cobweb-launcher 1.1.16__py3-none-any.whl → 1.1.17__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 cobweb-launcher might be problematic. Click here for more details.

Files changed (9) hide show
  1. cobweb/base/seed.py +5 -1
  2. cobweb/crawlers/base_crawler.py +2 -2
  3. cobweb/launchers/launcher_pro.py +1 -1
  4. cobweb/setting.py +1 -1
  5. {cobweb_launcher-1.1.16.dist-info → cobweb_launcher-1.1.17.dist-info}/METADATA +1 -1
  6. {cobweb_launcher-1.1.16.dist-info → cobweb_launcher-1.1.17.dist-info}/RECORD +9 -9
  7. {cobweb_launcher-1.1.16.dist-info → cobweb_launcher-1.1.17.dist-info}/LICENSE +0 -0
  8. {cobweb_launcher-1.1.16.dist-info → cobweb_launcher-1.1.17.dist-info}/WHEEL +0 -0
  9. {cobweb_launcher-1.1.16.dist-info → cobweb_launcher-1.1.17.dist-info}/top_level.txt +0 -0
cobweb/base/seed.py CHANGED
@@ -5,10 +5,11 @@ import hashlib
5
5
 
6
6
  class SeedParams:
7
7
 
8
- def __init__(self, retry, priority, seed_version):
8
+ def __init__(self, retry, priority, seed_version, seed_status=None):
9
9
  self.retry = retry or 0
10
10
  self.priority = priority or 300
11
11
  self.seed_version = seed_version or int(time.time())
12
+ self.seed_status = seed_status
12
13
 
13
14
 
14
15
  class Seed:
@@ -17,6 +18,7 @@ class Seed:
17
18
  "retry",
18
19
  "priority",
19
20
  "seed_version",
21
+ "seed_status"
20
22
  ]
21
23
 
22
24
  def __init__(
@@ -26,6 +28,7 @@ class Seed:
26
28
  retry=None,
27
29
  priority=None,
28
30
  seed_version=None,
31
+ seed_status=None,
29
32
  **kwargs
30
33
  ):
31
34
  if any(isinstance(seed, t) for t in (str, bytes)):
@@ -47,6 +50,7 @@ class Seed:
47
50
  "retry": retry,
48
51
  "priority": priority,
49
52
  "seed_version": seed_version,
53
+ "seed_status": seed_status,
50
54
  }
51
55
 
52
56
  if kwargs:
cobweb/crawlers/base_crawler.py CHANGED
@@ -56,7 +56,7 @@ class Crawler(threading.Thread):
56
56
  elif isinstance(item, str) and item == DealModel.done:
57
57
  self.launcher_queue['done'].push(seed)
58
58
  elif isinstance(item, str) and item == DealModel.fail:
59
- seed.identifier = DealModel.fail
59
+ seed.params.seed_status = DealModel.fail
60
60
  self.launcher_queue['done'].push(seed)
61
61
  else:
62
62
  raise TypeError("yield value type error!")
@@ -69,7 +69,7 @@ class Crawler(threading.Thread):
69
69
  continue
70
70
 
71
71
  elif seed.params.retry >= self.max_retries:
72
- seed.identifier = DealModel.fail
72
+ seed.params.identifier = DealModel.fail
73
73
  self.launcher_queue['done'].push(seed)
74
74
  continue
75
75
 
cobweb/launchers/launcher_pro.py CHANGED
@@ -117,7 +117,7 @@ class LauncherPro(Launcher):
117
117
  seed = self.__LAUNCHER_QUEUE__['done'].pop()
118
118
  if not seed:
119
119
  break
120
- if seed.identifier == DealModel.fail:
120
+ if seed.params.seed_status == DealModel.fail:
121
121
  f_seeds.append(seed.to_string)
122
122
  elif self._done_model == 1:
123
123
  s_seeds.append(seed.to_string)
cobweb/setting.py CHANGED
@@ -60,4 +60,4 @@ SPIDER_THREAD_NUM = 10
60
60
  SPIDER_MAX_RETRIES = 5
61
61
 
