tasks.py

import fnmatch
import glob
import hashlib
import json
import os
import platform
import re
import shutil
import stat
import subprocess
import sys
import zipfile
from datetime import datetime
from datetime import timezone
from pathlib import Path
from pathlib import PurePath
from tempfile import mkstemp
from tempfile import TemporaryDirectory

import boto3
import requests
from invoke import Collection
from invoke import task


# Below is from:
# https://stackoverflow.com/questions/3041986/apt-command-line-interface-like-yes-no-input
def query_yes_no(question, default="yes"):
    """Ask a yes/no question via input() and return answer.

    "question" is a string that is presented to the user.
    "default" is the presumed answer if the user just hits <Enter>.
        It must be "yes" (the default), "no" or None (meaning
        an answer is required of the user).

    The "answer" return value is True for "yes" or False for "no".
    """
    valid = {"yes": True, "y": True, "ye": True, "no": False, "n": False}

    if default is None:
        prompt = " [y/n] "
    elif default == "yes":
        prompt = " [Y/n] "
    elif default == "no":
        prompt = " [y/N] "
    else:
        raise ValueError("invalid default answer: '%s'" % default)

    while True:
        sys.stdout.write(question + prompt)
        choice = input().lower()

        if default is not None and choice == "":
            return valid[default]
        elif choice in valid:
            return valid[choice]
        else:
            sys.stdout.write("Please respond with 'yes' or 'no' " "(or 'y' or 'n').\n")


def get_version(c):
    with open(c.plugin.version_file_raw) as f:
        return f.readline().strip()


# Handle long filenames or readonly files on windows, see:
# http://bit.ly/2g58Yxu
def rmtree(top):
    for root, dirs, files in os.walk(top, topdown=False):
        for name in files:
            filename = os.path.join(root, name)
            os.chmod(filename, stat.S_IWUSR)
            try:
                os.remove(filename)
            except PermissionError:
                print(
                    "Permission error: unable to remove {}. Skipping that file.".format(
                        filename
                    )
                )

        for name in dirs:
            try:
                os.rmdir(os.path.join(root, name))
            except OSError:
                print(
                    "Unable to remove directory {}. Skipping removing that folder.".format(
                        os.path.join(root, name)
                    )
                )
    try:
        os.rmdir(top)
    except OSError:
        print(
            "Unable to remove directory {}. Skipping removing that folder.".format(top)
        )


# Function to find and replace in a file
def _replace(file_path, regex, subst):
    # Create temp file
    fh, abs_path = mkstemp()

    if sys.version_info[0] < 3:
        with os.fdopen(fh, "w") as new_file:
            with open(file_path) as old_file:
                for line in old_file:
                    new_file.write(regex.sub(subst, line))
    else:
        with open(fh, "w", encoding="Latin-1") as new_file:
            with open(file_path, encoding="Latin-1") as old_file:
                for line in old_file:
                    new_file.write(regex.sub(subst, line))
    os.remove(file_path)
    shutil.move(abs_path, file_path)


###############################################################################
# Misc development tasks (change version, deploy GEE scripts)
###############################################################################


@task(
    help={
        "v": "Version to set",
        "testing": "Set for requirements-testing.txt?",
        "modules": "Also set versions for any modules specified "
        "in ext_libs.local_modules",
        "tag": "Also set tag(s)",
        "gee": "Also set versions for gee scripts",
    }
)
def set_version(c, v=None, testing=False, modules=False, tag=False, gee=False):
    # Validate the version matches the regex

    if not v:
        version_update = False
        v = get_version(c)
        print(
            "No version specified, retaining version {}, but updating SHA and release date".format(
                v
            )
        )
    elif not re.match("[0-9]+([.][0-9]+)+", v):
        print("Must specify a valid version (example: 0.36)")

        return
    else:
        version_update = True

    revision = (
        subprocess.check_output(["git", "rev-parse", "HEAD"])
        .decode("utf-8")
        .strip("\n")[0:8]
    )
    release_date = datetime.now(timezone.utc).strftime("%Y/%m/%d %H:%M:%SZ")

    # Set in version.json
    print("Setting version to {} in version.json".format(v))
    with open(c.plugin.version_file_details, "w") as f:
        json.dump(
            {"version": v, "revision": revision, "release_date": release_date},
            f,
            indent=4,
        )

    if version_update:
        # Set in version.txt
        print("Setting version to {} in {}".format(v, c.plugin.version_file_raw))
        with open(c.plugin.version_file_raw, "w") as f:
            f.write(v)

        # Set in Sphinx docs in make.conf
        print("Setting version to {} in sphinx conf.py".format(v))
        sphinx_regex = re.compile(
            '(((version)|(release)) = ")[0-9]+([.][0-9]+)+(rc[0-9]*)?', re.IGNORECASE
        )
        _replace(
            os.path.join(c.sphinx.sourcedir, "conf.py"), sphinx_regex, r"\g<1>" + v
        )

        # Set in metadata.txt
        print("Setting version to {} in metadata.txt".format(v))
        sphinx_regex = re.compile("^(version=)[0-9]+([.][0-9]+)+(rc[0-9]*)?")
        _replace(
            os.path.join(c.plugin.source_dir, "metadata.txt"),
            sphinx_regex,
            r"\g<1>" + v,
        )

        requirements_txt_regex = re.compile(
            "((trends.earth-schemas.git@)|(trends.earth-algorithms.git@))([.0-9a-z]*)"
        )

        if gee:
            # For the GEE config files the version can't have a dot, so convert to
            # underscore
            v_gee = v.replace(".", "_")

            if not v or not re.match("[0-9]+(_[0-9]+)+(rc[0-9]*)?", v_gee):
                print("Must specify a valid version (example: 0.36)")

                return

            gee_id_regex = re.compile('(, )?"id": "[0-9a-z-]*"(, )?')
            gee_script_name_regex = re.compile(
                '("name": "[0-9a-zA-Z -]*)( [0-9]+(_[0-9]+)+)(rc[0-9]*)?"'
            )

            # Set version for GEE scripts

            for subdir, dirs, files in os.walk(c.gee.script_dir):
                for file in files:
                    filepath = os.path.join(subdir, file)

                    if file == "configuration.json":
                        # Validate the version matches the regex
                        print("Setting version to {} in {}".format(v, filepath))
                        # Update the version string
                        _replace(
                            filepath, gee_script_name_regex, r"\g<1> " + v_gee + '"'
                        )
                        # Clear the ID since a new one will be assigned due to the new name
                        _replace(filepath, gee_id_regex, "")
                    elif file == "requirements.txt":
                        print("Setting version to {} in {}".format(v, filepath))

                        if ("rc" in v.split(".")[-1]) or (
                            int(v.split(".")[-1]) % 2 == 0
                        ):
                            # Last number in version string is even (or this is an RC), so
                            # use a tagged version of schemas matching this version
                            _replace(filepath, requirements_txt_regex, r"\g<1>v" + v)
                        else:
                            # Last number in version string is odd, so this is a development
                            # version, so use development version of schemas
                            _replace(filepath, requirements_txt_regex, r"\g<1>develop")
                    elif file == "__init__.py":
                        print("Setting version to {} in {}".format(v, filepath))
                        init_version_regex = re.compile(
                            "^(__version__[ ]*=[ ]*[\"'])[0-9]+([.][0-9]+)+(rc[0-9]*)?"
                        )
                        _replace(filepath, init_version_regex, r"\g<1>" + v)

