crawlee 1.1.1b1__py3-none-any.whl → 1.2.1b7__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.
Potentially problematic release.
This version of crawlee might be problematic. Click here for more details.
- crawlee/__init__.py +2 -1
- crawlee/_request.py +29 -10
- crawlee/_types.py +42 -2
- crawlee/_utils/context.py +2 -2
- crawlee/_utils/file.py +7 -0
- crawlee/_utils/recurring_task.py +2 -1
- crawlee/_utils/time.py +41 -1
- crawlee/crawlers/__init__.py +2 -1
- crawlee/crawlers/_abstract_http/__init__.py +2 -1
- crawlee/crawlers/_abstract_http/_abstract_http_crawler.py +52 -14
- crawlee/crawlers/_adaptive_playwright/_adaptive_playwright_crawler.py +10 -33
- crawlee/crawlers/_adaptive_playwright/_adaptive_playwright_crawling_context.py +6 -2
- crawlee/crawlers/_basic/_basic_crawler.py +135 -118
- crawlee/crawlers/_basic/_context_utils.py +24 -0
- crawlee/crawlers/_basic/_logging_utils.py +23 -4
- crawlee/crawlers/_beautifulsoup/_beautifulsoup_crawler.py +2 -2
- crawlee/crawlers/_parsel/_parsel_crawler.py +2 -2
- crawlee/crawlers/_playwright/_playwright_crawler.py +58 -17
- crawlee/crawlers/_playwright/_playwright_http_client.py +7 -1
- crawlee/crawlers/_playwright/_playwright_pre_nav_crawling_context.py +4 -1
- crawlee/crawlers/_playwright/_types.py +12 -2
- crawlee/errors.py +4 -0
- crawlee/events/_event_manager.py +1 -3
- crawlee/http_clients/_base.py +4 -0
- crawlee/http_clients/_curl_impersonate.py +12 -0
- crawlee/http_clients/_httpx.py +16 -6
- crawlee/http_clients/_impit.py +25 -10
- crawlee/router.py +13 -3
- 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/storage_clients/_sql/_storage_client.py +0 -9
- {crawlee-1.1.1b1.dist-info → crawlee-1.2.1b7.dist-info}/METADATA +10 -16
- {crawlee-1.1.1b1.dist-info → crawlee-1.2.1b7.dist-info}/RECORD +37 -36
- {crawlee-1.1.1b1.dist-info → crawlee-1.2.1b7.dist-info}/WHEEL +1 -1
- {crawlee-1.1.1b1.dist-info → crawlee-1.2.1b7.dist-info}/entry_points.txt +0 -0
- {crawlee-1.1.1b1.dist-info → crawlee-1.2.1b7.dist-info}/licenses/LICENSE +0 -0
|
@@ -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
|
|
@@ -1,6 +1,5 @@
|
|
|
1
1
|
from __future__ import annotations
|
|
2
2
|
|
|
3
|
-
import sys
|
|
4
3
|
import warnings
|
|
5
4
|
from datetime import timedelta
|
|
6
5
|
from pathlib import Path
|
|
@@ -269,14 +268,6 @@ class SqlStorageClient(StorageClient):
|
|
|
269
268
|
'Unsupported database. Supported: sqlite, postgresql. Consider using a different database.'
|
|
270
269
|
)
|
|
271
270
|
|
|
272
|
-
# TODO: https://github.com/apify/crawlee-python/issues/1555
|
|
273
|
-
if 'postgresql' in connection_string and sys.version_info >= (3, 14):
|
|
274
|
-
raise ValueError(
|
|
275
|
-
'SqlStorageClient cannot use PostgreSQL with Python 3.14 '
|
|
276
|
-
'due to asyncpg compatibility limitations. '
|
|
277
|
-
'Please use Python 3.13 or earlier, or switch to SQLite.'
|
|
278
|
-
)
|
|
279
|
-
|
|
280
271
|
self._engine = create_async_engine(
|
|
281
272
|
connection_string,
|
|
282
273
|
future=True,
|
|
@@ -1,6 +1,6 @@
|
|
|
1
1
|
Metadata-Version: 2.4
|
|
2
2
|
Name: crawlee
|
|
3
|
-
Version: 1.
