnZkVzVrTFdWa2FYUTZJSFpoY2lndExXcHdMWFZwTFd4aGVXOTFkQzFqYjJ4dmNqRXBPMXh1SUNBdExXcHdMV05sYkd3dFpXUnBkRzl5TFdKdmNtUmxjaTFqYjJ4dmNpMWxaR2wwT2lCMllYSW9MUzFxY0MxaWNtRnVaQzFqYjJ4dmNqRXBPMXh1SUNBdExXcHdMV05sYkd3dGNISnZiWEIwTFhkcFpIUm9PaUF4TURCd2VEdGNiaUFnTFMxcWNDMWpaV3hzTFhCeWIyMXdkQzFtYjI1MExXWmhiV2xzZVRvZ0oxSnZZbTkwYnlCTmIyNXZKeXdnYlc5dWIzTndZV05sTzF4dUlDQXRMV3B3TFdObGJHd3RjSEp2YlhCMExXeGxkSFJsY2kxemNHRmphVzVuT2lBd2NIZzdYRzRnSUMwdGFuQXRZMlZzYkMxd2NtOXRjSFF0YjNCaFkybDBlVG9nTVM0d08xeHVJQ0F0TFdwd0xXTmxiR3d0Y0hKdmJYQjBMVzl3WVdOcGRIa3RibTkwTFdGamRHbDJaVG9nTUM0ME8xeHVJQ0F0TFdwd0xXTmxiR3d0Y0hKdmJYQjBMV1p2Ym5RdFkyOXNiM0l0Ym05MExXRmpkR2wyWlRvZ2RtRnlLQzB0YldRdFozSmxlUzAzTURBcE8xeHVJQ0F2S2lCQklHTjFjM1J2YlNCaWJHVnVaQ0J2WmlCTlJDQm5jbVY1SUdGdVpDQmliSFZsSURZd01GeHVJQ0FnS2lCVFpXVWdhSFIwY0hNNkx5OXRaWGxsY25kbFlpNWpiMjB2WlhKcFl5OTBiMjlzY3k5amIyeHZjaTFpYkdWdVpDOGpOVFEyUlRkQk9qRkZPRGhGTlRvMU9taGxlQ0FxTDF4dUlDQXRMV3B3TFdObGJHd3RhVzV3Y205dGNIUXRabTl1ZEMxamIyeHZjam9nSXpNd04wWkRNVHRjYmlBZ0x5b2dRU0JqZFhOMGIyMGdZbXhsYm1RZ2IyWWdUVVFnWjNKbGVTQmhibVFnYjNKaGJtZGxJRFl3TUZ4dUlDQWdLaUJvZEhSd2N6b3ZMMjFsZVdWeWQyVmlMbU52YlM5bGNtbGpMM1J2YjJ4ekwyTnZiRzl5TFdKc1pXNWtMeU0xTkRaRk4wRTZSalExTVRGRk9qVTZhR1Y0SUNvdlhHNGdJQzB0YW5BdFkyVnNiQzF2ZFhSd2NtOXRjSFF0Wm05dWRDMWpiMnh2Y2pvZ0kwSkdOVUl6UkR0Y2JseHVJQ0F2S2lCT2IzUmxZbTl2YXlCemNHVmphV1pwWXlCemRIbHNaWE1nS2k5Y2JseHVJQ0F0TFdwd0xXNXZkR1ZpYjI5ckxYQmhaR1JwYm1jNklERXdjSGc3WEc0Z0lDMHRhbkF0Ym05MFpXSnZiMnN0YzJOeWIyeHNMWEJoWkdScGJtYzZJREV3TUhCNE8xeHVYRzRnSUM4cUlFTnZibk52YkdVZ2MzQmxZMmxtYVdNZ2MzUjViR1Z6SUNvdlhHNWNiaUFnTFMxcWNDMWpiMjV6YjJ4bExXSmhZMnRuY205MWJtUTZJSFpoY2lndExXMWtMV2R5WlhrdE1UQXdLVHRjYmx4dUlDQXZLaUJVYjI5c1ltRnlJSE53WldOcFptbGpJSE4wZVd4bGN5QXFMMXh1WEc0Z0lDMHRhbkF0ZEc5dmJHSmhjaTFpYjNKa1pYSXRZMjlzYjNJNklIWmhjaWd0TFcxa0xXZHlaWGt0TkRBd0tUdGNiaUFnTFMxcWNDMTBiMjlzWW1GeUxXMXBZM0p2TFdobGFXZG9kRG9nT0hCNE8xeHVJQ0F0TFdwd0xYUnZiMnhpWVhJdFltRmphMmR5YjNWdVpEb2dkbUZ5S0MwdGFuQXRiR0Y1YjNWMExXTnZiRzl5TUNrN1hHNGdJQzB0YW5BdGRHOXZiR0poY2kxaWIzZ3RjMmhoWkc5M09pQXdjSGdnTUhCNElESndlQ0F3Y0hnZ2NtZGlZU2d3TERBc01Dd3dMakkwS1R0Y2JpQWdMUzFxY0MxMGIyOXNZbUZ5TFdobFlXUmxjaTF0WVhKbmFXNDZJRFJ3ZUNBMGNIZ2dNSEI0SURSd2VEdGNiaUFnTFMxcWNDMTBiMjlzWW1GeUxXRmpkR2wyWlMxaVlXTnJaM0p2ZFc1a09pQjJZWElvTFMxdFpDMW5jbVY1TFRNd01DazdYRzU5WEc0aUxDSXZLaXBjYmlBcUlGUm9aU0J0WVhSbGNtbGhiQ0JrWlhOcFoyNGdZMjlzYjNKeklHRnlaU0JoWkdGd2RHVmtJR1p5YjIwZ1oyOXZaMnhsTFcxaGRHVnlhV0ZzTFdOdmJHOXlJSFl4TGpJdU5seHVJQ29nYUhSMGNITTZMeTluYVhSb2RXSXVZMjl0TDJSaGJteGxkbUZ1TDJkdmIyZHNaUzF0WVhSbGNtbGhiQzFqYjJ4dmNseHVJQ29nYUhSMGNITTZMeTluYVhSb2RXSXVZMjl0TDJSaGJteGxkbUZ1TDJkdmIyZHNaUzF0WVhSbGNtbGhiQzFqYjJ4dmNpOWliRzlpTDJZMk4yTmhOV1kwTURJNFlqSm1NV0l6TkRnMk1tWTJOR0l3WTJFMk56TXlNMlk1TVdJd09EZ3ZaR2x6ZEM5d1lXeGxkSFJsTG5aaGNpNWpjM05jYmlBcVhHNGdLaUJVYUdVZ2JHbGpaVzV6WlNCbWIzSWdkR2hsSUcxaGRHVnlhV0ZzSUdSbGMybG5iaUJqYjJ4dmNpQkRVMU1nZG1GeWFXRmliR1Z6SUdseklHRnpJR1p2Ykd4dmQzTWdLSE5sWlZ4dUlDb2dhSFIwY0hNNkx5OW5hWFJvZFdJdVkyOXRMMlJoYm14bGRtRnVMMmR2YjJkc1pTMXRZWFJsY21saGJDMWpiMnh2Y2k5aWJHOWlMMlkyTjJOaE5XWTBNREk0WWpKbU1XSXpORGcyTW1ZMk5HSXdZMkUyTnpNeU0yWTVNV0l3T0RndlRFbERSVTVUUlNsY2JpQXFYRzRnS2lCVWFHVWdUVWxVSUV4cFkyVnVjMlVnS0UxSlZDbGNiaUFxWEc0Z0tpQkRiM0I1Y21sbmFIUWdLR01wSURJd01UUWdSR0Z1SUV4bElGWmhibHh1SUNwY2JpQXFJRkJsY20xcGMzTnBiMjRnYVhNZ2FHVnlaV0o1SUdkeVlXNTBaV1FzSUdaeVpXVWdiMllnWTJoaGNtZGxMQ0IwYnlCaGJua2djR1Z5YzI5dUlHOWlkR0ZwYm1sdVp5QmhJR052Y0hsY2JpQXFJRzltSUhSb2FYTWdjMjltZEhkaGNtVWdZVzVrSUdGemMyOWphV0YwWldRZ1pHOWpkVzFsYm5SaGRHbHZiaUJtYVd4bGN5QW9kR2hsSUZ3aVUyOW1kSGRoY21WY0lpa3NJSFJ2SUdSbFlXeGNiaUFxSUdsdUlIUm9aU0JUYjJaMGQyRnlaU0IzYVhSb2IzVjBJSEpsYzNSeWFXTjBhVzl1TENCcGJtTnNkV1JwYm1jZ2QybDBhRzkxZENCc2FXMXBkR0YwYVc5dUlIUm9aU0J5YVdkb2RITmNiaUFxSUhSdklIVnpaU3dnWTI5d2VTd2diVzlrYVdaNUxDQnRaWEpuWlN3Z2NIVmliR2x6YUN3Z1pHbHpkSEpwWW5WMFpTd2djM1ZpYkdsalpXNXpaU3dnWVc1a0wyOXlJSE5sYkd4Y2JpQXFJR052Y0dsbGN5QnZaaUIwYUdVZ1UyOW1kSGRoY21Vc0lHRnVaQ0IwYnlCd1pYSnRhWFFnY0dWeWMyOXVjeUIwYnlCM2FHOXRJSFJvWlNCVGIyWjBkMkZ5WlNCcGMxeHVJQ29nWm5WeWJtbHphR1ZrSUhSdklHUnZJSE52TENCemRXSnFaV04wSUhSdklIUm9aU0JtYjJ4c2IzZHBibWNnWTI5dVpHbDBhVzl1Y3pwY2JpQXFYRzRnS2lCVWFHVWdZV0p2ZG1VZ1kyOXdlWEpwWjJoMElHNXZkR2xqWlNCaGJtUWdkR2hwY3lCd1pYSnRhWE56YVc5dUlHNXZkR2xqWlNCemFHRnNiQ0JpWlNCcGJtTnNkV1JsWkNCcGJseHVJQ29nWVd4c0lHTnZjR2xsY3lCdmNpQnpkV0p6ZEdGdWRHbGhiQ0J3YjNKMGFXOXVjeUJ2WmlCMGFHVWdVMjltZEhkaGNtVXVYRzRnS2x4dUlDb2dWRWhGSUZOUFJsUlhRVkpGSUVsVElGQlNUMVpKUkVWRUlGd2lRVk1nU1ZOY0lpd2dWMGxVU0U5VlZDQlhRVkpTUVU1VVdTQlBSaUJCVGxrZ1MwbE9SQ3dnUlZoUVVrVlRVeUJQVWx4dUlDb2dTVTFRVEVsRlJDd2dTVTVEVEZWRVNVNUhJRUpWVkNCT1QxUWdURWxOU1ZSRlJDQlVUeUJVU0VVZ1YwRlNVa0ZPVkVsRlV5QlBSaUJOUlZKRFNFRk9WRUZDU1V4SlZGa3NYRzRnS2lCR1NWUk9SVk5USUVaUFVpQkJJRkJCVWxSSlExVk1RVklnVUZWU1VFOVRSU0JCVGtRZ1RrOU9TVTVHVWtsT1IwVk5SVTVVTGlCSlRpQk9UeUJGVmtWT1ZDQlRTRUZNVENCVVNFVmNiaUFxSUVGVlZFaFBVbE1nVDFJZ1EwOVFXVkpKUjBoVUlFaFBURVJGVWxNZ1FrVWdURWxCUWt4RklFWlBVaUJCVGxrZ1EweEJTVTBzSUVSQlRVRkhSVk1nVDFJZ1QxUklSVkpjYmlBcUlFeEpRVUpKVEVsVVdTd2dWMGhGVkVoRlVpQkpUaUJCVGlCQlExUkpUMDRnVDBZZ1EwOU9WRkpCUTFRc0lGUlBVbFFnVDFJZ1QxUklSVkpYU1ZORkxDQkJVa2xUU1U1SElFWlNUMDBzWEc0Z0tpQlBWVlFnVDBZZ1QxSWdTVTRnUTA5T1RrVkRWRWxQVGlCWFNWUklJRlJJUlNCVFQwWlVWMEZTUlNCUFVpQlVTRVVnVlZORklFOVNJRTlVU0VWU0lFUkZRVXhKVGtkVElFbE9JRlJJUlZ4dUlDb2dVMDlHVkZkQlVrVXVYRzRnS2k5Y2JqcHliMjkwSUh0Y2JpQWdMUzF0WkMxeVpXUXROVEE2SUNOR1JrVkNSVVU3WEc0Z0lDMHRiV1F0Y21Wa0xURXdNRG9nSTBaR1EwUkVNanRjYmlBZ0xTMXRaQzF5WldRdE1qQXdPaUFqUlVZNVFUbEJPMXh1SUNBdExXMWtMWEpsWkMwek1EQTZJQ05GTlRjek56TTdYRzRnSUMwdGJXUXRjbVZrTFRRd01Eb2dJMFZHTlRNMU1EdGNiaUFnTFMxdFpDMXlaV1F0TlRBd09pQWpSalEwTXpNMk8xeHVJQ0F0TFcxa0xYSmxaQzAyTURBNklDTkZOVE01TXpVN1hHNGdJQzB0YldRdGNtVmtMVGN3TURvZ0kwUXpNa1l5Ump0Y2JpQWdMUzF0WkMxeVpXUXRPREF3T2lBalF6WXlPREk0TzF4dUlDQXRMVzFrTFhKbFpDMDVNREE2SUNOQ056RkRNVU03WEc0Z0lDMHRiV1F0Y21Wa0xVRXhNREE2SUNOR1JqaEJPREE3WEc0Z0lDMHRiV1F0Y21Wa0xVRXlNREE2SUNOR1JqVXlOVEk3WEc0Z0lDMHRiV1F0Y21Wa0xVRTBNREE2SUNOR1JqRTNORFE3WEc0Z0lDMHRiV1F0Y21Wa0xVRTNNREE2SUNORU5UQXdNREE3WEc1Y2JpQWdMUzF0WkMxd2FXNXJMVFV3T2lBalJrTkZORVZETzF4dUlDQXRMVzFrTFhCcGJtc3RNVEF3T2lBalJqaENRa1F3TzF4dUlDQXRMVzFrTFhCcGJtc3RNakF3T2lBalJqUTRSa0l4TzF4dUlDQXRMVzFrTFhCcGJtc3RNekF3T2lBalJqQTJNamt5TzF4dUlDQXRMVzFrTFhCcGJtc3ROREF3T2lBalJVTTBNRGRCTzF4dUlDQXRMVzFrTFhCcGJtc3ROVEF3T2lBalJUa3hSVFl6TzF4dUlDQXRMVzFrTFhCcGJtc3ROakF3T2lBalJEZ3hRall3TzF4dUlDQXRMVzFrTFhCcGJtc3ROekF3T2lBalF6SXhPRFZDTzF4dUlDQXRMVzFrTFhCcGJtc3RPREF3T2lBalFVUXhORFUzTzF4dUlDQXRMVzFrTFhCcGJtc3RPVEF3T2lBak9EZ3dSVFJHTzF4dUlDQXRMVzFrTFhCcGJtc3RRVEV3TURvZ0kwWkdPREJCUWp0Y2JpQWdMUzF0WkMxd2FXNXJMVUV5TURBNklDTkdSalF3T0RFN1hHNGdJQzB0YldRdGNHbHVheTFCTkRBd09pQWpSalV3TURVM08xeHVJQ0F0TFcxa0xYQnBibXN0UVRjd01Eb2dJME0xTVRFMk1qdGNibHh1SUNBdExXMWtMWEIxY25Cc1pTMDFNRG9nSTBZelJUVkdOVHRjYmlBZ0xTMXRaQzF3ZFhKd2JHVXRNVEF3T2lBalJURkNSVVUzTzF4dUlDQXRMVzFrTFhCMWNuQnNaUzB5TURBNklDTkRSVGt6UkRnN1hHNGdJQzB0YldRdGNIVnljR3hsTFRNd01Eb2dJMEpCTmpoRE9EdGNiaUFnTFMxdFpDMXdkWEp3YkdVdE5EQXdPaUFqUVVJME4wSkRPMXh1SUNBdExXMWtMWEIxY25Cc1pTMDFNREE2SUNNNVF6STNRakE3WEc0Z0lDMHRiV1F0Y0hWeWNHeGxMVFl3TURvZ0l6aEZNalJCUVR0Y2JpQWdMUzF0WkMxd2RYSndiR1V0TnpBd09pQWpOMEl4UmtFeU8xeHVJQ0F0TFcxa0xYQjFjbkJzWlMwNE1EQTZJQ00yUVRGQ09VRTdYRzRnSUMwdGJXUXRjSFZ5Y0d4bExUa3dNRG9nSXpSQk1UUTRRenRjYmlBZ0xTMXRaQzF3ZFhKd2JHVXRRVEV3TURvZ0kwVkJPREJHUXp0Y2JpQWdMUzF0WkMxd2RYSndiR1V0UVRJd01Eb2dJMFV3TkRCR1FqdGNiaUFnTFMxdFpDMXdkWEp3YkdVdFFUUXdNRG9nSTBRMU1EQkdPVHRjYmlBZ0xTMXRaQzF3ZFhKd2JHVXRRVGN3TURvZ0kwRkJNREJHUmp0Y2JseHVJQ0F0TFcxa0xXUmxaWEF0Y0hWeWNHeGxMVFV3T2lBalJVUkZOMFkyTzF4dUlDQXRMVzFrTFdSbFpYQXRjSFZ5Y0d4bExURXdNRG9nSTBReFF6UkZPVHRjYmlBZ0xTMXRaQzFrWldWd0xYQjFjbkJzWlMweU1EQTZJQ05DTXpsRVJFSTdYRzRnSUMwdGJXUXRaR1ZsY0Mxd2RYSndiR1V0TXpBd09pQWpPVFUzTlVORU8xeHVJQ0F0TFcxa0xXUmxaWEF0Y0hWeWNHeGxMVFF3TURvZ0l6ZEZOVGRETWp0Y2JpQWdMUzF0WkMxa1pXVndMWEIxY25Cc1pTMDFNREE2SUNNMk56TkJRamM3WEc0Z0lDMHRiV1F0WkdWbGNDMXdkWEp3YkdVdE5qQXdPaUFqTlVVek5VSXhPMXh1SUNBdExXMWtMV1JsWlhBdGNIVnljR3hsTFRjd01Eb2dJelV4TWtSQk9EdGNiaUFnTFMxdFpDMWtaV1Z3TFhCMWNuQnNaUzA0TURBNklDTTBOVEkzUVRBN1hHNGdJQzB0YldRdFpHVmxjQzF3ZFhKd2JHVXRPVEF3T2lBak16RXhRamt5TzF4dUlDQXRMVzFrTFdSbFpYQXRjSFZ5Y0d4bExVRXhNREE2SUNOQ016ZzRSa1k3WEc0Z0lDMHRiV1F0WkdWbGNDMXdkWEp3YkdVdFFUSXdNRG9nSXpkRE5FUkdSanRjYmlBZ0xTMXRaQzFrWldWd0xYQjFjbkJzWlMxQk5EQXdPaUFqTmpVeFJrWkdPMXh1SUNBdExXMWtMV1JsWlhBdGNIVnljR3hsTFVFM01EQTZJQ00yTWpBd1JVRTdYRzVjYmlBZ0xTMXRaQzFwYm1ScFoyOHROVEE2SUNORk9FVkJSalk3WEc0Z0lDMHRiV1F0YVc1a2FXZHZMVEV3TURvZ0kwTTFRMEZGT1R0Y2JpQWdMUzF0WkMxcGJtUnBaMjh0TWpBd09pQWpPVVpCT0VSQk8xeHVJQ0F0TFcxa0xXbHVaR2xuYnkwek1EQTZJQ00zT1RnMlEwSTdYRzRnSUMwdGJXUXRhVzVrYVdkdkxUUXdNRG9nSXpWRE5rSkRNRHRjYmlBZ0xTMXRaQzFwYm1ScFoyOHROVEF3T2lBak0wWTFNVUkxTzF4dUlDQXRMVzFrTFdsdVpHbG5ieTAyTURBNklDTXpPVFE1UVVJN1hHNGdJQzB0YldRdGFXNWthV2R2TFRjd01Eb2dJek13TTBZNVJqdGNiaUFnTFMxdFpDMXBibVJwWjI4dE9EQXdPaUFqTWpnek5Ua3pPMXh1SUNBdExXMWtMV2x1WkdsbmJ5MDVNREE2SUNNeFFUSXpOMFU3WEc0Z0lDMHRiV1F0YVc1a2FXZHZMVUV4TURBNklDTTRRemxGUmtZN1hHNGdJQzB0YldRdGFXNWthV2R2TFVFeU1EQTZJQ00xTXpaRVJrVTdYRzRnSUMwdGJXUXRhVzVrYVdkdkxVRTBNREE2SUNNelJEVkJSa1U3WEc0Z0lDMHRiV1F0YVc1a2FXZHZMVUUzTURBNklDTXpNRFJHUmtVN1hHNWNiaUFnTFMxdFpDMWliSFZsTFRVd09pQWpSVE5HTWtaRU8xeHVJQ0F0TFcxa0xXSnNkV1V0TVRBd09pQWpRa0pFUlVaQ08xeHVJQ0F0TFcxa0xXSnNkV1V0TWpBd09pQWpPVEJEUVVZNU8xeHVJQ0F0TFcxa0xXSnNkV1V0TXpBd09pQWpOalJDTlVZMk8xeHVJQ0F0TFcxa0xXSnNkV1V0TkRBd09pQWpOREpCTlVZMU8xeHVJQ0F0TFcxa0xXSnNkV1V0TlRBd09pQWpNakU1TmtZek8xeHVJQ0F0TFcxa0xXSnNkV1V0TmpBd09pQWpNVVU0T0VVMU8xeHVJQ0F0TFcxa0xXSnNkV1V0TnpBd09pQWpNVGszTmtReU8xeHVJQ0F0TFcxa0xXSnNkV1V0T0RBd09pQWpNVFUyTlVNd08xeHVJQ0F0TFcxa0xXSnNkV1V0T1RBd09pQWpNRVEwTjBFeE8xeHVJQ0F0TFcxa0xXSnNkV1V0UVRFd01Eb2dJemd5UWpGR1JqdGNiaUFnTFMxdFpDMWliSFZsTFVFeU1EQTZJQ00wTkRoQlJrWTdYRzRnSUMwdGJXUXRZbXgxWlMxQk5EQXdPaUFqTWprM09VWkdPMXh1SUNBdExXMWtMV0pzZFdVdFFUY3dNRG9nSXpJNU5qSkdSanRjYmx4dUlDQXRMVzFrTFd4cFoyaDBMV0pzZFdVdE5UQTZJQ05GTVVZMVJrVTdYRzRnSUMwdGJXUXRiR2xuYUhRdFlteDFaUzB4TURBNklDTkNNMFUxUmtNN1hHNGdJQzB0YldRdGJHbG5hSFF0WW14MVpTMHlNREE2SUNNNE1VUTBSa0U3WEc0Z0lDMHRiV1F0YkdsbmFIUXRZbXgxWlMwek1EQTZJQ00wUmtNelJqYzdYRzRnSUMwdGJXUXRiR2xuYUhRdFlteDFaUzAwTURBNklDTXlPVUkyUmpZN1hHNGdJQzB0YldRdGJHbG5hSFF0WW14MVpTMDFNREE2SUNNd00wRTVSalE3WEc0Z0lDMHRiV1F0YkdsbmFIUXRZbXgxWlMwMk1EQTZJQ013TXpsQ1JUVTdYRzRnSUMwdGJXUXRiR2xuYUhRdFlteDFaUzAzTURBNklDTXdNamc0UkRFN1hHNGdJQzB0YldRdGJHbG5hSFF0WW14MVpTMDRNREE2SUNNd01qYzNRa1E3WEc0Z0lDMHRiV1F0YkdsbmFIUXRZbXgxWlMwNU1EQTZJQ013TVRVM09VSTdYRzRnSUMwdGJXUXRiR2xuYUhRdFlteDFaUzFCTVRBd09pQWpPREJFT0VaR08xeHVJQ0F0TFcxa0xXeHBaMmgwTFdKc2RXVXRRVEl3TURvZ0l6UXdRelJHUmp0Y2JpQWdMUzF0WkMxc2FXZG9kQzFpYkhWbExVRTBNREE2SUNNd01FSXdSa1k3WEc0Z0lDMHRiV1F0YkdsbmFIUXRZbXgxWlMxQk56QXdPaUFqTURBNU1VVkJPMXh1WEc0Z0lDMHRiV1F0WTNsaGJpMDFNRG9nSTBVd1JqZEdRVHRjYmlBZ0xTMXRaQzFqZVdGdUxURXdNRG9nSTBJeVJVSkdNanRjYmlBZ0xTMXRaQzFqZVdGdUxUSXdNRG9nSXpnd1JFVkZRVHRjYmlBZ0xTMXRaQzFqZVdGdUxUTXdNRG9nSXpSRVJEQkZNVHRjYmlBZ0xTMXRaQzFqZVdGdUxUUXdNRG9nSXpJMlF6WkVRVHRjYmlBZ0xTMXRaQzFqZVdGdUxUVXdNRG9nSXpBd1FrTkVORHRjYmlBZ0xTMXRaQzFqZVdGdUxUWXdNRG9nSXpBd1FVTkRNVHRjYmlBZ0xTMXRaQzFqZVdGdUxUY3dNRG9nSXpBd09UZEJOenRjYmlBZ0xTMXRaQzFqZVdGdUxUZ3dNRG9nSXpBd09ETTRSanRjYmlBZ0xTMXRaQzFqZVdGdUxUa3dNRG9nSXpBd05qQTJORHRjYmlBZ0xTMXRaQzFqZVdGdUxVRXhNREE2SUNNNE5FWkdSa1k3WEc0Z0lDMHRiV1F0WTNsaGJpMUJNakF3T2lBak1UaEdSa1pHTzF4dUlDQXRMVzFrTFdONVlXNHRRVFF3TURvZ0l6QXdSVFZHUmp0Y2JpQWdMUzF0WkMxamVXRnVMVUUzTURBNklDTXdNRUk0UkRRN1hHNWNiaUFnTFMxdFpDMTBaV0ZzTFRVd09pQWpSVEJHTWtZeE8xeHVJQ0F0TFcxa0xYUmxZV3d0TVRBd09pQWpRakpFUmtSQ08xeHVJQ0F0TFcxa0xYUmxZV3d0TWpBd09pQWpPREJEUWtNME8xeHVJQ0F0TFcxa0xYUmxZV3d0TXpBd09pQWpORVJDTmtGRE8xeHVJQ0F0TFcxa0xYUmxZV3d0TkRBd09pQWpNalpCTmpsQk8xeHVJQ0F0TFcxa0xYUmxZV3d0TlRBd09pQWpNREE1TmpnNE8xeHVJQ0F0TFcxa0xYUmxZV3d0TmpBd09pQWpNREE0T1RkQ08xeHVJQ0F0TFcxa0xYUmxZV3d0TnpBd09pQWpNREEzT1RaQ08xeHVJQ0F0TFcxa0xYUmxZV3d0T0RBd09pQWpNREEyT1RWRE8xeHVJQ0F0TFcxa0xYUmxZV3d0T1RBd09pQWpNREEwUkRRd08xeHVJQ0F0TFcxa0xYUmxZV3d0UVRFd01Eb2dJMEUzUmtaRlFqdGNiaUFnTFMxdFpDMTBaV0ZzTFVFeU1EQTZJQ00yTkVaR1JFRTdYRzRnSUMwdGJXUXRkR1ZoYkMxQk5EQXdPaUFqTVVSRk9VSTJPMXh1SUNBdExXMWtMWFJsWVd3dFFUY3dNRG9nSXpBd1FrWkJOVHRjYmx4dUlDQXRMVzFrTFdkeVpXVnVMVFV3T2lBalJUaEdOVVU1TzF4dUlDQXRMVzFrTFdkeVpXVnVMVEV3TURvZ0kwTTRSVFpET1R0Y2JpQWdMUzF0WkMxbmNtVmxiaTB5TURBNklDTkJOVVEyUVRjN1hHNGdJQzB0YldRdFozSmxaVzR0TXpBd09pQWpPREZETnpnME8xeHVJQ0F0TFcxa0xXZHlaV1Z1TFRRd01Eb2dJelkyUWtJMlFUdGNiaUFnTFMxdFpDMW5jbVZsYmkwMU1EQTZJQ00wUTBGR05UQTdYRzRnSUMwdGJXUXRaM0psWlc0dE5qQXdPaUFqTkROQk1EUTNPMXh1SUNBdExXMWtMV2R5WldWdUxUY3dNRG9nSXpNNE9FVXpRenRjYmlBZ0xTMXRaQzFuY21WbGJpMDRNREE2SUNNeVJUZEVNekk3WEc0Z0lDMHRiV1F0WjNKbFpXNHRPVEF3T2lBak1VSTFSVEl3TzF4dUlDQXRMVzFrTFdkeVpXVnVMVUV4TURBNklDTkNPVVkyUTBFN1hHNGdJQzB0YldRdFozSmxaVzR0UVRJd01Eb2dJelk1UmpCQlJUdGNiaUFnTFMxdFpDMW5jbVZsYmkxQk5EQXdPaUFqTURCRk5qYzJPMXh1SUNBdExXMWtMV2R5WldWdUxVRTNNREE2SUNNd01FTTROVE03WEc1Y2JpQWdMUzF0WkMxc2FXZG9kQzFuY21WbGJpMDFNRG9nSTBZeFJqaEZPVHRjYmlBZ0xTMXRaQzFzYVdkb2RDMW5jbVZsYmkweE1EQTZJQ05FUTBWRVF6ZzdYRzRnSUMwdGJXUXRiR2xuYUhRdFozSmxaVzR0TWpBd09pQWpRelZGTVVFMU8xeHVJQ0F0TFcxa0xXeHBaMmgwTFdkeVpXVnVMVE13TURvZ0kwRkZSRFU0TVR0Y2JpQWdMUzF0WkMxc2FXZG9kQzFuY21WbGJpMDBNREE2SUNNNVEwTkROalU3WEc0Z0lDMHRiV1F0YkdsbmFIUXRaM0psWlc0dE5UQXdPaUFqT0VKRE16UkJPMXh1SUNBdExXMWtMV3hwWjJoMExXZHlaV1Z1TFRZd01Eb2dJemREUWpNME1qdGNiaUFnTFMxdFpDMXNhV2RvZEMxbmNtVmxiaTAzTURBNklDTTJPRGxHTXpnN1hHNGdJQzB0YldRdGJHbG5hSFF0WjNKbFpXNHRPREF3T2lBak5UVTRRakpHTzF4dUlDQXRMVzFrTFd4cFoyaDBMV2R5WldWdUxUa3dNRG9nSXpNek5qa3hSVHRjYmlBZ0xTMXRaQzFzYVdkb2RDMW5jbVZsYmkxQk1UQXdPaUFqUTBOR1Jqa3dPMXh1SUNBdExXMWtMV3hwWjJoMExXZHlaV1Z1TFVFeU1EQTZJQ05DTWtaR05UazdYRzRnSUMwdGJXUXRiR2xuYUhRdFozSmxaVzR0UVRRd01Eb2dJemMyUmtZd016dGNiaUFnTFMxdFpDMXNhV2RvZEMxbmNtVmxiaTFCTnpBd09pQWpOalJFUkRFM08xeHVYRzRnSUMwdGJXUXRiR2x0WlMwMU1Eb2dJMFk1UmtKRk56dGNiaUFnTFMxdFpDMXNhVzFsTFRFd01Eb2dJMFl3UmpSRE16dGNiaUFnTFMxdFpDMXNhVzFsTFRJd01Eb2dJMFUyUlVVNVF6dGNiaUFnTFMxdFpDMXNhVzFsTFRNd01Eb2dJMFJEUlRjM05UdGNiaUFnTFMxdFpDMXNhVzFsTFRRd01Eb2dJMFEwUlRFMU56dGNiaUFnTFMxdFpDMXNhVzFsTFRVd01Eb2dJME5FUkVNek9UdGNiaUFnTFMxdFpDMXNhVzFsTFRZd01Eb2dJME13UTBFek16dGNiaUFnTFMxdFpDMXNhVzFsTFRjd01Eb2dJMEZHUWpReVFqdGNiaUFnTFMxdFpDMXNhVzFsTFRnd01Eb2dJemxGT1VReU5EdGNiaUFnTFMxdFpDMXNhVzFsTFRrd01Eb2dJemd5TnpjeE56dGNiaUFnTFMxdFpDMXNhVzFsTFVFeE1EQTZJQ05HTkVaR09ERTdYRzRnSUMwdGJXUXRiR2x0WlMxQk1qQXdPaUFqUlVWR1JqUXhPMXh1SUNBdExXMWtMV3hwYldVdFFUUXdNRG9nSTBNMlJrWXdNRHRjYmlBZ0xTMXRaQzFzYVcxbExVRTNNREE2SUNOQlJVVkJNREE3WEc1Y2JpQWdMUzF0WkMxNVpXeHNiM2N0TlRBNklDTkdSa1pFUlRjN1hHNGdJQzB0YldRdGVXVnNiRzkzTFRFd01Eb2dJMFpHUmpsRE5EdGNiaUFnTFMxdFpDMTVaV3hzYjNjdE1qQXdPaUFqUmtaR05UbEVPMXh1SUNBdExXMWtMWGxsYkd4dmR5MHpNREE2SUNOR1JrWXhOelk3WEc0Z0lDMHRiV1F0ZVdWc2JHOTNMVFF3TURvZ0kwWkdSVVUxT0R0Y2JpQWdMUzF0WkMxNVpXeHNiM2N0TlRBd09pQWpSa1pGUWpOQ08xeHVJQ0F0TFcxa0xYbGxiR3h2ZHkwMk1EQTZJQ05HUkVRNE16VTdYRzRnSUMwdGJXUXRlV1ZzYkc5M0xUY3dNRG9nSTBaQ1F6QXlSRHRjYmlBZ0xTMXRaQzE1Wld4c2IzY3RPREF3T2lBalJqbEJPREkxTzF4dUlDQXRMVzFrTFhsbGJHeHZkeTA1TURBNklDTkdOVGRHTVRjN1hHNGdJQzB0YldRdGVXVnNiRzkzTFVFeE1EQTZJQ05HUmtaR09FUTdYRzRnSUMwdGJXUXRlV1ZzYkc5M0xVRXlNREE2SUNOR1JrWkdNREE3WEc0Z0lDMHRiV1F0ZVdWc2JHOTNMVUUwTURBNklDTkdSa1ZCTURBN1hHNGdJQzB0YldRdGVXVnNiRzkzTFVFM01EQTZJQ05HUmtRMk1EQTdYRzVjYmlBZ0xTMXRaQzFoYldKbGNpMDFNRG9nSTBaR1JqaEZNVHRjYmlBZ0xTMXRaQzFoYldKbGNpMHhNREE2SUNOR1JrVkRRak03WEc0Z0lDMHRiV1F0WVcxaVpYSXRNakF3T2lBalJrWkZNRGd5TzF4dUlDQXRMVzFrTFdGdFltVnlMVE13TURvZ0kwWkdSRFUwUmp0Y2JpQWdMUzF0WkMxaGJXSmxjaTAwTURBNklDTkdSa05CTWpnN1hHNGdJQzB0YldRdFlXMWlaWEl0TlRBd09pQWpSa1pETVRBM08xeHVJQ0F0TFcxa0xXRnRZbVZ5TFRZd01Eb2dJMFpHUWpNd01EdGNiaUFnTFMxdFpDMWhiV0psY2kwM01EQTZJQ05HUmtFd01EQTdYRzRnSUMwdGJXUXRZVzFpWlhJdE9EQXdPaUFqUmtZNFJqQXdPMXh1SUNBdExXMWtMV0Z0WW1WeUxUa3dNRG9nSTBaR05rWXdNRHRjYmlBZ0xTMXRaQzFoYldKbGNpMUJNVEF3T2lBalJrWkZOVGRHTzF4dUlDQXRMVzFrTFdGdFltVnlMVUV5TURBNklDTkdSa1EzTkRBN1hHNGdJQzB0YldRdFlXMWlaWEl0UVRRd01Eb2dJMFpHUXpRd01EdGNiaUFnTFMxdFpDMWhiV0psY2kxQk56QXdPaUFqUmtaQlFqQXdPMXh1WEc0Z0lDMHRiV1F0YjNKaGJtZGxMVFV3T2lBalJrWkdNMFV3TzF4dUlDQXRMVzFrTFc5eVlXNW5aUzB4TURBNklDTkdSa1V3UWpJN1hHNGdJQzB0YldRdGIzSmhibWRsTFRJd01Eb2dJMFpHUTBNNE1EdGNiaUFnTFMxdFpDMXZjbUZ1WjJVdE16QXdPaUFqUmtaQ056UkVPMXh1SUNBdExXMWtMVzl5WVc1blpTMDBNREE2SUNOR1JrRTNNalk3WEc0Z0lDMHRiV1F0YjNKaGJtZGxMVFV3TURvZ0kwWkdPVGd3TUR0Y2JpQWdMUzF0WkMxdmNtRnVaMlV0TmpBd09pQWpSa0k0UXpBd08xeHVJQ0F0TFcxa0xXOXlZVzVuWlMwM01EQTZJQ05HTlRkRE1EQTdYRzRnSUMwdGJXUXRiM0poYm1kbExUZ3dNRG9nSTBWR05rTXdNRHRjYmlBZ0xTMXRaQzF2Y21GdVoyVXRPVEF3T2lBalJUWTFNVEF3TzF4dUlDQXRMVzFrTFc5eVlXNW5aUzFCTVRBd09pQWpSa1pFTVRnd08xeHVJQ0F0TFcxa0xXOXlZVzVuWlMxQk1qQXdPaUFqUmtaQlFqUXdPMXh1SUNBdExXMWtMVzl5WVc1blpTMUJOREF3T2lBalJrWTVNVEF3TzF4dUlDQXRMVzFrTFc5eVlXNW5aUzFCTnpBd09pQWpSa1kyUkRBd08xeHVYRzRnSUMwdGJXUXRaR1ZsY0MxdmNtRnVaMlV0TlRBNklDTkdRa1U1UlRjN1hHNGdJQzB0YldRdFpHVmxjQzF2Y21GdVoyVXRNVEF3T2lBalJrWkRRMEpETzF4dUlDQXRMVzFrTFdSbFpYQXRiM0poYm1kbExUSXdNRG9nSTBaR1FVSTVNVHRjYmlBZ0xTMXRaQzFrWldWd0xXOXlZVzVuWlMwek1EQTZJQ05HUmpoQk5qVTdYRzRnSUMwdGJXUXRaR1ZsY0MxdmNtRnVaMlV0TkRBd09pQWpSa1kzTURRek8xeHVJQ0F0TFcxa0xXUmxaWEF0YjNKaGJtZGxMVFV3TURvZ0kwWkdOVGN5TWp0Y2JpQWdMUzF0WkMxa1pXVndMVzl5WVc1blpTMDJNREE2SUNOR05EVXhNVVU3WEc0Z0lDMHRiV1F0WkdWbGNDMXZjbUZ1WjJVdE56QXdPaUFqUlRZMFFURTVPMXh1SUNBdExXMWtMV1JsWlhBdGIzSmhibWRsTFRnd01Eb2dJMFE0TkRNeE5UdGNiaUFnTFMxdFpDMWtaV1Z3TFc5eVlXNW5aUzA1TURBNklDTkNSak0yTUVNN1hHNGdJQzB0YldRdFpHVmxjQzF2Y21GdVoyVXRRVEV3TURvZ0kwWkdPVVU0TUR0Y2JpQWdMUzF0WkMxa1pXVndMVzl5WVc1blpTMUJNakF3T2lBalJrWTJSVFF3TzF4dUlDQXRMVzFrTFdSbFpYQXRiM0poYm1kbExVRTBNREE2SUNOR1JqTkVNREE3WEc0Z0lDMHRiV1F0WkdWbGNDMXZjbUZ1WjJVdFFUY3dNRG9nSTBSRU1rTXdNRHRjYmx4dUlDQXRMVzFrTFdKeWIzZHVMVFV3T2lBalJVWkZRa1U1TzF4dUlDQXRMVzFrTFdKeWIzZHVMVEV3TURvZ0kwUTNRME5ET0R0Y2JpQWdMUzF0WkMxaWNtOTNiaTB5TURBNklDTkNRMEZCUVRRN1hHNGdJQzB0YldRdFluSnZkMjR0TXpBd09pQWpRVEU0T0RkR08xeHVJQ0F0TFcxa0xXSnliM2R1TFRRd01Eb2dJemhFTmtVMk16dGNiaUFnTFMxdFpDMWljbTkzYmkwMU1EQTZJQ00zT1RVMU5EZzdYRzRnSUMwdGJXUXRZbkp2ZDI0dE5qQXdPaUFqTmtRMFF6UXhPMXh1SUNBdExXMWtMV0p5YjNkdUxUY3dNRG9nSXpWRU5EQXpOenRjYmlBZ0xTMXRaQzFpY205M2JpMDRNREE2SUNNMFJUTTBNa1U3WEc0Z0lDMHRiV1F0WW5KdmQyNHRPVEF3T2lBak0wVXlOekl6TzF4dVhHNGdJQzB0YldRdFozSmxlUzAxTURvZ0kwWkJSa0ZHUVR0Y2JpQWdMUzF0WkMxbmNtVjVMVEV3TURvZ0kwWTFSalZHTlR0Y2JpQWdMUzF0WkMxbmNtVjVMVEl3TURvZ0kwVkZSVVZGUlR0Y2JpQWdMUzF0WkMxbmNtVjVMVE13TURvZ0kwVXdSVEJGTUR0Y2JpQWdMUzF0WkMxbmNtVjVMVFF3TURvZ0kwSkVRa1JDUkR0Y2JpQWdMUzF0WkMxbmNtVjVMVFV3TURvZ0l6bEZPVVU1UlR0Y2JpQWdMUzF0WkMxbmNtVjVMVFl3TURvZ0l6YzFOelUzTlR0Y2JpQWdMUzF0WkMxbmNtVjVMVGN3TURvZ0l6WXhOakUyTVR0Y2JpQWdMUzF0WkMxbmNtVjVMVGd3TURvZ0l6UXlOREkwTWp0Y2JpQWdMUzF0WkMxbmNtVjVMVGt3TURvZ0l6SXhNakV5TVR0Y2JseHVJQ0F0TFcxa0xXSnNkV1V0WjNKbGVTMDFNRG9nSTBWRFJVWkdNVHRjYmlBZ0xTMXRaQzFpYkhWbExXZHlaWGt0TVRBd09pQWpRMFpFT0VSRE8xeHVJQ0F0TFcxa0xXSnNkV1V0WjNKbGVTMHlNREE2SUNOQ01FSkZRelU3WEc0Z0lDMHRiV1F0WW14MVpTMW5jbVY1TFRNd01Eb2dJemt3UVRSQlJUdGNiaUFnTFMxdFpDMWliSFZsTFdkeVpYa3ROREF3T2lBak56ZzVNRGxETzF4dUlDQXRMVzFrTFdKc2RXVXRaM0psZVMwMU1EQTZJQ00yTURkRU9FSTdYRzRnSUMwdGJXUXRZbXgxWlMxbmNtVjVMVFl3TURvZ0l6VTBOa1UzUVR0Y2JpQWdMUzF0WkMxaWJIVmxMV2R5WlhrdE56QXdPaUFqTkRVMVFUWTBPMXh1SUNBdExXMWtMV0pzZFdVdFozSmxlUzA0TURBNklDTXpOelEzTkVZN1hHNGdJQzB0YldRdFlteDFaUzFuY21WNUxUa3dNRG9nSXpJMk16SXpPRHRjYm4waUxDSXZLaUJEYjNCNWNtbG5hSFFnS0dNcElFcDFjSGwwWlhJZ1JHVjJaV3h2Y0cxbGJuUWdWR1ZoYlM1Y2JpQXFJRVJwYzNSeWFXSjFkR1ZrSUhWdVpHVnlJSFJvWlNCMFpYSnRjeUJ2WmlCMGFHVWdUVzlrYVdacFpXUWdRbE5FSUV4cFkyVnVjMlV1WEc0Z0tpOWNibHh1THlwY2JpQXFJRmRsSUdGemMzVnRaU0IwYUdGMElIUm9aU0JEVTFNZ2RtRnlhV0ZpYkdWeklHbHVYRzRnS2lCb2RIUndjem92TDJkcGRHaDFZaTVqYjIwdmFuVndlWFJsY214aFlpOXFkWEI1ZEdWeWJHRmlMMkpzYjJJdmJXRnpkR1Z5TDNOeVl5OWtaV1poZFd4MExYUm9aVzFsTDNaaGNtbGhZbXhsY3k1amMzTmNiaUFxSUdoaGRtVWdZbVZsYmlCa1pXWnBibVZrTGx4dUlDb3ZYRzVjYmtCcGJYQnZjblFnWENJdUwzQm9iM053YUc5eUxtTnpjMXdpTzF4dVhHNDZjbTl2ZENCN1hHNGdJQ0FnTFMxcWNDMTNhV1JuWlhSekxXTnZiRzl5T2lCMllYSW9MUzFxY0MxamIyNTBaVzUwTFdadmJuUXRZMjlzYjNJeEtUdGNiaUFnSUNBdExXcHdMWGRwWkdkbGRITXRiR0ZpWld3dFkyOXNiM0k2SUhaaGNpZ3RMV3B3TFhkcFpHZGxkSE10WTI5c2IzSXBPMXh1SUNBZ0lDMHRhbkF0ZDJsa1oyVjBjeTF5WldGa2IzVjBMV052Ykc5eU9pQjJZWElvTFMxcWNDMTNhV1JuWlhSekxXTnZiRzl5S1R0Y2JpQWdJQ0F0TFdwd0xYZHBaR2RsZEhNdFptOXVkQzF6YVhwbE9pQjJZWElvTFMxcWNDMTFhUzFtYjI1MExYTnBlbVV4S1R0Y2JpQWdJQ0F0TFdwd0xYZHBaR2RsZEhNdGJXRnlaMmx1T2lBeWNIZzdYRzRnSUNBZ0xTMXFjQzEzYVdSblpYUnpMV2x1YkdsdVpTMW9aV2xuYUhRNklESTRjSGc3WEc0Z0lDQWdMUzFxY0MxM2FXUm5aWFJ6TFdsdWJHbHVaUzEzYVdSMGFEb2dNekF3Y0hnN1hHNGdJQ0FnTFMxcWNDMTNhV1JuWlhSekxXbHViR2x1WlMxM2FXUjBhQzF6YUc5eWREb2dZMkZzWXloMllYSW9MUzFxY0MxM2FXUm5aWFJ6TFdsdWJHbHVaUzEzYVdSMGFDa2dMeUF5SUMwZ2RtRnlLQzB0YW5BdGQybGtaMlYwY3kxdFlYSm5hVzRwS1R0Y2JpQWdJQ0F0TFdwd0xYZHBaR2RsZEhNdGFXNXNhVzVsTFhkcFpIUm9MWFJwYm5rNklHTmhiR01vZG1GeUtDMHRhbkF0ZDJsa1oyVjBjeTFwYm14cGJtVXRkMmxrZEdndGMyaHZjblFwSUM4Z01pQXRJSFpoY2lndExXcHdMWGRwWkdkbGRITXRiV0Z5WjJsdUtTazdYRzRnSUNBZ0xTMXFjQzEzYVdSblpYUnpMV2x1YkdsdVpTMXRZWEpuYVc0NklEUndlRHNnTHlvZ2JXRnlaMmx1SUdKbGRIZGxaVzRnYVc1c2FXNWxJR1ZzWlcxbGJuUnpJQ292WEc0Z0lDQWdMUzFxY0MxM2FXUm5aWFJ6TFdsdWJHbHVaUzFzWVdKbGJDMTNhV1IwYURvZ09EQndlRHRjYmlBZ0lDQXRMV3B3TFhkcFpHZGxkSE10WW05eVpHVnlMWGRwWkhSb09pQjJZWElvTFMxcWNDMWliM0prWlhJdGQybGtkR2dwTzF4dUlDQWdJQzB0YW5BdGQybGtaMlYwY3kxMlpYSjBhV05oYkMxb1pXbG5hSFE2SURJd01IQjRPMXh1SUNBZ0lDMHRhbkF0ZDJsa1oyVjBjeTFvYjNKcGVtOXVkR0ZzTFhSaFlpMW9aV2xuYUhRNklESTBjSGc3WEc0Z0lDQWdMUzFxY0MxM2FXUm5aWFJ6TFdodmNtbDZiMjUwWVd3dGRHRmlMWGRwWkhSb09pQXhORFJ3ZUR0Y2JpQWdJQ0F0TFdwd0xYZHBaR2RsZEhNdGFHOXlhWHB2Ym5SaGJDMTBZV0l0ZEc5d0xXSnZjbVJsY2pvZ01uQjRPMXh1SUNBZ0lDMHRhbkF0ZDJsa1oyVjBjeTF3Y205bmNtVnpjeTEwYUdsamEyNWxjM002SURJd2NIZzdYRzRnSUNBZ0xTMXFjQzEzYVdSblpYUnpMV052Ym5SaGFXNWxjaTF3WVdSa2FXNW5PaUF4TlhCNE8xeHVJQ0FnSUMwdGFuQXRkMmxrWjJWMGN5MXBibkIxZEMxd1lXUmthVzVuT2lBMGNIZzdYRzRnSUNBZ0xTMXFjQzEzYVdSblpYUnpMWEpoWkdsdkxXbDBaVzB0YUdWcFoyaDBMV0ZrYW5WemRHMWxiblE2SURod2VEdGNiaUFnSUNBdExXcHdMWGRwWkdkbGRITXRjbUZrYVc4dGFYUmxiUzFvWldsbmFIUTZJR05oYkdNb2RtRnlLQzB0YW5BdGQybGtaMlYwY3kxcGJteHBibVV0YUdWcFoyaDBLU0F0SUhaaGNpZ3RMV3B3TFhkcFpHZGxkSE10Y21Ga2FXOHRhWFJsYlMxb1pXbG5hSFF0WVdScWRYTjBiV1Z1ZENrcE8xeHVJQ0FnSUMwdGFuQXRkMmxrWjJWMGN5MXpiR2xrWlhJdGRISmhZMnN0ZEdocFkydHVaWE56T2lBMGNIZzdYRzRnSUNBZ0xTMXFjQzEzYVdSblpYUnpMWE5zYVdSbGNpMWliM0prWlhJdGQybGtkR2c2SUhaaGNpZ3RMV3B3TFhkcFpHZGxkSE10WW05eVpHVnlMWGRwWkhSb0tUdGNiaUFnSUNBdExXcHdMWGRwWkdkbGRITXRjMnhwWkdWeUxXaGhibVJzWlMxemFYcGxPaUF4Tm5CNE8xeHVJQ0FnSUMwdGFuQXRkMmxrWjJWMGN5MXpiR2xrWlhJdGFHRnVaR3hsTFdKdmNtUmxjaTFqYjJ4dmNqb2dkbUZ5S0MwdGFuQXRZbTl5WkdWeUxXTnZiRzl5TVNrN1hHNGdJQ0FnTFMxcWNDMTNhV1JuWlhSekxYTnNhV1JsY2kxb1lXNWtiR1V0WW1GamEyZHliM1Z1WkMxamIyeHZjam9nZG1GeUtDMHRhbkF0YkdGNWIzVjBMV052Ykc5eU1TazdYRzRnSUNBZ0xTMXFjQzEzYVdSblpYUnpMWE5zYVdSbGNpMWhZM1JwZG1VdGFHRnVaR3hsTFdOdmJHOXlPaUIyWVhJb0xTMXFjQzFpY21GdVpDMWpiMnh2Y2pFcE8xeHVJQ0FnSUMwdGFuQXRkMmxrWjJWMGN5MXRaVzUxTFdsMFpXMHRhR1ZwWjJoME9pQXlOSEI0TzF4dUlDQWdJQzB0YW5BdGQybGtaMlYwY3kxa2NtOXdaRzkzYmkxaGNuSnZkem9nZFhKc0tGd2laR0YwWVRwcGJXRm5aUzl6ZG1jcmVHMXNPMkpoYzJVMk5DeFFSRGswWWxkM1oyUnRWbmxqTW14Mlltb3dhVTFUTkhkSmFVSnNZbTFPZGxwSGJIVmFlakJwWkZoU2JVeFVaMmxRZWpSTFVFTkZkRXhUUWtoYVZ6VnNZMjFHTUdJelNUWkpSVVpyWWpKS2JFbEZiSE5pU0ZaNlpFaEthR1JIT1hsSlJFVTFUR3BKZFUxVGQyZFZNVnBJU1VWV05HTkhPWGxrUTBKUllraFdia3hWYkhWSlF6Um5WVEZhU0VsR1dteGpiazV3WWpJME5rbEVXWFZOUkVGblVXNVdjR0pIVVdkTlEydG5TVU13ZEZCbmJ6aGpNMXB1U1VoYWJHTnVUbkJpTWpRNVNXcEZkVTFUU1dkaFYxRTVTV3Q0YUdWWFZubFlla1ZwU1Vob2RHSkhOWHBRVTBwdlpFaFNkMDlwT0haa00yUXpURzVqZWt4dE9YbGFlVGg1VFVSQmQwd3pUakphZVVsblpVY3hjMkp1VFRabFIzaHdZbTF6T1VsdGFEQmtTRUUyVEhrNU0yUXpZM1ZrZWsxMVlqTktia3g2UlRWUFZHdDJaVWQ0Y0dKdGMybEpTR2M1U1dwQ2QyVkRTV2RsVkRCcFRVaENORWxuYjBwSlNGcHdXbGhrUTJJelp6bEpha0ZuVFVOQmVFOURRWGhQUTBsbll6TlNOV0pIVlRsSmJWWjFXVmRLYzFwVE1XbFpWMDV5V2pOS2RtUlhOV3RQYlRWc1pIbEJkMGxFUVdkTlZHZG5UVlJuTjBscFFqUmlWM2MyWXpOQ2FGa3lWVGxKYmtKNVdsaE9iR051V214SmFqUkxVRWhPTUdWWGVHeEpTRkkxWTBkVk9VbHVVbXhsU0ZGMldUTk9la2xxTkV0RFV6VjZaRVJDTjFwdGJITmlSSEIxWWpJMWJFOHpNRXRRUXpsNlpFaHNjMXBVTkV0UVNFSm9aRWRuWjFwRU1HbFVWRlYxVFdsM01VeHFiRTFQVTNjMVRHcGtjMDE1TkRSTVZFMTFUMGQzZUV4cVNYTk5VelI1WWtNd01FeHFhM05PVjNkMFRrTTBOVXhVVmsxT1V6UjVURVJWZFU5WWIybE1lalJMVUVoQ2FHUkhaMmRaTW5ob1l6Tk5PVWx1VGpCTlEwbG5Xa1F3YVZSVVFYUk5RelF5WVVSRk5HUnFSVFJUUkVKWFRGUkJkVTV1YjJsTWVqUkxVRU01ZW1SdFl5dERaMXdpS1R0Y2JpQWdJQ0F0TFdwd0xYZHBaR2RsZEhNdGFXNXdkWFF0WTI5c2IzSTZJSFpoY2lndExXcHdMWFZwTFdadmJuUXRZMjlzYjNJeEtUdGNiaUFnSUNBdExXcHdMWGRwWkdkbGRITXRhVzV3ZFhRdFltRmphMmR5YjNWdVpDMWpiMnh2Y2pvZ2RtRnlLQzB0YW5BdGJHRjViM1YwTFdOdmJHOXlNU2s3WEc0Z0lDQWdMUzFxY0MxM2FXUm5aWFJ6TFdsdWNIVjBMV0p2Y21SbGNpMWpiMnh2Y2pvZ2RtRnlLQzB0YW5BdFltOXlaR1Z5TFdOdmJHOXlNU2s3WEc0Z0lDQWdMUzFxY0MxM2FXUm5aWFJ6TFdsdWNIVjBMV1p2WTNWekxXSnZjbVJsY2kxamIyeHZjam9nZG1GeUtDMHRhbkF0WW5KaGJtUXRZMjlzYjNJeUtUdGNiaUFnSUNBdExXcHdMWGRwWkdkbGRITXRhVzV3ZFhRdFltOXlaR1Z5TFhkcFpIUm9PaUIyWVhJb0xTMXFjQzEzYVdSblpYUnpMV0p2Y21SbGNpMTNhV1IwYUNrN1hHNGdJQ0FnTFMxcWNDMTNhV1JuWlhSekxXUnBjMkZpYkdWa0xXOXdZV05wZEhrNklEQXVOanRjYmx4dUlDQWdJQzhxSUVaeWIyMGdUV0YwWlhKcFlXd2dSR1Z6YVdkdUlFeHBkR1VnS2k5Y2JpQWdJQ0F0TFcxa0xYTm9ZV1J2ZHkxclpYa3RkVzFpY21FdGIzQmhZMmwwZVRvZ01DNHlPMXh1SUNBZ0lDMHRiV1F0YzJoaFpHOTNMV3RsZVMxd1pXNTFiV0p5WVMxdmNHRmphWFI1T2lBd0xqRTBPMXh1SUNBZ0lDMHRiV1F0YzJoaFpHOTNMV0Z0WW1sbGJuUXRjMmhoWkc5M0xXOXdZV05wZEhrNklEQXVNVEk3WEc1OVhHNWNiaTVxZFhCNWRHVnlMWGRwWkdkbGRITWdlMXh1SUNBZ0lHMWhjbWRwYmpvZ2RtRnlLQzB0YW5BdGQybGtaMlYwY3kxdFlYSm5hVzRwTzF4dUlDQWdJR0p2ZUMxemFYcHBibWM2SUdKdmNtUmxjaTFpYjNnN1hHNGdJQ0FnWTI5c2IzSTZJSFpoY2lndExXcHdMWGRwWkdkbGRITXRZMjlzYjNJcE8xeHVJQ0FnSUc5MlpYSm1iRzkzT2lCMmFYTnBZbXhsTzF4dWZWeHVYRzR1YW5Wd2VYUmxjaTEzYVdSblpYUnpMbXAxY0hsMFpYSXRkMmxrWjJWMGN5MWthWE5qYjI1dVpXTjBaV1E2T21KbFptOXlaU0I3WEc0Z0lDQWdiR2x1WlMxb1pXbG5hSFE2SUhaaGNpZ3RMV3B3TFhkcFpHZGxkSE10YVc1c2FXNWxMV2hsYVdkb2RDazdYRzRnSUNBZ2FHVnBaMmgwT2lCMllYSW9MUzFxY0MxM2FXUm5aWFJ6TFdsdWJHbHVaUzFvWldsbmFIUXBPMXh1ZlZ4dVhHNHVhbkF0VDNWMGNIVjBMWEpsYzNWc2RDQStJQzVxZFhCNWRHVnlMWGRwWkdkbGRITWdlMXh1SUNBZ0lHMWhjbWRwYmkxc1pXWjBPaUF3TzF4dUlDQWdJRzFoY21kcGJpMXlhV2RvZERvZ01EdGNibjFjYmx4dUx5b2dkbUp2ZUNCaGJtUWdhR0p2ZUNBcUwxeHVYRzR1ZDJsa1oyVjBMV2x1YkdsdVpTMW9ZbTk0SUh0Y2JpQWdJQ0F2S2lCSWIzSnBlbTl1ZEdGc0lIZHBaR2RsZEhNZ0tpOWNiaUFnSUNCaWIzZ3RjMmw2YVc1bk9pQmliM0prWlhJdFltOTRPMXh1SUNBZ0lHUnBjM0JzWVhrNklHWnNaWGc3WEc0Z0lDQWdabXhsZUMxa2FYSmxZM1JwYjI0NklISnZkenRjYmlBZ0lDQmhiR2xuYmkxcGRHVnRjem9nWW1GelpXeHBibVU3WEc1OVhHNWNiaTUzYVdSblpYUXRhVzVzYVc1bExYWmliM2dnZTF4dUlDQWdJQzhxSUZabGNuUnBZMkZzSUZkcFpHZGxkSE1nS2k5Y2JpQWdJQ0JpYjNndGMybDZhVzVuT2lCaWIzSmtaWEl0WW05NE8xeHVJQ0FnSUdScGMzQnNZWGs2SUdac1pYZzdYRzRnSUNBZ1pteGxlQzFrYVhKbFkzUnBiMjQ2SUdOdmJIVnRianRjYmlBZ0lDQmhiR2xuYmkxcGRHVnRjem9nWTJWdWRHVnlPMXh1ZlZ4dVhHNHVkMmxrWjJWMExXSnZlQ0I3WEc0Z0lDQWdZbTk0TFhOcGVtbHVaem9nWW05eVpHVnlMV0p2ZUR0Y2JpQWdJQ0JrYVhOd2JHRjVPaUJtYkdWNE8xeHVJQ0FnSUcxaGNtZHBiam9nTUR0Y2JpQWdJQ0J2ZG1WeVpteHZkem9nWVhWMGJ6dGNibjFjYmx4dUxuZHBaR2RsZEMxbmNtbGtZbTk0SUh0Y2JpQWdJQ0JpYjNndGMybDZhVzVuT2lCaWIzSmtaWEl0WW05NE8xeHVJQ0FnSUdScGMzQnNZWGs2SUdkeWFXUTdYRzRnSUNBZ2JXRnlaMmx1T2lBd08xeHVJQ0FnSUc5MlpYSm1iRzkzT2lCaGRYUnZPMXh1ZlZ4dVhHNHVkMmxrWjJWMExXaGliM2dnZTF4dUlDQWdJR1pzWlhndFpHbHlaV04wYVc5dU9pQnliM2M3WEc1OVhHNWNiaTUzYVdSblpYUXRkbUp2ZUNCN1hHNGdJQ0FnWm14bGVDMWthWEpsWTNScGIyNDZJR052YkhWdGJqdGNibjFjYmx4dUx5b2dSMlZ1WlhKaGJDQkNkWFIwYjI0Z1UzUjViR2x1WnlBcUwxeHVYRzR1YW5Wd2VYUmxjaTFpZFhSMGIyNGdlMXh1SUNBZ0lIQmhaR1JwYm1jdGJHVm1kRG9nTVRCd2VEdGNiaUFnSUNCd1lXUmthVzVuTFhKcFoyaDBPaUF4TUhCNE8xeHVJQ0FnSUhCaFpHUnBibWN0ZEc5d09pQXdjSGc3WEc0Z0lDQWdjR0ZrWkdsdVp5MWliM1IwYjIwNklEQndlRHRjYmlBZ0lDQmthWE53YkdGNU9pQnBibXhwYm1VdFlteHZZMnM3WEc0Z0lDQWdkMmhwZEdVdGMzQmhZMlU2SUc1dmQzSmhjRHRjYmlBZ0lDQnZkbVZ5Wm14dmR6b2dhR2xrWkdWdU8xeHVJQ0FnSUhSbGVIUXRiM1psY21ac2IzYzZJR1ZzYkdsd2MybHpPMXh1SUNBZ0lIUmxlSFF0WVd4cFoyNDZJR05sYm5SbGNqdGNiaUFnSUNCbWIyNTBMWE5wZW1VNklIWmhjaWd0TFdwd0xYZHBaR2RsZEhNdFptOXVkQzF6YVhwbEtUdGNiaUFnSUNCamRYSnpiM0k2SUhCdmFXNTBaWEk3WEc1Y2JpQWdJQ0JvWldsbmFIUTZJSFpoY2lndExXcHdMWGRwWkdkbGRITXRhVzVzYVc1bExXaGxhV2RvZENrN1hHNGdJQ0FnWW05eVpHVnlPaUF3Y0hnZ2MyOXNhV1E3WEc0Z0lDQWdiR2x1WlMxb1pXbG5hSFE2SUhaaGNpZ3RMV3B3TFhkcFpHZGxkSE10YVc1c2FXNWxMV2hsYVdkb2RDazdYRzRnSUNBZ1ltOTRMWE5vWVdSdmR6b2dibTl1WlR0Y2JseHVJQ0FnSUdOdmJHOXlPaUIyWVhJb0xTMXFjQzExYVMxbWIyNTBMV052Ykc5eU1TazdYRzRnSUNBZ1ltRmphMmR5YjNWdVpDMWpiMnh2Y2pvZ2RtRnlLQzB0YW5BdGJHRjViM1YwTFdOdmJHOXlNaWs3WEc0Z0lDQWdZbTl5WkdWeUxXTnZiRzl5T2lCMllYSW9MUzFxY0MxaWIzSmtaWEl0WTI5c2IzSXlLVHRjYmlBZ0lDQmliM0prWlhJNklHNXZibVU3WEc1OVhHNWNiaTVxZFhCNWRHVnlMV0oxZEhSdmJpQnBMbVpoSUh0Y2JpQWdJQ0J0WVhKbmFXNHRjbWxuYUhRNklIWmhjaWd0TFdwd0xYZHBaR2RsZEhNdGFXNXNhVzVsTFcxaGNtZHBiaWs3WEc0Z0lDQWdjRzlwYm5SbGNpMWxkbVZ1ZEhNNklHNXZibVU3WEc1OVhHNWNiaTVxZFhCNWRHVnlMV0oxZEhSdmJqcGxiWEIwZVRwaVpXWnZjbVVnZTF4dUlDQWdJR052Ym5SbGJuUTZJRndpWEZ3eU1EQmlYQ0k3SUM4cUlIcGxjbTh0ZDJsa2RHZ2djM0JoWTJVZ0tpOWNibjFjYmx4dUxtcDFjSGwwWlhJdGQybGtaMlYwY3k1cWRYQjVkR1Z5TFdKMWRIUnZianBrYVhOaFlteGxaQ0I3WEc0Z0lDQWdiM0JoWTJsMGVUb2dkbUZ5S0MwdGFuQXRkMmxrWjJWMGN5MWthWE5oWW14bFpDMXZjR0ZqYVhSNUtUdGNibjFjYmx4dUxtcDFjSGwwWlhJdFluVjBkRzl1SUdrdVptRXVZMlZ1ZEdWeUlIdGNiaUFnSUNCdFlYSm5hVzR0Y21sbmFIUTZJREE3WEc1OVhHNWNiaTVxZFhCNWRHVnlMV0oxZEhSdmJqcG9iM1psY2pwbGJtRmliR1ZrTENBdWFuVndlWFJsY2kxaWRYUjBiMjQ2Wm05amRYTTZaVzVoWW14bFpDQjdYRzRnSUNBZ0x5b2dUVVFnVEdsMFpTQXlaSEFnYzJoaFpHOTNJQ292WEc0Z0lDQWdZbTk0TFhOb1lXUnZkem9nTUNBeWNIZ2dNbkI0SURBZ2NtZGlZU2d3TENBd0xDQXdMQ0IyWVhJb0xTMXRaQzF6YUdGa2IzY3RhMlY1TFhCbGJuVnRZbkpoTFc5d1lXTnBkSGtwS1N4Y2JpQWdJQ0FnSUNBZ0lDQWdJQ0FnSUNBd0lETndlQ0F4Y0hnZ0xUSndlQ0J5WjJKaEtEQXNJREFzSURBc0lIWmhjaWd0TFcxa0xYTm9ZV1J2ZHkxclpYa3RkVzFpY21FdGIzQmhZMmwwZVNrcExGeHVJQ0FnSUNBZ0lDQWdJQ0FnSUNBZ0lEQWdNWEI0SURWd2VDQXdJSEpuWW1Fb01Dd2dNQ3dnTUN3Z2RtRnlLQzB0YldRdGMyaGhaRzkzTFdGdFltbGxiblF0YzJoaFpHOTNMVzl3WVdOcGRIa3BLVHRjYm4xY2JseHVMbXAxY0hsMFpYSXRZblYwZEc5dU9tRmpkR2wyWlN3Z0xtcDFjSGwwWlhJdFluVjBkRzl1TG0xdlpDMWhZM1JwZG1VZ2UxeHVJQ0FnSUM4cUlFMUVJRXhwZEdVZ05HUndJSE5vWVdSdmR5QXFMMXh1SUNBZ0lHSnZlQzF6YUdGa2IzYzZJREFnTkhCNElEVndlQ0F3SUhKblltRW9NQ3dnTUN3Z01Dd2dkbUZ5S0MwdGJXUXRjMmhoWkc5M0xXdGxlUzF3Wlc1MWJXSnlZUzF2Y0dGamFYUjVLU2tzWEc0Z0lDQWdJQ0FnSUNBZ0lDQWdJQ0FnTUNBeGNIZ2dNVEJ3ZUNBd0lISm5ZbUVvTUN3Z01Dd2dNQ3dnZG1GeUtDMHRiV1F0YzJoaFpHOTNMV0Z0WW1sbGJuUXRjMmhoWkc5M0xXOXdZV05wZEhrcEtTeGNiaUFnSUNBZ0lDQWdJQ0FnSUNBZ0lDQXdJREp3ZUNBMGNIZ2dMVEZ3ZUNCeVoySmhLREFzSURBc0lEQXNJSFpoY2lndExXMWtMWE5vWVdSdmR5MXJaWGt0ZFcxaWNtRXRiM0JoWTJsMGVTa3BPMXh1SUNBZ0lHTnZiRzl5T2lCMllYSW9MUzFxY0MxMWFTMW1iMjUwTFdOdmJHOXlNU2s3WEc0Z0lDQWdZbUZqYTJkeWIzVnVaQzFqYjJ4dmNqb2dkbUZ5S0MwdGFuQXRiR0Y1YjNWMExXTnZiRzl5TXlrN1hHNTlYRzVjYmk1cWRYQjVkR1Z5TFdKMWRIUnZianBtYjJOMWN6cGxibUZpYkdWa0lIdGNiaUFnSUNCdmRYUnNhVzVsT2lBeGNIZ2djMjlzYVdRZ2RtRnlLQzB0YW5BdGQybGtaMlYwY3kxcGJuQjFkQzFtYjJOMWN5MWliM0prWlhJdFkyOXNiM0lwTzF4dWZWeHVYRzR2S2lCQ2RYUjBiMjRnWENKUWNtbHRZWEo1WENJZ1UzUjViR2x1WnlBcUwxeHVYRzR1YW5Wd2VYUmxjaTFpZFhSMGIyNHViVzlrTFhCeWFXMWhjbmtnZTF4dUlDQWdJR052Ykc5eU9pQjJZWElvTFMxcWNDMXBiblpsY25ObExYVnBMV1p2Ym5RdFkyOXNiM0l4S1R0Y2JpQWdJQ0JpWVdOclozSnZkVzVrTFdOdmJHOXlPaUIyWVhJb0xTMXFjQzFpY21GdVpDMWpiMnh2Y2pFcE8xeHVmVnh1WEc0dWFuVndlWFJsY2kxaWRYUjBiMjR1Ylc5a0xYQnlhVzFoY25rdWJXOWtMV0ZqZEdsMlpTQjdYRzRnSUNBZ1kyOXNiM0k2SUhaaGNpZ3RMV3B3TFdsdWRtVnljMlV0ZFdrdFptOXVkQzFqYjJ4dmNqQXBPMXh1SUNBZ0lHSmhZMnRuY205MWJtUXRZMjlzYjNJNklIWmhjaWd0TFdwd0xXSnlZVzVrTFdOdmJHOXlNQ2s3WEc1OVhHNWNiaTVxZFhCNWRHVnlMV0oxZEhSdmJpNXRiMlF0Y0hKcGJXRnllVHBoWTNScGRtVWdlMXh1SUNBZ0lHTnZiRzl5T2lCMllYSW9MUzFxY0MxcGJuWmxjbk5sTFhWcExXWnZiblF0WTI5c2IzSXdLVHRjYmlBZ0lDQmlZV05yWjNKdmRXNWtMV052Ykc5eU9pQjJZWElvTFMxcWNDMWljbUZ1WkMxamIyeHZjakFwTzF4dWZWeHVYRzR2S2lCQ2RYUjBiMjRnWENKVGRXTmpaWE56WENJZ1UzUjViR2x1WnlBcUwxeHVYRzR1YW5Wd2VYUmxjaTFpZFhSMGIyNHViVzlrTFhOMVkyTmxjM01nZTF4dUlDQWdJR052Ykc5eU9pQjJZWElvTFMxcWNDMXBiblpsY25ObExYVnBMV1p2Ym5RdFkyOXNiM0l4S1R0Y2JpQWdJQ0JpWVdOclozSnZkVzVrTFdOdmJHOXlPaUIyWVhJb0xTMXFjQzF6ZFdOalpYTnpMV052Ykc5eU1TazdYRzU5WEc1Y2JpNXFkWEI1ZEdWeUxXSjFkSFJ2Ymk1dGIyUXRjM1ZqWTJWemN5NXRiMlF0WVdOMGFYWmxJSHRjYmlBZ0lDQmpiMnh2Y2pvZ2RtRnlLQzB0YW5BdGFXNTJaWEp6WlMxMWFTMW1iMjUwTFdOdmJHOXlNQ2s3WEc0Z0lDQWdZbUZqYTJkeWIzVnVaQzFqYjJ4dmNqb2dkbUZ5S0MwdGFuQXRjM1ZqWTJWemN5MWpiMnh2Y2pBcE8xeHVJSDFjYmx4dUxtcDFjSGwwWlhJdFluVjBkRzl1TG0xdlpDMXpkV05qWlhOek9tRmpkR2wyWlNCN1hHNGdJQ0FnWTI5c2IzSTZJSFpoY2lndExXcHdMV2x1ZG1WeWMyVXRkV2t0Wm05dWRDMWpiMnh2Y2pBcE8xeHVJQ0FnSUdKaFkydG5jbTkxYm1RdFkyOXNiM0k2SUhaaGNpZ3RMV3B3TFhOMVkyTmxjM010WTI5c2IzSXdLVHRjYmlCOVhHNWNiaUF2S2lCQ2RYUjBiMjRnWENKSmJtWnZYQ0lnVTNSNWJHbHVaeUFxTDF4dVhHNHVhblZ3ZVhSbGNpMWlkWFIwYjI0dWJXOWtMV2x1Wm04Z2UxeHVJQ0FnSUdOdmJHOXlPaUIyWVhJb0xTMXFjQzFwYm5abGNuTmxMWFZwTFdadmJuUXRZMjlzYjNJeEtUdGNiaUFnSUNCaVlXTnJaM0p2ZFc1a0xXTnZiRzl5T2lCMllYSW9MUzFxY0MxcGJtWnZMV052Ykc5eU1TazdYRzU5WEc1Y2JpNXFkWEI1ZEdWeUxXSjFkSFJ2Ymk1dGIyUXRhVzVtYnk1dGIyUXRZV04wYVhabElIdGNiaUFnSUNCamIyeHZjam9nZG1GeUtDMHRhbkF0YVc1MlpYSnpaUzExYVMxbWIyNTBMV052Ykc5eU1DazdYRzRnSUNBZ1ltRmphMmR5YjNWdVpDMWpiMnh2Y2pvZ2RtRnlLQzB0YW5BdGFXNW1ieTFqYjJ4dmNqQXBPMXh1ZlZ4dVhHNHVhblZ3ZVhSbGNpMWlkWFIwYjI0dWJXOWtMV2x1Wm04NllXTjBhWFpsSUh0Y2JpQWdJQ0JqYjJ4dmNqb2dkbUZ5S0MwdGFuQXRhVzUyWlhKelpTMTFhUzFtYjI1MExXTnZiRzl5TUNrN1hHNGdJQ0FnWW1GamEyZHliM1Z1WkMxamIyeHZjam9nZG1GeUtDMHRhbkF0YVc1bWJ5MWpiMnh2Y2pBcE8xeHVmVnh1WEc0dktpQkNkWFIwYjI0Z1hDSlhZWEp1YVc1blhDSWdVM1I1YkdsdVp5QXFMMXh1WEc0dWFuVndlWFJsY2kxaWRYUjBiMjR1Ylc5a0xYZGhjbTVwYm1jZ2UxeHVJQ0FnSUdOdmJHOXlPaUIyWVhJb0xTMXFjQzFwYm5abGNuTmxMWFZwTFdadmJuUXRZMjlzYjNJeEtUdGNiaUFnSUNCaVlXTnJaM0p2ZFc1a0xXTnZiRzl5T2lCMllYSW9MUzFxY0MxM1lYSnVMV052Ykc5eU1TazdYRzU5WEc1Y2JpNXFkWEI1ZEdWeUxXSjFkSFJ2Ymk1dGIyUXRkMkZ5Ym1sdVp5NXRiMlF0WVdOMGFYWmxJSHRjYmlBZ0lDQmpiMnh2Y2pvZ2RtRnlLQzB0YW5BdGFXNTJaWEp6WlMxMWFTMW1iMjUwTFdOdmJHOXlNQ2s3WEc0Z0lDQWdZbUZqYTJkeWIzVnVaQzFqYjJ4dmNqb2dkbUZ5S0MwdGFuQXRkMkZ5YmkxamIyeHZjakFwTzF4dWZWeHVYRzR1YW5Wd2VYUmxjaTFpZFhSMGIyNHViVzlrTFhkaGNtNXBibWM2WVdOMGFYWmxJSHRjYmlBZ0lDQmpiMnh2Y2pvZ2RtRnlLQzB0YW5BdGFXNTJaWEp6WlMxMWFTMW1iMjUwTFdOdmJHOXlNQ2s3WEc0Z0lDQWdZbUZqYTJkeWIzVnVaQzFqYjJ4dmNqb2dkbUZ5S0MwdGFuQXRkMkZ5YmkxamIyeHZjakFwTzF4dWZWeHVYRzR2S2lCQ2RYUjBiMjRnWENKRVlXNW5aWEpjSWlCVGRIbHNhVzVuSUNvdlhHNWNiaTVxZFhCNWRHVnlMV0oxZEhSdmJpNXRiMlF0WkdGdVoyVnlJSHRjYmlBZ0lDQmpiMnh2Y2pvZ2RtRnlLQzB0YW5BdGFXNTJaWEp6WlMxMWFTMW1iMjUwTFdOdmJHOXlNU2s3WEc0Z0lDQWdZbUZqYTJkeWIzVnVaQzFqYjJ4dmNqb2dkbUZ5S0MwdGFuQXRaWEp5YjNJdFkyOXNiM0l4S1R0Y2JuMWNibHh1TG1wMWNIbDBaWEl0WW5WMGRHOXVMbTF2WkMxa1lXNW5aWEl1Ylc5a0xXRmpkR2wyWlNCN1hHNGdJQ0FnWTI5c2IzSTZJSFpoY2lndExXcHdMV2x1ZG1WeWMyVXRkV2t0Wm05dWRDMWpiMnh2Y2pBcE8xeHVJQ0FnSUdKaFkydG5jbTkxYm1RdFkyOXNiM0k2SUhaaGNpZ3RMV3B3TFdWeWNtOXlMV052Ykc5eU1DazdYRzU5WEc1Y2JpNXFkWEI1ZEdWeUxXSjFkSFJ2Ymk1dGIyUXRaR0Z1WjJWeU9tRmpkR2wyWlNCN1hHNGdJQ0FnWTI5c2IzSTZJSFpoY2lndExXcHdMV2x1ZG1WeWMyVXRkV2t0Wm05dWRDMWpiMnh2Y2pBcE8xeHVJQ0FnSUdKaFkydG5jbTkxYm1RdFkyOXNiM0k2SUhaaGNpZ3RMV3B3TFdWeWNtOXlMV052Ykc5eU1DazdYRzU5WEc1Y2JpOHFJRmRwWkdkbGRDQkNkWFIwYjI0cUwxeHVYRzR1ZDJsa1oyVjBMV0oxZEhSdmJpd2dMbmRwWkdkbGRDMTBiMmRuYkdVdFluVjBkRzl1SUh0Y2JpQWdJQ0IzYVdSMGFEb2dkbUZ5S0MwdGFuQXRkMmxrWjJWMGN5MXBibXhwYm1VdGQybGtkR2d0YzJodmNuUXBPMXh1ZlZ4dVhHNHZLaUJYYVdSblpYUWdUR0ZpWld3Z1UzUjViR2x1WnlBcUwxeHVYRzR2S2lCUGRtVnljbWxrWlNCQ2IyOTBjM1J5WVhBZ2JHRmlaV3dnWTNOeklDb3ZYRzR1YW5Wd2VYUmxjaTEzYVdSblpYUnpJR3hoWW1Wc0lIdGNiaUFnSUNCdFlYSm5hVzR0WW05MGRHOXRPaUJwYm1sMGFXRnNPMXh1ZlZ4dVhHNHVkMmxrWjJWMExXeGhZbVZzTFdKaGMybGpJSHRjYmlBZ0lDQXZLaUJDWVhOcFl5Qk1ZV0psYkNBcUwxeHVJQ0FnSUdOdmJHOXlPaUIyWVhJb0xTMXFjQzEzYVdSblpYUnpMV3hoWW1Wc0xXTnZiRzl5S1R0Y2JpQWdJQ0JtYjI1MExYTnBlbVU2SUhaaGNpZ3RMV3B3TFhkcFpHZGxkSE10Wm05dWRDMXphWHBsS1R0Y2JpQWdJQ0J2ZG1WeVpteHZkem9nYUdsa1pHVnVPMXh1SUNBZ0lIUmxlSFF0YjNabGNtWnNiM2M2SUdWc2JHbHdjMmx6TzF4dUlDQWdJSGRvYVhSbExYTndZV05sT2lCdWIzZHlZWEE3WEc0Z0lDQWdiR2x1WlMxb1pXbG5hSFE2SUhaaGNpZ3RMV3B3TFhkcFpHZGxkSE10YVc1c2FXNWxMV2hsYVdkb2RDazdYRzU5WEc1Y2JpNTNhV1JuWlhRdGJHRmlaV3dnZTF4dUlDQWdJQzhxSUV4aFltVnNJQ292WEc0Z0lDQWdZMjlzYjNJNklIWmhjaWd0TFdwd0xYZHBaR2RsZEhNdGJHRmlaV3d0WTI5c2IzSXBPMXh1SUNBZ0lHWnZiblF0YzJsNlpUb2dkbUZ5S0MwdGFuQXRkMmxrWjJWMGN5MW1iMjUwTFhOcGVtVXBPMXh1SUNBZ0lHOTJaWEptYkc5M09pQm9hV1JrWlc0N1hHNGdJQ0FnZEdWNGRDMXZkbVZ5Wm14dmR6b2daV3hzYVhCemFYTTdYRzRnSUNBZ2QyaHBkR1V0YzNCaFkyVTZJRzV2ZDNKaGNEdGNiaUFnSUNCc2FXNWxMV2hsYVdkb2REb2dkbUZ5S0MwdGFuQXRkMmxrWjJWMGN5MXBibXhwYm1VdGFHVnBaMmgwS1R0Y2JuMWNibHh1TG5kcFpHZGxkQzFwYm14cGJtVXRhR0p2ZUNBdWQybGtaMlYwTFd4aFltVnNJSHRjYmlBZ0lDQXZLaUJJYjNKcGVtOXVkR0ZzSUZkcFpHZGxkQ0JNWVdKbGJDQXFMMXh1SUNBZ0lHTnZiRzl5T2lCMllYSW9MUzFxY0MxM2FXUm5aWFJ6TFd4aFltVnNMV052Ykc5eUtUdGNiaUFnSUNCMFpYaDBMV0ZzYVdkdU9pQnlhV2RvZER0Y2JpQWdJQ0J0WVhKbmFXNHRjbWxuYUhRNklHTmhiR01vSUhaaGNpZ3RMV3B3TFhkcFpHZGxkSE10YVc1c2FXNWxMVzFoY21kcGJpa2dLaUF5SUNrN1hHNGdJQ0FnZDJsa2RHZzZJSFpoY2lndExXcHdMWGRwWkdkbGRITXRhVzVzYVc1bExXeGhZbVZzTFhkcFpIUm9LVHRjYmlBZ0lDQm1iR1Y0TFhOb2NtbHVhem9nTUR0Y2JuMWNibHh1TG5kcFpHZGxkQzFwYm14cGJtVXRkbUp2ZUNBdWQybGtaMlYwTFd4aFltVnNJSHRjYmlBZ0lDQXZLaUJXWlhKMGFXTmhiQ0JYYVdSblpYUWdUR0ZpWld3Z0tpOWNiaUFnSUNCamIyeHZjam9nZG1GeUtDMHRhbkF0ZDJsa1oyVjBjeTFzWVdKbGJDMWpiMnh2Y2lrN1hHNGdJQ0FnZEdWNGRDMWhiR2xuYmpvZ1kyVnVkR1Z5TzF4dUlDQWdJR3hwYm1VdGFHVnBaMmgwT2lCMllYSW9MUzFxY0MxM2FXUm5aWFJ6TFdsdWJHbHVaUzFvWldsbmFIUXBPMXh1ZlZ4dVhHNHZLaUJYYVdSblpYUWdVbVZoWkc5MWRDQlRkSGxzYVc1bklDb3ZYRzVjYmk1M2FXUm5aWFF0Y21WaFpHOTFkQ0I3WEc0Z0lDQWdZMjlzYjNJNklIWmhjaWd0TFdwd0xYZHBaR2RsZEhNdGNtVmhaRzkxZEMxamIyeHZjaWs3WEc0Z0lDQWdabTl1ZEMxemFYcGxPaUIyWVhJb0xTMXFjQzEzYVdSblpYUnpMV1p2Ym5RdGMybDZaU2s3WEc0Z0lDQWdhR1ZwWjJoME9pQjJZWElvTFMxcWNDMTNhV1JuWlhSekxXbHViR2x1WlMxb1pXbG5hSFFwTzF4dUlDQWdJR3hwYm1VdGFHVnBaMmgwT2lCMllYSW9MUzFxY0MxM2FXUm5aWFJ6TFdsdWJHbHVaUzFvWldsbmFIUXBPMXh1SUNBZ0lHOTJaWEptYkc5M09pQm9hV1JrWlc0N1hHNGdJQ0FnZDJocGRHVXRjM0JoWTJVNklHNXZkM0poY0R0Y2JpQWdJQ0IwWlhoMExXRnNhV2R1T2lCalpXNTBaWEk3WEc1OVhHNWNiaTUzYVdSblpYUXRjbVZoWkc5MWRDNXZkbVZ5Wm14dmR5QjdYRzRnSUNBZ0x5b2dUM1psY21ac2IzZHBibWNnVW1WaFpHOTFkQ0FxTDF4dVhHNGdJQ0FnTHlvZ1JuSnZiU0JOWVhSbGNtbGhiQ0JFWlhOcFoyNGdUR2wwWlZ4dUlDQWdJQ0FnSUNCemFHRmtiM2N0YTJWNUxYVnRZbkpoTFc5d1lXTnBkSGs2SURBdU1qdGNiaUFnSUNBZ0lDQWdjMmhoWkc5M0xXdGxlUzF3Wlc1MWJXSnlZUzF2Y0dGamFYUjVPaUF3TGpFME8xeHVJQ0FnSUNBZ0lDQnphR0ZrYjNjdFlXMWlhV1Z1ZEMxemFHRmtiM2N0YjNCaFkybDBlVG9nTUM0eE1qdGNiaUFnSUNBZ0tpOWNiaUFnSUNBdGQyVmlhMmwwTFdKdmVDMXphR0ZrYjNjNklEQWdNbkI0SURKd2VDQXdJSEpuWW1Fb01Dd2dNQ3dnTUN3Z01DNHlLU3hjYmlBZ0lDQWdJQ0FnSUNBZ0lDQWdJQ0FnSUNBZ0lDQWdJREFnTTNCNElERndlQ0F0TW5CNElISm5ZbUVvTUN3Z01Dd2dNQ3dnTUM0eE5Da3NYRzRnSUNBZ0lDQWdJQ0FnSUNBZ0lDQWdJQ0FnSUNBZ0lDQXdJREZ3ZUNBMWNIZ2dNQ0J5WjJKaEtEQXNJREFzSURBc0lEQXVNVElwTzF4dVhHNGdJQ0FnTFcxdmVpMWliM2d0YzJoaFpHOTNPaUF3SURKd2VDQXljSGdnTUNCeVoySmhLREFzSURBc0lEQXNJREF1TWlrc1hHNGdJQ0FnSUNBZ0lDQWdJQ0FnSUNBZ0lDQWdJQ0F3SUROd2VDQXhjSGdnTFRKd2VDQnlaMkpoS0RBc0lEQXNJREFzSURBdU1UUXBMRnh1SUNBZ0lDQWdJQ0FnSUNBZ0lDQWdJQ0FnSUNBZ01DQXhjSGdnTlhCNElEQWdjbWRpWVNnd0xDQXdMQ0F3TENBd0xqRXlLVHRjYmx4dUlDQWdJR0p2ZUMxemFHRmtiM2M2SURBZ01uQjRJREp3ZUNBd0lISm5ZbUVvTUN3Z01Dd2dNQ3dnTUM0eUtTeGNiaUFnSUNBZ0lDQWdJQ0FnSUNBZ0lDQXdJRE53ZUNBeGNIZ2dMVEp3ZUNCeVoySmhLREFzSURBc0lEQXNJREF1TVRRcExGeHVJQ0FnSUNBZ0lDQWdJQ0FnSUNBZ0lEQWdNWEI0SURWd2VDQXdJSEpuWW1Fb01Dd2dNQ3dnTUN3Z01DNHhNaWs3WEc1OVhHNWNiaTUzYVdSblpYUXRhVzVzYVc1bExXaGliM2dnTG5kcFpHZGxkQzF5WldGa2IzVjBJSHRjYmlBZ0lDQXZLaUJJYjNKcGVtOXVkR0ZzSUZKbFlXUnZkWFFnS2k5Y2JpQWdJQ0IwWlhoMExXRnNhV2R1T2lCalpXNTBaWEk3WEc0Z0lDQWdiV0Y0TFhkcFpIUm9PaUIyWVhJb0xTMXFjQzEzYVdSblpYUnpMV2x1YkdsdVpTMTNhV1IwYUMxemFHOXlkQ2s3WEc0Z0lDQWdiV2x1TFhkcFpIUm9PaUIyWVhJb0xTMXFjQzEzYVdSblpYUnpMV2x1YkdsdVpTMTNhV1IwYUMxMGFXNTVLVHRjYmlBZ0lDQnRZWEpuYVc0dGJHVm1kRG9nZG1GeUtDMHRhbkF0ZDJsa1oyVjBjeTFwYm14cGJtVXRiV0Z5WjJsdUtUdGNibjFjYmx4dUxuZHBaR2RsZEMxcGJteHBibVV0ZG1KdmVDQXVkMmxrWjJWMExYSmxZV1J2ZFhRZ2UxeHVJQ0FnSUM4cUlGWmxjblJwWTJGc0lGSmxZV1J2ZFhRZ0tpOWNiaUFnSUNCdFlYSm5hVzR0ZEc5d09pQjJZWElvTFMxcWNDMTNhV1JuWlhSekxXbHViR2x1WlMxdFlYSm5hVzRwTzF4dUlDQWdJQzhxSUdGeklIZHBaR1VnWVhNZ2RHaGxJSGRwWkdkbGRDQXFMMXh1SUNBZ0lIZHBaSFJvT2lCcGJtaGxjbWwwTzF4dWZWeHVYRzR2S2lCWGFXUm5aWFFnUTJobFkydGliM2dnVTNSNWJHbHVaeUFxTDF4dVhHNHVkMmxrWjJWMExXTm9aV05yWW05NElIdGNiaUFnSUNCM2FXUjBhRG9nZG1GeUtDMHRhbkF0ZDJsa1oyVjBjeTFwYm14cGJtVXRkMmxrZEdncE8xeHVJQ0FnSUdobGFXZG9kRG9nZG1GeUtDMHRhbkF0ZDJsa1oyVjBjeTFwYm14cGJtVXRhR1ZwWjJoMEtUdGNiaUFnSUNCc2FXNWxMV2hsYVdkb2REb2dkbUZ5S0MwdGFuQXRkMmxrWjJWMGN5MXBibXhwYm1VdGFHVnBaMmgwS1R0Y2JuMWNibHh1TG5kcFpHZGxkQzFqYUdWamEySnZlQ0JwYm5CMWRGdDBlWEJsUFZ3aVkyaGxZMnRpYjNoY0lsMGdlMXh1SUNBZ0lHMWhjbWRwYmpvZ01IQjRJR05oYkdNb0lIWmhjaWd0TFdwd0xYZHBaR2RsZEhNdGFXNXNhVzVsTFcxaGNtZHBiaWtnS2lBeUlDa2dNSEI0SURCd2VEdGNiaUFnSUNCc2FXNWxMV2hsYVdkb2REb2dkbUZ5S0MwdGFuQXRkMmxrWjJWMGN5MXBibXhwYm1VdGFHVnBaMmgwS1R0Y2JpQWdJQ0JtYjI1MExYTnBlbVU2SUd4aGNtZGxPMXh1SUNBZ0lHWnNaWGd0WjNKdmR6b2dNVHRjYmlBZ0lDQm1iR1Y0TFhOb2NtbHVhem9nTUR0Y2JpQWdJQ0JoYkdsbmJpMXpaV3htT2lCalpXNTBaWEk3WEc1OVhHNWNiaThxSUZkcFpHZGxkQ0JXWVd4cFpDQlRkSGxzYVc1bklDb3ZYRzVjYmk1M2FXUm5aWFF0ZG1Gc2FXUWdlMXh1SUNBZ0lHaGxhV2RvZERvZ2RtRnlLQzB0YW5BdGQybGtaMlYwY3kxcGJteHBibVV0YUdWcFoyaDBLVHRjYmlBZ0lDQnNhVzVsTFdobGFXZG9kRG9nZG1GeUtDMHRhbkF0ZDJsa1oyVjBjeTFwYm14cGJtVXRhR1ZwWjJoMEtUdGNiaUFnSUNCM2FXUjBhRG9nZG1GeUtDMHRhbkF0ZDJsa1oyVjBjeTFwYm14cGJtVXRkMmxrZEdndGMyaHZjblFwTzF4dUlDQWdJR1p2Ym5RdGMybDZaVG9nZG1GeUtDMHRhbkF0ZDJsa1oyVjBjeTFtYjI1MExYTnBlbVVwTzF4dWZWeHVYRzR1ZDJsa1oyVjBMWFpoYkdsa0lHazZZbVZtYjNKbElIdGNiaUFnSUNCc2FXNWxMV2hsYVdkb2REb2dkbUZ5S0MwdGFuQXRkMmxrWjJWMGN5MXBibXhwYm1VdGFHVnBaMmgwS1R0Y2JpQWdJQ0J0WVhKbmFXNHRjbWxuYUhRNklIWmhjaWd0TFdwd0xYZHBaR2RsZEhNdGFXNXNhVzVsTFcxaGNtZHBiaWs3WEc0Z0lDQWdiV0Z5WjJsdUxXeGxablE2SUhaaGNpZ3RMV3B3TFhkcFpHZGxkSE10YVc1c2FXNWxMVzFoY21kcGJpazdYRzVjYmlBZ0lDQXZLaUJtY205dElIUm9aU0JtWVNCamJHRnpjeUJwYmlCR2IyNTBRWGRsYzI5dFpUb2dhSFIwY0hNNkx5OW5hWFJvZFdJdVkyOXRMMFp2Y25SQmQyVnpiMjFsTDBadmJuUXRRWGRsYzI5dFpTOWliRzlpTHpRNU1UQXdZemRqTTJFM1lqVTRaRFV3WW1GaE56RmxabVZtTVRGaFpqUXhZVFkyWWpBelpETXZZM056TDJadmJuUXRZWGRsYzI5dFpTNWpjM01qVERFMElDb3ZYRzRnSUNBZ1pHbHpjR3hoZVRvZ2FXNXNhVzVsTFdKc2IyTnJPMXh1SUNBZ0lHWnZiblE2SUc1dmNtMWhiQ0J1YjNKdFlXd2dibTl5YldGc0lERTBjSGd2TVNCR2IyNTBRWGRsYzI5dFpUdGNiaUFnSUNCbWIyNTBMWE5wZW1VNklHbHVhR1Z5YVhRN1hHNGdJQ0FnZEdWNGRDMXlaVzVrWlhKcGJtYzZJR0YxZEc4N1hHNGdJQ0FnTFhkbFltdHBkQzFtYjI1MExYTnRiMjkwYUdsdVp6b2dZVzUwYVdGc2FXRnpaV1E3WEc0Z0lDQWdMVzF2ZWkxdmMzZ3RabTl1ZEMxemJXOXZkR2hwYm1jNklHZHlZWGx6WTJGc1pUdGNibjFjYmx4dUxuZHBaR2RsZEMxMllXeHBaQzV0YjJRdGRtRnNhV1FnYVRwaVpXWnZjbVVnZTF4dUlDQWdJR052Ym5SbGJuUTZJRndpWEZ4bU1EQmpYQ0k3WEc0Z0lDQWdZMjlzYjNJNklHZHlaV1Z1TzF4dWZWeHVYRzR1ZDJsa1oyVjBMWFpoYkdsa0xtMXZaQzFwYm5aaGJHbGtJR2s2WW1WbWIzSmxJSHRjYmlBZ0lDQmpiMjUwWlc1ME9pQmNJbHhjWmpBd1pGd2lPMXh1SUNBZ0lHTnZiRzl5T2lCeVpXUTdYRzU5WEc1Y2JpNTNhV1JuWlhRdGRtRnNhV1F1Ylc5a0xYWmhiR2xrSUM1M2FXUm5aWFF0ZG1Gc2FXUXRjbVZoWkc5MWRDQjdYRzRnSUNBZ1pHbHpjR3hoZVRvZ2JtOXVaVHRjYm4xY2JseHVMeW9nVjJsa1oyVjBJRlJsZUhRZ1lXNWtJRlJsZUhSQmNtVmhJRk4wZVdsdVp5QXFMMXh1WEc0dWQybGtaMlYwTFhSbGVIUmhjbVZoTENBdWQybGtaMlYwTFhSbGVIUWdlMXh1SUNBZ0lIZHBaSFJvT2lCMllYSW9MUzFxY0MxM2FXUm5aWFJ6TFdsdWJHbHVaUzEzYVdSMGFDazdYRzU5WEc1Y2JpNTNhV1JuWlhRdGRHVjRkQ0JwYm5CMWRGdDBlWEJsUFZ3aWRHVjRkRndpWFN3Z0xuZHBaR2RsZEMxMFpYaDBJR2x1Y0hWMFczUjVjR1U5WENKdWRXMWlaWEpjSWwxN1hHNGdJQ0FnYUdWcFoyaDBPaUIyWVhJb0xTMXFjQzEzYVdSblpYUnpMV2x1YkdsdVpTMW9aV2xuYUhRcE8xeHVJQ0FnSUd4cGJtVXRhR1ZwWjJoME9pQjJZWElvTFMxcWNDMTNhV1JuWlhSekxXbHViR2x1WlMxb1pXbG5hSFFwTzF4dWZWeHVYRzR1ZDJsa1oyVjBMWFJsZUhRZ2FXNXdkWFJiZEhsd1pUMWNJblJsZUhSY0lsMDZaR2x6WVdKc1pXUXNJQzUzYVdSblpYUXRkR1Y0ZENCcGJuQjFkRnQwZVhCbFBWd2liblZ0WW1WeVhDSmRPbVJwYzJGaWJHVmtMQ0F1ZDJsa1oyVjBMWFJsZUhSaGNtVmhJSFJsZUhSaGNtVmhPbVJwYzJGaWJHVmtJSHRjYmlBZ0lDQnZjR0ZqYVhSNU9pQjJZWElvTFMxcWNDMTNhV1JuWlhSekxXUnBjMkZpYkdWa0xXOXdZV05wZEhrcE8xeHVmVnh1WEc0dWQybGtaMlYwTFhSbGVIUWdhVzV3ZFhSYmRIbHdaVDFjSW5SbGVIUmNJbDBzSUM1M2FXUm5aWFF0ZEdWNGRDQnBibkIxZEZ0MGVYQmxQVndpYm5WdFltVnlYQ0pkTENBdWQybGtaMlYwTFhSbGVIUmhjbVZoSUhSbGVIUmhjbVZoSUh0Y2JpQWdJQ0JpYjNndGMybDZhVzVuT2lCaWIzSmtaWEl0WW05NE8xeHVJQ0FnSUdKdmNtUmxjam9nZG1GeUtDMHRhbkF0ZDJsa1oyVjBjeTFwYm5CMWRDMWliM0prWlhJdGQybGtkR2dwSUhOdmJHbGtJSFpoY2lndExXcHdMWGRwWkdkbGRITXRhVzV3ZFhRdFltOXlaR1Z5TFdOdmJHOXlLVHRjYmlBZ0lDQmlZV05yWjNKdmRXNWtMV052Ykc5eU9pQjJZWElvTFMxcWNDMTNhV1JuWlhSekxXbHVjSFYwTFdKaFkydG5jbTkxYm1RdFkyOXNiM0lwTzF4dUlDQWdJR052Ykc5eU9pQjJZWElvTFMxcWNDMTNhV1JuWlhSekxXbHVjSFYwTFdOdmJHOXlLVHRjYmlBZ0lDQm1iMjUwTFhOcGVtVTZJSFpoY2lndExXcHdMWGRwWkdkbGRITXRabTl1ZEMxemFYcGxLVHRjYmlBZ0lDQndZV1JrYVc1bk9pQjJZWElvTFMxcWNDMTNhV1JuWlhSekxXbHVjSFYwTFhCaFpHUnBibWNwSUdOaGJHTW9JSFpoY2lndExXcHdMWGRwWkdkbGRITXRhVzV3ZFhRdGNHRmtaR2x1WnlrZ0tpQWdNaUFwTzF4dUlDQWdJR1pzWlhndFozSnZkem9nTVR0Y2JpQWdJQ0J0YVc0dGQybGtkR2c2SURBN0lDOHFJRlJvYVhNZ2JXRnJaWE1nYVhRZ2NHOXpjMmxpYkdVZ1ptOXlJSFJvWlNCbWJHVjRZbTk0SUhSdklITm9jbWx1YXlCMGFHbHpJR2x1Y0hWMElDb3ZYRzRnSUNBZ1pteGxlQzF6YUhKcGJtczZJREU3WEc0Z0lDQWdiM1YwYkdsdVpUb2dibTl1WlNBaGFXMXdiM0owWVc1ME8xeHVmVnh1WEc0dWQybGtaMlYwTFhSbGVIUmhjbVZoSUhSbGVIUmhjbVZoSUh0Y2JpQWdJQ0JvWldsbmFIUTZJR2x1YUdWeWFYUTdYRzRnSUNBZ2QybGtkR2c2SUdsdWFHVnlhWFE3WEc1OVhHNWNiaTUzYVdSblpYUXRkR1Y0ZENCcGJuQjFkRHBtYjJOMWN5d2dMbmRwWkdkbGRDMTBaWGgwWVhKbFlTQjBaWGgwWVhKbFlUcG1iMk4xY3lCN1hHNGdJQ0FnWW05eVpHVnlMV052Ykc5eU9pQjJZWElvTFMxcWNDMTNhV1JuWlhSekxXbHVjSFYwTFdadlkzVnpMV0p2Y21SbGNpMWpiMnh2Y2lrN1hHNTlYRzVjYmk4cUlGZHBaR2RsZENCVGJHbGtaWElnS2k5Y2JseHVMbmRwWkdkbGRDMXpiR2xrWlhJZ0xuVnBMWE5zYVdSbGNpQjdYRzRnSUNBZ0x5b2dVMnhwWkdWeUlGUnlZV05ySUNvdlhHNGdJQ0FnWW05eVpHVnlPaUIyWVhJb0xTMXFjQzEzYVdSblpYUnpMWE5zYVdSbGNpMWliM0prWlhJdGQybGtkR2dwSUhOdmJHbGtJSFpoY2lndExXcHdMV3hoZVc5MWRDMWpiMnh2Y2pNcE8xeHVJQ0FnSUdKaFkydG5jbTkxYm1RNklIWmhjaWd0TFdwd0xXeGhlVzkxZEMxamIyeHZjak1wTzF4dUlDQWdJR0p2ZUMxemFYcHBibWM2SUdKdmNtUmxjaTFpYjNnN1hHNGdJQ0FnY0c5emFYUnBiMjQ2SUhKbGJHRjBhWFpsTzF4dUlDQWdJR0p2Y21SbGNpMXlZV1JwZFhNNklEQndlRHRjYm4xY2JseHVMbmRwWkdkbGRDMXpiR2xrWlhJZ0xuVnBMWE5zYVdSbGNpQXVkV2t0YzJ4cFpHVnlMV2hoYm1Sc1pTQjdYRzRnSUNBZ0x5b2dVMnhwWkdWeUlFaGhibVJzWlNBcUwxeHVJQ0FnSUc5MWRHeHBibVU2SUc1dmJtVWdJV2x0Y0c5eWRHRnVkRHNnTHlvZ1ptOWpkWE5sWkNCemJHbGtaWElnYUdGdVpHeGxjeUJoY21VZ1kyOXNiM0psWkNBdElITmxaU0JpWld4dmR5QXFMMXh1SUNBZ0lIQnZjMmwwYVc5dU9pQmhZbk52YkhWMFpUdGNiaUFnSUNCaVlXTnJaM0p2ZFc1a0xXTnZiRzl5T2lCMllYSW9MUzFxY0MxM2FXUm5aWFJ6TFhOc2FXUmxjaTFvWVc1a2JHVXRZbUZqYTJkeWIzVnVaQzFqYjJ4dmNpazdYRzRnSUNBZ1ltOXlaR1Z5T2lCMllYSW9MUzFxY0MxM2FXUm5aWFJ6TFhOc2FXUmxjaTFpYjNKa1pYSXRkMmxrZEdncElITnZiR2xrSUhaaGNpZ3RMV3B3TFhkcFpHZGxkSE10YzJ4cFpHVnlMV2hoYm1Sc1pTMWliM0prWlhJdFkyOXNiM0lwTzF4dUlDQWdJR0p2ZUMxemFYcHBibWM2SUdKdmNtUmxjaTFpYjNnN1hHNGdJQ0FnZWkxcGJtUmxlRG9nTVR0Y2JpQWdJQ0JpWVdOclozSnZkVzVrTFdsdFlXZGxPaUJ1YjI1bE95QXZLaUJQZG1WeWNtbGtaU0JxY1hWbGNua3RkV2tnS2k5Y2JuMWNibHh1THlvZ1QzWmxjbkpwWkdVZ2FuRjFaWEo1TFhWcElDb3ZYRzR1ZDJsa1oyVjBMWE5zYVdSbGNpQXVkV2t0YzJ4cFpHVnlJQzUxYVMxemJHbGtaWEl0YUdGdVpHeGxPbWh2ZG1WeUxDQXVkMmxrWjJWMExYTnNhV1JsY2lBdWRXa3RjMnhwWkdWeUlDNTFhUzF6Ykdsa1pYSXRhR0Z1Wkd4bE9tWnZZM1Z6SUh0Y2JpQWdJQ0JpWVdOclozSnZkVzVrTFdOdmJHOXlPaUIyWVhJb0xTMXFjQzEzYVdSblpYUnpMWE5zYVdSbGNpMWhZM1JwZG1VdGFHRnVaR3hsTFdOdmJHOXlLVHRjYmlBZ0lDQmliM0prWlhJNklIWmhjaWd0TFdwd0xYZHBaR2RsZEhNdGMyeHBaR1Z5TFdKdmNtUmxjaTEzYVdSMGFDa2djMjlzYVdRZ2RtRnlLQzB0YW5BdGQybGtaMlYwY3kxemJHbGtaWEl0WVdOMGFYWmxMV2hoYm1Sc1pTMWpiMnh2Y2lrN1hHNTlYRzVjYmk1M2FXUm5aWFF0YzJ4cFpHVnlJQzUxYVMxemJHbGtaWElnTG5WcExYTnNhV1JsY2kxb1lXNWtiR1U2WVdOMGFYWmxJSHRjYmlBZ0lDQmlZV05yWjNKdmRXNWtMV052Ykc5eU9pQjJZWElvTFMxcWNDMTNhV1JuWlhSekxYTnNhV1JsY2kxaFkzUnBkbVV0YUdGdVpHeGxMV052Ykc5eUtUdGNiaUFnSUNCaWIzSmtaWEl0WTI5c2IzSTZJSFpoY2lndExXcHdMWGRwWkdkbGRITXRjMnhwWkdWeUxXRmpkR2wyWlMxb1lXNWtiR1V0WTI5c2IzSXBPMXh1SUNBZ0lIb3RhVzVrWlhnNklESTdYRzRnSUNBZ2RISmhibk5tYjNKdE9pQnpZMkZzWlNneExqSXBPMXh1ZlZ4dVhHNHVkMmxrWjJWMExYTnNhV1JsY2lBZ0xuVnBMWE5zYVdSbGNpQXVkV2t0YzJ4cFpHVnlMWEpoYm1kbElIdGNiaUFnSUNBdktpQkpiblJsY25aaGJDQmlaWFIzWldWdUlIUm9aU0IwZDI4Z2MzQmxZMmxtYVdWa0lIWmhiSFZsSUc5bUlHRWdaRzkxWW14bElITnNhV1JsY2lBcUwxeHVJQ0FnSUhCdmMybDBhVzl1T2lCaFluTnZiSFYwWlR0Y2JpQWdJQ0JpWVdOclozSnZkVzVrT2lCMllYSW9MUzFxY0MxM2FXUm5aWFJ6TFhOc2FXUmxjaTFoWTNScGRtVXRhR0Z1Wkd4bExXTnZiRzl5S1R0Y2JpQWdJQ0I2TFdsdVpHVjRPaUF3TzF4dWZWeHVYRzR2S2lCVGFHRndaWE1nYjJZZ1UyeHBaR1Z5SUVoaGJtUnNaWE1nS2k5Y2JseHVMbmRwWkdkbGRDMW9jMnhwWkdWeUlDNTFhUzF6Ykdsa1pYSWdMblZwTFhOc2FXUmxjaTFvWVc1a2JHVWdlMXh1SUNBZ0lIZHBaSFJvT2lCMllYSW9MUzFxY0MxM2FXUm5aWFJ6TFhOc2FXUmxjaTFvWVc1a2JHVXRjMmw2WlNrN1hHNGdJQ0FnYUdWcFoyaDBPaUIyWVhJb0xTMXFjQzEzYVdSblpYUnpMWE5zYVdSbGNpMW9ZVzVrYkdVdGMybDZaU2s3WEc0Z0lDQWdiV0Z5WjJsdUxYUnZjRG9nWTJGc1l5Z29kbUZ5S0MwdGFuQXRkMmxrWjJWMGN5MXpiR2xrWlhJdGRISmhZMnN0ZEdocFkydHVaWE56S1NBdElIWmhjaWd0TFdwd0xYZHBaR2RsZEhNdGMyeHBaR1Z5TFdoaGJtUnNaUzF6YVhwbEtTa2dMeUF5SUMwZ2RtRnlLQzB0YW5BdGQybGtaMlYwY3kxemJHbGtaWEl0WW05eVpHVnlMWGRwWkhSb0tTazdYRzRnSUNBZ2JXRnlaMmx1TFd4bFpuUTZJR05oYkdNb2RtRnlLQzB0YW5BdGQybGtaMlYwY3kxemJHbGtaWEl0YUdGdVpHeGxMWE5wZW1VcElDOGdMVElnS3lCMllYSW9MUzFxY0MxM2FXUm5aWFJ6TFhOc2FXUmxjaTFpYjNKa1pYSXRkMmxrZEdncEtUdGNiaUFnSUNCaWIzSmtaWEl0Y21Ga2FYVnpPaUExTUNVN1hHNGdJQ0FnZEc5d09pQXdPMXh1ZlZ4dVhHNHVkMmxrWjJWMExYWnpiR2xrWlhJZ0xuVnBMWE5zYVdSbGNpQXVkV2t0YzJ4cFpHVnlMV2hoYm1Sc1pTQjdYRzRnSUNBZ2QybGtkR2c2SUhaaGNpZ3RMV3B3TFhkcFpHZGxkSE10YzJ4cFpHVnlMV2hoYm1Sc1pTMXphWHBsS1R0Y2JpQWdJQ0JvWldsbmFIUTZJSFpoY2lndExXcHdMWGRwWkdkbGRITXRjMnhwWkdWeUxXaGhibVJzWlMxemFYcGxLVHRjYmlBZ0lDQnRZWEpuYVc0dFltOTBkRzl0T2lCallXeGpLSFpoY2lndExXcHdMWGRwWkdkbGRITXRjMnhwWkdWeUxXaGhibVJzWlMxemFYcGxLU0F2SUMweUlDc2dkbUZ5S0MwdGFuQXRkMmxrWjJWMGN5MXpiR2xrWlhJdFltOXlaR1Z5TFhkcFpIUm9LU2s3WEc0Z0lDQWdiV0Z5WjJsdUxXeGxablE2SUdOaGJHTW9LSFpoY2lndExXcHdMWGRwWkdkbGRITXRjMnhwWkdWeUxYUnlZV05yTFhSb2FXTnJibVZ6Y3lrZ0xTQjJZWElvTFMxcWNDMTNhV1JuWlhSekxYTnNhV1JsY2kxb1lXNWtiR1V0YzJsNlpTa3BJQzhnTWlBdElIWmhjaWd0TFdwd0xYZHBaR2RsZEhNdGMyeHBaR1Z5TFdKdmNtUmxjaTEzYVdSMGFDa3BPMXh1SUNBZ0lHSnZjbVJsY2kxeVlXUnBkWE02SURVd0pUdGNiaUFnSUNCc1pXWjBPaUF3TzF4dWZWeHVYRzR1ZDJsa1oyVjBMV2h6Ykdsa1pYSWdMblZwTFhOc2FXUmxjaUF1ZFdrdGMyeHBaR1Z5TFhKaGJtZGxJSHRjYmlBZ0lDQm9aV2xuYUhRNklHTmhiR01vSUhaaGNpZ3RMV3B3TFhkcFpHZGxkSE10YzJ4cFpHVnlMWFJ5WVdOckxYUm9hV05yYm1WemN5a2dLaUF5SUNrN1hHNGdJQ0FnYldGeVoybHVMWFJ2Y0RvZ1kyRnNZeWdvZG1GeUtDMHRhbkF0ZDJsa1oyVjBjeTF6Ykdsa1pYSXRkSEpoWTJzdGRHaHBZMnR1WlhOektTQXRJSFpoY2lndExXcHdMWGRwWkdkbGRITXRjMnhwWkdWeUxYUnlZV05yTFhSb2FXTnJibVZ6Y3lrZ0tpQXlJQ2tnTHlBeUlDMGdkbUZ5S0MwdGFuQXRkMmxrWjJWMGN5MXpiR2xrWlhJdFltOXlaR1Z5TFhkcFpIUm9LU2s3WEc1OVhHNWNiaTUzYVdSblpYUXRkbk5zYVdSbGNpQXVkV2t0YzJ4cFpHVnlJQzUxYVMxemJHbGtaWEl0Y21GdVoyVWdlMXh1SUNBZ0lIZHBaSFJvT2lCallXeGpLQ0IyWVhJb0xTMXFjQzEzYVdSblpYUnpMWE5zYVdSbGNpMTBjbUZqYXkxMGFHbGphMjVsYzNNcElDb2dNaUFwTzF4dUlDQWdJRzFoY21kcGJpMXNaV1owT2lCallXeGpLQ2gyWVhJb0xTMXFjQzEzYVdSblpYUnpMWE5zYVdSbGNpMTBjbUZqYXkxMGFHbGphMjVsYzNNcElDMGdkbUZ5S0MwdGFuQXRkMmxrWjJWMGN5MXpiR2xrWlhJdGRISmhZMnN0ZEdocFkydHVaWE56S1NBcUlESWdLU0F2SURJZ0xTQjJZWElvTFMxcWNDMTNhV1JuWlhSekxYTnNhV1JsY2kxaWIzSmtaWEl0ZDJsa2RHZ3BLVHRjYm4xY2JseHVMeW9nU0c5eWFYcHZiblJoYkNCVGJHbGtaWElnS2k5Y2JseHVMbmRwWkdkbGRDMW9jMnhwWkdWeUlIdGNiaUFnSUNCM2FXUjBhRG9nZG1GeUtDMHRhbkF0ZDJsa1oyVjBjeTFwYm14cGJtVXRkMmxrZEdncE8xeHVJQ0FnSUdobGFXZG9kRG9nZG1GeUtDMHRhbkF0ZDJsa1oyVjBjeTFwYm14cGJtVXRhR1ZwWjJoMEtUdGNiaUFnSUNCc2FXNWxMV2hsYVdkb2REb2dkbUZ5S0MwdGFuQXRkMmxrWjJWMGN5MXBibXhwYm1VdGFHVnBaMmgwS1R0Y2JseHVJQ0FnSUM4cUlFOTJaWEp5YVdSbElIUm9aU0JoYkdsbmJpMXBkR1Z0Y3lCaVlYTmxiR2x1WlM0Z1ZHaHBjeUIzWVhrc0lIUm9aU0JrWlhOamNtbHdkR2x2YmlCaGJtUWdjbVZoWkc5MWRGeHVJQ0FnSUhOMGFXeHNJSE5sWlcwZ2RHOGdZV3hwWjI0Z2RHaGxhWElnWW1GelpXeHBibVVnY0hKdmNHVnliSGtzSUdGdVpDQjNaU0JrYjI0bmRDQm9ZWFpsSUhSdklHaGhkbVZjYmlBZ0lDQmhiR2xuYmkxelpXeG1PaUJ6ZEhKbGRHTm9JR2x1SUhSb1pTQXVjMnhwWkdWeUxXTnZiblJoYVc1bGNpNGdLaTljYmlBZ0lDQmhiR2xuYmkxcGRHVnRjem9nWTJWdWRHVnlPMXh1ZlZ4dVhHNHVkMmxrWjJWMGN5MXpiR2xrWlhJZ0xuTnNhV1JsY2kxamIyNTBZV2x1WlhJZ2UxeHVJQ0FnSUc5MlpYSm1iRzkzT2lCMmFYTnBZbXhsTzF4dWZWeHVYRzR1ZDJsa1oyVjBMV2h6Ykdsa1pYSWdMbk5zYVdSbGNpMWpiMjUwWVdsdVpYSWdlMXh1SUNBZ0lHaGxhV2RvZERvZ2RtRnlLQzB0YW5BdGQybGtaMlYwY3kxcGJteHBibVV0YUdWcFoyaDBLVHRjYmlBZ0lDQnRZWEpuYVc0dGJHVm1kRG9nWTJGc1l5aDJZWElvTFMxcWNDMTNhV1JuWlhSekxYTnNhV1JsY2kxb1lXNWtiR1V0YzJsNlpTa2dMeUF5SUMwZ01pQXFJSFpoY2lndExXcHdMWGRwWkdkbGRITXRjMnhwWkdWeUxXSnZjbVJsY2kxM2FXUjBhQ2twTzF4dUlDQWdJRzFoY21kcGJpMXlhV2RvZERvZ1kyRnNZeWgyWVhJb0xTMXFjQzEzYVdSblpYUnpMWE5zYVdSbGNpMW9ZVzVrYkdVdGMybDZaU2tnTHlBeUlDMGdNaUFxSUhaaGNpZ3RMV3B3TFhkcFpHZGxkSE10YzJ4cFpHVnlMV0p2Y21SbGNpMTNhV1IwYUNrcE8xeHVJQ0FnSUdac1pYZzZJREVnTVNCMllYSW9MUzFxY0MxM2FXUm5aWFJ6TFdsdWJHbHVaUzEzYVdSMGFDMXphRzl5ZENrN1hHNTlYRzVjYmk1M2FXUm5aWFF0YUhOc2FXUmxjaUF1ZFdrdGMyeHBaR1Z5SUh0Y2JpQWdJQ0F2S2lCSmJtNWxjaXdnYVc1MmFYTnBZbXhsSUhOc2FXUmxJR1JwZGlBcUwxeHVJQ0FnSUdobGFXZG9kRG9nZG1GeUtDMHRhbkF0ZDJsa1oyVjBjeTF6Ykdsa1pYSXRkSEpoWTJzdGRHaHBZMnR1WlhOektUdGNiaUFnSUNCdFlYSm5hVzR0ZEc5d09pQmpZV3hqS0NoMllYSW9MUzFxY0MxM2FXUm5aWFJ6TFdsdWJHbHVaUzFvWldsbmFIUXBJQzBnZG1GeUtDMHRhbkF0ZDJsa1oyVjBjeTF6Ykdsa1pYSXRkSEpoWTJzdGRHaHBZMnR1WlhOektTa2dMeUF5S1R0Y2JpQWdJQ0IzYVdSMGFEb2dNVEF3SlR0Y2JuMWNibHh1THlvZ1ZtVnlkR2xqWVd3Z1UyeHBaR1Z5SUNvdlhHNWNiaTUzYVdSblpYUXRkbUp2ZUNBdWQybGtaMlYwTFd4aFltVnNJSHRjYmlBZ0lDQm9aV2xuYUhRNklIWmhjaWd0TFdwd0xYZHBaR2RsZEhNdGFXNXNhVzVsTFdobGFXZG9kQ2s3WEc0Z0lDQWdiR2x1WlMxb1pXbG5hSFE2SUhaaGNpZ3RMV3B3TFhkcFpHZGxkSE10YVc1c2FXNWxMV2hsYVdkb2RDazdYRzU5WEc1Y2JpNTNhV1JuWlhRdGRuTnNhV1JsY2lCN1hHNGdJQ0FnTHlvZ1ZtVnlkR2xqWVd3Z1UyeHBaR1Z5SUNvdlhHNGdJQ0FnYUdWcFoyaDBPaUIyWVhJb0xTMXFjQzEzYVdSblpYUnpMWFpsY25ScFkyRnNMV2hsYVdkb2RDazdYRzRnSUNBZ2QybGtkR2c2SUhaaGNpZ3RMV3B3TFhkcFpHZGxkSE10YVc1c2FXNWxMWGRwWkhSb0xYUnBibmtwTzF4dWZWeHVYRzR1ZDJsa1oyVjBMWFp6Ykdsa1pYSWdMbk5zYVdSbGNpMWpiMjUwWVdsdVpYSWdlMXh1SUNBZ0lHWnNaWGc2SURFZ01TQjJZWElvTFMxcWNDMTNhV1JuWlhSekxXbHViR2x1WlMxM2FXUjBhQzF6YUc5eWRDazdYRzRnSUNBZ2JXRnlaMmx1TFd4bFpuUTZJR0YxZEc4N1hHNGdJQ0FnYldGeVoybHVMWEpwWjJoME9pQmhkWFJ2TzF4dUlDQWdJRzFoY21kcGJpMWliM1IwYjIwNklHTmhiR01vZG1GeUtDMHRhbkF0ZDJsa1oyVjBjeTF6Ykdsa1pYSXRhR0Z1Wkd4bExYTnBlbVVwSUM4Z01pQXRJRElnS2lCMllYSW9MUzFxY0MxM2FXUm5aWFJ6TFhOc2FXUmxjaTFpYjNKa1pYSXRkMmxrZEdncEtUdGNiaUFnSUNCdFlYSm5hVzR0ZEc5d09pQmpZV3hqS0haaGNpZ3RMV3B3TFhkcFpHZGxkSE10YzJ4cFpHVnlMV2hoYm1Sc1pTMXphWHBsS1NBdklESWdMU0F5SUNvZ2RtRnlLQzB0YW5BdGQybGtaMlYwY3kxemJHbGtaWEl0WW05eVpHVnlMWGRwWkhSb0tTazdYRzRnSUNBZ1pHbHpjR3hoZVRvZ1pteGxlRHRjYmlBZ0lDQm1iR1Y0TFdScGNtVmpkR2x2YmpvZ1kyOXNkVzF1TzF4dWZWeHVYRzR1ZDJsa1oyVjBMWFp6Ykdsa1pYSWdMblZwTFhOc2FXUmxjaTEyWlhKMGFXTmhiQ0I3WEc0Z0lDQWdMeW9nU1c1dVpYSXNJR2x1ZG1semFXSnNaU0J6Ykdsa1pTQmthWFlnS2k5Y2JpQWdJQ0IzYVdSMGFEb2dkbUZ5S0MwdGFuQXRkMmxrWjJWMGN5MXpiR2xrWlhJdGRISmhZMnN0ZEdocFkydHVaWE56S1R0Y2JpQWdJQ0JtYkdWNExXZHliM2M2SURFN1hHNGdJQ0FnYldGeVoybHVMV3hsWm5RNklHRjFkRzg3WEc0Z0lDQWdiV0Z5WjJsdUxYSnBaMmgwT2lCaGRYUnZPMXh1ZlZ4dVhHNHZLaUJYYVdSblpYUWdVSEp2WjNKbGMzTWdVM1I1YkdsdVp5QXFMMXh1WEc0dWNISnZaM0psYzNNdFltRnlJSHRjYmlBZ0lDQXRkMlZpYTJsMExYUnlZVzV6YVhScGIyNDZJRzV2Ym1VN1hHNGdJQ0FnTFcxdmVpMTBjbUZ1YzJsMGFXOXVPaUJ1YjI1bE8xeHVJQ0FnSUMxdGN5MTBjbUZ1YzJsMGFXOXVPaUJ1YjI1bE8xeHVJQ0FnSUMxdkxYUnlZVzV6YVhScGIyNDZJRzV2Ym1VN1hHNGdJQ0FnZEhKaGJuTnBkR2x2YmpvZ2JtOXVaVHRjYm4xY2JseHVMbkJ5YjJkeVpYTnpMV0poY2lCN1hHNGdJQ0FnYUdWcFoyaDBPaUIyWVhJb0xTMXFjQzEzYVdSblpYUnpMV2x1YkdsdVpTMW9aV2xuYUhRcE8xeHVmVnh1WEc0dWNISnZaM0psYzNNdFltRnlJSHRjYmlBZ0lDQmlZV05yWjNKdmRXNWtMV052Ykc5eU9pQjJZWElvTFMxcWNDMWljbUZ1WkMxamIyeHZjakVwTzF4dWZWeHVYRzR1Y0hKdlozSmxjM010WW1GeUxYTjFZMk5sYzNNZ2UxeHVJQ0FnSUdKaFkydG5jbTkxYm1RdFkyOXNiM0k2SUhaaGNpZ3RMV3B3TFhOMVkyTmxjM010WTI5c2IzSXhLVHRjYm4xY2JseHVMbkJ5YjJkeVpYTnpMV0poY2kxcGJtWnZJSHRjYmlBZ0lDQmlZV05yWjNKdmRXNWtMV052Ykc5eU9pQjJZWElvTFMxcWNDMXBibVp2TFdOdmJHOXlNU2s3WEc1OVhHNWNiaTV3Y205bmNtVnpjeTFpWVhJdGQyRnlibWx1WnlCN1hHNGdJQ0FnWW1GamEyZHliM1Z1WkMxamIyeHZjam9nZG1GeUtDMHRhbkF0ZDJGeWJpMWpiMnh2Y2pFcE8xeHVmVnh1WEc0dWNISnZaM0psYzNNdFltRnlMV1JoYm1kbGNpQjdYRzRnSUNBZ1ltRmphMmR5YjNWdVpDMWpiMnh2Y2pvZ2RtRnlLQzB0YW5BdFpYSnliM0l0WTI5c2IzSXhLVHRjYm4xY2JseHVMbkJ5YjJkeVpYTnpJSHRjYmlBZ0lDQmlZV05yWjNKdmRXNWtMV052Ykc5eU9pQjJZWElvTFMxcWNDMXNZWGx2ZFhRdFkyOXNiM0l5S1R0Y2JpQWdJQ0JpYjNKa1pYSTZJRzV2Ym1VN1hHNGdJQ0FnWW05NExYTm9ZV1J2ZHpvZ2JtOXVaVHRjYm4xY2JseHVMeW9nU0c5eWFYTnZiblJoYkNCUWNtOW5jbVZ6Y3lBcUwxeHVYRzR1ZDJsa1oyVjBMV2h3Y205bmNtVnpjeUI3WEc0Z0lDQWdMeW9nVUhKdlozSmxjM01nUW1GeUlDb3ZYRzRnSUNBZ2FHVnBaMmgwT2lCMllYSW9MUzFxY0MxM2FXUm5aWFJ6TFdsdWJHbHVaUzFvWldsbmFIUXBPMXh1SUNBZ0lHeHBibVV0YUdWcFoyaDBPaUIyWVhJb0xTMXFjQzEzYVdSblpYUnpMV2x1YkdsdVpTMW9aV2xuYUhRcE8xeHVJQ0FnSUhkcFpIUm9PaUIyWVhJb0xTMXFjQzEzYVdSblpYUnpMV2x1YkdsdVpTMTNhV1IwYUNrN1hHNGdJQ0FnWVd4cFoyNHRhWFJsYlhNNklHTmxiblJsY2p0Y2JseHVmVnh1WEc0dWQybGtaMlYwTFdod2NtOW5jbVZ6Y3lBdWNISnZaM0psYzNNZ2UxeHVJQ0FnSUdac1pYZ3RaM0p2ZHpvZ01UdGNiaUFnSUNCdFlYSm5hVzR0ZEc5d09pQjJZWElvTFMxcWNDMTNhV1JuWlhSekxXbHVjSFYwTFhCaFpHUnBibWNwTzF4dUlDQWdJRzFoY21kcGJpMWliM1IwYjIwNklIWmhjaWd0TFdwd0xYZHBaR2RsZEhNdGFXNXdkWFF0Y0dGa1pHbHVaeWs3WEc0Z0lDQWdZV3hwWjI0dGMyVnNaam9nYzNSeVpYUmphRHRjYmlBZ0lDQXZLaUJQZG1WeWNtbGtaU0JpYjI5MGMzUnlZWEFnYzNSNWJHVWdLaTljYmlBZ0lDQm9aV2xuYUhRNklHbHVhWFJwWVd3N1hHNTlYRzVjYmk4cUlGWmxjblJwWTJGc0lGQnliMmR5WlhOeklDb3ZYRzVjYmk1M2FXUm5aWFF0ZG5CeWIyZHlaWE56SUh0Y2JpQWdJQ0JvWldsbmFIUTZJSFpoY2lndExXcHdMWGRwWkdkbGRITXRkbVZ5ZEdsallXd3RhR1ZwWjJoMEtUdGNiaUFnSUNCM2FXUjBhRG9nZG1GeUtDMHRhbkF0ZDJsa1oyVjBjeTFwYm14cGJtVXRkMmxrZEdndGRHbHVlU2s3WEc1OVhHNWNiaTUzYVdSblpYUXRkbkJ5YjJkeVpYTnpJQzV3Y205bmNtVnpjeUI3WEc0Z0lDQWdabXhsZUMxbmNtOTNPaUF4TzF4dUlDQWdJSGRwWkhSb09pQjJZWElvTFMxcWNDMTNhV1JuWlhSekxYQnliMmR5WlhOekxYUm9hV05yYm1WemN5azdYRzRnSUNBZ2JXRnlaMmx1TFd4bFpuUTZJR0YxZEc4N1hHNGdJQ0FnYldGeVoybHVMWEpwWjJoME9pQmhkWFJ2TzF4dUlDQWdJRzFoY21kcGJpMWliM1IwYjIwNklEQTdYRzU5WEc1Y2JpOHFJRk5sYkdWamRDQlhhV1JuWlhRZ1UzUjViR2x1WnlBcUwxeHVYRzR1ZDJsa1oyVjBMV1J5YjNCa2IzZHVJSHRjYmlBZ0lDQm9aV2xuYUhRNklIWmhjaWd0TFdwd0xYZHBaR2RsZEhNdGFXNXNhVzVsTFdobGFXZG9kQ2s3WEc0Z0lDQWdkMmxrZEdnNklIWmhjaWd0TFdwd0xYZHBaR2RsZEhNdGFXNXNhVzVsTFhkcFpIUm9LVHRjYmlBZ0lDQnNhVzVsTFdobGFXZG9kRG9nZG1GeUtDMHRhbkF0ZDJsa1oyVjBjeTFwYm14cGJtVXRhR1ZwWjJoMEtUdGNibjFjYmx4dUxuZHBaR2RsZEMxa2NtOXdaRzkzYmlBK0lITmxiR1ZqZENCN1hHNGdJQ0FnY0dGa1pHbHVaeTF5YVdkb2REb2dNakJ3ZUR0Y2JpQWdJQ0JpYjNKa1pYSTZJSFpoY2lndExXcHdMWGRwWkdkbGRITXRhVzV3ZFhRdFltOXlaR1Z5TFhkcFpIUm9LU0J6YjJ4cFpDQjJZWElvTFMxcWNDMTNhV1JuWlhSekxXbHVjSFYwTFdKdmNtUmxjaTFqYjJ4dmNpazdYRzRnSUNBZ1ltOXlaR1Z5TFhKaFpHbDFjem9nTUR0Y2JpQWdJQ0JvWldsbmFIUTZJR2x1YUdWeWFYUTdYRzRnSUNBZ1pteGxlRG9nTVNBeElIWmhjaWd0TFdwd0xYZHBaR2RsZEhNdGFXNXNhVzVsTFhkcFpIUm9MWE5vYjNKMEtUdGNiaUFnSUNCdGFXNHRkMmxrZEdnNklEQTdJQzhxSUZSb2FYTWdiV0ZyWlhNZ2FYUWdjRzl6YzJsaWJHVWdabTl5SUhSb1pTQm1iR1Y0WW05NElIUnZJSE5vY21sdWF5QjBhR2x6SUdsdWNIVjBJQ292WEc0Z0lDQWdZbTk0TFhOcGVtbHVaem9nWW05eVpHVnlMV0p2ZUR0Y2JpQWdJQ0J2ZFhSc2FXNWxPaUJ1YjI1bElDRnBiWEJ2Y25SaGJuUTdYRzRnSUNBZ1ltOTRMWE5vWVdSdmR6b2dibTl1WlR0Y2JpQWdJQ0JpWVdOclozSnZkVzVrTFdOdmJHOXlPaUIyWVhJb0xTMXFjQzEzYVdSblpYUnpMV2x1Y0hWMExXSmhZMnRuY205MWJtUXRZMjlzYjNJcE8xeHVJQ0FnSUdOdmJHOXlPaUIyWVhJb0xTMXFjQzEzYVdSblpYUnpMV2x1Y0hWMExXTnZiRzl5S1R0Y2JpQWdJQ0JtYjI1MExYTnBlbVU2SUhaaGNpZ3RMV3B3TFhkcFpHZGxkSE10Wm05dWRDMXphWHBsS1R0Y2JpQWdJQ0IyWlhKMGFXTmhiQzFoYkdsbmJqb2dkRzl3TzF4dUlDQWdJSEJoWkdScGJtY3RiR1ZtZERvZ1kyRnNZeWdnZG1GeUtDMHRhbkF0ZDJsa1oyVjBjeTFwYm5CMWRDMXdZV1JrYVc1bktTQXFJRElwTzF4dVhIUmhjSEJsWVhKaGJtTmxPaUJ1YjI1bE8xeHVYSFF0ZDJWaWEybDBMV0Z3Y0dWaGNtRnVZMlU2SUc1dmJtVTdYRzVjZEMxdGIzb3RZWEJ3WldGeVlXNWpaVG9nYm05dVpUdGNiaUFnSUNCaVlXTnJaM0p2ZFc1a0xYSmxjR1ZoZERvZ2JtOHRjbVZ3WldGME8xeHVYSFJpWVdOclozSnZkVzVrTFhOcGVtVTZJREl3Y0hnN1hHNWNkR0poWTJ0bmNtOTFibVF0Y0c5emFYUnBiMjQ2SUhKcFoyaDBJR05sYm5SbGNqdGNiaUFnSUNCaVlXTnJaM0p2ZFc1a0xXbHRZV2RsT2lCMllYSW9MUzFxY0MxM2FXUm5aWFJ6TFdSeWIzQmtiM2R1TFdGeWNtOTNLVHRjYm4xY2JpNTNhV1JuWlhRdFpISnZjR1J2ZDI0Z1BpQnpaV3hsWTNRNlptOWpkWE1nZTF4dUlDQWdJR0p2Y21SbGNpMWpiMnh2Y2pvZ2RtRnlLQzB0YW5BdGQybGtaMlYwY3kxcGJuQjFkQzFtYjJOMWN5MWliM0prWlhJdFkyOXNiM0lwTzF4dWZWeHVYRzR1ZDJsa1oyVjBMV1J5YjNCa2IzZHVJRDRnYzJWc1pXTjBPbVJwYzJGaWJHVmtJSHRjYmlBZ0lDQnZjR0ZqYVhSNU9pQjJZWElvTFMxcWNDMTNhV1JuWlhSekxXUnBjMkZpYkdWa0xXOXdZV05wZEhrcE8xeHVmVnh1WEc0dktpQlVieUJrYVhOaFlteGxJSFJvWlNCa2IzUjBaV1FnWW05eVpHVnlJR2x1SUVacGNtVm1iM2dnWVhKdmRXNWtJSE5sYkdWamRDQmpiMjUwY205c2N5NWNiaUFnSUZObFpTQm9kSFJ3T2k4dmMzUmhZMnR2ZG1WeVpteHZkeTVqYjIwdllTOHhPRGcxTXpBd01pQXFMMXh1TG5kcFpHZGxkQzFrY205d1pHOTNiaUErSUhObGJHVmpkRG90Ylc5NkxXWnZZM1Z6Y21sdVp5QjdYRzRnSUNBZ1kyOXNiM0k2SUhSeVlXNXpjR0Z5Wlc1ME8xeHVJQ0FnSUhSbGVIUXRjMmhoWkc5M09pQXdJREFnTUNBak1EQXdPMXh1ZlZ4dVhHNHZLaUJUWld4bFkzUWdZVzVrSUZObGJHVmpkRTExYkhScGNHeGxJQ292WEc1Y2JpNTNhV1JuWlhRdGMyVnNaV04wSUh0Y2JpQWdJQ0IzYVdSMGFEb2dkbUZ5S0MwdGFuQXRkMmxrWjJWMGN5MXBibXhwYm1VdGQybGtkR2dwTzF4dUlDQWdJR3hwYm1VdGFHVnBaMmgwT2lCMllYSW9MUzFxY0MxM2FXUm5aWFJ6TFdsdWJHbHVaUzFvWldsbmFIUXBPMXh1WEc0Z0lDQWdMeW9nUW1WallYVnpaU0JHYVhKbFptOTRJR1JsWm1sdVpYTWdkR2hsSUdKaGMyVnNhVzVsSUc5bUlHRWdjMlZzWldOMElHRnpJSFJvWlNCaWIzUjBiMjBnYjJZZ2RHaGxYRzRnSUNBZ1kyOXVkSEp2YkN3Z2QyVWdZV3hwWjI0Z2RHaGxJR1Z1ZEdseVpTQmpiMjUwY205c0lIUnZJSFJvWlNCMGIzQWdZVzVrSUdGa1pDQndZV1JrYVc1bklIUnZJSFJvWlZ4dUlDQWdJSE5sYkdWamRDQjBieUJuWlhRZ1lXNGdZWEJ3Y205NGFXMWhkR1VnWm1seWMzUWdiR2x1WlNCaVlYTmxiR2x1WlNCaGJHbG5ibTFsYm5RdUlDb3ZYRzRnSUNBZ1lXeHBaMjR0YVhSbGJYTTZJR1pzWlhndGMzUmhjblE3WEc1OVhHNWNiaTUzYVdSblpYUXRjMlZzWldOMElENGdjMlZzWldOMElIdGNiaUFnSUNCaWIzSmtaWEk2SUhaaGNpZ3RMV3B3TFhkcFpHZGxkSE10YVc1d2RYUXRZbTl5WkdWeUxYZHBaSFJvS1NCemIyeHBaQ0IyWVhJb0xTMXFjQzEzYVdSblpYUnpMV2x1Y0hWMExXSnZjbVJsY2kxamIyeHZjaWs3WEc0Z0lDQWdZbUZqYTJkeWIzVnVaQzFqYjJ4dmNqb2dkbUZ5S0MwdGFuQXRkMmxrWjJWMGN5MXBibkIxZEMxaVlXTnJaM0p2ZFc1a0xXTnZiRzl5S1R0Y2JpQWdJQ0JqYjJ4dmNqb2dkbUZ5S0MwdGFuQXRkMmxrWjJWMGN5MXBibkIxZEMxamIyeHZjaWs3WEc0Z0lDQWdabTl1ZEMxemFYcGxPaUIyWVhJb0xTMXFjQzEzYVdSblpYUnpMV1p2Ym5RdGMybDZaU2s3WEc0Z0lDQWdabXhsZURvZ01TQXhJSFpoY2lndExXcHdMWGRwWkdkbGRITXRhVzVzYVc1bExYZHBaSFJvTFhOb2IzSjBLVHRjYmlBZ0lDQnZkWFJzYVc1bE9pQnViMjVsSUNGcGJYQnZjblJoYm5RN1hHNGdJQ0FnYjNabGNtWnNiM2M2SUdGMWRHODdYRzRnSUNBZ2FHVnBaMmgwT2lCcGJtaGxjbWwwTzF4dVhHNGdJQ0FnTHlvZ1FtVmpZWFZ6WlNCR2FYSmxabTk0SUdSbFptbHVaWE1nZEdobElHSmhjMlZzYVc1bElHOW1JR0VnYzJWc1pXTjBJR0Z6SUhSb1pTQmliM1IwYjIwZ2IyWWdkR2hsWEc0Z0lDQWdZMjl1ZEhKdmJDd2dkMlVnWVd4cFoyNGdkR2hsSUdWdWRHbHlaU0JqYjI1MGNtOXNJSFJ2SUhSb1pTQjBiM0FnWVc1a0lHRmtaQ0J3WVdSa2FXNW5JSFJ2SUhSb1pWeHVJQ0FnSUhObGJHVmpkQ0IwYnlCblpYUWdZVzRnWVhCd2NtOTRhVzFoZEdVZ1ptbHljM1FnYkdsdVpTQmlZWE5sYkdsdVpTQmhiR2xuYm0xbGJuUXVJQ292WEc0Z0lDQWdjR0ZrWkdsdVp5MTBiM0E2SURWd2VEdGNibjFjYmx4dUxuZHBaR2RsZEMxelpXeGxZM1FnUGlCelpXeGxZM1E2Wm05amRYTWdlMXh1SUNBZ0lHSnZjbVJsY2kxamIyeHZjam9nZG1GeUtDMHRhbkF0ZDJsa1oyVjBjeTFwYm5CMWRDMW1iMk4xY3kxaWIzSmtaWEl0WTI5c2IzSXBPMXh1ZlZ4dVhHNHVkMmxuWlhRdGMyVnNaV04wSUQ0Z2MyVnNaV04wSUQ0Z2IzQjBhVzl1SUh0Y2JpQWdJQ0J3WVdSa2FXNW5MV3hsWm5RNklIWmhjaWd0TFdwd0xYZHBaR2RsZEhNdGFXNXdkWFF0Y0dGa1pHbHVaeWs3WEc0Z0lDQWdiR2x1WlMxb1pXbG5hSFE2SUhaaGNpZ3RMV3B3TFhkcFpHZGxkSE10YVc1c2FXNWxMV2hsYVdkb2RDazdYRzRnSUNBZ0x5b2diR2x1WlMxb1pXbG5hSFFnWkc5bGMyNG5kQ0IzYjNKcklHOXVJSE52YldVZ1luSnZkM05sY25NZ1ptOXlJSE5sYkdWamRDQnZjSFJwYjI1eklDb3ZYRzRnSUNBZ2NHRmtaR2x1WnkxMGIzQTZJR05oYkdNb2RtRnlLQzB0YW5BdGQybGtaMlYwY3kxcGJteHBibVV0YUdWcFoyaDBLUzEyWVhJb0xTMXFjQzEzYVdSblpYUnpMV1p2Ym5RdGMybDZaU2t2TWlrN1hHNGdJQ0FnY0dGa1pHbHVaeTFpYjNSMGIyMDZJR05oYkdNb2RtRnlLQzB0YW5BdGQybGtaMlYwY3kxcGJteHBibVV0YUdWcFoyaDBLUzEyWVhJb0xTMXFjQzEzYVdSblpYUnpMV1p2Ym5RdGMybDZaU2t2TWlrN1hHNTlYRzVjYmx4dVhHNHZLaUJVYjJkbmJHVWdRblYwZEc5dWN5QlRkSGxzYVc1bklDb3ZYRzVjYmk1M2FXUm5aWFF0ZEc5bloyeGxMV0oxZEhSdmJuTWdlMXh1SUNBZ0lHeHBibVV0YUdWcFoyaDBPaUIyWVhJb0xTMXFjQzEzYVdSblpYUnpMV2x1YkdsdVpTMW9aV2xuYUhRcE8xeHVmVnh1WEc0dWQybGtaMlYwTFhSdloyZHNaUzFpZFhSMGIyNXpJQzUzYVdSblpYUXRkRzluWjJ4bExXSjFkSFJ2YmlCN1hHNGdJQ0FnYldGeVoybHVMV3hsWm5RNklIWmhjaWd0TFdwd0xYZHBaR2RsZEhNdGJXRnlaMmx1S1R0Y2JpQWdJQ0J0WVhKbmFXNHRjbWxuYUhRNklIWmhjaWd0TFdwd0xYZHBaR2RsZEhNdGJXRnlaMmx1S1R0Y2JuMWNibHh1TG5kcFpHZGxkQzEwYjJkbmJHVXRZblYwZEc5dWN5QXVhblZ3ZVhSbGNpMWlkWFIwYjI0NlpHbHpZV0pzWldRZ2UxeHVJQ0FnSUc5d1lXTnBkSGs2SUhaaGNpZ3RMV3B3TFhkcFpHZGxkSE10WkdsellXSnNaV1F0YjNCaFkybDBlU2s3WEc1OVhHNWNiaThxSUZKaFpHbHZJRUoxZEhSdmJuTWdVM1I1YkdsdVp5QXFMMXh1WEc0dWQybGtaMlYwTFhKaFpHbHZJSHRjYmlBZ0lDQjNhV1IwYURvZ2RtRnlLQzB0YW5BdGQybGtaMlYwY3kxcGJteHBibVV0ZDJsa2RHZ3BPMXh1SUNBZ0lHeHBibVV0YUdWcFoyaDBPaUIyWVhJb0xTMXFjQzEzYVdSblpYUnpMV2x1YkdsdVpTMW9aV2xuYUhRcE8xeHVmVnh1WEc0dWQybGtaMlYwTFhKaFpHbHZMV0p2ZUNCN1hHNGdJQ0FnWkdsemNHeGhlVG9nWm14bGVEdGNiaUFnSUNCbWJHVjRMV1JwY21WamRHbHZiam9nWTI5c2RXMXVPMXh1SUNBZ0lHRnNhV2R1TFdsMFpXMXpPaUJ6ZEhKbGRHTm9PMXh1SUNBZ0lHSnZlQzF6YVhwcGJtYzZJR0p2Y21SbGNpMWliM2c3WEc0Z0lDQWdabXhsZUMxbmNtOTNPaUF4TzF4dUlDQWdJRzFoY21kcGJpMWliM1IwYjIwNklIWmhjaWd0TFdwd0xYZHBaR2RsZEhNdGNtRmthVzh0YVhSbGJTMW9aV2xuYUhRdFlXUnFkWE4wYldWdWRDazdYRzU5WEc1Y2JpNTNhV1JuWlhRdGNtRmthVzh0WW05NElHeGhZbVZzSUh0Y2JpQWdJQ0JvWldsbmFIUTZJSFpoY2lndExXcHdMWGRwWkdkbGRITXRjbUZrYVc4dGFYUmxiUzFvWldsbmFIUXBPMXh1SUNBZ0lHeHBibVV0YUdWcFoyaDBPaUIyWVhJb0xTMXFjQzEzYVdSblpYUnpMWEpoWkdsdkxXbDBaVzB0YUdWcFoyaDBLVHRjYmlBZ0lDQm1iMjUwTFhOcGVtVTZJSFpoY2lndExXcHdMWGRwWkdkbGRITXRabTl1ZEMxemFYcGxLVHRjYm4xY2JseHVMbmRwWkdkbGRDMXlZV1JwYnkxaWIzZ2dhVzV3ZFhRZ2UxeHVJQ0FnSUdobGFXZG9kRG9nZG1GeUtDMHRhbkF0ZDJsa1oyVjBjeTF5WVdScGJ5MXBkR1Z0TFdobGFXZG9kQ2s3WEc0Z0lDQWdiR2x1WlMxb1pXbG5hSFE2SUhaaGNpZ3RMV3B3TFhkcFpHZGxkSE10Y21Ga2FXOHRhWFJsYlMxb1pXbG5hSFFwTzF4dUlDQWdJRzFoY21kcGJqb2dNQ0JqWVd4aktDQjJZWElvTFMxcWNDMTNhV1JuWlhSekxXbHVjSFYwTFhCaFpHUnBibWNwSUNvZ01pQXBJREFnTVhCNE8xeHVJQ0FnSUdac2IyRjBPaUJzWldaME8xeHVmVnh1WEc0dktpQkRiMnh2Y2lCUWFXTnJaWElnVTNSNWJHbHVaeUFxTDF4dVhHNHVkMmxrWjJWMExXTnZiRzl5Y0dsamEyVnlJSHRjYmlBZ0lDQjNhV1IwYURvZ2RtRnlLQzB0YW5BdGQybGtaMlYwY3kxcGJteHBibVV0ZDJsa2RHZ3BPMXh1SUNBZ0lHaGxhV2RvZERvZ2RtRnlLQzB0YW5BdGQybGtaMlYwY3kxcGJteHBibVV0YUdWcFoyaDBLVHRjYmlBZ0lDQnNhVzVsTFdobGFXZG9kRG9nZG1GeUtDMHRhbkF0ZDJsa1oyVjBjeTFwYm14cGJtVXRhR1ZwWjJoMEtUdGNibjFjYmx4dUxuZHBaR2RsZEMxamIyeHZjbkJwWTJ0bGNpQStJQzUzYVdSblpYUXRZMjlzYjNKd2FXTnJaWEl0YVc1d2RYUWdlMXh1SUNBZ0lHWnNaWGd0WjNKdmR6b2dNVHRjYmlBZ0lDQm1iR1Y0TFhOb2NtbHVhem9nTVR0Y2JpQWdJQ0J0YVc0dGQybGtkR2c2SUhaaGNpZ3RMV3B3TFhkcFpHZGxkSE10YVc1c2FXNWxMWGRwWkhSb0xYUnBibmtwTzF4dWZWeHVYRzR1ZDJsa1oyVjBMV052Ykc5eWNHbGphMlZ5SUdsdWNIVjBXM1I1Y0dVOVhDSmpiMnh2Y2x3aVhTQjdYRzRnSUNBZ2QybGtkR2c2SUhaaGNpZ3RMV3B3TFhkcFpHZGxkSE10YVc1c2FXNWxMV2hsYVdkb2RDazdYRzRnSUNBZ2FHVnBaMmgwT2lCMllYSW9MUzFxY0MxM2FXUm5aWFJ6TFdsdWJHbHVaUzFvWldsbmFIUXBPMXh1SUNBZ0lIQmhaR1JwYm1jNklEQWdNbkI0T3lBdktpQnRZV3RsSUhSb1pTQmpiMnh2Y2lCemNYVmhjbVVnWVdOMGRXRnNiSGtnYzNGMVlYSmxJRzl1SUVOb2NtOXRaU0J2YmlCUFV5QllJQ292WEc0Z0lDQWdZbUZqYTJkeWIzVnVaRG9nZG1GeUtDMHRhbkF0ZDJsa1oyVjBjeTFwYm5CMWRDMWlZV05yWjNKdmRXNWtMV052Ykc5eUtUdGNiaUFnSUNCamIyeHZjam9nZG1GeUtDMHRhbkF0ZDJsa1oyVjBjeTFwYm5CMWRDMWpiMnh2Y2lrN1hHNGdJQ0FnWW05eVpHVnlPaUIyWVhJb0xTMXFjQzEzYVdSblpYUnpMV2x1Y0hWMExXSnZjbVJsY2kxM2FXUjBhQ2tnYzI5c2FXUWdkbUZ5S0MwdGFuQXRkMmxrWjJWMGN5MXBibkIxZEMxaWIzSmtaWEl0WTI5c2IzSXBPMXh1SUNBZ0lHSnZjbVJsY2kxc1pXWjBPaUJ1YjI1bE8xeHVJQ0FnSUdac1pYZ3RaM0p2ZHpvZ01EdGNiaUFnSUNCbWJHVjRMWE5vY21sdWF6b2dNRHRjYmlBZ0lDQmliM2d0YzJsNmFXNW5PaUJpYjNKa1pYSXRZbTk0TzF4dUlDQWdJR0ZzYVdkdUxYTmxiR1k2SUhOMGNtVjBZMmc3WEc0Z0lDQWdiM1YwYkdsdVpUb2dibTl1WlNBaGFXMXdiM0owWVc1ME8xeHVmVnh1WEc0dWQybGtaMlYwTFdOdmJHOXljR2xqYTJWeUxtTnZibU5wYzJVZ2FXNXdkWFJiZEhsd1pUMWNJbU52Ykc5eVhDSmRJSHRjYmlBZ0lDQmliM0prWlhJdGJHVm1kRG9nZG1GeUtDMHRhbkF0ZDJsa1oyVjBjeTFwYm5CMWRDMWliM0prWlhJdGQybGtkR2dwSUhOdmJHbGtJSFpoY2lndExXcHdMWGRwWkdkbGRITXRhVzV3ZFhRdFltOXlaR1Z5TFdOdmJHOXlLVHRjYm4xY2JseHVMbmRwWkdkbGRDMWpiMnh2Y25CcFkydGxjaUJwYm5CMWRGdDBlWEJsUFZ3aVkyOXNiM0pjSWwwNlptOWpkWE1zSUM1M2FXUm5aWFF0WTI5c2IzSndhV05yWlhJZ2FXNXdkWFJiZEhsd1pUMWNJblJsZUhSY0lsMDZabTlqZFhNZ2UxeHVJQ0FnSUdKdmNtUmxjaTFqYjJ4dmNqb2dkbUZ5S0MwdGFuQXRkMmxrWjJWMGN5MXBibkIxZEMxbWIyTjFjeTFpYjNKa1pYSXRZMjlzYjNJcE8xeHVmVnh1WEc0dWQybGtaMlYwTFdOdmJHOXljR2xqYTJWeUlHbHVjSFYwVzNSNWNHVTlYQ0owWlhoMFhDSmRJSHRjYmlBZ0lDQm1iR1Y0TFdkeWIzYzZJREU3WEc0Z0lDQWdiM1YwYkdsdVpUb2dibTl1WlNBaGFXMXdiM0owWVc1ME8xeHVJQ0FnSUdobGFXZG9kRG9nZG1GeUtDMHRhbkF0ZDJsa1oyVjBjeTFwYm14cGJtVXRhR1ZwWjJoMEtUdGNiaUFnSUNCc2FXNWxMV2hsYVdkb2REb2dkbUZ5S0MwdGFuQXRkMmxrWjJWMGN5MXBibXhwYm1VdGFHVnBaMmgwS1R0Y2JpQWdJQ0JpWVdOclozSnZkVzVrT2lCMllYSW9MUzFxY0MxM2FXUm5aWFJ6TFdsdWNIVjBMV0poWTJ0bmNtOTFibVF0WTI5c2IzSXBPMXh1SUNBZ0lHTnZiRzl5T2lCMllYSW9MUzFxY0MxM2FXUm5aWFJ6TFdsdWNIVjBMV052Ykc5eUtUdGNiaUFnSUNCaWIzSmtaWEk2SUhaaGNpZ3RMV3B3TFhkcFpHZGxkSE10YVc1d2RYUXRZbTl5WkdWeUxYZHBaSFJvS1NCemIyeHBaQ0IyWVhJb0xTMXFjQzEzYVdSblpYUnpMV2x1Y0hWMExXSnZjbVJsY2kxamIyeHZjaWs3WEc0Z0lDQWdabTl1ZEMxemFYcGxPaUIyWVhJb0xTMXFjQzEzYVdSblpYUnpMV1p2Ym5RdGMybDZaU2s3WEc0Z0lDQWdjR0ZrWkdsdVp6b2dkbUZ5S0MwdGFuQXRkMmxrWjJWMGN5MXBibkIxZEMxd1lXUmthVzVuS1NCallXeGpLQ0IyWVhJb0xTMXFjQzEzYVdSblpYUnpMV2x1Y0hWMExYQmhaR1JwYm1jcElDb2dJRElnS1R0Y2JpQWdJQ0J0YVc0dGQybGtkR2c2SURBN0lDOHFJRlJvYVhNZ2JXRnJaWE1nYVhRZ2NHOXpjMmxpYkdVZ1ptOXlJSFJvWlNCbWJHVjRZbTk0SUhSdklITm9jbWx1YXlCMGFHbHpJR2x1Y0hWMElDb3ZYRzRnSUNBZ1pteGxlQzF6YUhKcGJtczZJREU3WEc0Z0lDQWdZbTk0TFhOcGVtbHVaem9nWW05eVpHVnlMV0p2ZUR0Y2JuMWNibHh1TG5kcFpHZGxkQzFqYjJ4dmNuQnBZMnRsY2lCcGJuQjFkRnQwZVhCbFBWd2lkR1Y0ZEZ3aVhUcGthWE5oWW14bFpDQjdYRzRnSUNBZ2IzQmhZMmwwZVRvZ2RtRnlLQzB0YW5BdGQybGtaMlYwY3kxa2FYTmhZbXhsWkMxdmNHRmphWFI1S1R0Y2JuMWNibHh1THlvZ1JHRjBaU0JRYVdOclpYSWdVM1I1YkdsdVp5QXFMMXh1WEc0dWQybGtaMlYwTFdSaGRHVndhV05yWlhJZ2UxeHVJQ0FnSUhkcFpIUm9PaUIyWVhJb0xTMXFjQzEzYVdSblpYUnpMV2x1YkdsdVpTMTNhV1IwYUNrN1hHNGdJQ0FnYUdWcFoyaDBPaUIyWVhJb0xTMXFjQzEzYVdSblpYUnpMV2x1YkdsdVpTMW9aV2xuYUhRcE8xeHVJQ0FnSUd4cGJtVXRhR1ZwWjJoME9pQjJZWElvTFMxcWNDMTNhV1JuWlhSekxXbHViR2x1WlMxb1pXbG5hSFFwTzF4dWZWeHVYRzR1ZDJsa1oyVjBMV1JoZEdWd2FXTnJaWElnYVc1d2RYUmJkSGx3WlQxY0ltUmhkR1ZjSWwwZ2UxeHVJQ0FnSUdac1pYZ3RaM0p2ZHpvZ01UdGNiaUFnSUNCbWJHVjRMWE5vY21sdWF6b2dNVHRjYmlBZ0lDQnRhVzR0ZDJsa2RHZzZJREE3SUM4cUlGUm9hWE1nYldGclpYTWdhWFFnY0c5emMybGliR1VnWm05eUlIUm9aU0JtYkdWNFltOTRJSFJ2SUhOb2NtbHVheUIwYUdseklHbHVjSFYwSUNvdlhHNGdJQ0FnYjNWMGJHbHVaVG9nYm05dVpTQWhhVzF3YjNKMFlXNTBPMXh1SUNBZ0lHaGxhV2RvZERvZ2RtRnlLQzB0YW5BdGQybGtaMlYwY3kxcGJteHBibVV0YUdWcFoyaDBLVHRjYmlBZ0lDQmliM0prWlhJNklIWmhjaWd0TFdwd0xYZHBaR2RsZEhNdGFXNXdkWFF0WW05eVpHVnlMWGRwWkhSb0tTQnpiMnhwWkNCMllYSW9MUzFxY0MxM2FXUm5aWFJ6TFdsdWNIVjBMV0p2Y21SbGNpMWpiMnh2Y2lrN1hHNGdJQ0FnWW1GamEyZHliM1Z1WkMxamIyeHZjam9nZG1GeUtDMHRhbkF0ZDJsa1oyVjBjeTFwYm5CMWRDMWlZV05yWjNKdmRXNWtMV052Ykc5eUtUdGNiaUFnSUNCamIyeHZjam9nZG1GeUtDMHRhbkF0ZDJsa1oyVjBjeTFwYm5CMWRDMWpiMnh2Y2lrN1hHNGdJQ0FnWm05dWRDMXphWHBsT2lCMllYSW9MUzFxY0MxM2FXUm5aWFJ6TFdadmJuUXRjMmw2WlNrN1hHNGdJQ0FnY0dGa1pHbHVaem9nZG1GeUtDMHRhbkF0ZDJsa1oyVjBjeTFwYm5CMWRDMXdZV1JrYVc1bktTQmpZV3hqS0NCMllYSW9MUzFxY0MxM2FXUm5aWFJ6TFdsdWNIVjBMWEJoWkdScGJtY3BJQ29nSURJZ0tUdGNiaUFnSUNCaWIzZ3RjMmw2YVc1bk9pQmliM0prWlhJdFltOTRPMXh1ZlZ4dVhHNHVkMmxrWjJWMExXUmhkR1Z3YVdOclpYSWdhVzV3ZFhSYmRIbHdaVDFjSW1SaGRHVmNJbDA2Wm05amRYTWdlMXh1SUNBZ0lHSnZjbVJsY2kxamIyeHZjam9nZG1GeUtDMHRhbkF0ZDJsa1oyVjBjeTFwYm5CMWRDMW1iMk4xY3kxaWIzSmtaWEl0WTI5c2IzSXBPMXh1ZlZ4dVhHNHVkMmxrWjJWMExXUmhkR1Z3YVdOclpYSWdhVzV3ZFhSYmRIbHdaVDFjSW1SaGRHVmNJbDA2YVc1MllXeHBaQ0I3WEc0Z0lDQWdZbTl5WkdWeUxXTnZiRzl5T2lCMllYSW9MUzFxY0MxM1lYSnVMV052Ykc5eU1TazdYRzU5WEc1Y2JpNTNhV1JuWlhRdFpHRjBaWEJwWTJ0bGNpQnBibkIxZEZ0MGVYQmxQVndpWkdGMFpWd2lYVHBrYVhOaFlteGxaQ0I3WEc0Z0lDQWdiM0JoWTJsMGVUb2dkbUZ5S0MwdGFuQXRkMmxrWjJWMGN5MWthWE5oWW14bFpDMXZjR0ZqYVhSNUtUdGNibjFjYmx4dUx5b2dVR3hoZVNCWGFXUm5aWFFnS2k5Y2JseHVMbmRwWkdkbGRDMXdiR0Y1SUh0Y2JpQWdJQ0IzYVdSMGFEb2dkbUZ5S0MwdGFuQXRkMmxrWjJWMGN5MXBibXhwYm1VdGQybGtkR2d0YzJodmNuUXBPMXh1SUNBZ0lHUnBjM0JzWVhrNklHWnNaWGc3WEc0Z0lDQWdZV3hwWjI0dGFYUmxiWE02SUhOMGNtVjBZMmc3WEc1OVhHNWNiaTUzYVdSblpYUXRjR3hoZVNBdWFuVndlWFJsY2kxaWRYUjBiMjRnZTF4dUlDQWdJR1pzWlhndFozSnZkem9nTVR0Y2JpQWdJQ0JvWldsbmFIUTZJR0YxZEc4N1hHNTlYRzVjYmk1M2FXUm5aWFF0Y0d4aGVTQXVhblZ3ZVhSbGNpMWlkWFIwYjI0NlpHbHpZV0pzWldRZ2UxeHVJQ0FnSUc5d1lXTnBkSGs2SUhaaGNpZ3RMV3B3TFhkcFpHZGxkSE10WkdsellXSnNaV1F0YjNCaFkybDBlU2s3WEc1OVhHNWNiaThxSUZSaFlpQlhhV1JuWlhRZ0tpOWNibHh1TG1wMWNIbDBaWEl0ZDJsa1oyVjBjeTUzYVdSblpYUXRkR0ZpSUh0Y2JpQWdJQ0JrYVhOd2JHRjVPaUJtYkdWNE8xeHVJQ0FnSUdac1pYZ3RaR2x5WldOMGFXOXVPaUJqYjJ4MWJXNDdYRzU5WEc1Y2JpNXFkWEI1ZEdWeUxYZHBaR2RsZEhNdWQybGtaMlYwTFhSaFlpQStJQzV3TFZSaFlrSmhjaUI3WEc0Z0lDQWdMeW9nVG1WalpYTnpZWEo1SUhOdklIUm9ZWFFnWVNCMFlXSWdZMkZ1SUdKbElITm9hV1owWldRZ1pHOTNiaUIwYnlCdmRtVnliR0Y1SUhSb1pTQmliM0prWlhJZ2IyWWdkR2hsSUdKdmVDQmlaV3h2ZHk0Z0tpOWNiaUFnSUNCdmRtVnlabXh2ZHkxNE9pQjJhWE5wWW14bE8xeHVJQ0FnSUc5MlpYSm1iRzkzTFhrNklIWnBjMmxpYkdVN1hHNTlYRzVjYmk1cWRYQjVkR1Z5TFhkcFpHZGxkSE11ZDJsa1oyVjBMWFJoWWlBK0lDNXdMVlJoWWtKaGNpQStJQzV3TFZSaFlrSmhjaTFqYjI1MFpXNTBJSHRjYmlBZ0lDQXZLaUJOWVd0bElITjFjbVVnZEdoaGRDQjBhR1VnZEdGaUlHZHliM2R6SUdaeWIyMGdZbTkwZEc5dElIVndJQ292WEc0Z0lDQWdZV3hwWjI0dGFYUmxiWE02SUdac1pYZ3RaVzVrTzF4dUlDQWdJRzFwYmkxM2FXUjBhRG9nTUR0Y2JpQWdJQ0J0YVc0dGFHVnBaMmgwT2lBd08xeHVmVnh1WEc0dWFuVndlWFJsY2kxM2FXUm5aWFJ6TG5kcFpHZGxkQzEwWVdJZ1BpQXVkMmxrWjJWMExYUmhZaTFqYjI1MFpXNTBjeUI3WEc0Z0lDQWdkMmxrZEdnNklERXdNQ1U3WEc0Z0lDQWdZbTk0TFhOcGVtbHVaem9nWW05eVpHVnlMV0p2ZUR0Y2JpQWdJQ0J0WVhKbmFXNDZJREE3WEc0Z0lDQWdZbUZqYTJkeWIzVnVaRG9nZG1GeUtDMHRhbkF0YkdGNWIzVjBMV052Ykc5eU1TazdYRzRnSUNBZ1kyOXNiM0k2SUhaaGNpZ3RMV3B3TFhWcExXWnZiblF0WTI5c2IzSXhLVHRjYmlBZ0lDQmliM0prWlhJNklIWmhjaWd0TFdwd0xXSnZjbVJsY2kxM2FXUjBhQ2tnYzI5c2FXUWdkbUZ5S0MwdGFuQXRZbTl5WkdWeUxXTnZiRzl5TVNrN1hHNGdJQ0FnY0dGa1pHbHVaem9nZG1GeUtDMHRhbkF0ZDJsa1oyVjBjeTFqYjI1MFlXbHVaWEl0Y0dGa1pHbHVaeWs3WEc0Z0lDQWdabXhsZUMxbmNtOTNPaUF4TzF4dUlDQWdJRzkyWlhKbWJHOTNPaUJoZFhSdk8xeHVmVnh1WEc0dWFuVndlWFJsY2kxM2FXUm5aWFJ6TG5kcFpHZGxkQzEwWVdJZ1BpQXVjQzFVWVdKQ1lYSWdlMXh1SUNBZ0lHWnZiblE2SUhaaGNpZ3RMV3B3TFhkcFpHZGxkSE10Wm05dWRDMXphWHBsS1NCSVpXeDJaWFJwWTJFc0lFRnlhV0ZzTENCellXNXpMWE5sY21sbU8xeHVJQ0FnSUcxcGJpMW9aV2xuYUhRNklHTmhiR01vZG1GeUtDMHRhbkF0ZDJsa1oyVjBjeTFvYjNKcGVtOXVkR0ZzTFhSaFlpMW9aV2xuYUhRcElDc2dkbUZ5S0MwdGFuQXRZbTl5WkdWeUxYZHBaSFJvS1NrN1hHNTlYRzVjYmk1cWRYQjVkR1Z5TFhkcFpHZGxkSE11ZDJsa1oyVjBMWFJoWWlBK0lDNXdMVlJoWWtKaGNpQXVjQzFVWVdKQ1lYSXRkR0ZpSUh0Y2JpQWdJQ0JtYkdWNE9pQXdJREVnZG1GeUtDMHRhbkF0ZDJsa1oyVjBjeTFvYjNKcGVtOXVkR0ZzTFhSaFlpMTNhV1IwYUNrN1hHNGdJQ0FnYldsdUxYZHBaSFJvT2lBek5YQjRPMXh1SUNBZ0lHMXBiaTFvWldsbmFIUTZJR05oYkdNb2RtRnlLQzB0YW5BdGQybGtaMlYwY3kxb2IzSnBlbTl1ZEdGc0xYUmhZaTFvWldsbmFIUXBJQ3NnZG1GeUtDMHRhbkF0WW05eVpHVnlMWGRwWkhSb0tTazdYRzRnSUNBZ2JHbHVaUzFvWldsbmFIUTZJSFpoY2lndExXcHdMWGRwWkdkbGRITXRhRzl5YVhwdmJuUmhiQzEwWVdJdGFHVnBaMmgwS1R0Y2JpQWdJQ0J0WVhKbmFXNHRiR1ZtZERvZ1kyRnNZeWd0TVNBcUlIWmhjaWd0TFdwd0xXSnZjbVJsY2kxM2FXUjBhQ2twTzF4dUlDQWdJSEJoWkdScGJtYzZJREJ3ZUNBeE1IQjRPMXh1SUNBZ0lHSmhZMnRuY205MWJtUTZJSFpoY2lndExXcHdMV3hoZVc5MWRDMWpiMnh2Y2pJcE8xeHVJQ0FnSUdOdmJHOXlPaUIyWVhJb0xTMXFjQzExYVMxbWIyNTBMV052Ykc5eU1pazdYRzRnSUNBZ1ltOXlaR1Z5T2lCMllYSW9MUzFxY0MxaWIzSmtaWEl0ZDJsa2RHZ3BJSE52Ykdsa0lIWmhjaWd0TFdwd0xXSnZjbVJsY2kxamIyeHZjakVwTzF4dUlDQWdJR0p2Y21SbGNpMWliM1IwYjIwNklHNXZibVU3WEc0Z0lDQWdjRzl6YVhScGIyNDZJSEpsYkdGMGFYWmxPMXh1ZlZ4dVhHNHVhblZ3ZVhSbGNpMTNhV1JuWlhSekxuZHBaR2RsZEMxMFlXSWdQaUF1Y0MxVVlXSkNZWElnTG5BdFZHRmlRbUZ5TFhSaFlpNXdMVzF2WkMxamRYSnlaVzUwSUh0Y2JpQWdJQ0JqYjJ4dmNqb2dkbUZ5S0MwdGFuQXRkV2t0Wm05dWRDMWpiMnh2Y2pBcE8xeHVJQ0FnSUM4cUlGZGxJSGRoYm5RZ2RHaGxJR0poWTJ0bmNtOTFibVFnZEc4Z2JXRjBZMmdnZEdobElIUmhZaUJqYjI1MFpXNTBJR0poWTJ0bmNtOTFibVFnS2k5Y2JpQWdJQ0JpWVdOclozSnZkVzVrT2lCMllYSW9MUzFxY0Mxc1lYbHZkWFF0WTI5c2IzSXhLVHRjYmlBZ0lDQnRhVzR0YUdWcFoyaDBPaUJqWVd4aktIWmhjaWd0TFdwd0xYZHBaR2RsZEhNdGFHOXlhWHB2Ym5SaGJDMTBZV0l0YUdWcFoyaDBLU0FySURJZ0tpQjJZWElvTFMxcWNDMWliM0prWlhJdGQybGtkR2dwS1R0Y2JpQWdJQ0IwY21GdWMyWnZjbTA2SUhSeVlXNXpiR0YwWlZrb2RtRnlLQzB0YW5BdFltOXlaR1Z5TFhkcFpIUm9LU2s3WEc0Z0lDQWdiM1psY21ac2IzYzZJSFpwYzJsaWJHVTdYRzU5WEc1Y2JpNXFkWEI1ZEdWeUxYZHBaR2RsZEhNdWQybGtaMlYwTFhSaFlpQStJQzV3TFZSaFlrSmhjaUF1Y0MxVVlXSkNZWEl0ZEdGaUxuQXRiVzlrTFdOMWNuSmxiblE2WW1WbWIzSmxJSHRjYmlBZ0lDQndiM05wZEdsdmJqb2dZV0p6YjJ4MWRHVTdYRzRnSUNBZ2RHOXdPaUJqWVd4aktDMHhJQ29nZG1GeUtDMHRhbkF0WW05eVpHVnlMWGRwWkhSb0tTazdYRzRnSUNBZ2JHVm1kRG9nWTJGc1l5Z3RNU0FxSUhaaGNpZ3RMV3B3TFdKdmNtUmxjaTEzYVdSMGFDa3BPMXh1SUNBZ0lHTnZiblJsYm5RNklDY25PMXh1SUNBZ0lHaGxhV2RvZERvZ2RtRnlLQzB0YW5BdGQybGtaMlYwY3kxb2IzSnBlbTl1ZEdGc0xYUmhZaTEwYjNBdFltOXlaR1Z5S1R0Y2JpQWdJQ0IzYVdSMGFEb2dZMkZzWXlneE1EQWxJQ3NnTWlBcUlIWmhjaWd0TFdwd0xXSnZjbVJsY2kxM2FXUjBhQ2twTzF4dUlDQWdJR0poWTJ0bmNtOTFibVE2SUhaaGNpZ3RMV3B3TFdKeVlXNWtMV052Ykc5eU1TazdYRzU5WEc1Y2JpNXFkWEI1ZEdWeUxYZHBaR2RsZEhNdWQybGtaMlYwTFhSaFlpQStJQzV3TFZSaFlrSmhjaUF1Y0MxVVlXSkNZWEl0ZEdGaU9tWnBjbk4wTFdOb2FXeGtJSHRjYmlBZ0lDQnRZWEpuYVc0dGJHVm1kRG9nTUR0Y2JuMWNibHh1TG1wMWNIbDBaWEl0ZDJsa1oyVjBjeTUzYVdSblpYUXRkR0ZpSUQ0Z0xuQXRWR0ZpUW1GeUlDNXdMVlJoWWtKaGNpMTBZV0k2YUc5MlpYSTZibTkwS0M1d0xXMXZaQzFqZFhKeVpXNTBLU0I3WEc0Z0lDQWdZbUZqYTJkeWIzVnVaRG9nZG1GeUtDMHRhbkF0YkdGNWIzVjBMV052Ykc5eU1TazdYRzRnSUNBZ1kyOXNiM0k2SUhaaGNpZ3RMV3B3TFhWcExXWnZiblF0WTI5c2IzSXhLVHRjYm4xY2JseHVMbXAxY0hsMFpYSXRkMmxrWjJWMGN5NTNhV1JuWlhRdGRHRmlJRDRnTG5BdFZHRmlRbUZ5SUM1d0xXMXZaQzFqYkc5ellXSnNaU0ErSUM1d0xWUmhZa0poY2kxMFlXSkRiRzl6WlVsamIyNGdlMXh1SUNBZ0lHMWhjbWRwYmkxc1pXWjBPaUEwY0hnN1hHNTlYRzVjYmk1cWRYQjVkR1Z5TFhkcFpHZGxkSE11ZDJsa1oyVjBMWFJoWWlBK0lDNXdMVlJoWWtKaGNpQXVjQzF0YjJRdFkyeHZjMkZpYkdVZ1BpQXVjQzFVWVdKQ1lYSXRkR0ZpUTJ4dmMyVkpZMjl1T21KbFptOXlaU0I3WEc0Z0lDQWdabTl1ZEMxbVlXMXBiSGs2SUVadmJuUkJkMlZ6YjIxbE8xeHVJQ0FnSUdOdmJuUmxiblE2SUNkY1hHWXdNR1FuT3lBdktpQmpiRzl6WlNBcUwxeHVmVnh1WEc0dWFuVndlWFJsY2kxM2FXUm5aWFJ6TG5kcFpHZGxkQzEwWVdJZ1BpQXVjQzFVWVdKQ1lYSWdMbkF0VkdGaVFtRnlMWFJoWWtsamIyNHNYRzR1YW5Wd2VYUmxjaTEzYVdSblpYUnpMbmRwWkdkbGRDMTBZV0lnUGlBdWNDMVVZV0pDWVhJZ0xuQXRWR0ZpUW1GeUxYUmhZa3hoWW1Wc0xGeHVMbXAxY0hsMFpYSXRkMmxrWjJWMGN5NTNhV1JuWlhRdGRHRmlJRDRnTG5BdFZHRmlRbUZ5SUM1d0xWUmhZa0poY2kxMFlXSkRiRzl6WlVsamIyNGdlMXh1SUNBZ0lHeHBibVV0YUdWcFoyaDBPaUIyWVhJb0xTMXFjQzEzYVdSblpYUnpMV2h2Y21sNmIyNTBZV3d0ZEdGaUxXaGxhV2RvZENrN1hHNTlYRzVjYmk4cUlFRmpZMjl5WkdsdmJpQlhhV1JuWlhRZ0tpOWNibHh1TG5BdFEyOXNiR0Z3YzJVZ2UxeHVJQ0FnSUdScGMzQnNZWGs2SUdac1pYZzdYRzRnSUNBZ1pteGxlQzFrYVhKbFkzUnBiMjQ2SUdOdmJIVnRianRjYmlBZ0lDQmhiR2xuYmkxcGRHVnRjem9nYzNSeVpYUmphRHRjYm4xY2JseHVMbkF0UTI5c2JHRndjMlV0YUdWaFpHVnlJSHRjYmlBZ0lDQndZV1JrYVc1bk9pQjJZWElvTFMxcWNDMTNhV1JuWlhSekxXbHVjSFYwTFhCaFpHUnBibWNwTzF4dUlDQWdJR04xY25OdmNqb2djRzlwYm5SbGNqdGNiaUFnSUNCamIyeHZjam9nZG1GeUtDMHRhbkF0ZFdrdFptOXVkQzFqYjJ4dmNqSXBPMXh1SUNBZ0lHSmhZMnRuY205MWJtUXRZMjlzYjNJNklIWmhjaWd0TFdwd0xXeGhlVzkxZEMxamIyeHZjaklwTzF4dUlDQWdJR0p2Y21SbGNqb2dkbUZ5S0MwdGFuQXRkMmxrWjJWMGN5MWliM0prWlhJdGQybGtkR2dwSUhOdmJHbGtJSFpoY2lndExXcHdMV0p2Y21SbGNpMWpiMnh2Y2pFcE8xeHVJQ0FnSUhCaFpHUnBibWM2SUdOaGJHTW9kbUZ5S0MwdGFuQXRkMmxrWjJWMGN5MWpiMjUwWVdsdVpYSXRjR0ZrWkdsdVp5a2dLaUF5SUM4Z015a2dkbUZ5S0MwdGFuQXRkMmxrWjJWMGN5MWpiMjUwWVdsdVpYSXRjR0ZrWkdsdVp5azdYRzRnSUNBZ1ptOXVkQzEzWldsbmFIUTZJR0p2YkdRN1hHNTlYRzVjYmk1d0xVTnZiR3hoY0hObExXaGxZV1JsY2pwb2IzWmxjaUI3WEc0Z0lDQWdZbUZqYTJkeWIzVnVaQzFqYjJ4dmNqb2dkbUZ5S0MwdGFuQXRiR0Y1YjNWMExXTnZiRzl5TVNrN1hHNGdJQ0FnWTI5c2IzSTZJSFpoY2lndExXcHdMWFZwTFdadmJuUXRZMjlzYjNJeEtUdGNibjFjYmx4dUxuQXRRMjlzYkdGd2MyVXRiM0JsYmlBK0lDNXdMVU52Ykd4aGNITmxMV2hsWVdSbGNpQjdYRzRnSUNBZ1ltRmphMmR5YjNWdVpDMWpiMnh2Y2pvZ2RtRnlLQzB0YW5BdGJHRjViM1YwTFdOdmJHOXlNU2s3WEc0Z0lDQWdZMjlzYjNJNklIWmhjaWd0TFdwd0xYVnBMV1p2Ym5RdFkyOXNiM0l3S1R0Y2JpQWdJQ0JqZFhKemIzSTZJR1JsWm1GMWJIUTdYRzRnSUNBZ1ltOXlaR1Z5TFdKdmRIUnZiVG9nYm05dVpUdGNibjFjYmx4dUxuQXRRMjlzYkdGd2MyVWdMbkF0UTI5c2JHRndjMlV0YUdWaFpHVnlPanBpWldadmNtVWdlMXh1SUNBZ0lHTnZiblJsYm5RNklDZGNYR1l3WkdGY1hEQXdRVEFuT3lBZ0x5b2dZMkZ5WlhRdGNtbG5hSFFzSUc1dmJpMWljbVZoYTJsdVp5QnpjR0ZqWlNBcUwxeHVJQ0FnSUdScGMzQnNZWGs2SUdsdWJHbHVaUzFpYkc5amF6dGNiaUFnSUNCbWIyNTBPaUJ1YjNKdFlXd2dibTl5YldGc0lHNXZjbTFoYkNBeE5IQjRMekVnUm05dWRFRjNaWE52YldVN1hHNGdJQ0FnWm05dWRDMXphWHBsT2lCcGJtaGxjbWwwTzF4dUlDQWdJSFJsZUhRdGNtVnVaR1Z5YVc1bk9pQmhkWFJ2TzF4dUlDQWdJQzEzWldKcmFYUXRabTl1ZEMxemJXOXZkR2hwYm1jNklHRnVkR2xoYkdsaGMyVmtPMXh1SUNBZ0lDMXRiM290YjNONExXWnZiblF0YzIxdmIzUm9hVzVuT2lCbmNtRjVjMk5oYkdVN1hHNTlYRzVjYmk1d0xVTnZiR3hoY0hObExXOXdaVzRnUGlBdWNDMURiMnhzWVhCelpTMW9aV0ZrWlhJNk9tSmxabTl5WlNCN1hHNGdJQ0FnWTI5dWRHVnVkRG9nSjF4Y1pqQmtOMXhjTURCQk1DYzdJQzhxSUdOaGNtVjBMV1J2ZDI0c0lHNXZiaTFpY21WaGEybHVaeUJ6Y0dGalpTQXFMMXh1ZlZ4dVhHNHVjQzFEYjJ4c1lYQnpaUzFqYjI1MFpXNTBjeUI3WEc0Z0lDQWdjR0ZrWkdsdVp6b2dkbUZ5S0MwdGFuQXRkMmxrWjJWMGN5MWpiMjUwWVdsdVpYSXRjR0ZrWkdsdVp5azdYRzRnSUNBZ1ltRmphMmR5YjNWdVpDMWpiMnh2Y2pvZ2RtRnlLQzB0YW5BdGJHRjViM1YwTFdOdmJHOXlNU2s3WEc0Z0lDQWdZMjlzYjNJNklIWmhjaWd0TFdwd0xYVnBMV1p2Ym5RdFkyOXNiM0l4S1R0Y2JpQWdJQ0JpYjNKa1pYSXRiR1ZtZERvZ2RtRnlLQzB0YW5BdGQybGtaMlYwY3kxaWIzSmtaWEl0ZDJsa2RHZ3BJSE52Ykdsa0lIWmhjaWd0TFdwd0xXSnZjbVJsY2kxamIyeHZjakVwTzF4dUlDQWdJR0p2Y21SbGNpMXlhV2RvZERvZ2RtRnlLQzB0YW5BdGQybGtaMlYwY3kxaWIzSmtaWEl0ZDJsa2RHZ3BJSE52Ykdsa0lIWmhjaWd0TFdwd0xXSnZjbVJsY2kxamIyeHZjakVwTzF4dUlDQWdJR0p2Y21SbGNpMWliM1IwYjIwNklIWmhjaWd0TFdwd0xYZHBaR2RsZEhNdFltOXlaR1Z5TFhkcFpIUm9LU0J6YjJ4cFpDQjJZWElvTFMxcWNDMWliM0prWlhJdFkyOXNiM0l4S1R0Y2JpQWdJQ0J2ZG1WeVpteHZkem9nWVhWMGJ6dGNibjFjYmx4dUxuQXRRV05qYjNKa2FXOXVJSHRjYmlBZ0lDQmthWE53YkdGNU9pQm1iR1Y0TzF4dUlDQWdJR1pzWlhndFpHbHlaV04wYVc5dU9pQmpiMngxYlc0N1hHNGdJQ0FnWVd4cFoyNHRhWFJsYlhNNklITjBjbVYwWTJnN1hHNTlYRzVjYmk1d0xVRmpZMjl5WkdsdmJpQXVjQzFEYjJ4c1lYQnpaU0I3WEc0Z0lDQWdiV0Z5WjJsdUxXSnZkSFJ2YlRvZ01EdGNibjFjYmx4dUxuQXRRV05qYjNKa2FXOXVJQzV3TFVOdmJHeGhjSE5sSUNzZ0xuQXRRMjlzYkdGd2MyVWdlMXh1SUNBZ0lHMWhjbWRwYmkxMGIzQTZJRFJ3ZUR0Y2JuMWNibHh1WEc1Y2JpOHFJRWhVVFV3Z2QybGtaMlYwSUNvdlhHNWNiaTUzYVdSblpYUXRhSFJ0YkN3Z0xuZHBaR2RsZEMxb2RHMXNiV0YwYUNCN1hHNGdJQ0FnWm05dWRDMXphWHBsT2lCMllYSW9MUzFxY0MxM2FXUm5aWFJ6TFdadmJuUXRjMmw2WlNrN1hHNTlYRzVjYmk1M2FXUm5aWFF0YUhSdGJDQStJQzUzYVdSblpYUXRhSFJ0YkMxamIyNTBaVzUwTENBdWQybGtaMlYwTFdoMGJXeHRZWFJvSUQ0Z0xuZHBaR2RsZEMxb2RHMXNMV052Ym5SbGJuUWdlMXh1SUNBZ0lDOHFJRVpwYkd3Z2IzVjBJSFJvWlNCaGNtVmhJR2x1SUhSb1pTQklWRTFNSUhkcFpHZGxkQ0FxTDF4dUlDQWdJR0ZzYVdkdUxYTmxiR1k2SUhOMGNtVjBZMmc3WEc0Z0lDQWdabXhsZUMxbmNtOTNPaUF4TzF4dUlDQWdJR1pzWlhndGMyaHlhVzVyT2lBeE8xeHVJQ0FnSUM4cUlFMWhhMlZ6SUhOMWNtVWdkR2hsSUdKaGMyVnNhVzVsSUdseklITjBhV3hzSUdGc2FXZHVaV1FnZDJsMGFDQnZkR2hsY2lCbGJHVnRaVzUwY3lBcUwxeHVJQ0FnSUd4cGJtVXRhR1ZwWjJoME9pQjJZWElvTFMxcWNDMTNhV1JuWlhSekxXbHViR2x1WlMxb1pXbG5hSFFwTzF4dUlDQWdJQzhxSUUxaGEyVWdhWFFnY0c5emMybGliR1VnZEc4Z2FHRjJaU0JoWW5OdmJIVjBaV3g1TFhCdmMybDBhVzl1WldRZ1pXeGxiV1Z1ZEhNZ2FXNGdkR2hsSUdoMGJXd2dLaTljYmlBZ0lDQndiM05wZEdsdmJqb2djbVZzWVhScGRtVTdYRzU5WEc0aUxDSXZLaUJVYUdseklHWnBiR1VnYUdGeklHTnZaR1VnWkdWeWFYWmxaQ0JtY205dElGQm9iM053YUc5eVNsTWdRMU5USUdacGJHVnpMQ0JoY3lCdWIzUmxaQ0JpWld4dmR5NGdWR2hsSUd4cFkyVnVjMlVnWm05eUlIUm9hWE1nVUdodmMzQm9iM0pLVXlCamIyUmxJR2x6T2x4dVhHNURiM0I1Y21sbmFIUWdLR01wSURJd01UUXRNakF4Tnl3Z1VHaHZjM0JvYjNKS1V5QkRiMjUwY21saWRYUnZjbk5jYmtGc2JDQnlhV2RvZEhNZ2NtVnpaWEoyWldRdVhHNWNibEpsWkdsemRISnBZblYwYVc5dUlHRnVaQ0IxYzJVZ2FXNGdjMjkxY21ObElHRnVaQ0JpYVc1aGNua2dabTl5YlhNc0lIZHBkR2dnYjNJZ2QybDBhRzkxZEZ4dWJXOWthV1pwWTJGMGFXOXVMQ0JoY21VZ2NHVnliV2wwZEdWa0lIQnliM1pwWkdWa0lIUm9ZWFFnZEdobElHWnZiR3h2ZDJsdVp5QmpiMjVrYVhScGIyNXpJR0Z5WlNCdFpYUTZYRzVjYmlvZ1VtVmthWE4wY21saWRYUnBiMjV6SUc5bUlITnZkWEpqWlNCamIyUmxJRzExYzNRZ2NtVjBZV2x1SUhSb1pTQmhZbTkyWlNCamIzQjVjbWxuYUhRZ2JtOTBhV05sTENCMGFHbHpYRzRnSUd4cGMzUWdiMllnWTI5dVpHbDBhVzl1Y3lCaGJtUWdkR2hsSUdadmJHeHZkMmx1WnlCa2FYTmpiR0ZwYldWeUxseHVYRzRxSUZKbFpHbHpkSEpwWW5WMGFXOXVjeUJwYmlCaWFXNWhjbmtnWm05eWJTQnRkWE4wSUhKbGNISnZaSFZqWlNCMGFHVWdZV0p2ZG1VZ1kyOXdlWEpwWjJoMElHNXZkR2xqWlN4Y2JpQWdkR2hwY3lCc2FYTjBJRzltSUdOdmJtUnBkR2x2Ym5NZ1lXNWtJSFJvWlNCbWIyeHNiM2RwYm1jZ1