crawlee 1.1.1b6__py3-none-any.whl → 1.1.1b7__py3-none-any.whl
This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
- crawlee/storage_clients/_file_system/_dataset_client.py +2 -2
- crawlee/storage_clients/_file_system/_key_value_store_client.py +3 -3
- crawlee/storage_clients/_file_system/_request_queue_client.py +3 -3
- {crawlee-1.1.1b6.dist-info → crawlee-1.1.1b7.dist-info}/METADATA +1 -1
- {crawlee-1.1.1b6.dist-info → crawlee-1.1.1b7.dist-info}/RECORD +8 -8
- {crawlee-1.1.1b6.dist-info → crawlee-1.1.1b7.dist-info}/WHEEL +1 -1
- {crawlee-1.1.1b6.dist-info → crawlee-1.1.1b7.dist-info}/entry_points.txt +0 -0
- {crawlee-1.1.1b6.dist-info → crawlee-1.1.1b7.dist-info}/licenses/LICENSE +0 -0
|
@@ -134,7 +134,7 @@ class FileSystemDatasetClient(DatasetClient):
|
|
|
134
134
|
continue
|
|
135
135
|
|
|
136
136
|
try:
|
|
137
|
-
file = await asyncio.to_thread(path_to_metadata.open)
|
|
137
|
+
file = await asyncio.to_thread(path_to_metadata.open, 'r', encoding='utf-8')
|
|
138
138
|
try:
|
|
139
139
|
file_content = json.load(file)
|
|
140
140
|
metadata = DatasetMetadata(**file_content)
|
|
@@ -163,7 +163,7 @@ class FileSystemDatasetClient(DatasetClient):
|
|
|
163
163
|
|
|
164
164
|
# If the dataset directory exists, reconstruct the client from the metadata file.
|
|
165
165
|
if path_to_dataset.exists() and path_to_metadata.exists():
|
|
166
|
-
file = await asyncio.to_thread(open, path_to_metadata)
|
|
166
|
+
file = await asyncio.to_thread(open, path_to_metadata, 'r', encoding='utf-8')
|
|
167
167
|
try:
|
|
168
168
|
file_content = json.load(file)
|
|
169
169
|
finally:
|
|
@@ -133,7 +133,7 @@ class FileSystemKeyValueStoreClient(KeyValueStoreClient):
|
|
|
133
133
|
continue
|
|
134
134
|
|
|
135
135
|
try:
|
|
136
|
-
file = await asyncio.to_thread(path_to_metadata.open)
|
|
136
|
+
file = await asyncio.to_thread(path_to_metadata.open, 'r', encoding='utf-8')
|
|
137
137
|
try:
|
|
138
138
|
file_content = json.load(file)
|
|
139
139
|
metadata = KeyValueStoreMetadata(**file_content)
|
|
@@ -162,7 +162,7 @@ class FileSystemKeyValueStoreClient(KeyValueStoreClient):
|
|
|
162
162
|
|
|
163
163
|
# If the key-value store directory exists, reconstruct the client from the metadata file.
|
|
164
164
|
if path_to_kvs.exists() and path_to_metadata.exists():
|
|
165
|
-
file = await asyncio.to_thread(open, path_to_metadata)
|
|
165
|
+
file = await asyncio.to_thread(open, path_to_metadata, 'r', encoding='utf-8')
|
|
166
166
|
try:
|
|
167
167
|
file_content = json.load(file)
|
|
168
168
|
finally:
|
|
@@ -239,7 +239,7 @@ class FileSystemKeyValueStoreClient(KeyValueStoreClient):
|
|
|
239
239
|
# Read the metadata file
|
|
240
240
|
async with self._lock:
|
|
241
241
|
try:
|
|
242
|
-
file = await asyncio.to_thread(open, record_metadata_filepath)
|
|
242
|
+
file = await asyncio.to_thread(open, record_metadata_filepath, 'r', encoding='utf-8')
|
|
243
243
|
except FileNotFoundError:
|
|
244
244
|
logger.warning(f'Metadata file disappeared for key "{key}", aborting get_value')
|
|
245
245
|
return None
|
|
@@ -197,7 +197,7 @@ class FileSystemRequestQueueClient(RequestQueueClient):
|
|
|
197
197
|
continue
|
|
198
198
|
|
|
199
199
|
try:
|
|
200
|
-
file = await asyncio.to_thread(path_to_metadata.open)
|
|
200
|
+
file = await asyncio.to_thread(path_to_metadata.open, 'r', encoding='utf-8')
|
|
201
201
|
try:
|
|
202
202
|
file_content = json.load(file)
|
|
203
203
|
metadata = RequestQueueMetadata(**file_content)
|
|
@@ -232,7 +232,7 @@ class FileSystemRequestQueueClient(RequestQueueClient):
