sunholo 0.100.2__tar.gz → 0.101.0__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 (165) hide show
  1. {sunholo-0.100.2 → sunholo-0.101.0}/PKG-INFO +2 -2
  2. {sunholo-0.100.2 → sunholo-0.101.0}/setup.py +1 -1
  3. {sunholo-0.100.2 → sunholo-0.101.0}/sunholo/gcs/add_file.py +35 -8
  4. {sunholo-0.100.2 → sunholo-0.101.0}/sunholo/gcs/download_url.py +1 -0
  5. sunholo-0.101.0/sunholo/gcs/extract_and_sign.py +41 -0
  6. sunholo-0.101.0/sunholo/genai/images.py +38 -0
  7. {sunholo-0.100.2 → sunholo-0.101.0}/sunholo/genai/process_funcs_cls.py +24 -2
  8. sunholo-0.101.0/sunholo/invoke/async_class.py +178 -0
  9. {sunholo-0.100.2 → sunholo-0.101.0}/sunholo/langfuse/prompts.py +8 -4
  10. {sunholo-0.100.2 → sunholo-0.101.0}/sunholo/streaming/content_buffer.py +0 -1
  11. sunholo-0.101.0/sunholo/utils/mime.py +63 -0
  12. {sunholo-0.100.2 → sunholo-0.101.0}/sunholo.egg-info/PKG-INFO +2 -2
  13. {sunholo-0.100.2 → sunholo-0.101.0}/sunholo.egg-info/SOURCES.txt +4 -0
  14. sunholo-0.101.0/tests/test_async.py +93 -0
  15. sunholo-0.100.2/sunholo/invoke/async_class.py +0 -117
  16. {sunholo-0.100.2 → sunholo-0.101.0}/LICENSE.txt +0 -0
  17. {sunholo-0.100.2 → sunholo-0.101.0}/MANIFEST.in +0 -0
  18. {sunholo-0.100.2 → sunholo-0.101.0}/README.md +0 -0
  19. {sunholo-0.100.2 → sunholo-0.101.0}/setup.cfg +0 -0
  20. {sunholo-0.100.2 → sunholo-0.101.0}/sunholo/__init__.py +0 -0
  21. {sunholo-0.100.2 → sunholo-0.101.0}/sunholo/agents/__init__.py +0 -0
  22. {sunholo-0.100.2 → sunholo-0.101.0}/sunholo/agents/chat_history.py +0 -0
  23. {sunholo-0.100.2 → sunholo-0.101.0}/sunholo/agents/dispatch_to_qa.py +0 -0
  24. {sunholo-0.100.2 → sunholo-0.101.0}/sunholo/agents/fastapi/__init__.py +0 -0
  25. {sunholo-0.100.2 → sunholo-0.101.0}/sunholo/agents/fastapi/base.py +0 -0
  26. {sunholo-0.100.2 → sunholo-0.101.0}/sunholo/agents/fastapi/qna_routes.py +0 -0
  27. {sunholo-0.100.2 → sunholo-0.101.0}/sunholo/agents/flask/__init__.py +0 -0
  28. {sunholo-0.100.2 → sunholo-0.101.0}/sunholo/agents/flask/base.py +0 -0
  29. {sunholo-0.100.2 → sunholo-0.101.0}/sunholo/agents/flask/qna_routes.py +0 -0
  30. {sunholo-0.100.2 → sunholo-0.101.0}/sunholo/agents/flask/vac_routes.py +0 -0
  31. {sunholo-0.100.2 → sunholo-0.101.0}/sunholo/agents/langserve.py +0 -0
  32. {sunholo-0.100.2 → sunholo-0.101.0}/sunholo/agents/pubsub.py +0 -0
  33. {sunholo-0.100.2 → sunholo-0.101.0}/sunholo/agents/route.py +0 -0
  34. {sunholo-0.100.2 → sunholo-0.101.0}/sunholo/agents/special_commands.py +0 -0
  35. {sunholo-0.100.2 → sunholo-0.101.0}/sunholo/agents/swagger.py +0 -0
  36. {sunholo-0.100.2 → sunholo-0.101.0}/sunholo/archive/__init__.py +0 -0
  37. {sunholo-0.100.2 → sunholo-0.101.0}/sunholo/archive/archive.py +0 -0
  38. {sunholo-0.100.2 → sunholo-0.101.0}/sunholo/auth/__init__.py +0 -0
  39. {sunholo-0.100.2 → sunholo-0.101.0}/sunholo/auth/gcloud.py +0 -0
  40. {sunholo-0.100.2 → sunholo-0.101.0}/sunholo/auth/refresh.py +0 -0
  41. {sunholo-0.100.2 → sunholo-0.101.0}/sunholo/auth/run.py +0 -0
  42. {sunholo-0.100.2 → sunholo-0.101.0}/sunholo/azure/__init__.py +0 -0
  43. {sunholo-0.100.2 → sunholo-0.101.0}/sunholo/azure/auth.py +0 -0
  44. {sunholo-0.100.2 → sunholo-0.101.0}/sunholo/azure/blobs.py +0 -0
  45. {sunholo-0.100.2 → sunholo-0.101.0}/sunholo/azure/event_grid.py +0 -0
  46. {sunholo-0.100.2 → sunholo-0.101.0}/sunholo/bots/__init__.py +0 -0
  47. {sunholo-0.100.2 → sunholo-0.101.0}/sunholo/bots/discord.py +0 -0
  48. {sunholo-0.100.2 → sunholo-0.101.0}/sunholo/bots/github_webhook.py +0 -0
