cobweb-launcher 3.1.23__tar.gz → 3.1.24__tar.gz

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.
Files changed (45) hide show
  1. {cobweb-launcher-3.1.23 → cobweb-launcher-3.1.24}/PKG-INFO +1 -1
  2. {cobweb-launcher-3.1.23 → cobweb-launcher-3.1.24}/cobweb/launchers/uploader.py +13 -30
  3. {cobweb-launcher-3.1.23 → cobweb-launcher-3.1.24}/cobweb/schedulers/scheduler_with_redis.py +2 -1
  4. {cobweb-launcher-3.1.23 → cobweb-launcher-3.1.24}/cobweb_launcher.egg-info/PKG-INFO +1 -1
  5. {cobweb-launcher-3.1.23 → cobweb-launcher-3.1.24}/setup.py +1 -1
  6. {cobweb-launcher-3.1.23 → cobweb-launcher-3.1.24}/LICENSE +0 -0
  7. {cobweb-launcher-3.1.23 → cobweb-launcher-3.1.24}/README.md +0 -0
  8. {cobweb-launcher-3.1.23 → cobweb-launcher-3.1.24}/cobweb/__init__.py +0 -0
  9. {cobweb-launcher-3.1.23 → cobweb-launcher-3.1.24}/cobweb/base/__init__.py +0 -0
  10. {cobweb-launcher-3.1.23 → cobweb-launcher-3.1.24}/cobweb/base/common_queue.py +0 -0
  11. {cobweb-launcher-3.1.23 → cobweb-launcher-3.1.24}/cobweb/base/item.py +0 -0
  12. {cobweb-launcher-3.1.23 → cobweb-launcher-3.1.24}/cobweb/base/logger.py +0 -0
  13. {cobweb-launcher-3.1.23 → cobweb-launcher-3.1.24}/cobweb/base/request.py +0 -0
  14. {cobweb-launcher-3.1.23 → cobweb-launcher-3.1.24}/cobweb/base/response.py +0 -0
  15. {cobweb-launcher-3.1.23 → cobweb-launcher-3.1.24}/cobweb/base/seed.py +0 -0
  16. {cobweb-launcher-3.1.23 → cobweb-launcher-3.1.24}/cobweb/base/task_queue.py +0 -0
  17. {cobweb-launcher-3.1.23 → cobweb-launcher-3.1.24}/cobweb/constant.py +0 -0
  18. {cobweb-launcher-3.1.23 → cobweb-launcher-3.1.24}/cobweb/crawlers/__init__.py +0 -0
  19. {cobweb-launcher-3.1.23 → cobweb-launcher-3.1.24}/cobweb/crawlers/crawler.py +0 -0
  20. {cobweb-launcher-3.1.23 → cobweb-launcher-3.1.24}/cobweb/db/__init__.py +0 -0
  21. {cobweb-launcher-3.1.23 → cobweb-launcher-3.1.24}/cobweb/db/api_db.py +0 -0
  22. {cobweb-launcher-3.1.23 → cobweb-launcher-3.1.24}/cobweb/db/redis_db.py +0 -0
  23. {cobweb-launcher-3.1.23 → cobweb-launcher-3.1.24}/cobweb/exceptions/__init__.py +0 -0
  24. {cobweb-launcher-3.1.23 → cobweb-launcher-3.1.24}/cobweb/exceptions/oss_db_exception.py +0 -0
  25. {cobweb-launcher-3.1.23 → cobweb-launcher-3.1.24}/cobweb/launchers/__init__.py +0 -0
  26. {cobweb-launcher-3.1.23 → cobweb-launcher-3.1.24}/cobweb/launchers/distributor.py +0 -0
  27. {cobweb-launcher-3.1.23 → cobweb-launcher-3.1.24}/cobweb/launchers/launcher.py +0 -0
  28. {cobweb-launcher-3.1.23 → cobweb-launcher-3.1.24}/cobweb/pipelines/__init__.py +0 -0
  29. {cobweb-launcher-3.1.23 → cobweb-launcher-3.1.24}/cobweb/pipelines/pipeline.py +0 -0
  30. {cobweb-launcher-3.1.23 → cobweb-launcher-3.1.24}/cobweb/pipelines/pipeline_csv.py +0 -0
  31. {cobweb-launcher-3.1.23 → cobweb-launcher-3.1.24}/cobweb/pipelines/pipeline_loghub.py +0 -0
  32. {cobweb-launcher-3.1.23 → cobweb-launcher-3.1.24}/cobweb/schedulers/__init__.py +0 -0
  33. {cobweb-launcher-3.1.23 → cobweb-launcher-3.1.24}/cobweb/schedulers/scheduler.py +0 -0
  34. {cobweb-launcher-3.1.23 → cobweb-launcher-3.1.24}/cobweb/setting.py +0 -0
  35. {cobweb-launcher-3.1.23 → cobweb-launcher-3.1.24}/cobweb/utils/__init__.py +0 -0
  36. {cobweb-launcher-3.1.23 → cobweb-launcher-3.1.24}/cobweb/utils/bloom.py +0 -0
  37. {cobweb-launcher-3.1.23 → cobweb-launcher-3.1.24}/cobweb/utils/decorators.py +0 -0
  38. {cobweb-launcher-3.1.23 → cobweb-launcher-3.1.24}/cobweb/utils/dotting.py +0 -0
  39. {cobweb-launcher-3.1.23 → cobweb-launcher-3.1.24}/cobweb/utils/oss.py +0 -0
  40. {cobweb-launcher-3.1.23 → cobweb-launcher-3.1.24}/cobweb/utils/tools.py +0 -0
  41. {cobweb-launcher-3.1.23 → cobweb-launcher-3.1.24}/cobweb_launcher.egg-info/SOURCES.txt +0 -0
  42. {cobweb-launcher-3.1.23 → cobweb-launcher-3.1.24}/cobweb_launcher.egg-info/dependency_links.txt +0 -0
  43. {cobweb-launcher-3.1.23 → cobweb-launcher-3.1.24}/cobweb_launcher.egg-info/requires.txt +0 -0
  44. {cobweb-launcher-3.1.23 → cobweb-launcher-3.1.24}/cobweb_launcher.egg-info/top_level.txt +0 -0
  45. {cobweb-launcher-3.1.23 → cobweb-launcher-3.1.24}/setup.cfg +0 -0
@@ -1,6 +1,6 @@
1
1
  Metadata-Version: 2.1
2
2
  Name: cobweb-launcher
3
- Version: 3.1.23
3
+ Version: 3.1.24
4
4
  Summary: spider_hole
5
5
  Home-page: https://github.com/Juannie-PP/cobweb
6
6
  Author: Juannie-PP
@@ -39,45 +39,28 @@ class Uploader(threading.Thread):
39
39
 