62
62
  # 文件下载响应类型过滤
63
- FILE_FILTER_CONTENT_TYPE = ["text/html", "application/xhtml+xml"]
63
+ FILE_FILTER_CONTENT_TYPE = ["text/html", "application/xhtml+xml"]
{cobweb_launcher-1.1.16.dist-info → cobweb_launcher-1.1.17.dist-info}/METADATA RENAMED
@@ -1,6 +1,6 @@
1
1
  Metadata-Version: 2.1
2
2
  Name: cobweb-launcher
3
- Version: 1.1.16
3
+ Version: 1.1.17
4
4
  Summary: spider_hole
5
5
  Home-page: https://github.com/Juannie-PP/cobweb
6
6
  Author: Juannie-PP
{cobweb_launcher-1.1.16.dist-info → cobweb_launcher-1.1.17.dist-info}/RECORD RENAMED
@@ -1,6 +1,6 @@
1
1
  cobweb/__init__.py,sha256=1vB4yfJvrBVnixRRVNHLrJm8L0l5p_zOzLol5vKwP9M,80
2
2
  cobweb/constant.py,sha256=dMzvtsPje_5VgtMoOn-EvDtgycUWTKBKmDzX-CDfHDY,2025
3
- cobweb/setting.py,sha256=EKLY6WLIEhHSZ7Q27uccP1Knaucr72wZ_gWMD9Tieyo,1952
3
+ cobweb/setting.py,sha256=bg_oO85xsxwKqNUJScoVUI2v4e2lzSNOpsoW0tIaOrg,1953
4
4
  cobweb/base/__init__.py,sha256=diiK5MygQaWjlWNLbW6eUIg-93O6glMGC9WLNM5jyOc,209
5
5
  cobweb/base/common_queue.py,sha256=W7PPZZFl52j3Mc916T0imHj7oAUelA6aKJwW-FecDPE,872
6
6
  cobweb/base/decorators.py,sha256=wDCaQ94aAZGxks9Ljc0aXq6omDXT1_yzFy83ZW6VbVI,930
@@ -8,9 +8,9 @@ cobweb/base/item.py,sha256=pMriHStzUXtSvIf5Z3KXsP-bCvjlG1gM3z33wWeuoH8,966
8
8
  cobweb/base/log.py,sha256=L01hXdk3L2qEm9X1FOXQ9VmWIoHSELe0cyZvrdAN61A,2003
9
9
  cobweb/base/request.py,sha256=tEkgMVUfdQI-kZuzWuiit9P_q4Q9-_RZh9aXXpc0314,2352
10
10
  cobweb/base/response.py,sha256=eB1DWMXFCpn3cJ3yzgCRU1WeZAdayGDohRgdjdMUFN4,406
11
- cobweb/base/seed.py,sha256=QxlXztWjV8VvcHu-cTzyoDHaDvoX26iyoJKiWk64HnE,2759
11
+ cobweb/base/seed.py,sha256=Uz_VBRlAxNYQcFHk3tsZFMlU96yPOedHaWGTvk-zKd8,2908
12
12
  cobweb/crawlers/__init__.py,sha256=Rr3DTjD-abMA1_FYcQJZYNvQvcWMuEVcsIU6duqHrw4,75
13
- cobweb/crawlers/base_crawler.py,sha256=XZZIi0zOyGax1T9TkKGJ0Enw7927VJOcrya2DgR_fV8,5174
13
+ cobweb/crawlers/base_crawler.py,sha256=X7RdgKCcMcYI8NSyGVXdT03uP9uHsEiX137IeBlkg00,5189
14
14
  cobweb/crawlers/file_crawler.py,sha256=2Sjbdgxzqd41WykKUQE3QQlGai3T8k-pmHNmPlTchjQ,4454
15
15
  cobweb/db/__init__.py,sha256=ut0iEyBLjcJL06WNG_5_d4hO5PJWvDrKWMkDOdmgh2M,30
16
