orbitkit 0.8.39__tar.gz → 0.8.41__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 (78) hide show
  1. {orbitkit-0.8.39/orbitkit.egg-info → orbitkit-0.8.41}/PKG-INFO +1 -1
  2. orbitkit-0.8.41/orbitkit/VERSION +1 -0
  3. {orbitkit-0.8.39 → orbitkit-0.8.41}/orbitkit/airflow_handler/data_preprocessing.py +10 -1
  4. {orbitkit-0.8.39 → orbitkit-0.8.41}/orbitkit/airflow_handler/file_handler_v2.py +7 -6
  5. {orbitkit-0.8.39 → orbitkit-0.8.41/orbitkit.egg-info}/PKG-INFO +1 -1
  6. orbitkit-0.8.39/orbitkit/VERSION +0 -1
  7. {orbitkit-0.8.39 → orbitkit-0.8.41}/LICENSE +0 -0
  8. {orbitkit-0.8.39 → orbitkit-0.8.41}/MANIFEST.in +0 -0
  9. {orbitkit-0.8.39 → orbitkit-0.8.41}/README.md +0 -0
  10. {orbitkit-0.8.39 → orbitkit-0.8.41}/orbitkit/__init__.py +0 -0
  11. {orbitkit-0.8.39 → orbitkit-0.8.41}/orbitkit/airflow_handler/__init__.py +0 -0
  12. {orbitkit-0.8.39 → orbitkit-0.8.41}/orbitkit/airflow_handler/file_flow_entry_process.py +0 -0
  13. {orbitkit-0.8.39 → orbitkit-0.8.41}/orbitkit/airflow_handler/file_flow_exit_process.py +0 -0
  14. {orbitkit-0.8.39 → orbitkit-0.8.41}/orbitkit/airflow_handler/file_handler.py +0 -0
  15. {orbitkit-0.8.39 → orbitkit-0.8.41}/orbitkit/audio_transcoder/__init__.py +0 -0
  16. {orbitkit-0.8.39 → orbitkit-0.8.41}/orbitkit/audio_transcoder/netmind_extract_v1.py +0 -0
  17. {orbitkit-0.8.39 → orbitkit-0.8.41}/orbitkit/constant/__init__.py +0 -0
  18. {orbitkit-0.8.39 → orbitkit-0.8.41}/orbitkit/constant/report_schema.py +0 -0
  19. {orbitkit-0.8.39 → orbitkit-0.8.41}/orbitkit/id_srv/__init__.py +0 -0
  20. {orbitkit-0.8.39 → orbitkit-0.8.41}/orbitkit/id_srv/id_gen.py +0 -0
  21. {orbitkit-0.8.39 → orbitkit-0.8.41}/orbitkit/id_srv/id_perm_like.py +0 -0
  22. {orbitkit-0.8.39 → orbitkit-0.8.41}/orbitkit/lark_send/__init__.py +0 -0
  23. {orbitkit-0.8.39 → orbitkit-0.8.41}/orbitkit/lark_send/lark.py +0 -0
  24. {orbitkit-0.8.39 → orbitkit-0.8.41}/orbitkit/llm_tools/__init__.py +0 -0
  25. {orbitkit-0.8.39 → orbitkit-0.8.41}/orbitkit/llm_tools/quick_rag_chat.py +0 -0
  26. {orbitkit-0.8.39 → orbitkit-0.8.41}/orbitkit/orbit_type/__init__.py +0 -0
  27. {orbitkit-0.8.39 → orbitkit-0.8.41}/orbitkit/orbit_type/doc_4_compile_rule.py +0 -0
  28. {orbitkit-0.8.39 → orbitkit-0.8.41}/orbitkit/orbit_type/orbit_type_simple.py +0 -0