|
|
3
|
+
Version: 1.2.1b7
|
|
4
4
|
Summary: Crawlee for Python
|
|
5
5
|
Project-URL: Apify Homepage, https://apify.com
|
|
6
6
|
Project-URL: Changelog, https://crawlee.dev/python/docs/changelog
|
|
@@ -226,6 +226,7 @@ Classifier: Programming Language :: Python :: 3.13
|
|
|
226
226
|
Classifier: Programming Language :: Python :: 3.14
|
|
227
227
|
Classifier: Topic :: Software Development :: Libraries
|
|
228
228
|
Requires-Python: >=3.10
|
|
229
|
+
Requires-Dist: async-timeout>=5.0.1
|
|
229
230
|
Requires-Dist: cachetools>=5.5.0
|
|
230
231
|
Requires-Dist: colorama>=0.4.0
|
|
231
232
|
Requires-Dist: impit>=0.8.0
|
|
@@ -247,7 +248,7 @@ Requires-Dist: scikit-learn>=1.6.0; extra == 'adaptive-crawler'
|
|
|
247
248
|
Provides-Extra: all
|
|
248
249
|
Requires-Dist: aiosqlite>=0.21.0; extra == 'all'
|
|
249
250
|
Requires-Dist: apify-fingerprint-datapoints>=0.0.2; extra == 'all'
|
|
250
|
-
Requires-Dist: asyncpg>=0.24.0;
|
|
251
|
+
Requires-Dist: asyncpg>=0.24.0; extra == 'all'
|
|
251
252
|
Requires-Dist: beautifulsoup4[lxml]>=4.12.0; extra == 'all'
|
|
252
253
|
Requires-Dist: browserforge>=1.2.3; extra == 'all'
|
|
253
254
|
Requires-Dist: cookiecutter>=2.6.0; extra == 'all'
|
|
@@ -301,7 +302,7 @@ Requires-Dist: playwright>=1.27.0; extra == 'playwright'
|
|
|
301
302
|
Provides-Extra: redis
|
|
302
303
|
Requires-Dist: redis[hiredis]>=7.0.0; extra == 'redis'
|
|
303
304
|
Provides-Extra: sql-postgres
|
|
304
|
-
Requires-Dist: asyncpg>=0.24.0;
|
|
305
|
+
Requires-Dist: asyncpg>=0.24.0; extra == 'sql-postgres'
|
|
305
306
|
Requires-Dist: sqlalchemy[asyncio]<3.0.0,>=2.0.0; extra == 'sql-postgres'
|
|
306
307
|
Provides-Extra: sql-sqlite
|
|
307
308
|
Requires-Dist: aiosqlite>=0.21.0; extra == 'sql-sqlite'
|
|
@@ -323,19 +324,12 @@ Description-Content-Type: text/markdown
|
|
|
323
324
|
<a href="https://trendshift.io/repositories/11169" target="_blank"><img src="https://trendshift.io/api/badge/repositories/11169" alt="apify%2Fcrawlee-python | Trendshift" style="width: 250px; height: 55px;" width="250" height="55"/></a>
|
|
324
325
|
</p>
|
|
325
326
|
|
|
326
|
-
<p align=center>
|
|
327
|
-
|
|
328
|
-
|
|
329
|
-
|
|
330
|
-
|
|
331
|
-
|
|
332
|
-
</a>
|
|
333
|
-
<a href="https://pypi.org/project/crawlee/" rel="nofollow">
|
|
334
|
-
<img src="https://img.shields.io/pypi/pyversions/crawlee" alt="PyPI - Python Version" style="max-width: 100%;">
|
|
335
|
-
</a>
|
|
336
|
-
<a href="https://discord.gg/jyEM2PRvMU" rel="nofollow">
|
|
337
|
-
<img src="https://img.shields.io/discord/801163717915574323?label=discord" alt="Chat on discord" style="max-width: 100%;">
|
|
338
|
-
</a>
|
|
327
|
+
<p align="center">
|
|
328
|
+