            # Set in scripts.json
            print("Setting version to {} in scripts.json".format(v))
            scripts_regex = re.compile(
                '("version": ")[0-9]+([-._][0-9]+)+(rc[0-9]*)?', re.IGNORECASE
            )
            _replace(
                os.path.join(c.plugin.source_dir, "data", "scripts.json"),
                scripts_regex,
                r"\g<1>" + v,
            )

        requirements_file = (
            "requirements.txt" if not testing else "requirements-testing.txt"
        )
        print("Setting version to {} in package {}".format(v, requirements_file))

        if ("rc" in v.split(".")[-1]) or (int(v.split(".")[-1]) % 2 == 0):
            # Last number in version string is even (or an RC), so use a tagged version
            # of schemas matching this version
            _replace(requirements_file, requirements_txt_regex, r"\g<1>v" + v)
        else:
            # Last number in version string is odd, so this is a development
            # version, so use development version of schemas
            _replace(requirements_file, requirements_txt_regex, r"\g<1>develop")

    if tag:
        set_tag(c)

    if modules:
        for module in c.plugin.ext_libs.local_modules:
            module_path = Path(module["path"]).parent
            ret = query_yes_no(
                f"Also set version {'and tag ' if tag else ''}for {module['name']}?"
            )
            if ret:
                if tag:
                    subprocess.check_call(
                        ["invoke", "set-version", "-v", v, "-t"], cwd=module_path
                    )
                else:
                    subprocess.check_call(
                        ["invoke", "set-version", "-v", v], cwd=module_path
                    )


@task()
def release_github(c):
    v = get_version(c)

    # TODO: Add zipfile as an asset
    # https://docs.github.com/en/rest/reference/repos#upload-a-release-asset

    # Make release
    payload = {
        "tag_name": "v{}".format(v),
        "name": "Version {}".format(v),
        "body": """To install this release, download the LDMP.zip file below and then follow [the instructions for installing a release from Github](https://github.com/ConservationInternational/trends.earth#stable-version-from-zipfile).""",
    }

    s = requests.Session()
    res = s.get("https://github.com")
    cookies = dict(res.cookies)

    r = requests.post(
        "{}/repos/{}/{}/releases".format(
            c.github.api_url, c.github.repo_owner, c.github.repo_name
        ),
        json=payload,
        headers={"Authorization": "token {}".format(c.github.token)},
        cookies=cookies,
    )
    r.raise_for_status()
    # TODO: Link asset to release. See:
    # https://docs.github.com/en/rest/reference/repos#update-a-release-asset


@task(
    help={
        "modules": "Also set tag for any modules specified " "in ext_libs.local_modules"
    }
)
def set_tag(c, modules=False):
    v = get_version(c)
    ret = subprocess.run(
        ["git", "diff-index", "HEAD", "--"], capture_output=True, text=True
    )

    if ret.stdout != "":
        ret = query_yes_no("Uncommitted changes exist in repository. Commit these?")

        if ret:
            ret = subprocess.run(
                ["git", "commit", "-m", "Updating version tags for v{}".format(v)]
            )
            ret.check_returncode()
        else:
            print("Changes not committed - VERSION TAG NOT SET".format(v))

    print("Tagging version {} and pushing tag to origin".format(v))
    ret = subprocess.run(
        ["git", "tag", "-l", "v{}".format(v)], capture_output=True, text=True
    )
    ret.check_returncode()

    if "v{}".format(v) in ret.stdout:
        # Try to delete this tag on remote in case it exists there
        ret = subprocess.run(["git", "push", "origin", "--delete", "v{}".format(v)])

        if ret.returncode == 0:
            print("Deleted tag v{} on origin".format(v))
    subprocess.check_call(
        ["git", "tag", "-f", "-a", "v{}".format(v), "-m", "Version {}".format(v)]
    )
    subprocess.check_call(["git", "push", "origin", "v{}".format(v)])

    if modules:
        for module in c.plugin.ext_libs.local_modules:
            module_path = Path(module["path"]).parent
            print(f"Also setting tag for {module['name']}")
            subprocess.check_call(["invoke", "set-tag"], cwd=module_path)


def check_tecli_python_version():
    if sys.version_info[0] < 3:
        print(
            "ERROR: tecli tasks require Python version > 2 (you are running Python version {}.{})".format(
                sys.version_info[0], sys.version_info[1]
            )
        )

        return False
    else:
        return True


@task
def tecli_login(c):
    if not check_tecli_python_version():
        return
    subprocess.check_call(["python", os.path.abspath(c.gee.tecli), "login"])


@task
def tecli_clear(c):
    if not check_tecli_python_version():
        return
    subprocess.check_call(["python", os.path.abspath(c.gee.tecli), "clear"])


@task(help={"key": "GEE key in JSON format (base64 encoded)"})
def tecli_config(c, key):
    if not check_tecli_python_version():
        return
    subprocess.check_call(
        [
            "python",
            os.path.abspath(c.gee.tecli),
            "config",
            "set",
            "EE_SERVICE_ACCOUNT_JSON",
            key,
        ]
    )


@task(help={"script": "Script name", "overwrite": "Overwrite scripts if existing?"})
def tecli_publish(c, script=None, overwrite=False):
    if not check_tecli_python_version():
        return

    if not script and not overwrite:
        ret = query_yes_no(
            "WARNING: this will overwrite all scripts on the server with version {}.\nDo you wish to continue?".format(
                get_version(c)
            )
        )

        if not ret:
            return

    dirs = next(os.walk(c.gee.script_dir))[1]
    n = 0

    for dir in dirs:
        script_dir = os.path.join(c.gee.script_dir, dir)

        if os.path.exists(os.path.join(script_dir, "configuration.json")) and (
            script == None or script == dir
        ):
            print("Publishing {}...".format(dir))
            subprocess.check_call(
                [
                    "python",
                    os.path.abspath(c.gee.tecli),
                    "publish",
                    "--public=True",
                    "--overwrite=True",
                ],
                cwd=script_dir,
            )
            n += 1

    if script and n == 0:
        print('Script "{}" not found.'.format(script))

    # Updating GEE script IDs in config
    subprocess.check_call(["invoke", "update-script-ids"])


@task(
    help={
        "script": "Script name",
        "queryParams": "Parameters",
        "payload": "Parameters (as json",
    }
)
def tecli_run(c, script, queryParams=None, payload=None):
    if not check_tecli_python_version():
        return
    dirs = next(os.walk(c.gee.script_dir))[1]
    n = 0
    script_dir = None

    for dir in dirs:
        script_dir = os.path.join(c.gee.script_dir, dir)

        if (
            os.path.exists(os.path.join(script_dir, "configuration.json"))
            and script == dir
        ):
            print("Running {}...".format(dir))

            if queryParams:
                print("Using given query parameters as input to script.")
                subprocess.check_call(
                    [
                        "python",
                        os.path.abspath(c.gee.tecli),
                        "start",
                        "--queryParams={}".format(queryParams),
                    ],
                    cwd=script_dir,
                )
            elif payload:
                print("Using given payload as input to script.")
                subprocess.check_call(
                    [
                        "python",
                        os.path.abspath(c.gee.tecli),
                        "start",
                        "--payload={}".format(os.path.abspath(payload)),
                    ],
                    cwd=script_dir,
                )
            else:
                print("Running script without any input parameters.")
                subprocess.check_call(
                    ["python", os.path.abspath(c.gee.tecli), "start"], cwd=script_dir
                )

            n += 1

            break

    if script and n == 0:
        print('Script "{}" not found.'.format(script))