  29. {orbitkit-0.8.39 → orbitkit-0.8.41}/orbitkit/orbit_type/tools.py +0 -0
  30. {orbitkit-0.8.39 → orbitkit-0.8.41}/orbitkit/pdf_embedding/__init__.py +0 -0
  31. {orbitkit-0.8.39 → orbitkit-0.8.41}/orbitkit/pdf_embedding/pdf_txt_embedding.py +0 -0
  32. {orbitkit-0.8.39 → orbitkit-0.8.41}/orbitkit/pdf_embedding/pdf_txt_embedding_v2.py +0 -0
  33. {orbitkit-0.8.39 → orbitkit-0.8.41}/orbitkit/pdf_extractor/__init__.py +0 -0
  34. {orbitkit-0.8.39 → orbitkit-0.8.41}/orbitkit/pdf_extractor/a_stock_extractor_v1.py +0 -0
  35. {orbitkit-0.8.39 → orbitkit-0.8.41}/orbitkit/pdf_extractor/exceptions.py +0 -0
  36. {orbitkit-0.8.39 → orbitkit-0.8.41}/orbitkit/pdf_extractor/pdf_block_extractor_base.py +0 -0
  37. {orbitkit-0.8.39 → orbitkit-0.8.41}/orbitkit/pdf_extractor/pdf_block_extractor_v1.py +0 -0
  38. {orbitkit-0.8.39 → orbitkit-0.8.41}/orbitkit/pdf_extractor/pdf_block_extractor_v2.py +0 -0
  39. {orbitkit-0.8.39 → orbitkit-0.8.41}/orbitkit/pdf_extractor/pdf_extractor_azure.py +0 -0
  40. {orbitkit-0.8.39 → orbitkit-0.8.41}/orbitkit/pdf_extractor/pdf_extractor_minerU_v1.py +0 -0
  41. {orbitkit-0.8.39 → orbitkit-0.8.41}/orbitkit/pdf_extractor/pdf_extractor_netmind_v1.py +0 -0
  42. {orbitkit-0.8.39 → orbitkit-0.8.41}/orbitkit/pdf_extractor/pdf_extractor_netmind_v2.py +0 -0
  43. {orbitkit-0.8.39 → orbitkit-0.8.41}/orbitkit/pdf_extractor/pdf_extractor_netmind_v3.py +0 -0
  44. {orbitkit-0.8.39 → orbitkit-0.8.41}/orbitkit/pdf_extractor/pdf_extractor_orbit.py +0 -0
  45. {orbitkit-0.8.39 → orbitkit-0.8.41}/orbitkit/pdf_extractor_simple/__init__.py +0 -0
  46. {orbitkit-0.8.39 → orbitkit-0.8.41}/orbitkit/pdf_extractor_simple/base.py +0 -0
  47. {orbitkit-0.8.39 → orbitkit-0.8.41}/orbitkit/pdf_extractor_simple/cloud_provider.py +0 -0
  48. {orbitkit-0.8.39 → orbitkit-0.8.41}/orbitkit/pdf_extractor_simple/core.py +0 -0
  49. {orbitkit-0.8.39 → orbitkit-0.8.41}/orbitkit/pdf_extractor_simple/exceptions.py +0 -0
  50. {orbitkit-0.8.39 → orbitkit-0.8.41}/orbitkit/pdf_extractor_simple/extractors.py +0 -0
  51. {orbitkit-0.8.39 → orbitkit-0.8.41}/orbitkit/pdf_extractor_simple/utils.py +0 -0
  52. {orbitkit-0.8.39 → orbitkit-0.8.41}/orbitkit/pdf_writer/__init__.py +0 -0
  53. {orbitkit-0.8.39 → orbitkit-0.8.41}/orbitkit/pdf_writer/pdf_writer_simple.py +0 -0