<a href="https://badge.fury.io/py/crawlee" rel="nofollow"><img src="https://badge.fury.io/py/crawlee.svg" alt="PyPI package version"></a>
|
|
329
|
+
<a href="https://pypi.org/project/crawlee/" rel="nofollow"><img src="https://img.shields.io/pypi/dm/crawlee" alt="PyPI package downloads"></a>
|
|
330
|
+
<a href="https://codecov.io/gh/apify/crawlee-python"><img src="https://codecov.io/gh/apify/crawlee-python/graph/badge.svg?token=cCju61iPQG" alt="Codecov report"></a>
|
|
331
|
+
<a href="https://pypi.org/project/crawlee/" rel="nofollow"><img src="https://img.shields.io/pypi/pyversions/crawlee" alt="PyPI Python version"></a>
|
|
332
|
+
<a href="https://discord.gg/jyEM2PRvMU" rel="nofollow"><img src="https://img.shields.io/discord/801163717915574323?label=discord" alt="Chat on Discord"></a>
|
|
339
333
|
</p>
|
|
340
334
|
|
|
341
335
|
Crawlee covers your crawling and scraping end-to-end and **helps you build reliable scrapers. Fast.**
|
|
@@ -1,16 +1,16 @@
|
|
|
1
|
-
crawlee/__init__.py,sha256=
|
|
1
|
+
crawlee/__init__.py,sha256=ECFcNbLQp3HX-o6K4eMo38rZQ5NnZg7udvEEkjkqnuw,548
|
|
2
2
|
crawlee/_browserforge_workaround.py,sha256=FYQaqpqfZGYkx-A8evF9nsHnj4KK4IMtjNq3LtmX_vA,1664
|
|
3
3
|
crawlee/_cli.py,sha256=czuEsGD8QYEiq5gtMcBxrL08hQ5OJQQkMVhAr1pvDaQ,10353
|
|
4
4
|
crawlee/_consts.py,sha256=RQ96gx7V-WPH91cVsMUz76X5UZUNDNhCudtlyGkxFVk,133
|
|
5
5
|
crawlee/_log_config.py,sha256=VyxoEfWCq_9fyicmmJbjiZ5KC91onMcAtX2L4oKX4m4,5999
|
|
6
|
-
crawlee/_request.py,sha256=
|
|
6
|
+
crawlee/_request.py,sha256=M8hTSs5dJTBBW0JIDh0QSUhWyEWarEg86Un9kX12qy4,17374
|
|
7
7
|
crawlee/_service_locator.py,sha256=SJ8ABYtclBl7rz8kfZ2jZkIgKq5oNIoGT7WmN8ApTzo,5058
|
|
8
|
-
crawlee/_types.py,sha256=
|
|
8
|
+
crawlee/_types.py,sha256=_CQyq1BmvuHr0p25NFn6rHbgsiuR65o8gLxCCuQWfAg,30534
|
|
9
9
|
crawlee/configuration.py,sha256=DWS2z1FC6Ua93W2tStK3R1ZKZbZjVQYWGiGFbZFaRtA,8064
|
|
10
|
-
crawlee/errors.py,sha256=
|
|
10
|
+
crawlee/errors.py,sha256=fnAFpyvJKMDq3VDGr1iq1E-JqnfoOEI7cd8YjDaqb9s,4062
|
|
11
11
|
crawlee/proxy_configuration.py,sha256=rqf67yerXvLvraBaAHW04nvf5ECze3wMQbK7LlqXucM,10386
|
|
12
12
|
crawlee/py.typed,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
|
|
13
|
-
crawlee/router.py,sha256=
|
|
13
|
+
crawlee/router.py,sha256=HbKxE22r8ZVu93tIxBdGObMa3fGPcuSvKthqibimekU,4252
|
|
14
14
|
crawlee/_autoscaling/__init__.py,sha256=t6Z44gU488C0UmkBCTtwsgAR8iqJcv2g4ZlC4NYh0ZI,182
|
|
15
15
|
crawlee/_autoscaling/_types.py,sha256=xnrRHXYOVn7GwELLVHi_y7B-Ic7u3hPkYl3P-LT3Fhk,5453
|
|
16
16
|
crawlee/_autoscaling/autoscaled_pool.py,sha256=Bcu2jDgK2SYMnZN5xfjs8Oxti0ZxrktjydWv3J0Hz48,12214
|
|
@@ -21,21 +21,21 @@ crawlee/_utils/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