@task
def update_script_ids(c):
    with open(c.gee.scripts_json_file) as fin:
        scripts = json.load(fin)

    dirs = next(os.walk(c.gee.script_dir))[1]
    n = 0
    script_dir = None

    for dir in dirs:
        script_dir = os.path.join(c.gee.script_dir, dir)

        if os.path.exists(os.path.join(script_dir, "configuration.json")):
            with open(os.path.join(script_dir, "configuration.json")) as fin:
                config = json.load(fin)
            try:
                script_name = re.compile("( [0-9]+(_[0-9]+)+$)").sub("", config["name"])

                # Find location of this script in the list of scripts in the
                # script configuration JSON
                script_index = [
                    index
                    for index, script in enumerate(scripts)
                    if script["name"] == script_name
                ]
                assert len(script_index) <= 1
                script_index = script_index[0]

                try:
                    scripts[script_index]["id"] = config["id"]
                except KeyError:
                    print(f"No id found in config for {script_name}")
                script_version = (
                    re.compile("^[a-zA-Z0-9-]* ")
                    .sub("", config["name"])
                    .replace("_", ".")
                )
                scripts[script_index]["version"] = script_version
            except IndexError:
                print(
                    f"Skipping {script_name} as not found in scripts.json - maybe need to publish?"
                )
    with open(c.gee.scripts_json_file, "w") as f_out:
        json.dump(scripts, f_out, sort_keys=True, indent=4)


@task(help={"script": "Script name"})
def tecli_info(c, script=None):
    if not check_tecli_python_version():
        return
    dirs = next(os.walk(c.gee.script_dir))[1]
    n = 0
    script_dir = None

    for dir in dirs:
        script_dir = os.path.join(c.gee.script_dir, dir)

        if os.path.exists(os.path.join(script_dir, "configuration.json")) and (
            script == None or script == dir
        ):
            print("Checking info on {}...".format(dir))
            subprocess.check_call(
                ["python", os.path.abspath(c.gee.tecli), "info"], cwd=script_dir
            )
            n += 1

    if script and n == 0:
        print('Script "{}" not found.'.format(script))


@task(
    help={
        "script": "Script name",
        "since": "Print logs since (number of hours = default 1)",
    }
)
def tecli_logs(c, script, since=1):
    if not check_tecli_python_version():
        return
    dirs = next(os.walk(c.gee.script_dir))[1]
    n = 0
    script_dir = None

    for dir in dirs:
        script_dir = os.path.join(c.gee.script_dir, dir)

        if (
            os.path.exists(os.path.join(script_dir, "configuration.json"))
            and script == dir
        ):
            print("Checking logs for {}...".format(dir))
            subprocess.check_call(
                [
                    "python",
                    os.path.abspath(c.gee.tecli),
                    "logs",
                    f"--since={since}",
                ],
                cwd=script_dir,
            )
            n += 1

            break

    if script and n == 0:
        print('Script "{}" not found.'.format(script))


###############################################################################
# Setup dependencies and install package
###############################################################################


def not_comments(lines, s, e):
    return [line for line in lines[s:e] if line[0] != "#"]


def read_requirements():
    """Return a list of runtime and list of test requirements"""
    with open("requirements.txt") as f:
        lines = f.readlines()
    lines = [line for line in [line.strip() for line in lines] if line]
    divider = "# test requirements"

    try:
        idx = lines.index(divider)
    except ValueError:
        raise Exception('Expected to find "{}" in requirements.txt'.format(divider))

    return not_comments(lines, 0, idx), not_comments(lines, idx + 1, None)


def _safe_remove_folder(rootdir):
    """
    Supports removing a folder that may have symlinks in it

    Needed on windows to avoid removing the original files linked to within
    each folder
    """
    rootdir = Path(rootdir)
    if rootdir.is_symlink():
        rootdir.rmdir()
    else:
        folders = [path for path in Path(rootdir).iterdir() if path.is_dir()]
        for folder in folders:
            if folder.is_symlink():
                folder.rmdir()
            else:
                shutil.rmtree(folder)
        files = [path for path in Path(rootdir).iterdir()]
        for file in files:
            file.unlink()
        shutil.rmtree(rootdir)


@task(
    help={
        "clean": "Clean out dependencies first",
        "link": "Symlink dependendencies to their local repos",
        "pip": 'Path to pip (usually "pip" or "pip3"',
    }
)
def plugin_setup(c, clean=True, link=False, pip="pip"):
    """install dependencies"""
    ext_libs = os.path.abspath(c.plugin.ext_libs.path)

    if clean and os.path.exists(ext_libs):
        _safe_remove_folder(ext_libs)

    if sys.version_info[0] < 3:
        if not os.path.exists(ext_libs):
            os.makedirs(ext_libs)
    else:
        os.makedirs(ext_libs, exist_ok=True)
    runtime, test = read_requirements()

    os.environ["PYTHONPATH"] = ext_libs

    for req in runtime + test:
        # Don't install numpy with pyqtgraph as QGIS already has numpy. So use
        # the --no-deps flag (-N for short) with that package only.

        if "pyqtgraph" in req:
            subprocess.check_call(
                [pip, "install", "--upgrade", "--no-deps", "-t", ext_libs, req]
            )
        else:
            subprocess.check_call([pip, "install", "--upgrade", "-t", ext_libs, req])

    if link:
        for module in c.plugin.ext_libs.local_modules:
            l = os.path.abspath(c.plugin.ext_libs.path) + os.path.sep + module["name"]

            if os.path.islink(l):
                print(f"{l} is already a link (to {os.readlink(l)})")
            else:
                print(
                    "Linking local repo of {} to plugin ext_libs".format(module["name"])
                )
                shutil.rmtree(l)
                os.symlink(module["path"], l)


@task(
    help={
        "clean": "remove existing install folder first",
        "version": "what version of QGIS to install to",
        "profile": "what profile to install to (only applies to QGIS3",
        "fast": "Skip compiling numba files",
        "link": "Symlink folder to QGIS profile directory",
    }
)
def plugin_install(
    c, clean=False, version=3, profile="default", fast=False, link=False
):
    """install plugin to qgis"""
    set_version(c)
    plugin_name = c.plugin.name
    src = os.path.join(os.path.dirname(__file__), plugin_name)

    if version == 2:
        folder = ".qgis2"
    elif version == 3:
        if platform.system() == "Darwin":
            folder = "Library/Application Support/QGIS/QGIS3/profiles/"

        if platform.system() == "Linux":
            folder = ".local/share/QGIS/QGIS3/profiles/"

        if platform.system() == "Windows":
            folder = "AppData\\Roaming\\QGIS\\QGIS3\\profiles\\"
        folder = os.path.join(folder, profile)
    else:
        print("ERROR: unknown qgis version {}".format(version))

        return

    dst_plugins = os.path.join(os.path.expanduser("~"), folder, "python", "plugins")
    dst_this_plugin = os.path.join(dst_plugins, plugin_name)
    src = os.path.abspath(src)
    dst_this_plugin = os.path.abspath(dst_this_plugin)

    if not hasattr(os, "symlink") or not link:
        if clean and os.path.exists(dst_this_plugin):
            print(f"Removing folder {dst_this_plugin}")
            _safe_remove_folder(dst_this_plugin)

        print(
            "Copying plugin to QGIS version {} plugin folder at {}".format(
                version, dst_this_plugin
            )
        )
        for root, dirs, files in os.walk(src):
            relpath = os.path.relpath(root)

            if not os.path.exists(os.path.join(dst_plugins, relpath)):
                os.makedirs(os.path.join(dst_plugins, relpath))

            for f in _filter_excludes(root, files, c):
                try:
                    shutil.copy(
                        os.path.join(root, f), os.path.join(dst_plugins, relpath, f)
                    )
                except PermissionError:
                    print(
                        "Permission error: unable to copy {} to {}. Skipping that file.".format(
                            f, os.path.join(dst_plugins, relpath, f)
                        )
                    )
            _filter_excludes(root, dirs, c)
    else:
        if clean and os.path.exists(dst_this_plugin):
            print(f"Removing folder {dst_this_plugin}")
            _safe_remove_folder(dst_this_plugin)

        if os.path.exists(dst_this_plugin):
            print(
                f"Not linking - plugin folder for QGIS version {version} already "
                f"exists at {dst_this_plugin}. Use '-c' to clean that folder if "
                "desired."
            )
        else:
            print(
                "Linking plugin development folder to QGIS version {} plugin folder at {}".format(
                    version, dst_this_plugin
                )
            )
            os.symlink(src, dst_this_plugin)


def file_changed(infile, outfile):
    try:
        infile_s = os.stat(infile)
        outfile_s = os.stat(outfile)

        return infile_s.st_mtime > outfile_s.st_mtime
    except:
        return True


def _filter_excludes(root, items, c):
    excludes = set(c.plugin.excludes)
    skips = c.plugin.skip_exclude

    exclude = lambda p: any([fnmatch.fnmatch(p, e) for e in excludes])

    if not items:
        return []

    # to prevent descending into dirs, modify the list in place

    for item in list(items):  # copy list or iteration values change
        itempath = os.path.join(os.path.relpath(root), item)

        if exclude(itempath) and item not in skips:
            # debug('Excluding {}'.format(itempath))
            items.remove(item)

    return items


# Compile all ui and resource files - only used on CI so that gui can be loaded
# on docker image
@task(
    help={
        "clean": "remove existing install folder first",
    }
)
def compile_files(c, clean=False):
    pyrcc = "pyrcc5"
    pyrcc_path = check_path(pyrcc)

    if not pyrcc:
        print(
            "ERROR: {} is not in your path---unable to compile resource file(s)".format(
                pyrcc
            )
        )
        return
    else:
        res_files = c.plugin.resource_files
        res_count = 0
        skip_count = 0
        for res in res_files:
            if os.path.exists(res):
                (base, ext) = os.path.splitext(res)
                output = "{}.py".format(base)
                if clean or file_changed(res, output):
                    print("Compiling {} to {}".format(res, output))
                    subprocess.check_call([pyrcc_path, "-o", output, res])
                    res_count += 1
                else:
                    skip_count += 1
            else:
                print("{} does not exist---skipped".format(res))
        print("Compiled {} resource files. Skipped {}.".format(res_count, skip_count))


# Below is based on pb_tool:
# https://github.com/g-sherman/plugin_build_tool
def check_path(app):
    """Adapted from StackExchange:
    http://stackoverflow.com/questions/377017
    """

    def is_exe(fpath):
        return os.path.exists(fpath) and os.access(fpath, os.X_OK)

    def ext_candidates(fpath):
        yield fpath

        for ext in os.environ.get("PATHEXT", "").split(os.pathsep):
            yield fpath + ext

    # Also look in folders under this python versions lib path
    folders = os.environ["PATH"].split(os.pathsep)
    folders.extend(
        [x[0] for x in os.walk(os.path.join(os.path.dirname(sys.executable), "Lib"))]
    )
    fpath, fname = os.path.split(app)

    if fpath:
        if is_exe(app):
            return app
    else:
        for path in folders:
            exe_file = os.path.join(path, app)

            for candidate in ext_candidates(exe_file):
                if is_exe(candidate):
                    return candidate

    return None


###############################################################################
# Translation
###############################################################################


@task
def lrelease(c):
    print("Releasing translations using lrelease...")
    lrelease = check_path("lrelease")

    if not lrelease:
        print(
            "ERROR: lrelease is not in your path---unable to release translation files"
        )

        return

    for translation in c.plugin.translations:
        subprocess.check_call(
            [
                lrelease,
                os.path.join(c.plugin.i18n_dir, "LDMP_{}.ts".format(translation)),
            ]
        )


@task(
    help={
        "force": "Force the download of the translations files regardless of whether "
        "timestamps on the local computer are newer than those on the server"
    }
)
def translate_pull(c, force=False):
    print("Pulling transifex translations...")

    if force:
        subprocess.check_call([c.sphinx.tx_path, "pull", "-f"])
    else:
        subprocess.check_call([c.sphinx.tx_path, "pull"])

    lrelease(c)


# @task
# def translate_update_resources(c):
#     print("Updating transifex...")
#     subprocess.check_call("sphinx-intl update-txconfig-resources --pot-dir {docroot}/i18n/pot --transifex-project-name {transifex_name}".format(docroot=c.sphinx.docroot, transifex_name=c.sphinx.transifex_name))
#


@task(
    help={
        "force": "Push source files to transifex without checking modification times",
        "version": "what version of QGIS to install to",
    }
)
def translate_push(c, force=False, version=3):
    print("Building changelog...")
    changelog_build(c)

    print("Building download page...")
    build_download_page(c)

    # Below is necessary just to avoid warning messages regarding missing image
    # files when Sphinx is used later on
    print("Localizing resources...")
    localize_resources(c, "en")

    print("Gathering strings...")
    gettext(c)
    print("Generating the pot files for the LDMP toolbox help files...")

    for translation in c.plugin.translations:
        subprocess.check_call(
            c.sphinx.sphinx_intl.split()
            + [
                "--config",
                f"{c.sphinx.sourcedir}/conf.py",
                "update",
                "-p",
                f"{c.sphinx.docroot}/i18n/pot",
                "-l",
                f"{translation}",
            ]
        )

    print("Gathering strings for translation using pylupdate...")