40
40
  @check_pause
41
41
  def upload_data(self):
42
- if task_list := self.task_queue.get_task_by_status(
43
- status=Status.UPLOAD, limit=self.upload_size
44
- ):
45
- try:
46
- data_info, task_ids = dict(), set()
42
+ try:
43
+ data_info, task_ids = dict(), set()
44
+ if task_list := self.task_queue.get_task_by_status(
45
+ status=Status.UPLOAD, limit=self.upload_size
46
+ ):
47
47
  for task_item in task_list:
48
48
  upload_data = self.pipeline.build(task_item.data)
49
49
  data_info.setdefault(task_item.data.table, []).append(upload_data)
50
+ task_ids.add(task_item.task_id)
50
51
 
51
52
  for table, datas in data_info.items():
52
53
  try:
53
54
  self.pipeline.upload(table, datas)
54
55
  except Exception as e:
55
56
  logger.info(e)
56
- except Exception as e:
57
- logger.info(e)
58
-
59
- if self.task_queue.status_length(status=Status.UPLOAD) < self.upload_size:
60
- time.sleep(self.wait_seconds)
61
-
62
- # data_info, seeds = {}, []
63
- # try:
64
- # for _ in range(self.upload_size):
65
- # item = self.upload.pop()
66
- # if not item:
67
- # break
68
- # # seeds.append(item.seed)
69
- # data = self.pipeline.build(item)
70
- # data_info.setdefault(item.table, []).append(data)
71
- # for table, datas in data_info.items():
72
- # try:
73
- # self.pipeline.upload(table, datas)
74
- # # TODO: 上传打点
75
- # except Exception as e:
76
- # logger.info(e)
77
- # except Exception as e:
78
- # logger.info(e)
79
- # if self.upload.length < self.upload_size:
80
- # time.sleep(self.wait_seconds)
57
+
58
+ self.task_queue.remove(task_ids)
59
+ except Exception as e:
60
+ logger.info(e)
61
+
62
+ if self.task_queue.status_length(status=Status.UPLOAD) < self.upload_size:
63
+ time.sleep(self.wait_seconds)
81
64
 
82
65
  def run(self):
83
66
  self.callback_register(self.upload_data, tag="Uploader")
@@ -52,7 +52,8 @@ class RedisScheduler(Scheduler):
52
52
  time.sleep(self.scheduler_wait_seconds)
53
53
  return
54
54
 
55
- if self.task_queue.status_length(Status.PENDING) >= self.todo_queue_size:
55
+ if self.task_queue.status_length(Status.PENDING) >= self.todo_queue_size\
56
+ or self.task_queue.length() > 5 * self.todo_queue_size:
56
57
  time.sleep(self.todo_queue_full_wait_seconds)
57
58
  return
58
59
 
@@ -1,6 +1,6 @@
1
1
  Metadata-Version: 2.1
2
2
  Name: cobweb-launcher
3
- Version: 3.1.23
3
+ Version: 3.1.24
4
4
  Summary: spider_hole
5
5
  Home-page: https://github.com/Juannie-PP/cobweb
6
6
  Author: Juannie-PP
@@ -5,7 +5,7 @@ with open("README.md", "r", encoding="utf-8") as fh:
5
5
 
6
6
  setup(
7
7
  name="cobweb-launcher",
8
- version="3.1.23",
8
+ version="3.1.24",
9
9
  packages=find_packages(),
10
10
  url="https://github.com/Juannie-PP/cobweb",
11
11
  license="MIT",