|
|
|
21
21
|
crawlee/_utils/blocked.py,sha256=sxN99AouFXMoe6uG1EvCTCmKMGk73DBMUk9nOkWK86I,863
|
|
22
22
|
crawlee/_utils/byte_size.py,sha256=zs4qWUEDgTGDqYfUJ7t5edWNYYJCG8Y1EyJ9GASfRL4,3744
|
|
23
23
|
crawlee/_utils/console.py,sha256=vAIM8AO7cT-HdXg44eR8zQyHAHk8X8G7J1KKFCBL2LY,2242
|
|
24
|
-
crawlee/_utils/context.py,sha256=
|
|
24
|
+
crawlee/_utils/context.py,sha256=LFIXjJQBhv94j1prbK-2yjH3EXg5jPOfVqW8P6cwNIY,1726
|
|
25
25
|
crawlee/_utils/crypto.py,sha256=tYzn2z91KgV3ugxz4CKtSTcCjW-3FC8un7hpDNCl6rs,757
|
|
26
26
|
crawlee/_utils/docs.py,sha256=S09-3xAQAlUvrmPpBXVJpE8wblB8LtS6QduLNncfqdQ,1130
|
|
27
|
-
crawlee/_utils/file.py,sha256=
|
|
27
|
+
crawlee/_utils/file.py,sha256=FJHTC25qSWQs3ZhCZrLgs0cUwA9K81MlQRGEmcWKAQU,5758
|
|
28
28
|
crawlee/_utils/globs.py,sha256=SGX2J35Kqw7yZnSS5c4mLz9UD8c77PF0IoCgXQM5uiw,5310
|
|
29
29
|
crawlee/_utils/html_to_text.py,sha256=1iykT-OXd2xXNy7isHVWHqPxe23X82CGQBHIfbZbZkY,902
|
|
30
30
|
crawlee/_utils/models.py,sha256=EqM50Uc-xvxKlLCLA2lPpRduzfKvT0z_-Q-UWG8aTRQ,1955
|
|
31
31
|
crawlee/_utils/raise_if_too_many_kwargs.py,sha256=J2gaUJmsmNwexohuehXw_mdYKv-eWiui6WUHFsQ3qTQ,597
|
|
32
32
|
crawlee/_utils/recoverable_state.py,sha256=c1D2ZecxEliGZzhqYz9_oU5CF2Hm0UKvpOHqO6CDJRE,9032
|
|
33
|
-
crawlee/_utils/recurring_task.py,sha256=
|
|
33
|
+
crawlee/_utils/recurring_task.py,sha256=_injmSsvG4p0xS4nBtoZZIR02syBG8JcLkuwgNDL8Nc,2143
|
|
34
34
|
crawlee/_utils/requests.py,sha256=yOjai7bHR9_duPJ0ck-L76y9AnKZr49JBfSOQv9kvJc,5048
|
|
35
35
|
crawlee/_utils/robots.py,sha256=DBU5ni4Y-p7bIKMbLd_ws8wgHSFc4K8zPVF3JvH_pkw,4661
|
|
36
36
|
crawlee/_utils/sitemap.py,sha256=UI9EJiFiyFvV5_flVUtdsEVz8ZsJeRERPtcx8ZsqjTU,16632
|
|
37
37
|
crawlee/_utils/system.py,sha256=tA8AP__9vsJ9OTLTnAYAKkxc8U5-IEna0N_hqYBybUo,4294
|
|
38
|
-
crawlee/_utils/time.py,sha256=
|
|
38
|
+
crawlee/_utils/time.py,sha256=awyzUqMvoC3BrEDINHaGPaM-u98AW0UtQqQsVJPAnsA,3899
|
|
39
39
|
crawlee/_utils/try_import.py,sha256=QI_58ifc2l0Rxehzu6xcofQrRAVeLzZuBTTTHttLl8s,1310
|
|
40
40
|
crawlee/_utils/urls.py,sha256=fEYXJxBT02f-DIYKF_h7PdaKAShfXBs99-dHDjDX03A,1725
|
|
41
41
|
crawlee/_utils/wait.py,sha256=RfiXhp5VUBxOEtEMtru7_jNfKDr2BJCcFge5qGg2gxk,2848
|
|
@@ -49,29 +49,30 @@ crawlee/browsers/_playwright_browser_controller.py,sha256=W6G5MjZpg9IcZoHts6lTML
|
|
|
49
49
|
crawlee/browsers/_playwright_browser_plugin.py,sha256=A1qa1nJhTSKNP9uOiO-oGzR7VGlnOMo0A0YNedccb2A,8869
|
|
50
50
|
crawlee/browsers/_types.py,sha256=ZnDgJHeQNSd_s_mXhgQnAf09c2smuiXC31VbawHHXUM,436
|
|
51
51
|
crawlee/browsers/py.typed,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
|
|
52
|
-
crawlee/crawlers/__init__.py,sha256=
|
|
52
|
+