    if version == 2:
        pylupdate = "pylupdate4"
    elif version == 3:
        pylupdate = "pylupdate5"
    else:
        print("ERROR: unknown qgis version {}".format(version))

        return
    pylupdate = check_path(pylupdate)

    if not pylupdate:
        print(
            "ERROR: pylupdate4/pylupdate5 is not in your path---unable to gather strings for translation"
        )

        return
    else:
        subprocess.check_call(
            [pylupdate, os.path.join(c.plugin.i18n_dir, "i18n.pro"), "-noobsolete"]
        )

    if force:
        subprocess.check_call([c.sphinx.tx_path, "push", "-f", "-s"])
    else:
        subprocess.check_call([c.sphinx.tx_path, "push", "-s"])


@task(help={"language": "language"})
def gettext(c, language=None):
    if not language:
        language = c.sphinx.base_language
    script_folder = str(Path(__file__).parent)
    SPHINX_OPTS = (
        f"-t language_{language} -A language={language} "
        f"{script_folder}/{c.sphinx.sourcedir}"
    )
    I18N_SPHINX_OPTS = f"{SPHINX_OPTS} {script_folder}/{c.sphinx.docroot}/i18n/pot"

    subprocess.check_call(
        c.sphinx.sphinx_build.split()
        + ["-b", "gettext", "-a"]
        + I18N_SPHINX_OPTS.split()
    )


###############################################################################
# Build documentation
###############################################################################


@task(
    help={
        "ignore_errors": "ignore documentation errors",
        "language": "which language to build (all are built by default)",
        "fast": "only check english docs",
    }
)
def docs_spellcheck(c, ignore_errors=False, language=None, fast=False):
    if language:
        languages = [language]
    else:
        languages = [c.sphinx.base_language]
        languages.extend(c.plugin.translations)

    for language in languages:
        print(f"\nBuilding {language} documentation...")
        SPHINX_OPTS = (
            f"-t language_{language} -A language={language} {c.sphinx.sourcedir}"
        )

        if language != "en" or ignore_errors:
            subprocess.check_call(
                c.sphinx.sphinx_intl.split()
                + ["-b", "spelling", "-a"]
                + SPHINX_OPTS.split()
                + [f"{c.sphinx.builddir}/html/{language}"]
            )
        else:
            subprocess.check_call(
                c.sphinx.sphinx_build.split()
                + ["-n", "-W", "-b", "spelling", "-a"]
                + SPHINX_OPTS.split()
                + [f"{c.sphinx.builddir}/html/{language}"]
            )

        if fast:
            break


@task(
    help={
        "clean": "clean out built artifacts first",
        "ignore_errors": "ignore documentation errors",
        "language": "which language to build (all are built by default)",
        "fast": "only build english html docs",
        "pdf": "build pdf docs",
        "upload": "upload pdfs of docs to S3",
    }
)
def docs_build(
    c,
    clean=False,
    ignore_errors=False,
    language=None,
    fast=False,
    pdf=False,
    upload=False,
):
    if clean:
        rmtree(c.sphinx.builddir)

    if language:
        languages = [language]
    else:
        languages = [c.sphinx.base_language]
        languages.extend(c.plugin.translations)

    if fast:
        pdf = False
        languages = ["en"]

    print("\nBuilding changelog...")
    changelog_build(c)

    print("\nBuilding download page...")
    build_download_page(c)

    client = _get_s3_client()

    for language in languages:
        print(f"\nBuilding {language} documentation...")
        SPHINX_OPTS = (
            f"-t language_{language} -A language={language} {c.sphinx.sourcedir}"
        )

        print(f"\nLocalizing resources for {language} documentation...")

        localize_resources(c, language)

        subprocess.check_call(
            c.sphinx.sphinx_intl.split()
            + [
                "--config",
                f"{c.sphinx.sourcedir}/conf.py",
                "build",
                f"--language={language}",
            ]
        )

        if language != "en" or ignore_errors:
            subprocess.check_call(
                c.sphinx.sphinx_build.split()
                + ["-b", "html", "-a"]
                + SPHINX_OPTS.split()
                + [f"{c.sphinx.builddir}/html/{language}"]
            )
        else:
            subprocess.check_call(
                c.sphinx.sphinx_build.split()
                + ["-n", "-W", "-b", "html", "-a"]
                + SPHINX_OPTS.split()
                + [f"{c.sphinx.builddir}/html/{language}"]
            )
        print(
            f"HTML Build finished. The HTML pages for '{language}' "
            f"are in {c.sphinx.builddir}/html/{language}."
        )

        if pdf:
            # Build PDF, by first making latex from sphinx, then pdf from that
            tex_dir = f"{c.sphinx.builddir}/latex/{language}"
            subprocess.check_call(
                c.sphinx.sphinx_build.split()
                + ["-b", "latex", "-a"]
                + SPHINX_OPTS.split()
                + [f"{tex_dir}"]
            )

            tex_files = [
                Path(tex_file).name for tex_file in glob.glob(f"{tex_dir}/*.tex")
            ]
            for tex_file in tex_files:
                for _ in range(3):
                    # Run multiple times to ensure crossreferences are right
                    subprocess.check_call(["xelatex", tex_file], cwd=tex_dir)
                # Move the PDF to the html folder so it will be uploaded with the
                # site
                pdf_file = os.path.splitext(tex_file)[0] + ".pdf"
                out_dir = f"{c.sphinx.builddir}/html/{language}/pdfs"

                if not os.path.exists(out_dir):
                    os.makedirs(out_dir)
                shutil.move(f"{tex_dir}/{pdf_file}", f"{out_dir}/{pdf_file}")

                if upload:
                    data = open(f"{out_dir}/{pdf_file}", "rb")
                    key = f"documentation/{pdf_file}"
                    client.put_object(
                        Key=key,
                        Body=data,
                        Bucket=c.sphinx.documentation_deploy_s3_bucket,
                    )
                    client.put_object_acl(
                        ACL="public-read",
                        Key=key,
                        Bucket=c.sphinx.documentation_deploy_s3_bucket,
                    )
                    data.close()
                    print(f"{pdf_file} uploaded to S3")
            print(
                f"PDF Build finished. The PDF pages for '{language}' are in {out_dir}."
            )


def localize_resources(c, language=None):
    if language is None:
        language = c.sphinx.base_language

    print(
        "Removing all static content from {sourcedir}/static.".format(
            sourcedir=c.sphinx.sourcedir
        )
    )

    if os.path.exists("{sourcedir}/static".format(sourcedir=c.sphinx.sourcedir)):
        rmtree("{sourcedir}/static".format(sourcedir=c.sphinx.sourcedir))
    print(
        "Copy 'en' (base) static content to {sourcedir}/static.".format(
            sourcedir=c.sphinx.sourcedir
        )
    )

    if os.path.exists("{resourcedir}/en".format(resourcedir=c.sphinx.resourcedir)):
        shutil.copytree(
            "{resourcedir}/en".format(resourcedir=c.sphinx.resourcedir),
            "{sourcedir}/static".format(sourcedir=c.sphinx.sourcedir),
        )
    print(
        "Copy localized '{lang}' static content to {sourcedir}/static.".format(
            lang=language, sourcedir=c.sphinx.sourcedir
        )
    )

