orbitkit 0.8.56__tar.gz → 0.8.58__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 (79) hide show
  1. {orbitkit-0.8.56/orbitkit.egg-info → orbitkit-0.8.58}/PKG-INFO +1 -1
  2. orbitkit-0.8.58/orbitkit/VERSION +1 -0
  3. {orbitkit-0.8.56 → orbitkit-0.8.58}/orbitkit/airflow_handler/data_preprocessing.py +4 -1
  4. {orbitkit-0.8.56 → orbitkit-0.8.58}/orbitkit/airflow_handler/file_flow_exit_process.py +5 -0
  5. {orbitkit-0.8.56 → orbitkit-0.8.58/orbitkit.egg-info}/PKG-INFO +1 -1
  6. orbitkit-0.8.56/orbitkit/VERSION +0 -1
  7. {orbitkit-0.8.56 → orbitkit-0.8.58}/LICENSE +0 -0
  8. {orbitkit-0.8.56 → orbitkit-0.8.58}/MANIFEST.in +0 -0
  9. {orbitkit-0.8.56 → orbitkit-0.8.58}/README.md +0 -0
  10. {orbitkit-0.8.56 → orbitkit-0.8.58}/orbitkit/__init__.py +0 -0
  11. {orbitkit-0.8.56 → orbitkit-0.8.58}/orbitkit/airflow_handler/__init__.py +0 -0
  12. {orbitkit-0.8.56 → orbitkit-0.8.58}/orbitkit/airflow_handler/file_flow_entry_process.py +0 -0
  13. {orbitkit-0.8.56 → orbitkit-0.8.58}/orbitkit/airflow_handler/file_handler.py +0 -0
  14. {orbitkit-0.8.56 → orbitkit-0.8.58}/orbitkit/airflow_handler/file_handler_v2.py +0 -0
  15. {orbitkit-0.8.56 → orbitkit-0.8.58}/orbitkit/audio_transcoder/__init__.py +0 -0
  16. {orbitkit-0.8.56 → orbitkit-0.8.58}/orbitkit/audio_transcoder/netmind_extract_v1.py +0 -0
  17. {orbitkit-0.8.56 → orbitkit-0.8.58}/orbitkit/constant/__init__.py +0 -0
  18. {orbitkit-0.8.56 → orbitkit-0.8.58}/orbitkit/constant/report_schema.py +0 -0
  19. {orbitkit-0.8.56 → orbitkit-0.8.58}/orbitkit/id_srv/__init__.py +0 -0
  20. {orbitkit-0.8.56 → orbitkit-0.8.58}/orbitkit/id_srv/id_gen.py +0 -0
  21. {orbitkit-0.8.56 → orbitkit-0.8.58}/orbitkit/id_srv/id_perm_like.py +0 -0
  22. {orbitkit-0.8.56 → orbitkit-0.8.58}/orbitkit/lark_send/__init__.py +0 -0
  23. {orbitkit-0.8.56 → orbitkit-0.8.58}/orbitkit/lark_send/lark.py +0 -0
  24. {orbitkit-0.8.56 → orbitkit-0.8.58}/orbitkit/llm_tools/__init__.py +0 -0
  25. {orbitkit-0.8.56 → orbitkit-0.8.58}/orbitkit/llm_tools/quick_rag_chat.py +0 -0
  26. {orbitkit-0.8.56 → orbitkit-0.8.58}/orbitkit/orbit_type/__init__.py +0 -0
  27. {orbitkit-0.8.56 → orbitkit-0.8.58}/orbitkit/orbit_type/doc_4_compile_rule.py +0 -0
  28. {orbitkit-0.8.56 → orbitkit-0.8.58}/orbitkit/orbit_type/orbit_type_simple.py +0 -0
  29. {orbitkit-0.8.56 → orbitkit-0.8.58}/orbitkit/orbit_type/tools.py +0 -0
  30. {orbitkit-0.8.56 → orbitkit-0.8.58}/orbitkit/pdf_embedding/__init__.py +0 -0
  31. {orbitkit-0.8.56 → orbitkit-0.8.58}/orbitkit/pdf_embedding/pdf_txt_embedding.py +0 -0
  32. {orbitkit-0.8.56 → orbitkit-0.8.58}/orbitkit/pdf_embedding/pdf_txt_embedding_v2.py +0 -0
  33. {orbitkit-0.8.56 → orbitkit-0.8.58}/orbitkit/pdf_extractor/__init__.py +0 -0
  34. {orbitkit-0.8.56 → orbitkit-0.8.58}/orbitkit/pdf_extractor/a_stock_extractor_v1.py +0 -0
  35. {orbitkit-0.8.56 → orbitkit-0.8.58}/orbitkit/pdf_extractor/exceptions.py +0 -0