crawlee/crawlers/__init__.py,sha256=jNFMsPizSgCN0ARYSmHs9Ppk8yvGgjUH5PxUeDchljE,2386
|
|
53
53
|
crawlee/crawlers/_types.py,sha256=xbGTJQirgz5wUbfr12afMR4q-_5AWP7ngF2e8K5P8l0,355
|
|
54
54
|
crawlee/crawlers/py.typed,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
|
|
55
|
-
crawlee/crawlers/_abstract_http/__init__.py,sha256=
|
|
56
|
-
crawlee/crawlers/_abstract_http/_abstract_http_crawler.py,sha256=
|
|
55
|
+
crawlee/crawlers/_abstract_http/__init__.py,sha256=h8jVWcPbDXzWHill1Vm7J7iliJW0hIrea0gkg-Hkb-M,319
|
|
56
|
+
crawlee/crawlers/_abstract_http/_abstract_http_crawler.py,sha256=Y12SBNAiF8QNJH83s6pPoao1W5ZSUhxHRHHKjE0qZhk,13174
|
|
57
57
|
crawlee/crawlers/_abstract_http/_abstract_http_parser.py,sha256=Y5o_hiW_0mQAte5GFqkUxscwKEFpWrBYRsLKP1cfBwE,3521
|
|
58
58
|
crawlee/crawlers/_abstract_http/_http_crawling_context.py,sha256=Rno_uJ8ivmyRxFQv2MyY_z9B5WPHSEd5MAPz31_1ZIo,2179
|
|
59
59
|
crawlee/crawlers/_abstract_http/py.typed,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
|
|
60
60
|
crawlee/crawlers/_adaptive_playwright/__init__.py,sha256=LREq9WR9BKsE8S8lSsEhlCoNjQaLhlJ9yo8y_6a8o4c,1072
|
|
61
|
-
crawlee/crawlers/_adaptive_playwright/_adaptive_playwright_crawler.py,sha256=
|
|
61
|
+
crawlee/crawlers/_adaptive_playwright/_adaptive_playwright_crawler.py,sha256=qAtZUwzGMwASwl5NKLAOsYnVA03IpZkk-BLKm3SwHoM,21588
|
|
62
62
|
crawlee/crawlers/_adaptive_playwright/_adaptive_playwright_crawler_statistics.py,sha256=_At8T8S3JLGPA-1AeCFGrpE-FuCDW9sazrXt9U0tK6U,1048
|
|
63
|
-
crawlee/crawlers/_adaptive_playwright/_adaptive_playwright_crawling_context.py,sha256=
|
|
63
|
+
crawlee/crawlers/_adaptive_playwright/_adaptive_playwright_crawling_context.py,sha256=tejw-yfA8zVR8L-shIZOTFoMUQOI5Kt7FBJa8H0q4H0,10664
|
|
64
64
|
crawlee/crawlers/_adaptive_playwright/_rendering_type_predictor.py,sha256=TM4mkbIN_059jUyCG8Z6XAb_FBLClIKw7z-aDvjon2I,10834
|
|
65
65
|
crawlee/crawlers/_adaptive_playwright/_result_comparator.py,sha256=NAfw5VKzTnkvARtLr_zrZj6UGeMp05Voc6Oi8oPxU3w,1747
|
|
66
66
|
crawlee/crawlers/_adaptive_playwright/_utils.py,sha256=EUYVz5i2YkLpL_gbVRp9BAD5u6w1xJ_AFzc_qB9bdDQ,1102
|
|
67
67
|
crawlee/crawlers/_basic/__init__.py,sha256=LPln8SiBBXSMqrApiFUfpqz3hvqxN5HUa1cHQXMVKgU,280
|
|
68
|
-
crawlee/crawlers/_basic/_basic_crawler.py,sha256
|
|
68
|
+
crawlee/crawlers/_basic/_basic_crawler.py,sha256=wTZW_1vM2A1x14VADRBsUr0TJzKfGoJODeHX0gOZnnY,73914
|
|
69
69
|
crawlee/crawlers/_basic/_basic_crawling_context.py,sha256=fjxm2RQXMDkDlWu38dQ3xn5rrGUOhJXkXiqkgbFJFk4,155
|
|
70
70
|
crawlee/crawlers/_basic/_context_pipeline.py,sha256=vM8EEvnCoguERjRV3oyrxUq2Ln2F9DzY7P5dAEiuMHo,5869
|
|
71
|
-
crawlee/crawlers/_basic/
|
|
71
|
+