    if language != "en" and os.path.exists(
        "{resourcedir}/{lang}".format(resourcedir=c.sphinx.resourcedir, lang=language)
    ):
        src = "{resourcedir}/{lang}".format(
            resourcedir=c.sphinx.resourcedir, lang=language
        )
        dst = "{sourcedir}/static".format(sourcedir=c.sphinx.sourcedir)

        for item in os.listdir(src):
            s = os.path.join(src, item)
            d = os.path.join(dst, item)

            if os.path.isdir(s):
                shutil.copytree(s, d)
            else:
                shutil.copy2(s, d)


def check_docs_image_ext(c):
    """
    Check the capitalization of *.png images for docs. They should be in
    lowercase otherwise they will be skipped from the build process as Linux
    is case-sensitive w.r.t. file extensions.
    """
    res_dir = c.sphinx.resourcedir
    if not os.path.exists(res_dir):
        print("doc's resource directory does not exist. Unable to browse image files.")
        return

    for p in Path(res_dir).rglob("*.PNG"):
        if p.suffix.isupper():
            np = Path(f"{p.with_suffix('').as_posix()}.png")
            os.rename(p.as_posix(), np.as_posix())


@task
def rtd_pre_build(c):
    """
    Checks capitalization and copies docs resources based on language
    prior to build in RTD.
    """
    print("Checking case of image extensions...")
    check_docs_image_ext(c)
    print("Copying docs resources based on language...")
    localize_resources(c)
    if os.environ.get("READTHEDOCS_PROJECT") == "trends.earth" and os.environ.get(
        "READTHEDOCS_VERSION_TYPE"
    ) in ["branch", "tag"]:
        print("Building download page...")
        build_download_page(c)
    print("Building changelog...")
    changelog_build(c)


@task
def changelog_build(c):
    out_txt = [
        "Changelog\n",
        "======================\n",
        "\n",
        "This page lists the version history of |trends.earth|.\n",
    ]

    with open(os.path.join(c.plugin.source_dir, "metadata.txt")) as fin:
        metadata = fin.readlines()

    changelog_header_re = re.compile("^changelog=", re.IGNORECASE)
    version_header_re = re.compile(
        r"^[ ]*[0-9]+(\.[0-9]+){1,2}(rc[0-9]*)?", re.IGNORECASE
    )

    at_changelog = False

    for line in metadata:
        if not at_changelog and not changelog_header_re.match(line):
            continue
        elif changelog_header_re.match(line):
            line = changelog_header_re.sub("  ", line)
            at_changelog = True
        version_header = version_header_re.match(line)

        if version_header:
            version_number = version_header.group(0)
            version_number = version_number.strip(" \n")
            line = line.strip(" \n")
            line = "\n`{} <https://github.com/ConservationInternational/trends.earth/releases/tag/{}>`_\n".format(
                line, version_number
            )
            line = [
                line,
                "-----------------------------------------------------------------------------------------------------------------------------\n\n",
            ]
        out_txt.extend(line)

    out_file = "{docroot}/source/for_developers/changelog.rst".format(
        docroot=c.sphinx.docroot
    )
    with open(out_file, "w") as fout:
        metadata = fout.writelines(out_txt)


def _make_download_link(c, title, key, data):
    filename = data.get(key, "")
    if filename:
        return (
            f"[{title}]"
            f"(https://{c.data_downloads.s3_bucket}/"
            f"{c.data_downloads.s3_prefix}{filename})"
        )
    else:
        return ""


def _make_sdg_download_row(c, iso, data):
    return (
        f"| {iso} | "
        + f'{_make_download_link(c, f"{iso} (JRC LPD)", "JRC-LPD-5", data)} | '
        + f'{_make_download_link(c, f"{iso} (Trends.Earth LPD)", "TrendsEarth-LPD-5", data)} | '
        + f'{_make_download_link(c, f"{iso} (FAO-WOCAT LPD)", "FAO-WOCAT-LPD-5", data)} |\n'
    )


def _make_drought_download_row(c, iso, data):
    return (
        f"| {iso} | "
        + f'{_make_download_link(c, f"{iso} (Drought)", "Drought", data)} |\n'
    )


@task
def build_download_page(c):
    out_txt = """# Downloads

This page lists data packages containing default datasets that can be used in
Trends.Earth.

**This site and the products of Trends.Earth are made available under the terms of the
Creative Commons Attribution 4.0 International License (CC BY 4.0). The boundaries and
names used, and the designations used, do not imply official endorsement or acceptance
by Conservation International Foundation, or its partner organizations and
contributors.**

## SDG Indicator 15.3.1 (UNCCD Strategic Objectives 1 and 2)

The below datasets can be used to support assessing SDG Indicator 15.3.1, and include
indicators of change in land productivity dynamics (LPD), land cover, and soil organic
carbon. These datasets can be used to support reporting on UNCCD Strategic Objectives 1
and 2. Note that there are three different LPD datasets available (from JRC, from
the default Trends.Earth method, and from FAO-WOCAT).

| Country | SDG 15.3.1 using JRC LPD | SDG 15.3.1 using Trends.Earth LPD  | SDG 15.3.1 using FAO-WOCAT LPD |
|---------|---------|--------------------|---------------|
"""

    client = _get_s3_client()

    objects = client.list_objects(
        Bucket=c.data_downloads.s3_bucket, Prefix=c.data_downloads.s3_prefix
    )["Contents"]

    sdg_links = {}
    for item in objects:
        if item["Key"] == c.data_downloads.s3_prefix:
            # Skip the key that is just the parent folder itself
            continue
        filename = PurePath(item["Key"]).name
        iso = re.match("[A-Z]{3}", filename)[0]

        if iso not in sdg_links:
            sdg_links[iso] = {}

        if re.search("SDG15_JRC-LPD-5", filename):
            sdg_links[iso]["JRC-LPD-5"] = filename
        elif re.search("SDG15_TrendsEarth-LPD-5", filename):
            sdg_links[iso]["TrendsEarth-LPD-5"] = filename
        elif re.search("SDG15_FAO-WOCAT-LPD-5", filename):
            sdg_links[iso]["FAO-WOCAT-LPD-5"] = filename
        else:
            continue

    for iso, values in sdg_links.items():
        out_txt += _make_sdg_download_row(c, iso, values)

    out_txt += """

## Drought hazard, vulnerability and exposure (UNCCD Strategic Objective 3)

The below datasets can be used to support assessing drought hazard, vulnerability, and
exposure, and for reporting on UNCCD Strategic Objective 3.

| Country | Drought indicators (2000-2019) |
|---------|--------------------------------|
"""

    drought_links = {}
    for item in objects:
        if item["Key"] == c.data_downloads.s3_prefix:
            # Skip the key that is just the parent folder itself
            continue
        filename = PurePath(item["Key"]).name
        iso = re.match("[A-Z]{3}", filename)[0]

        if iso not in drought_links:
            drought_links[iso] = {}

        if re.search("Drought", filename):
            drought_links[iso]["Drought"] = filename
        else:
            continue

    for iso, values in drought_links.items():
        out_txt += _make_drought_download_row(c, iso, values)

    with open(
        os.path.join(os.path.dirname(__file__), c.data_downloads.downloads_page), "w"
    ) as fout:
        fout.writelines(out_txt)