  54. {orbitkit-0.8.39 → orbitkit-0.8.41}/orbitkit/util/__init__.py +0 -0
  55. {orbitkit-0.8.39 → orbitkit-0.8.41}/orbitkit/util/cache_asset_downloader.py +0 -0
  56. {orbitkit-0.8.39 → orbitkit-0.8.41}/orbitkit/util/common.py +0 -0
  57. {orbitkit-0.8.39 → orbitkit-0.8.41}/orbitkit/util/customize_regix_manager.py +0 -0
  58. {orbitkit-0.8.39 → orbitkit-0.8.41}/orbitkit/util/secret_manager.py +0 -0
  59. {orbitkit-0.8.39 → orbitkit-0.8.41}/orbitkit/util/util_aliyun.py +0 -0
  60. {orbitkit-0.8.39 → orbitkit-0.8.41}/orbitkit/util/util_aliyun_oss_simple.py +0 -0
  61. {orbitkit-0.8.39 → orbitkit-0.8.41}/orbitkit/util/util_aws.py +0 -0
  62. {orbitkit-0.8.39 → orbitkit-0.8.41}/orbitkit/util/util_aws_s3_wrapper.py +0 -0
  63. {orbitkit-0.8.39 → orbitkit-0.8.41}/orbitkit/util/util_date.py +0 -0
  64. {orbitkit-0.8.39 → orbitkit-0.8.41}/orbitkit/util/util_html.py +0 -0
  65. {orbitkit-0.8.39 → orbitkit-0.8.41}/orbitkit/util/util_kafka.py +0 -0
  66. {orbitkit-0.8.39 → orbitkit-0.8.41}/orbitkit/util/util_md5.py +0 -0
  67. {orbitkit-0.8.39 → orbitkit-0.8.41}/orbitkit/util/util_selenium.py +0 -0
  68. {orbitkit-0.8.39 → orbitkit-0.8.41}/orbitkit/util/util_simple_timer.py +0 -0
  69. {orbitkit-0.8.39 → orbitkit-0.8.41}/orbitkit/util/util_str.py +0 -0
  70. {orbitkit-0.8.39 → orbitkit-0.8.41}/orbitkit/util/util_type_mapping.py +0 -0
  71. {orbitkit-0.8.39 → orbitkit-0.8.41}/orbitkit/util/util_url.py +0 -0
  72. {orbitkit-0.8.39 → orbitkit-0.8.41}/orbitkit.egg-info/SOURCES.txt +0 -0
  73. {orbitkit-0.8.39 → orbitkit-0.8.41}/orbitkit.egg-info/dependency_links.txt +0 -0
  74. {orbitkit-0.8.39 → orbitkit-0.8.41}/orbitkit.egg-info/not-zip-safe +0 -0
  75. {orbitkit-0.8.39 → orbitkit-0.8.41}/orbitkit.egg-info/requires.txt +0 -0
  76. {orbitkit-0.8.39 → orbitkit-0.8.41}/orbitkit.egg-info/top_level.txt +0 -0
  77. {orbitkit-0.8.39 → orbitkit-0.8.41}/setup.cfg +0 -0
  78. {orbitkit-0.8.39 → orbitkit-0.8.41}/setup.py +0 -0
@@ -1,6 +1,6 @@
1
1
  Metadata-Version: 2.1
2
2
  Name: orbitkit
3
- Version: 0.8.39
3
+ Version: 0.8.41
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
+ 0.8.41
@@ -1,9 +1,11 @@
1
1
  import os
2
2
  import datetime
3
3
  from collections import defaultdict
4
+ from googletrans import Translator
4
5
 