crawlee/crawlers/_basic/_context_utils.py,sha256=U1s0nl7EW9k-JrZA2VM7d_aWnE7Je3lXK04RFrXvRC4,655
|
|
72
|
+
crawlee/crawlers/_basic/_logging_utils.py,sha256=6Q206Sv0RzHztwu5y5XSdUpZhpqQ5-zSapQzUY9GxCo,4014
|
|
72
73
|
crawlee/crawlers/_basic/py.typed,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
|
|
73
74
|
crawlee/crawlers/_beautifulsoup/__init__.py,sha256=7pL273ashA7yYDrH6nokYZ7SAMUAezilGIWdfThi_Co,822
|
|
74
|
-
crawlee/crawlers/_beautifulsoup/_beautifulsoup_crawler.py,sha256=
|
|
75
|
+
crawlee/crawlers/_beautifulsoup/_beautifulsoup_crawler.py,sha256=Q8Sb_lflpdYIwDZ1fIeuquPzdDG2zCnKsrcj8fe8n6k,3056
|
|
75
76
|
crawlee/crawlers/_beautifulsoup/_beautifulsoup_crawling_context.py,sha256=CMHQbYmXdhnXSo3hjtmAqWPH2lEaoVHzwPY2Ka85s70,1149
|
|
76
77
|
crawlee/crawlers/_beautifulsoup/_beautifulsoup_parser.py,sha256=egCBcaE6RaQoJOFuxex_McItCd4Ddd3ZDE-2DVGictA,1661
|
|
77
78
|
crawlee/crawlers/_beautifulsoup/_utils.py,sha256=xUJM7Y0RGlTEe0TJe252uuIdjk8gwBFjuhQnRG7-jv4,3127
|
|
@@ -80,19 +81,19 @@ crawlee/crawlers/_http/__init__.py,sha256=BY6KKY1eD8CWDFfurJDIgcnFIfgXAVuMHKSL3w
|
|
|
80
81
|
crawlee/crawlers/_http/_http_crawler.py,sha256=HoCTku4FL9EN3WMlzbZNbb2BtXedP487CN0JlObsoGg,2041
|
|
81
82
|
crawlee/crawlers/_http/_http_parser.py,sha256=Bfe3TEz6i5UhONO2qfyw2p31QoVoXbvGci3T3HeObiU,1495
|
|
82
83
|
crawlee/crawlers/_parsel/__init__.py,sha256=c_FveWxCuaEaG2HAYlaN9N_aPEaVdpaYVMsIcCcTdnU,617
|
|
83
|
-
crawlee/crawlers/_parsel/_parsel_crawler.py,sha256=
|
|
84
|
+
crawlee/crawlers/_parsel/_parsel_crawler.py,sha256=Aolo96FU_U3nsEjGoxEothgF7pVuoXLkhRmNWgKzGYg,2717
|
|
84
85
|
crawlee/crawlers/_parsel/_parsel_crawling_context.py,sha256=sZB26RcRLjSoD15myEOMPeolIN7apG76aqRmKQvEep8,1142
|
|
85
86
|
crawlee/crawlers/_parsel/_parsel_parser.py,sha256=yWBfuXUHMriK4DRnyrXTQoGeqX5WV9bOEkBp_g0YCvQ,1540
|
|
86
87
|
crawlee/crawlers/_parsel/_utils.py,sha256=MbRwx-cdjlq1zLzFYf64M3spOGQ6yxum4FvP0sdqA_Q,2693
|
|
87
88
|
crawlee/crawlers/_playwright/__init__.py,sha256=6Cahe6VEF82o8CYiP8Cmp58Cmb6Rb8uMeyy7wnwe5ms,837
|
|
88
|
-
crawlee/crawlers/_playwright/_playwright_crawler.py,sha256=
|
|
89
|
+
crawlee/crawlers/_playwright/_playwright_crawler.py,sha256=WappMIb0w-AnS745vlJpQNxwibKS7ok6_5a6iAcoTDs,26207
|
|
89
90
|
crawlee/crawlers/_playwright/_playwright_crawling_context.py,sha256=Oi0tMBXHaEDlFjqG01DzgB7Ck52bjVjz-X__eMioxas,1249
|
|
90
|
-
crawlee/crawlers/_playwright/_playwright_http_client.py,sha256=
|
|
91
|
-
crawlee/crawlers/_playwright/_playwright_pre_nav_crawling_context.py,sha256=
|
|
92
|
-
crawlee/crawlers/_playwright/_types.py,sha256=
|
|
91
|
+
crawlee/crawlers/_playwright/_playwright_http_client.py,sha256=4mvaCI9Zum7znbm0F-ZZ6T1FEqZ-N-cvPOk1iqtcUSo,4164
|
|
92
|
+