  36. {orbitkit-0.8.56 → orbitkit-0.8.58}/orbitkit/pdf_extractor/pdf_block_extractor_base.py +0 -0
  37. {orbitkit-0.8.56 → orbitkit-0.8.58}/orbitkit/pdf_extractor/pdf_block_extractor_v1.py +0 -0
  38. {orbitkit-0.8.56 → orbitkit-0.8.58}/orbitkit/pdf_extractor/pdf_block_extractor_v2.py +0 -0
  39. {orbitkit-0.8.56 → orbitkit-0.8.58}/orbitkit/pdf_extractor/pdf_extractor_azure.py +0 -0
  40. {orbitkit-0.8.56 → orbitkit-0.8.58}/orbitkit/pdf_extractor/pdf_extractor_minerU_v1.py +0 -0
  41. {orbitkit-0.8.56 → orbitkit-0.8.58}/orbitkit/pdf_extractor/pdf_extractor_netmind_v1.py +0 -0
  42. {orbitkit-0.8.56 → orbitkit-0.8.58}/orbitkit/pdf_extractor/pdf_extractor_netmind_v2.py +0 -0
  43. {orbitkit-0.8.56 → orbitkit-0.8.58}/orbitkit/pdf_extractor/pdf_extractor_netmind_v3.py +0 -0
  44. {orbitkit-0.8.56 → orbitkit-0.8.58}/orbitkit/pdf_extractor/pdf_extractor_orbit.py +0 -0
  45. {orbitkit-0.8.56 → orbitkit-0.8.58}/orbitkit/pdf_extractor_simple/__init__.py +0 -0
  46. {orbitkit-0.8.56 → orbitkit-0.8.58}/orbitkit/pdf_extractor_simple/base.py +0 -0
  47. {orbitkit-0.8.56 → orbitkit-0.8.58}/orbitkit/pdf_extractor_simple/cloud_provider.py +0 -0
  48. {orbitkit-0.8.56 → orbitkit-0.8.58}/orbitkit/pdf_extractor_simple/core.py +0 -0
  49. {orbitkit-0.8.56 → orbitkit-0.8.58}/orbitkit/pdf_extractor_simple/exceptions.py +0 -0
  50. {orbitkit-0.8.56 → orbitkit-0.8.58}/orbitkit/pdf_extractor_simple/extractors.py +0 -0
  51. {orbitkit-0.8.56 → orbitkit-0.8.58}/orbitkit/pdf_extractor_simple/utils.py +0 -0
  52. {orbitkit-0.8.56 → orbitkit-0.8.58}/orbitkit/pdf_writer/__init__.py +0 -0
  53. {orbitkit-0.8.56 → orbitkit-0.8.58}/orbitkit/pdf_writer/pdf_writer_simple.py +0 -0
  54. {orbitkit-0.8.56 → orbitkit-0.8.58}/orbitkit/util/__init__.py +0 -0
  55. {orbitkit-0.8.56 → orbitkit-0.8.58}/orbitkit/util/cache_asset_downloader.py +0 -0
  56. {orbitkit-0.8.56 → orbitkit-0.8.58}/orbitkit/util/common.py +0 -0
  57. {orbitkit-0.8.56 → orbitkit-0.8.58}/orbitkit/util/customize_regix_manager.py +0 -0
  58. {orbitkit-0.8.56 → orbitkit-0.8.58}/orbitkit/util/secret_manager.py +0 -0
  59. {orbitkit-0.8.56 → orbitkit-0.8.58}/orbitkit/util/universal_extractor.py +0 -0
  60. {orbitkit-0.8.56 → orbitkit-0.8.58}/orbitkit/util/util_aliyun.py +0 -0
  61. {orbitkit-0.8.56 → orbitkit-0.8.58}/orbitkit/util/util_aliyun_oss_simple.py +0 -0
  62. {orbitkit-0.8.56 → orbitkit-0.8.58}/orbitkit/util/util_aws.py +0 -0
  63. {orbitkit-0.8.56 → orbitkit-0.8.58}/orbitkit/util/util_aws_s3_wrapper.py +0 -0
  64. {orbitkit-0.8.56 → orbitkit-0.8.58}/orbitkit/util/util_date.py +0 -0
  65. {orbitkit-0.8.56 → orbitkit-0.8.58}/orbitkit/util/util_html.py +0 -0
  66. {orbitkit-0.8.56 → orbitkit-0.8.58}/orbitkit/util/util_kafka.py +0 -0
  67. {orbitkit-0.8.56 → orbitkit-0.8.58}/orbitkit/util/util_md5.py +0 -0
  68. {orbitkit-0.8.56 → orbitkit-0.8.58}/orbitkit/util/util_selenium.py +0 -0
  69. {orbitkit-0.8.56 → orbitkit-0.8.58}/orbitkit/util/util_simple_timer.py +0 -0