###############################################################################
# Package plugin zipfile
###############################################################################


@task(
    help={
        "clean": "Clean out dependencies before packaging",
        "version": "what version of QGIS to prepare ZIP file for",
        "tests": "Package tests with plugin",
        "filename": "Name for output file",
        #'python': 'Python to use for setup and compiling',
        "pip": 'Path to pip (usually "pip" or "pip3"',
        "tag": "Whether to tag on Github",
    }
)
def zipfile_build(
    c, clean=True, version=3, tests=False, filename=None, pip="pip", tag=False
):
    """Create plugin package"""
    set_version(c, modules=True)

    if tag:
        set_tag(c)
    else:
        print(
            "***Not setting tag on github***\nIf this is a "
            "production deployment you MUST tag this version on github, "
            "so cancel this process and re-run with tag=True"
        )

    plugin_setup(c, clean=clean, pip=pip)

    # Make sure compiled versions of translation files are included
    lrelease(c)

    package_dir = c.plugin.package_dir

    if sys.version_info[0] < 3:
        if not os.path.exists(package_dir):
            os.makedirs(package_dir)
    else:
        os.makedirs(package_dir, exist_ok=True)
    # package_file =  os.path.join(package_dir, '{}_{}.zip'.format(c.plugin.name, get_version()))

    if not filename:
        filename = os.path.join(
            package_dir, "{}_QGIS{}.zip".format(c.plugin.name, version)
        )

    print(f"Removing untracked datafiles from {c.plugin.data_dir}...")
    subprocess.check_call(["git", "clean", "-f", "-x", c.plugin.data_dir])

    print("Building zipfile...")
    with zipfile.ZipFile(filename, "w", zipfile.ZIP_DEFLATED) as zf:
        if not tests:
            c.plugin.excludes.extend(c.plugin.tests)
        _make_zip(zf, c)

    return filename


def _make_zip(zipFile, c):
    src_dir = c.plugin.source_dir

    for root, dirs, files in os.walk(src_dir):
        for f in _filter_excludes(root, files, c):
            relpath = os.path.relpath(root)
            zipFile.write(os.path.join(root, f), os.path.join(relpath, f))
        _filter_excludes(root, dirs, c)


@task(
    help={
        "qgis": "QGIS version to target",
        "clean": "Clean out dependencies and untracked data files before packaging",
        "pip": 'Path to pip (usually "pip" or "pip3"',
        "tag": "Whether to tag on Github",
        "filename": "Name for output file",
    }
)
def zipfile_deploy(c, qgis, clean=True, pip="pip", tag=False, filename=None):
    binaries_sync(c)
    binaries_deploy(c, qgis=qgis)
    print("Binaries uploaded")

    filename = zipfile_build(c, pip=pip, clean=clean, tag=tag, filename=filename)
    client = _get_s3_client()

    print("Uploading package to S3")
    data = open(filename, "rb")
    client.put_object(
        Key="sharing/{}".format(os.path.basename(filename)),
        Body=data,
        Bucket=c.sphinx.zipfile_deploy_s3_bucket,
    )
    data.close()
    print("Package uploaded")


# Function
def _recursive_dir_create(d):
    if sys.version_info[0] < 3:
        if not os.path.exists(d):
            os.makedirs(os.path.join(os.path.abspath(os.path.dirname(d)), ""))
    else:
        os.makedirs(
            os.path.join(os.path.abspath(os.path.dirname(d)), ""), exist_ok=True
        )


def _get_s3_client():
    try:
        with open(
            os.path.join(os.path.dirname(__file__), "aws_credentials.json")
        ) as fin:
            keys = json.load(fin)
        client = boto3.client(
            "s3",
            aws_access_key_id=keys["access_key_id"],
            aws_secret_access_key=keys["secret_access_key"],
        )
    except OSError:
        print(
            "Warning: AWS credentials file not found. Credentials must be in environment variable or in default AWS credentials location."
        )
        client = boto3.client("s3")

    return client


def _s3_sync(c, bucket, s3_prefix, local_folder, patterns=["*"]):
    client = _get_s3_client()

    objects = client.list_objects(Bucket=bucket, Prefix="{}/".format(s3_prefix))[
        "Contents"
    ]

    for obj in objects:
        filename = os.path.basename(obj["Key"])

        if filename == "":
            # Catch the case of the key pointing to the root of the bucket and
            # skip it

            continue
        local_path = os.path.join(local_folder, filename)

        # First ensure all the files that are on S3 are up to date relative to
        # the local files, copying files in either direction as necessary

        if os.path.exists(local_path):
            if not _check_hash(obj["ETag"].strip('"'), local_path):
                lm_s3 = obj["LastModified"]
                lm_local = datetime.fromtimestamp(
                    os.path.getmtime(local_path), lm_s3.tzinfo
                )

                if lm_local > lm_s3:
                    print(
                        "Local version of {} is newer than on S3 - copying to S3.".format(
                            filename
                        )
                    )
                    data = open(local_path, "rb")
                    client.put_object(
                        Key="{}/{}".format(s3_prefix, os.path.basename(filename)),
                        Body=data,
                        Bucket=bucket,
                    )
                    data.close()
                else:
                    print(
                        "S3 version of {} is newer than local - copying to local.".format(
                            filename
                        )
                    )
                    _recursive_dir_create(local_path)
                    client.download_file(
                        Key="{}/{}".format(s3_prefix, os.path.basename(filename)),
                        Bucket=bucket,
                        Filename=local_path,
                    )
        else:
            print("Local version of {} is missing - copying to local.".format(filename))
            _recursive_dir_create(local_path)
            client.download_file(
                Key="{}/{}".format(s3_prefix, os.path.basename(filename)),
                Bucket=bucket,
                Filename=local_path,
            )

    # Now copy back to S3 any files that aren't yet there
    files = [glob.glob(pattern) for pattern in patterns]
    files = [item for sublist in files for item in sublist]
    s3_objects = client.list_objects(Bucket=bucket, Prefix="{}/".format(s3_prefix))[
        "Contents"
    ]
    s3_object_names = [os.path.basename(obj["Key"]) for obj in s3_objects]

    for f in files:
        if os.path.isdir(f):
            continue

        if not os.path.basename(f) in s3_object_names:
            print("S3 is missing {} - copying to S3.".format(f))
            data = open(f, "rb")
            client.put_object(
                Key="{}/{}".format(s3_prefix, os.path.basename(f)),
                Body=data,
                Bucket=bucket,
            )
            data.close()


def _check_hash(expected, filename):
    md5hash = hashlib.md5(open(filename, "rb").read()).hexdigest()

    if md5hash == expected:
        return True
    else:
        return False


@task(help={"extensions": "Which file extensions to sync"})
def binaries_sync(c, extensions=None):
    if not extensions:
        extensions = c.plugin.numba.binary_extensions
    client = _get_s3_client()
    patterns = [os.path.join(c.plugin.numba.binary_folder, "*" + p) for p in extensions]
    _s3_sync(
        c,
        c.sphinx.zipfile_deploy_s3_bucket,
        "plugin_binaries",
        c.plugin.numba.binary_folder,
        patterns,
    )