16
  cobweb/db/redis_db.py,sha256=NNI2QkRV1hEZI-z-COEncXt88z3pZN6wusKlcQzc8V4,4304
@@ -18,15 +18,15 @@ cobweb/exceptions/__init__.py,sha256=E9SHnJBbhD7fOgPFMswqyOf8SKRDrI_i25L0bSpohvk
18
18
  cobweb/exceptions/oss_db_exception.py,sha256=iP_AImjNHT3-Iv49zCFQ3rdLnlvuHa3h2BXApgrOYpA,636
19
19
  cobweb/launchers/__init__.py,sha256=qwlkEJVri7dvCgi45aX3lqAmQS0HrPicAipDvH75kew,69
20
20
  cobweb/launchers/launcher.py,sha256=LQdlpaF4fafEX01_3B9CB6hD-6YMPag6QOGl9rDprNE,5707
21
- cobweb/launchers/launcher_pro.py,sha256=20o-Yd_JAdzJPchpilCtCd61hrf90LKkxXcCjm2azhQ,6586
21
+ cobweb/launchers/launcher_pro.py,sha256=lBHj5OaLNRbHwohI3tRRL3WZQ0uOolZAb8HdoTftq40,6594
22
22
  cobweb/pipelines/__init__.py,sha256=xanY-Z1d7zRR5JhCdW2htzrAywnKBkigiaUlTFa6of0,80
23
23
  cobweb/pipelines/base_pipeline.py,sha256=fYnWf79GmhufXpcnMa3te18SbmnVeYLwxfyo-zLd9CY,1577
24
24
  cobweb/pipelines/loghub_pipeline.py,sha256=cjPO6w6UJ0jNw2fVvdX0BCdlm58T7dmYXlxzXOBpvfY,1027
25
25
  cobweb/utils/__init__.py,sha256=JTE4sBfHnKHhD6w9Auk0MIT7O9BMOamCeryhlHNx3Zg,47
26
26
  cobweb/utils/oss.py,sha256=qAl05ybL2Jp6KFjHDHVMfmeBHQmDKPtZleHjHAY7LZc,3277
27
27
  cobweb/utils/tools.py,sha256=bVd3iRGBvwhohQAH7AXTTjbmQ54Z35K0O-fatEyhePU,1249
28
- cobweb_launcher-1.1.16.dist-info/LICENSE,sha256=z1rxSIGOyzcSb3orZxFPxzx-0C1vTocmswqBNxpKfEk,1063
29
- cobweb_launcher-1.1.16.dist-info/METADATA,sha256=jMP0dSD4lbeLtjDjQuZSlfIRacdVMDMLPDCjWpPBKI8,1246
30
- cobweb_launcher-1.1.16.dist-info/WHEEL,sha256=ewwEueio1C2XeHTvT17n8dZUJgOvyCWCt0WVNLClP9o,92
31
- cobweb_launcher-1.1.16.dist-info/top_level.txt,sha256=4GETBGNsKqiCUezmT-mJn7tjhcDlu7nLIV5gGgHBW4I,7
32
- cobweb_launcher-1.1.16.dist-info/RECORD,,
28
+ cobweb_launcher-1.1.17.dist-info/LICENSE,sha256=z1rxSIGOyzcSb3orZxFPxzx-0C1vTocmswqBNxpKfEk,1063
29
+ cobweb_launcher-1.1.17.dist-info/METADATA,sha256=j-Mf27c0yJOKyZWxGEg3nOvWJE2fQaKZnyB6-MBEMis,1246
30
+ cobweb_launcher-1.1.17.dist-info/WHEEL,sha256=ewwEueio1C2XeHTvT17n8dZUJgOvyCWCt0WVNLClP9o,92
31
+ cobweb_launcher-1.1.17.dist-info/top_level.txt,sha256=4GETBGNsKqiCUezmT-mJn7tjhcDlu7nLIV5gGgHBW4I,7
32
+ cobweb_launcher-1.1.17.dist-info/RECORD,,