  70. {orbitkit-0.8.56 → orbitkit-0.8.58}/orbitkit/util/util_str.py +0 -0
  71. {orbitkit-0.8.56 → orbitkit-0.8.58}/orbitkit/util/util_type_mapping.py +0 -0
  72. {orbitkit-0.8.56 → orbitkit-0.8.58}/orbitkit/util/util_url.py +0 -0
  73. {orbitkit-0.8.56 → orbitkit-0.8.58}/orbitkit.egg-info/SOURCES.txt +0 -0
  74. {orbitkit-0.8.56 → orbitkit-0.8.58}/orbitkit.egg-info/dependency_links.txt +0 -0
  75. {orbitkit-0.8.56 → orbitkit-0.8.58}/orbitkit.egg-info/not-zip-safe +0 -0
  76. {orbitkit-0.8.56 → orbitkit-0.8.58}/orbitkit.egg-info/requires.txt +0 -0
  77. {orbitkit-0.8.56 → orbitkit-0.8.58}/orbitkit.egg-info/top_level.txt +0 -0
  78. {orbitkit-0.8.56 → orbitkit-0.8.58}/setup.cfg +0 -0
  79. {orbitkit-0.8.56 → orbitkit-0.8.58}/setup.py +0 -0
@@ -1,6 +1,6 @@
1
1
  Metadata-Version: 2.4
2
2
  Name: orbitkit
3
- Version: 0.8.56
3
+ Version: 0.8.58
4
4
  Summary: This project is only for Orbit Tech internal use.
5
5
  Home-page: https://github.com/clown-0726/orbitkit
6
6
  Author: Lilu Cao
@@ -0,0 +1 @@
1
+ 0.8.58
@@ -3,6 +3,7 @@ import datetime
3
3
  from collections import defaultdict
4
4
  from importlib.metadata import version
5
5
  import googletrans
6
+ from dateutil.relativedelta import relativedelta
6
7
 
7
8
 
8
9
  class DocumentProcessor:
@@ -122,11 +123,13 @@ class DocumentProcessor:
122
123
  @staticmethod
123
124
  def update_target_stage_by_reported_at(doc, target_stage):
124
125
  date_str = doc.get('x_reported_at_utc_date', '1970-01-01')
126
+ now = datetime.datetime.now()
127
+ one_year_ago = now - relativedelta(years=1)
125
128
  try:
126
129
  reported_date = datetime.datetime.strptime(date_str, "%Y-%m-%d")
127
130
  except ValueError:
128
131
  reported_date = datetime.datetime(1970, 1, 1)
129
- return "extract" if reported_date < datetime.datetime(2023, 1, 1) else target_stage
132
+ return "extract" if reported_date < one_year_ago else target_stage
130
133
 
131
134
  @staticmethod
132
135
  def update_target_stage_by_perm_match(doc, target_stage):
@@ -133,6 +133,11 @@ class FlowUpdater:
133
133
  db_doc = self._check_and_create_collection(data_source).find_one({'_id': report_id},
134
134
  {'_id': 1, 'x_attachments': 1,
135
135
  'x_status_list': 1})
136
+ if not db_doc and data_source == 'filing_data':
137
+ data_source = 'filing_data_staging'
138
+ db_doc = self._check_and_create_collection(data_source).find_one({'_id': report_id},
139
+ {'_id': 1, 'x_attachments': 1,
140
+ 'x_status_list': 1})
136
141
  if not db_doc:
137
142
  logger.warning(f"No document found with id {report_id}.")
138
143
  return
@@ -1,6 +1,6 @@
1
1
  Metadata-Version: 2.4
2
2
  Name: orbitkit
3
- Version: 0.8.56
3
+ Version: 0.8.58
4
4
  Summary: This project is only for Orbit Tech internal use.
5
5
  Home-page: https://github.com/clown-0726/orbitkit
6
6
  Author: Lilu Cao
@@ -1 +0,0 @@
1
- 0.8.56
File without changes
File without changes
File without changes
File without changes
File without changes