|
|
|
232
232
|
|
|
233
233
|
# If the RQ directory exists, reconstruct the client from the metadata file.
|
|
234
234
|
if path_to_rq.exists() and path_to_metadata.exists():
|
|
235
|
-
file = await asyncio.to_thread(open, path_to_metadata)
|
|
235
|
+
file = await asyncio.to_thread(open, path_to_metadata, 'r', encoding='utf-8')
|
|
236
236
|
try:
|
|
237
237
|
file_content = json.load(file)
|
|
238
238
|
finally:
|
|
@@ -775,7 +775,7 @@ class FileSystemRequestQueueClient(RequestQueueClient):
|
|
|
775
775
|
"""
|
|
776
776
|
# Open the request file.
|
|
777
777
|
try:
|
|
778
|
-
file = await asyncio.to_thread(open, file_path)
|
|
778
|
+
file = await asyncio.to_thread(open, file_path, 'r', encoding='utf-8')
|
|
779
779
|
except FileNotFoundError:
|
|
780
780
|
logger.warning(f'Request file "{file_path}" not found.')
|
|
781
781
|
return None
|
|
@@ -159,9 +159,9 @@ crawlee/storage_clients/_base/_request_queue_client.py,sha256=cgM4yk6xJwgfzP-xaN
|
|
|
159
159
|
crawlee/storage_clients/_base/_storage_client.py,sha256=RvmKCV1U9_KxyG7n8xhClm2vwD2SKChWIiBLk6cuqw0,3523
|
|
160
160
|
crawlee/storage_clients/_base/py.typed,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
|
|
161
161
|
crawlee/storage_clients/_file_system/__init__.py,sha256=w3twfwz5YeLYeu_70pNPBRINS2wXRvzOMvA1hUDYgf0,387
|
|
162
|
-
crawlee/storage_clients/_file_system/_dataset_client.py,sha256=
|
|
163
|
-
crawlee/storage_clients/_file_system/_key_value_store_client.py,sha256=
|
|
164
|
-
crawlee/storage_clients/_file_system/_request_queue_client.py,sha256=
|
|
162
|
+
crawlee/storage_clients/_file_system/_dataset_client.py,sha256=DTRYlm37VV7FuowenG0JoqiQdH5AMg9G0O1PPJJO-u0,17781
|
|
163
|
+
crawlee/storage_clients/_file_system/_key_value_store_client.py,sha256=zPXCKPm6w8UYLYwSOuAoc4uoFswJjTAoWMulucvFBiI,18745
|
|
164
|
+
crawlee/storage_clients/_file_system/_request_queue_client.py,sha256=3dn9DM750ftuUzDCp_Uj56tNakYb93nhmeSo2LjPeV0,34039
|
|
165
165
|
crawlee/storage_clients/_file_system/_storage_client.py,sha256=My63uc513kfUPe5X-PTYWBRe9xUGnkLqJN7IcsQd2yw,3293
|
|
166
166
|
crawlee/storage_clients/_file_system/_utils.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
|
|
167
167
|
crawlee/storage_clients/_file_system/py.typed,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
|
|
@@ -199,8 +199,8 @@ crawlee/storages/_request_queue.py,sha256=bjBOGbpMaGUsqJPVB-JD2VShziPAYMI-GvWKKp
|
|
|
199
199
|
crawlee/storages/_storage_instance_manager.py,sha256=72n0YlPwNpSQDJSPf4TxnI2GvIK6L-ZiTmHRbFcoVU0,8164
|
|
200
200
|
crawlee/storages/_utils.py,sha256=Yz-5tEBYKYCFJemYT29--uGJqoJLApLDLgPcsnbifRw,439
|
|
201
201
|
crawlee/storages/py.typed,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
|
|
202
|
-
crawlee-1.1.
|
|
203
|
-
crawlee-1.1.
|
|
204
|
-
crawlee-1.1.
|
|
205
|
-
crawlee-1.1.
|
|
206
|
-
crawlee-1.1.
|
|
202
|
+
crawlee-1.1.1b7.dist-info/METADATA,sha256=Zcoo5OQ-typ3bxMOO5ph-_YuEuPKXAq_VD6fJdHgVGE,29532
|
|
203
|
+
crawlee-1.1.1b7.dist-info/WHEEL,sha256=WLgqFyCfm_KASv4WHyYy0P3pM_m7J5L9k2skdKLirC8,87
|
|
204
|
+
crawlee-1.1.1b7.dist-info/entry_points.txt,sha256=1p65X3dA-cYvzjtlxLL6Kn1wpY-3uEDVqJLp53uNPeo,45
|
|
205
|
+
crawlee-1.1.1b7.dist-info/licenses/LICENSE,sha256=AsFjHssKjj4LGd2ZCqXn6FBzMqcWdjQre1byPPSypVw,11355
|
|
206
|
+
crawlee-1.1.1b7.dist-info/RECORD,,
|
|
File without changes
|
|
File without changes
|