@task
def testdata_sync(c):
    _s3_sync(
        c,
        c.sphinx.zipfile_deploy_s3_bucket,
        "plugin_testdata",
        "LDMP/test/integration/fixtures",
        c.plugin.testdata_patterns,
    )


def find_binaries(c, folder, version=None):
    files = []

    for pattern in c.plugin.numba.binary_extensions:
        if version:
            files.append(
                [
                    f
                    for f in os.listdir(folder)
                    if re.search(f"{version}.*{pattern}$", f)
                ]
            )
        else:
            files.append(
                [
                    f
                    for f in os.listdir(folder)
                    if re.search(r".*{}$".format(pattern), f)
                ]
            )
    # Return a flattened list

    return [item for sublist in files for item in sublist]


@task(help={"qgis": "QGIS version to target"})
def binaries_deploy(c, qgis):
    # Copy down any missing binaries
    binaries_sync(c)

    v = get_version(c).replace(".", "-")
    qgis = qgis.replace(".", "-")
    zipfile_basename = f"trends_earth_binaries_{v}_{qgis}"
    files = find_binaries(c, c.plugin.numba.binary_folder, v)
    with TemporaryDirectory() as tmpdir:
        # Make module dir within the tmp dir, inside a folder containing the
        # version string - this is needed to allow clean unzipping of the
        # modules later on machines that might have multiple versions of the
        # binaries installed in the same folder
        moduledir = os.path.join(tmpdir, zipfile_basename, "trends_earth_binaries")
        os.makedirs(moduledir)
        with open(os.path.join(moduledir, "__init__.py"), "w") as fp:
            pass
        # Copy binaries to temp folder for later zipping

        for f in files:
            # Strip version string from files before placing them in zipfile
            out_path_no_v = os.path.join(
                moduledir, re.sub("_[0-9]+-[0-9]+(-[0-9]+)?", "", os.path.basename(f))
            )
            shutil.copy(os.path.join(c.plugin.numba.binary_folder, f), out_path_no_v)

        # Save to zipfile
        out_zip = os.path.join(c.plugin.numba.binary_folder, zipfile_basename + ".zip")

        if os.path.exists(out_zip):
            os.remove(out_zip)
        with zipfile.ZipFile(out_zip, "w", zipfile.ZIP_DEFLATED) as zf:
            for root, dirs, files in os.walk(tmpdir):
                for f in files:
                    zf.write(
                        os.path.join(root, f),
                        os.path.relpath(os.path.join(root, f), tmpdir),
                    )

    # Upload the newly generated zipfile
    binaries_sync(c, [".zip"])


@task(
    help={
        "clean": "Clean out dependencies before packaging",
        "python": "Python to use for setup and compiling",
    }
)
def binaries_compile(c, clean=False, python="python"):
    print("Compiling exported numba functions...")
    n = 0

    if not os.path.exists(c.plugin.numba.binary_folder):
        os.makedirs(c.plugin.numba.binary_folder)

    for numba_file in c.plugin.numba.aot_files:
        (base, ext) = os.path.splitext(os.path.basename(numba_file))
        subprocess.check_call([python, numba_file])
        n += 1

    v = get_version(c).replace(".", "-")

    for folder in {os.path.dirname(f) for f in c.plugin.numba.aot_files}:
        files = find_binaries(c, folder)

        for f in files:
            # Add version strings to the compiled files so they won't overwrite
            # files from other Trends.Earth versions when synced to S3
            module_name_regex = re.compile(r"([a-zA-Z0-9_])\.(.*)")
            out_file = module_name_regex.sub(rf"\g<1>_{v}.\g<2>", os.path.basename(f))
            out_path_with_v = os.path.join(c.plugin.numba.binary_folder, out_file)
            shutil.move(os.path.join(folder, f), out_path_with_v)

    print("Compiled {} numba files.".format(n))


###############################################################################
# Options
###############################################################################

ns = Collection(
    set_version,
    set_tag,
    plugin_setup,
    plugin_install,
    compile_files,
    docs_build,
    docs_spellcheck,
    translate_pull,
    translate_push,
    lrelease,
    changelog_build,
    tecli_login,
    tecli_clear,
    tecli_config,
    tecli_publish,
    tecli_run,
    tecli_info,
    tecli_logs,
    zipfile_build,
    zipfile_deploy,
    binaries_compile,
    binaries_sync,
    binaries_deploy,
    release_github,
    update_script_ids,
    testdata_sync,
    rtd_pre_build,
    build_download_page,
)

ns.configure(
    {
        "plugin": {
            "name": "LDMP",
            "version_file_raw": "version.txt",
            "version_file_details": "LDMP/version.json",
            "ext_libs": {"path": "LDMP/ext-libs", "local_modules": []},
            "gui_dir": "LDMP/gui",
            "source_dir": "LDMP",
            "data_dir": "LDMP/data",
            "i18n_dir": "LDMP/i18n",
            "translations": ["fr", "es", "sw", "pt", "ar", "ru", "zh", "fa"],
            "numba": {
                "aot_files": [
                    "LDMP/localexecution/ldn_numba.py",
                    "LDMP/localexecution/drought_numba.py",
                    "LDMP/localexecution/util_numba.py",
                ],
                "binary_extensions": [".so", ".pyd"],
                "binary_folder": "LDMP/binaries",
                "binary_list": "LDMP/data/binaries.txt",
            },
            "testdata_patterns": ["LDMP/test/integration/fixtures/*"],
            "package_dir": "build",
            "tests": ["LDMP/test"],
            "excludes": [
                "LDMP/data_prep_scripts",
                "LDMP/binaries",
                "docs",
                "gee",
                "util",
                "*.pyc",
                "*.ts",
                "*.pro",
            ],
            # skip certain files inadvertently found by exclude pattern globbing
            "skip_exclude": [],
        },
        "schemas": {
            "setup_dir": "LDMP/schemas",
        },
        "gee": {
            "script_dir": "gee",
            "scripts_json_file": "LDMP/data/scripts.json",
            "tecli": "../trends.earth-CLI/tecli",
        },
        "sphinx": {
            "sphinx_build": f"{sys.executable} -m sphinx.cmd.build",
            "sphinx_intl": "sphinx-intl",
            "docroot": "docs",
            "sourcedir": "docs/source",
            "builddir": "docs/build",
            "resourcedir": "docs/resources",
            "zipfile_deploy_s3_bucket": "trends.earth",
            "documentation_deploy_s3_bucket": "data.trends.earth",
            "docs_s3_prefix": "docs/",
            "transifex_name": "trendsearth-v2",
            "tx_path": f"{os.path.dirname(__file__)}/tx",
            "base_language": "en",
        },
        "data_downloads": {
            "downloads_page": "docs/source/for_users/downloads/index.md",
            "s3_bucket": "data.trends.earth",
            "s3_prefix": "unccd_reporting/2016-2019/packages/",
        },
        "github": {
            "api_url": "https://api.github.com",
            "repo_owner": "ConservationInternational",
            "repo_name": "trends.earth",
            "token": None,
        },
    }
)