5
6
 
6
7
  class DocumentProcessor:
8
+
7
9
  AUDIO_SUFFIXES = [".mp3", ".wav", ".aac", ".wma", ".m4a"]
8
10
  VIDEO_SUFFIXES = [".mp4", ".avi", ".mkv", ".mov", ".wmv", ".flv", ".webm", ".mpeg", ".3gp", ".hevc"]
9
11
  PDF_SUFFIXES = [".pdf"]
@@ -17,6 +19,12 @@ class DocumentProcessor:
17
19
  def get_file_suffix(file_path):
18
20
  return f".{file_path.split('.')[-1]}".lower()
19
21
 
22
+ @staticmethod
23
+ def translate_text(text, dest='en'):
24
+ translator = Translator()
25
+ result = translator.translate(text, dest=dest)
26
+ return result.text
27
+
20
28
  @staticmethod
21
29
  def create_xbrl_template():
22
30
  return {
@@ -124,7 +132,8 @@ class DocumentProcessor:
124
132
  _, ext = os.path.splitext(att['store_path'])
125
133
  if ext in cls.AUDIO_SUFFIXES or ext in cls.VIDEO_SUFFIXES:
126
134
  add_extends = {
127
- "title": doc['x_orbit_data']['report_title'],
135
+ "original_title": doc['x_orbit_data']['report_title'],
136
+ "title": cls.translate_text(doc['x_orbit_data']['report_title']),
128
137
  "published": doc['x_reported_at_utc_date'],
129
138
  "tickers": [],
130
139
  "perm_id_list": doc['x_orbit_data']['perm_id_list'],
@@ -5,7 +5,8 @@ from contextlib import contextmanager
5
5
  from typing import List, Dict, Tuple, Any, Optional, Union
6
6
  from sqlalchemy.dialects.postgresql import array
7
7
  from sqlalchemy.orm import sessionmaker, scoped_session
8
- from sqlalchemy import create_engine, MetaData, Table, select, Column, Integer, insert, and_, text, delete, update, desc
8
+ from sqlalchemy import create_engine, MetaData, Table, select, Column, Integer, insert, and_, text, delete, update, \
9
+ desc, LABEL_STYLE_TABLENAME_PLUS_COL
9
10
 
10
11
  logger = logging.getLogger(__name__)
11
12
 
@@ -397,9 +398,9 @@ class FileFlowHandleV2:
397
398
  select(op_meta)
398
399
  .where(and_(*conditions)) if conditions else select(op_meta)
399
400
  )
400
- meta_subquery = meta_subquery.order_by(desc(op_meta.c.priority)).limit(limit_size).subquery()
401
+ meta_subquery = meta_subquery.order_by(desc(op_meta.c.priority)).limit(limit_size).subquery(name="public_op_meta")
401
402
  join_stmt = meta_subquery.outerjoin(op_attachment, meta_subquery.c.id == op_attachment.c.meta_id)
402
- stmt = select(meta_subquery, op_attachment).select_from(join_stmt)
403
+ stmt = (select(meta_subquery, op_attachment).select_from(join_stmt).set_label_style(LABEL_STYLE_TABLENAME_PLUS_COL))
403
404
 
404
405
  with self.session_scope() as session:
405
406
  result = session.execute(stmt).fetchall()
@@ -407,10 +408,10 @@ class FileFlowHandleV2:
407
408
  meta_map = {}
408
409
  for row in result:
409
410
  row_dict = dict(row._mapping)
410
- meta_id = row_dict['id']
411
+ meta_id = row_dict['public_op_meta_id']
411
412
 
412
- meta_fields: Dict[str, Any] = {k: v for k, v in row_dict.items() if k in op_meta.c}
413
- attachment_fields: Dict[str, Any] = {k: v for k, v in row_dict.items() if k in op_attachment.c}
413
+ meta_fields: Dict[str, Any] = {k.replace('public_op_meta_', ''): v for k, v in row_dict.items() if k.startswith('public_op_meta_')}
414
+ attachment_fields: Dict[str, Any] = {k.replace('public_op_attachment_', ''): v for k, v in row_dict.items() if k.startswith('public_op_attachment_')}
414
415
 
415
416
  meta_fields['created_at'] = meta_fields['created_at'].isoformat()
416
417
  meta_fields['updated_at'] = meta_fields['updated_at'].isoformat()
@@ -1,6 +1,6 @@
1
1
  Metadata-Version: 2.1
2
2
  Name: orbitkit
3
- Version: 0.8.39
3
+ Version: 0.8.41
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
- 0.8.39
File without changes
File without changes
File without changes
File without changes
File without changes