pHbHpZMnhoYVcxbGNpQnBiaUIwYUdVZ1pHOWpkVzFsYm5SaGRHbHZibHh1SUNCaGJtUXZiM0lnYjNSb1pYSWdiV0YwWlhKcFlXeHpJSEJ5YjNacFpHVmtJSGRwZEdnZ2RHaGxJR1JwYzNSeWFXSjFkR2x2Ymk1Y2JseHVLaUJPWldsMGFHVnlJSFJvWlNCdVlXMWxJRzltSUhSb1pTQmpiM0I1Y21sbmFIUWdhRzlzWkdWeUlHNXZjaUIwYUdVZ2JtRnRaWE1nYjJZZ2FYUnpYRzRnSUdOdmJuUnlhV0oxZEc5eWN5QnRZWGtnWW1VZ2RYTmxaQ0IwYnlCbGJtUnZjbk5sSUc5eUlIQnliMjF2ZEdVZ2NISnZaSFZqZEhNZ1pHVnlhWFpsWkNCbWNtOXRYRzRnSUhSb2FYTWdjMjltZEhkaGNtVWdkMmwwYUc5MWRDQnpjR1ZqYVdacFl5QndjbWx2Y2lCM2NtbDBkR1Z1SUhCbGNtMXBjM05wYjI0dVhHNWNibFJJU1ZNZ1UwOUdWRmRCVWtVZ1NWTWdVRkpQVmtsRVJVUWdRbGtnVkVoRklFTlBVRmxTU1VkSVZDQklUMHhFUlZKVElFRk9SQ0JEVDA1VVVrbENWVlJQVWxNZ1hDSkJVeUJKVTF3aVhHNUJUa1FnUVU1WklFVllVRkpGVTFNZ1QxSWdTVTFRVEVsRlJDQlhRVkpTUVU1VVNVVlRMQ0JKVGtOTVZVUkpUa2NzSUVKVlZDQk9UMVFnVEVsTlNWUkZSQ0JVVHl3Z1ZFaEZYRzVKVFZCTVNVVkVJRmRCVWxKQlRsUkpSVk1nVDBZZ1RVVlNRMGhCVGxSQlFrbE1TVlJaSUVGT1JDQkdTVlJPUlZOVElFWlBVaUJCSUZCQlVsUkpRMVZNUVZJZ1VGVlNVRTlUUlNCQlVrVmNia1JKVTBOTVFVbE5SVVF1SUVsT0lFNVBJRVZXUlU1VUlGTklRVXhNSUZSSVJTQkRUMUJaVWtsSFNGUWdTRTlNUkVWU0lFOVNJRU5QVGxSU1NVSlZWRTlTVXlCQ1JTQk1TVUZDVEVWY2JrWlBVaUJCVGxrZ1JFbFNSVU5VTENCSlRrUkpVa1ZEVkN3Z1NVNURTVVJGVGxSQlRDd2dVMUJGUTBsQlRDd2dSVmhGVFZCTVFWSlpMQ0JQVWlCRFQwNVRSVkZWUlU1VVNVRk1YRzVFUVUxQlIwVlRJQ2hKVGtOTVZVUkpUa2NzSUVKVlZDQk9UMVFnVEVsTlNWUkZSQ0JVVHl3Z1VGSlBRMVZTUlUxRlRsUWdUMFlnVTFWQ1UxUkpWRlZVUlNCSFQwOUVVeUJQVWx4dVUwVlNWa2xEUlZNN0lFeFBVMU1nVDBZZ1ZWTkZMQ0JFUVZSQkxDQlBVaUJRVWs5R1NWUlRPeUJQVWlCQ1ZWTkpUa1ZUVXlCSlRsUkZVbEpWVUZSSlQwNHBJRWhQVjBWV1JWSmNia05CVlZORlJDQkJUa1FnVDA0Z1FVNVpJRlJJUlU5U1dTQlBSaUJNU1VGQ1NVeEpWRmtzSUZkSVJWUklSVklnU1U0Z1EwOU9WRkpCUTFRc0lGTlVVa2xEVkNCTVNVRkNTVXhKVkZrc1hHNVBVaUJVVDFKVUlDaEpUa05NVlVSSlRrY2dUa1ZIVEVsSFJVNURSU0JQVWlCUFZFaEZVbGRKVTBVcElFRlNTVk5KVGtjZ1NVNGdRVTVaSUZkQldTQlBWVlFnVDBZZ1ZFaEZJRlZUUlZ4dVQwWWdWRWhKVXlCVFQwWlVWMEZTUlN3Z1JWWkZUaUJKUmlCQlJGWkpVMFZFSUU5R0lGUklSU0JRVDFOVFNVSkpURWxVV1NCUFJpQlRWVU5JSUVSQlRVRkhSUzVjYmx4dUtpOWNibHh1THlwY2JpQXFJRlJvWlNCbWIyeHNiM2RwYm1jZ2MyVmpkR2x2YmlCcGN5QmtaWEpwZG1Wa0lHWnliMjBnYUhSMGNITTZMeTluYVhSb2RXSXVZMjl0TDNCb2IzTndhRzl5YW5NdmNHaHZjM0JvYjNJdllteHZZaTh5TTJJNVpEQTNOV1ZpWXpWaU56TmhZakUwT0dJMlpXSm1Zekl3WVdZNU4yWTROVGN4TkdNMEwzQmhZMnRoWjJWekwzZHBaR2RsZEhNdmMzUjViR1V2ZEdGaVltRnlMbU56Y3lCY2JpQXFJRmRsSjNabElITmpiM0JsWkNCMGFHVWdjblZzWlhNZ2MyOGdkR2hoZENCMGFHVjVJR0Z5WlNCamIyNXphWE4wWlc1MElIZHBkR2dnWlhoaFkzUnNlU0J2ZFhJZ1kyOWtaUzVjYmlBcUwxeHVYRzR1YW5Wd2VYUmxjaTEzYVdSblpYUnpMbmRwWkdkbGRDMTBZV0lnUGlBdWNDMVVZV0pDWVhJZ2UxeHVJQ0JrYVhOd2JHRjVPaUJtYkdWNE8xeHVJQ0F0ZDJWaWEybDBMWFZ6WlhJdGMyVnNaV04wT2lCdWIyNWxPMXh1SUNBdGJXOTZMWFZ6WlhJdGMyVnNaV04wT2lCdWIyNWxPMXh1SUNBdGJYTXRkWE5sY2kxelpXeGxZM1E2SUc1dmJtVTdYRzRnSUhWelpYSXRjMlZzWldOME9pQnViMjVsTzF4dWZWeHVYRzVjYmk1cWRYQjVkR1Z5TFhkcFpHZGxkSE11ZDJsa1oyVjBMWFJoWWlBK0lDNXdMVlJoWWtKaGNsdGtZWFJoTFc5eWFXVnVkR0YwYVc5dVBTZG9iM0pwZW05dWRHRnNKMTBnZTF4dUlDQm1iR1Y0TFdScGNtVmpkR2x2YmpvZ2NtOTNPMXh1ZlZ4dVhHNWNiaTVxZFhCNWRHVnlMWGRwWkdkbGRITXVkMmxrWjJWMExYUmhZaUErSUM1d0xWUmhZa0poY2x0a1lYUmhMVzl5YVdWdWRHRjBhVzl1UFNkMlpYSjBhV05oYkNkZElIdGNiaUFnWm14bGVDMWthWEpsWTNScGIyNDZJR052YkhWdGJqdGNibjFjYmx4dVhHNHVhblZ3ZVhSbGNpMTNhV1JuWlhSekxuZHBaR2RsZEMxMFlXSWdQaUF1Y0MxVVlXSkNZWElnUGlBdWNDMVVZV0pDWVhJdFkyOXVkR1Z1ZENCN1hHNGdJRzFoY21kcGJqb2dNRHRjYmlBZ2NHRmtaR2x1WnpvZ01EdGNiaUFnWkdsemNHeGhlVG9nWm14bGVEdGNiaUFnWm14bGVEb2dNU0F4SUdGMWRHODdYRzRnSUd4cGMzUXRjM1I1YkdVdGRIbHdaVG9nYm05dVpUdGNibjFjYmx4dVhHNHVhblZ3ZVhSbGNpMTNhV1JuWlhSekxuZHBaR2RsZEMxMFlXSWdQaUF1Y0MxVVlXSkNZWEpiWkdGMFlTMXZjbWxsYm5SaGRHbHZiajBuYUc5eWFYcHZiblJoYkNkZElENGdMbkF0VkdGaVFtRnlMV052Ym5SbGJuUWdlMXh1SUNCbWJHVjRMV1JwY21WamRHbHZiam9nY205M08xeHVmVnh1WEc1Y2JpNXFkWEI1ZEdWeUxYZHBaR2RsZEhNdWQybGtaMlYwTFhSaFlpQStJQzV3TFZSaFlrSmhjbHRrWVhSaExXOXlhV1Z1ZEdGMGFXOXVQU2QyWlhKMGFXTmhiQ2RkSUQ0Z0xuQXRWR0ZpUW1GeUxXTnZiblJsYm5RZ2UxeHVJQ0JtYkdWNExXUnBjbVZqZEdsdmJqb2dZMjlzZFcxdU8xeHVmVnh1WEc1Y2JpNXFkWEI1ZEdWeUxYZHBaR2RsZEhNdWQybGtaMlYwTFhSaFlpQStJQzV3TFZSaFlrSmhjaUF1Y0MxVVlXSkNZWEl0ZEdGaUlIdGNiaUFnWkdsemNHeGhlVG9nWm14bGVEdGNiaUFnWm14bGVDMWthWEpsWTNScGIyNDZJSEp2ZHp0Y2JpQWdZbTk0TFhOcGVtbHVaem9nWW05eVpHVnlMV0p2ZUR0Y2JpQWdiM1psY21ac2IzYzZJR2hwWkdSbGJqdGNibjFjYmx4dVhHNHVhblZ3ZVhSbGNpMTNhV1JuWlhSekxuZHBaR2RsZEMxMFlXSWdQaUF1Y0MxVVlXSkNZWElnTG5BdFZHRmlRbUZ5TFhSaFlrbGpiMjRzWEc0dWFuVndlWFJsY2kxM2FXUm5aWFJ6TG5kcFpHZGxkQzEwWVdJZ1BpQXVjQzFVWVdKQ1lYSWdMbkF0VkdGaVFtRnlMWFJoWWtOc2IzTmxTV052YmlCN1hHNGdJR1pzWlhnNklEQWdNQ0JoZFhSdk8xeHVmVnh1WEc1Y2JpNXFkWEI1ZEdWeUxYZHBaR2RsZEhNdWQybGtaMlYwTFhSaFlpQStJQzV3TFZSaFlrSmhjaUF1Y0MxVVlXSkNZWEl0ZEdGaVRHRmlaV3dnZTF4dUlDQm1iR1Y0T2lBeElERWdZWFYwYnp0Y2JpQWdiM1psY21ac2IzYzZJR2hwWkdSbGJqdGNiaUFnZDJocGRHVXRjM0JoWTJVNklHNXZkM0poY0R0Y2JuMWNibHh1WEc0dWFuVndlWFJsY2kxM2FXUm5aWFJ6TG5kcFpHZGxkQzEwWVdJZ1BpQXVjQzFVWVdKQ1lYSWdMbkF0VkdGaVFtRnlMWFJoWWk1d0xXMXZaQzFvYVdSa1pXNGdlMXh1SUNCa2FYTndiR0Y1T2lCdWIyNWxJQ0ZwYlhCdmNuUmhiblE3WEc1OVhHNWNibHh1TG1wMWNIbDBaWEl0ZDJsa1oyVjBjeTUzYVdSblpYUXRkR0ZpSUQ0Z0xuQXRWR0ZpUW1GeUxuQXRiVzlrTFdSeVlXZG5hVzVuSUM1d0xWUmhZa0poY2kxMFlXSWdlMXh1SUNCd2IzTnBkR2x2YmpvZ2NtVnNZWFJwZG1VN1hHNTlYRzVjYmx4dUxtcDFjSGwwWlhJdGQybGtaMlYwY3k1M2FXUm5aWFF0ZEdGaUlENGdMbkF0VkdGaVFtRnlMbkF0Ylc5a0xXUnlZV2RuYVc1blcyUmhkR0V0YjNKcFpXNTBZWFJwYjI0OUoyaHZjbWw2YjI1MFlXd25YU0F1Y0MxVVlXSkNZWEl0ZEdGaUlIdGNiaUFnYkdWbWREb2dNRHRjYmlBZ2RISmhibk5wZEdsdmJqb2diR1ZtZENBeE5UQnRjeUJsWVhObE8xeHVmVnh1WEc1Y2JpNXFkWEI1ZEdWeUxYZHBaR2RsZEhNdWQybGtaMlYwTFhSaFlpQStJQzV3TFZSaFlrSmhjaTV3TFcxdlpDMWtjbUZuWjJsdVoxdGtZWFJoTFc5eWFXVnVkR0YwYVc5dVBTZDJaWEowYVdOaGJDZGRJQzV3TFZSaFlrSmhjaTEwWVdJZ2UxeHVJQ0IwYjNBNklEQTdYRzRnSUhSeVlXNXphWFJwYjI0NklIUnZjQ0F4TlRCdGN5QmxZWE5sTzF4dWZWeHVYRzVjYmk1cWRYQjVkR1Z5TFhkcFpHZGxkSE11ZDJsa1oyVjBMWFJoWWlBK0lDNXdMVlJoWWtKaGNpNXdMVzF2WkMxa2NtRm5aMmx1WnlBdWNDMVVZV0pDWVhJdGRHRmlMbkF0Ylc5a0xXUnlZV2RuYVc1bklIdGNiaUFnZEhKaGJuTnBkR2x2YmpvZ2JtOXVaVHRjYm4xY2JseHVMeW9nUlc1a0lIUmhZbUpoY2k1amMzTWdLaTljYmlKZGZRPT0gKi8=",
       "headers": [
        [
         "content-type",
         "text/css"
        ]
       ],
       "ok": true,
       "status": 200,
       "status_text": ""
      }
     }
    },
    "colab_type": "code",
    "id": "_ObmMvYCJSDV",
    "outputId": "2af85b3f-6133-485a-cb8d-f17c4aec8241"
   },
   "outputs": [
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "b35a969e1bf844eb9cf45497258fbb8c",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "HBox(children=(IntProgress(value=0, max=250), HTML(value='')))"
      ]
     },
     "metadata": {
      "tags": []
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "gan_model = CycleGan(train_loader,val_loader)\n",
    "gan_model.train()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "gyThAcWdtrIk"
   },
   "outputs": [],
   "source": [
    "## Results SIR 11.670795746936527 SAR 11.670795746936527 SDR 6.5125854585956136\n",
    "\n",
    "import mir_eval\n",
    "import museval\n",
    "import gc\n",
    "def _any_source_silent(sources):\n",
    "    \"\"\"Returns true if the parameter sources has any silent first dimensions\"\"\"\n",
    "    return np.any(np.all(np.sum(\n",
    "        sources, axis=tuple(range(2, sources.ndim))) == 0, axis=1))\n",
    "\n",
    "test_dataset = LMDBWavLoader('musdb_test',True)\n",
    "test_data_len = len(test_dataset) \n",
    "eps = 1e-15\n",
    "with torch.no_grad():\n",
    "    # to get 1 min output\n",
    "    reconstructed_signals = []\n",
    "    mixed_signals = []\n",
    "    original_signals = []\n",
    "    out_sdr = []\n",
    "    out_sir = []\n",
    "    out_sar = []\n",
    "    for i in tqdm(range(test_data_len)):\n",
    "        #sample_rate = 20480\n",
    "        val_data  = test_dataset[i] \n",
    "        mixed_signal = []\n",
    "        wav_iter = list(audio_generator(val_data[1]))\n",
    "        subsample_n_samples = len(wav_iter)\n",
    "        for j in range(subsample_n_samples):\n",
    "            mixed_signal.append(wav_iter[j])\n",
    "        mixed_signal = torch.squeeze(torch.from_numpy(np.stack(mixed_signal,axis=0)).float()).to(device)\n",
    "        mixed_signal = torch.unsqueeze(mixed_signal, dim=1)\n",
    "        reconstructed = gan_model.generator(mixed_signal) \n",
    "        single_source = val_data[0]\n",
    "        \n",
    "        n_items_per_eval = 3\n",
    "        data_len = (single_source.shape[-1]//n_items_per_eval) * n_items_per_eval\n",
    "        reconstructed = torch.squeeze(reconstructed).detach().cpu().numpy().reshape(1,-1)[:,:data_len]\n",
    "\n",
    "        reconstructed = reconstructed[:data_len].reshape( -1)\n",
    "        single_source = single_source[:data_len].reshape( -1) \n",
    "        mixed_source = val_data[1][:data_len].reshape( -1)\n",
    "        clean_inference = mixed_source - single_source\n",
    "        predicted_inference = mixed_source - reconstructed\n",
    "        #single_source = librosa.resample(single_source, sample_rate, window_length) \n",
    "        #reconstructed = librosa.resample(reconstructed, sample_rate, window_length)\n",
    "        reference_music =single_source#, mixed_signal[:data_len].reshape(n_items_per_eval, -1) - single_source[:data_len].reshape(n_items_per_eval, -1)])\n",
    "        estimates_music = reconstructed#, mixed_signal[:data_len].reshape(n_items_per_eval, -1) - reconstructed])\n",
    "        del reconstructed\n",
    "        del single_source\n",
    "        del mixed_signal\n",
    "        gc.collect()\n",
    "        sdr_b,  sir_b, sar_b, _ =mir_eval.separation.bss_eval_sources_framewise(np.array([reference_music, clean_inference]), np.array([estimates_music, predicted_inference]))\n",
    "        sdr, sir, sar = sdr_b, sir_b, sar_b\n",
    "        #sdr_inter, sir_inter, sar_inter = sdr_b[1], sir_b[1], sar_b[1]\n",
    "        out_sdr.append(np.mean(sdr[~np.isnan(sdr)]))\n",
    "        out_sir.append(np.mean(sir[~np.isnan(sir)]))\n",
    "        out_sar.append(np.mean(sir[~np.isnan(sar)]))\n",
    "    sdr  = np.median(out_sdr)\n",
    "    sir = np.median(out_sir)\n",
    "    sar = np.median(out_sar)\n",
    "    print('SIR {} SAR {} SDR {}'.format(str(sir),str(sar),str(sdr)))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "PwITOkhsSPgD"
   },
   "outputs": [],
   "source": [
    "def preprocess(x):\n",
    "    output_array = np.copy(x)\n",
    "    audio_len = output_array.shape[-1]\n",
    "    for i in range(window_length, audio_len, window_length):\n",
    "        output_array[0,0,i-1] = 0\n",
    "    return output_array\n",
    "    \n",
    "def predict_long_wav(wav_file_path, out_folder_name):\n",
    "    wav_data = load_wav(wav_file_path)\n",
    "    wav_iter = list(audio_generator(wav_data))\n",
    "    wav_iter = np.stack(wav_iter)\n",
    "    wav_iter = torch.from_numpy(wav_iter).float().to(device)\n",
    "    wav_iter = torch.unsqueeze(wav_iter,dim=1)\n",
    "    reconstructed = gan_model.generator(wav_iter)\n",
    "    x = reconstructed.view(1,1,-1).detach().cpu().numpy()\n",
    "    x = preprocess(x)\n",
    "    save_samples(x,  out_folder_name )\n",
    "predict_long_wav('mixture_blue.wav', 'blue')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "Mi8SaE4_9r-1"
   },
   "outputs": [],
   "source": [
    "torch.cuda.empty_cache()"
   ]
  }
 ],
 "metadata": {
  "accelerator": "GPU",
  "colab": {
   "collapsed_sections": [],
   "name": "Music_Generator_CycleGan_single_gen Muse DB.ipynb",
   "provenance": [],
   "toc_visible": true,
   "version": "0.3.2"
  },
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.6.8"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 1
}