crawlee/crawlers/_playwright/_playwright_pre_nav_crawling_context.py,sha256=NFenJKgXcPuifaVYc2sdU5AV2BX6836GUuqFTE2Q0lU,1545
|
|
93
|
+
crawlee/crawlers/_playwright/_types.py,sha256=D4MaRWgYdps1CwgNWURJRLKkJk_9Oyue70jvkHAxnEU,2534
|
|
93
94
|
crawlee/crawlers/_playwright/_utils.py,sha256=FQ_-LYo7DGHsNHRrTtWt3mC06VzQvQ2wkGqpA2wBzYU,3441
|
|
94
95
|
crawlee/events/__init__.py,sha256=YMgOXKI0LsXfImKQy06PZ2Vdjy-uD_-acioagHft1do,577
|
|
95
|
-
crawlee/events/_event_manager.py,sha256=
|
|
96
|
+
crawlee/events/_event_manager.py,sha256=M8nKPc2BJo8RIBVHaG9BYuks0jwt5v3BFYQLA7IvolI,11380
|
|
96
97
|
crawlee/events/_local_event_manager.py,sha256=CSiMJ6a_BwX0PPwtffEOtHm21dmALJz1zifo3AuMAk8,3708
|
|
97
98
|
crawlee/events/_types.py,sha256=MKsI014OOKKhjPJRrvWYrezIDGoLjGGhWXrkqYw26Ns,3313
|
|
98
99
|
crawlee/events/py.typed,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
|
|
@@ -104,10 +105,10 @@ crawlee/fingerprint_suite/_header_generator.py,sha256=9X9FbStehXdw-FZc_D0y-nLk1B
|
|
|
104
105
|
crawlee/fingerprint_suite/_types.py,sha256=7n2LJTiL2XvL-H4G-Y26Uoq5-ZXzH07Dq4o50uhMa-w,2423
|
|
105
106
|
crawlee/fingerprint_suite/py.typed,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
|
|
106
107
|
crawlee/http_clients/__init__.py,sha256=OQFhR9F8BrdlIaS5aRS7hvgQ0tKJPQ8FiyYPualyQcU,890
|
|
107
|
-
crawlee/http_clients/_base.py,sha256=
|
|
108
|
-
crawlee/http_clients/_curl_impersonate.py,sha256=
|
|
109
|
-
crawlee/http_clients/_httpx.py,sha256=
|
|
110
|
-
crawlee/http_clients/_impit.py,sha256=
|
|
108
|
+
crawlee/http_clients/_base.py,sha256=sz-NCgGyRKhNz2tPR7Zq10iYQ8Itig-fls4h7GMHhNM,7593
|
|
109
|
+
crawlee/http_clients/_curl_impersonate.py,sha256=a_1QQ04Tk_d5t2mIjxEKPvjF7zTn3q74QosuBjG8Wkc,11847
|
|
110
|
+
crawlee/http_clients/_httpx.py,sha256=abRe2YjQUfIrtmOwOAB72viwd8CF0V2mAXQceLuIWvo,12375
|
|
111
|
+
crawlee/http_clients/_impit.py,sha256=wEfDSOGIOmfLUkmLXAmZkiM5wp4PltUv-oH7zq2NMfQ,9434
|
|
111
112
|
crawlee/otel/__init__.py,sha256=g5y1tJfpDKfcIPGcKBztMgP6sptum-vJrtemeR8_-co,108
|
|
112
113
|
crawlee/otel/crawler_instrumentor.py,sha256=yC367A1NnAdhOanvym2zfiu4H4BskUslrib0GcHiVJs,6865
|
|
113
114
|
crawlee/project_template/cookiecutter.json,sha256=dJeYxLx5QEy2DCzXsDpqJQJlIJ3nw42lJrclZFoSZ8w,622
|
|
@@ -159,9 +160,9 @@ crawlee/storage_clients/_base/_request_queue_client.py,sha256=cgM4yk6xJwgfzP-xaN
|
|
|
159
160
|
crawlee/storage_clients/_base/_storage_client.py,sha256=RvmKCV1U9_KxyG7n8xhClm2vwD2SKChWIiBLk6cuqw0,3523
|
|
160
161
|
crawlee/storage_clients/_base/py.typed,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
|
|
161
162
|
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=
|
|
163
|
+
