/
convert_to_libdeeplake.py
251 lines (215 loc) · 8.22 KB
/
convert_to_libdeeplake.py
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
from deeplake.core.dataset import Dataset
from deeplake.constants import MB
from deeplake.core.storage.gcs import GCSProvider
from deeplake.enterprise.util import raise_indra_installation_error # type: ignore
from deeplake.core.storage import S3Provider
from deeplake.core.storage.indra import IndraProvider
from deeplake.core.storage.azure import AzureProvider
from deeplake.util.remove_cache import get_base_storage
from deeplake.util.exceptions import EmptyTokenException
from deeplake.util.dataset import try_flushing # type: ignore
import importlib
import jwt
# Load lazy to avoid cycylic import.
INDRA_API = None
def import_indra_api_silent():
global INDRA_API
if INDRA_API:
return INDRA_API
if not importlib.util.find_spec("indra"):
return None
try:
from indra import api # type: ignore
INDRA_API = api
return api
except Exception as e:
return e
def import_indra_api():
api = import_indra_api_silent()
if api is None:
raise_indra_installation_error() # type: ignore
elif isinstance(api, Exception):
raise_indra_installation_error(api)
else:
return api
def _get_indra_ds_from_native_provider(provider: IndraProvider):
api = import_indra_api()
return api.dataset(provider.core)
def _get_indra_ds_from_azure_provider(
path: str,
token: str,
provider: AzureProvider,
):
if provider is None:
return None
api = import_indra_api()
account_name = provider.account_name
account_key = provider.account_key
sas_token = provider.get_sas_token()
expiration = str(provider.expiration) if provider.expiration else None
storage = IndraProvider(
path,
read_only=provider.read_only,
token=token,
account_name=account_name,
account_key=account_key,
sas_token=sas_token,
expiration=expiration,
)
return _get_indra_ds_from_native_provider(storage)
def _get_indra_ds_from_gcp_provider(
path: str,
token: str,
provider: GCSProvider,
):
if provider is None:
return None
api = import_indra_api()
creds = provider.get_creds()
anon = creds.get("anon", "")
expiration = creds.get("expiration", "")
access_token = creds.get("access_token", "")
json_credentials = creds.get("json_credentials", "")
endpoint_override = creds.get("endpoint_override", "")
scheme = creds.get("scheme", "")
retry_limit_seconds = creds.get("retry_limit_seconds", "")
storage = IndraProvider(
path,
read_only=provider.read_only,
token=token,
origin_path=provider.root,
anon=anon,
expiration=expiration,
access_token=access_token,
json_credentials=json_credentials,
endpoint_override=endpoint_override,
scheme=scheme,
retry_limit_seconds=retry_limit_seconds,
)
return _get_indra_ds_from_native_provider(storage)
def _get_indra_ds_from_s3_provider(
path: str,
token: str,
provider: S3Provider,
):
if provider is None:
return None
api = import_indra_api()
creds_used = provider.creds_used
if creds_used == "PLATFORM":
provider._check_update_creds()
storage = IndraProvider(
path,
read_only=provider.read_only,
token=token,
origin_path=provider.root,
aws_access_key_id=provider.aws_access_key_id,
aws_secret_access_key=provider.aws_secret_access_key,
aws_session_token=provider.aws_session_token,
region_name=provider.aws_region,
endpoint_url=provider.endpoint_url,
expiration=str(provider.expiration),
)
return _get_indra_ds_from_native_provider(storage)
elif creds_used == "ENV":
storage = IndraProvider(
path,
read_only=provider.read_only,
token=token,
origin_path=provider.root,
profile_name=provider.profile_name,
)
return _get_indra_ds_from_native_provider(storage)
elif creds_used == "DICT":
storage = IndraProvider(
path,
read_only=provider.read_only,
token=token,
origin_path=provider.root,
aws_access_key_id=provider.aws_access_key_id,
aws_secret_access_key=provider.aws_secret_access_key,
aws_session_token=provider.aws_session_token,
region_name=provider.aws_region,
endpoint_url=provider.endpoint_url,
)
return _get_indra_ds_from_native_provider(storage)
def dataset_to_libdeeplake(hub2_dataset: Dataset):
"""Convert a hub 2.x dataset object to a libdeeplake dataset object."""
try_flushing(hub2_dataset)
api = import_indra_api()
path: str = hub2_dataset.path
token = (
hub2_dataset.client.get_token()
if (hub2_dataset.token is None or hub2_dataset._token == "")
and hub2_dataset.client
else hub2_dataset.token
)
if token is None or token == "":
raise EmptyTokenException
if hub2_dataset.libdeeplake_dataset is not None:
libdeeplake_dataset = hub2_dataset.libdeeplake_dataset
elif isinstance(hub2_dataset.storage.next_storage, IndraProvider):
libdeeplake_dataset = api.dataset(hub2_dataset.storage.next_storage.core)
else:
libdeeplake_dataset = None
if path.startswith("gdrive://"):
raise ValueError("Gdrive datasets are not supported for libdeeplake")
elif path.startswith("mem://"):
raise ValueError("In memory datasets are not supported for libdeeplake")
elif path.startswith("hub://"):
provider = hub2_dataset.storage.next_storage
if isinstance(provider, S3Provider):
libdeeplake_dataset = _get_indra_ds_from_s3_provider(
path=path, token=token, provider=provider
)
elif isinstance(provider, GCSProvider):
libdeeplake_dataset = _get_indra_ds_from_gcp_provider(
path=path, token=token, provider=provider
)
elif isinstance(provider, AzureProvider):
libdeeplake_dataset = _get_indra_ds_from_azure_provider(
path=path, token=token, provider=provider
)
elif isinstance(provider, IndraProvider):
libdeeplake_dataset = _get_indra_ds_from_native_provider(
provider=provider
)
else:
raise ValueError("Unknown storage provider for hub:// dataset")
elif path.startswith("s3://"):
libdeeplake_dataset = _get_indra_ds_from_s3_provider(
path=path, token=token, provider=hub2_dataset.storage.next_storage
)
elif path.startswith(("gcs://", "gs://", "gcp://")):
provider = get_base_storage(hub2_dataset.storage)
libdeeplake_dataset = _get_indra_ds_from_gcp_provider(
path=path, token=token, provider=provider
)
elif path.startswith(("az://", "azure://")):
az_provider = get_base_storage(hub2_dataset.storage)
libdeeplake_dataset = _get_indra_ds_from_azure_provider(
path=path, token=token, provider=az_provider
)
else:
org_id = hub2_dataset.org_id
org_id = (
org_id or jwt.decode(token, options={"verify_signature": False})["id"]
)
storage = IndraProvider(
path, read_only=hub2_dataset.read_only, token=token, org_id=org_id
)
libdeeplake_dataset = api.dataset(storage.core)
hub2_dataset.libdeeplake_dataset = libdeeplake_dataset
assert libdeeplake_dataset is not None
if hasattr(hub2_dataset.storage, "cache_size"):
libdeeplake_dataset._max_cache_size = max(
hub2_dataset.storage.cache_size, libdeeplake_dataset._max_cache_size
)
commit_id = hub2_dataset.pending_commit_id
libdeeplake_dataset.checkout(commit_id)
slice_ = hub2_dataset.index.values[0].value
if slice_ != slice(None):
if isinstance(slice_, tuple):
slice_ = list(slice_)
libdeeplake_dataset = libdeeplake_dataset[slice_]
return libdeeplake_dataset