crawlee/storage_clients/_file_system/_dataset_client.py,sha256=DTRYlm37VV7FuowenG0JoqiQdH5AMg9G0O1PPJJO-u0,17781
|
|
164
|
+
crawlee/storage_clients/_file_system/_key_value_store_client.py,sha256=zPXCKPm6w8UYLYwSOuAoc4uoFswJjTAoWMulucvFBiI,18745
|
|
165
|
+
crawlee/storage_clients/_file_system/_request_queue_client.py,sha256=3dn9DM750ftuUzDCp_Uj56tNakYb93nhmeSo2LjPeV0,34039
|
|
165
166
|
crawlee/storage_clients/_file_system/_storage_client.py,sha256=My63uc513kfUPe5X-PTYWBRe9xUGnkLqJN7IcsQd2yw,3293
|
|
166
167
|
crawlee/storage_clients/_file_system/_utils.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
|
|
167
168
|
crawlee/storage_clients/_file_system/py.typed,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
|
|
@@ -189,7 +190,7 @@ crawlee/storage_clients/_sql/_dataset_client.py,sha256=tiJVvOPZgc7cy4kGfWnun-g2T
|
|
|
189
190
|
crawlee/storage_clients/_sql/_db_models.py,sha256=KzA-R_L6zv9gqQg7B27mF-fERNJuMUEnewV9iofmTnI,9812
|
|
190
191
|
crawlee/storage_clients/_sql/_key_value_store_client.py,sha256=LnVLWhOjo4LdvtCac4fwuf__DgEQjlqSxz8KkjY3Qx4,11311
|
|
191
192
|
crawlee/storage_clients/_sql/_request_queue_client.py,sha256=OlvAOwEoYY5f4NO7BdhLFRT_i_E3YzJDb_ptKKK2huY,29478
|
|
192
|
-
crawlee/storage_clients/_sql/_storage_client.py,sha256=
|
|
193
|
+
crawlee/storage_clients/_sql/_storage_client.py,sha256=ITtMpwfotIW4SZjO4rycB5wfMKaqTAJgMvzcUZxckrk,10905
|
|
193
194
|
crawlee/storage_clients/_sql/py.typed,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
|
|
194
195
|
crawlee/storages/__init__.py,sha256=wc2eioyCKAAYrg4N7cshpjC-UbE23OzGar9nK_kteSY,186
|
|
195
196
|
crawlee/storages/_base.py,sha256=zUOcMJTg8MAzq-m9X1NJcWncCfxzI5mb5MyY35WAkMk,2310
|
|
@@ -199,8 +200,8 @@ crawlee/storages/_request_queue.py,sha256=bjBOGbpMaGUsqJPVB-JD2VShziPAYMI-GvWKKp
|
|
|
199
200
|
crawlee/storages/_storage_instance_manager.py,sha256=72n0YlPwNpSQDJSPf4TxnI2GvIK6L-ZiTmHRbFcoVU0,8164
|
|
200
201
|
crawlee/storages/_utils.py,sha256=Yz-5tEBYKYCFJemYT29--uGJqoJLApLDLgPcsnbifRw,439
|
|
201
202
|
crawlee/storages/py.typed,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
|
|
202
|
-
crawlee-1.
|
|
203
|
-
crawlee-1.
|
|
204
|
-
crawlee-1.
|
|
205
|
-
crawlee-1.
|
|
206
|
-
crawlee-1.
|
|
203
|
+
crawlee-1.2.1b7.dist-info/METADATA,sha256=YPjeW0r_pqD_lHRtFfJ8GL84Z4t1IvEgu1uBtxc4IuY,29526
|
|
204
|
+
crawlee-1.2.1b7.dist-info/WHEEL,sha256=WLgqFyCfm_KASv4WHyYy0P3pM_m7J5L9k2skdKLirC8,87
|
|
205
|
+
crawlee-1.2.1b7.dist-info/entry_points.txt,sha256=1p65X3dA-cYvzjtlxLL6Kn1wpY-3uEDVqJLp53uNPeo,45
|
|
206
|
+
crawlee-1.2.1b7.dist-info/licenses/LICENSE,sha256=AsFjHssKjj4LGd2ZCqXn6FBzMqcWdjQre1byPPSypVw,11355
|
|
207
|
+
crawlee-1.2.1b7.dist-info/RECORD,,
|
|
File without changes
|
|
File without changes
|