  49. {sunholo-0.100.2 → sunholo-0.101.0}/sunholo/bots/webapp.py +0 -0
  50. {sunholo-0.100.2 → sunholo-0.101.0}/sunholo/chunker/__init__.py +0 -0
  51. {sunholo-0.100.2 → sunholo-0.101.0}/sunholo/chunker/azure.py +0 -0
  52. {sunholo-0.100.2 → sunholo-0.101.0}/sunholo/chunker/doc_handling.py +0 -0
  53. {sunholo-0.100.2 → sunholo-0.101.0}/sunholo/chunker/encode_metadata.py +0 -0
  54. {sunholo-0.100.2 → sunholo-0.101.0}/sunholo/chunker/images.py +0 -0
  55. {sunholo-0.100.2 → sunholo-0.101.0}/sunholo/chunker/loaders.py +0 -0
  56. {sunholo-0.100.2 → sunholo-0.101.0}/sunholo/chunker/message_data.py +0 -0
  57. {sunholo-0.100.2 → sunholo-0.101.0}/sunholo/chunker/pdfs.py +0 -0
  58. {sunholo-0.100.2 → sunholo-0.101.0}/sunholo/chunker/process_chunker_data.py +0 -0
  59. {sunholo-0.100.2 → sunholo-0.101.0}/sunholo/chunker/publish.py +0 -0
  60. {sunholo-0.100.2 → sunholo-0.101.0}/sunholo/chunker/pubsub.py +0 -0
  61. {sunholo-0.100.2 → sunholo-0.101.0}/sunholo/chunker/splitter.py +0 -0
  62. {sunholo-0.100.2 → sunholo-0.101.0}/sunholo/cli/__init__.py +0 -0
  63. {sunholo-0.100.2 → sunholo-0.101.0}/sunholo/cli/chat_vac.py +0 -0
  64. {sunholo-0.100.2 → sunholo-0.101.0}/sunholo/cli/cli.py +0 -0
  65. {sunholo-0.100.2 → sunholo-0.101.0}/sunholo/cli/cli_init.py +0 -0
  66. {sunholo-0.100.2 → sunholo-0.101.0}/sunholo/cli/configs.py +0 -0
  67. {sunholo-0.100.2 → sunholo-0.101.0}/sunholo/cli/deploy.py +0 -0
  68. {sunholo-0.100.2 → sunholo-0.101.0}/sunholo/cli/embedder.py +0 -0
  69. {sunholo-0.100.2 → sunholo-0.101.0}/sunholo/cli/merge_texts.py +0 -0
  70. {sunholo-0.100.2 → sunholo-0.101.0}/sunholo/cli/run_proxy.py +0 -0
  71. {sunholo-0.100.2 → sunholo-0.101.0}/sunholo/cli/sun_rich.py +0 -0
  72. {sunholo-0.100.2 → sunholo-0.101.0}/sunholo/cli/swagger.py +0 -0
  73. {sunholo-0.100.2 → sunholo-0.101.0}/sunholo/cli/vertex.py +0 -0
  74. {sunholo-0.100.2 → sunholo-0.101.0}/sunholo/components/__init__.py +0 -0
  75. {sunholo-0.100.2 → sunholo-0.101.0}/sunholo/components/llm.py +0 -0
  76. {sunholo-0.100.2 → sunholo-0.101.0}/sunholo/components/retriever.py +0 -0
  77. {sunholo-0.100.2 → sunholo-0.101.0}/sunholo/components/vectorstore.py +0 -0
  78. {sunholo-0.100.2 → sunholo-0.101.0}/sunholo/custom_logging.py +0 -0
  79. {sunholo-0.100.2 → sunholo-0.101.0}/sunholo/database/__init__.py +0 -0
  80. {sunholo-0.100.2 → sunholo-0.101.0}/sunholo/database/alloydb.py +0 -0
  81. {sunholo-0.100.2 → sunholo-0.101.0}/sunholo/database/alloydb_client.py +0 -0
  82. {sunholo-0.100.2 → sunholo-0.101.0}/sunholo/database/database.py +0 -0
  83. {sunholo-0.100.2 → sunholo-0.101.0}/sunholo/database/lancedb.py +0 -0
  84. {sunholo-0.100.2 → sunholo-0.101.0}/sunholo/database/sql/sb/create_function.sql +0 -0
  85. {sunholo-0.100.2 → sunholo-0.101.0}/sunholo/database/sql/sb/create_function_time.sql +0 -0
  86. {sunholo-0.100.2 → sunholo-0.101.0}/sunholo/database/sql/sb/create_table.sql +0 -0
  87. {sunholo-0.100.2 → sunholo-0.101.0}/sunholo/database/sql/sb/delete_source_row.sql +0 -0
  88. {sunholo-0.100.2 → sunholo-0.101.0}/sunholo/database/sql/sb/return_sources.sql +0 -0
  89. {sunholo-0.100.2 → sunholo-0.101.0}/sunholo/database/sql/sb/setup.sql +0 -0
  90. {sunholo-0.100.2 → sunholo-0.101.0}/sunholo/database/static_dbs.py +0 -0
  91. {sunholo-0.100.2 → sunholo-0.101.0}/sunholo/database/uuid.py +0 -0
  92. {sunholo-0.100.2 → sunholo-0.101.0}/sunholo/discovery_engine/__init__.py +0 -0
  93. {sunholo-0.100.2 → sunholo-0.101.0}/sunholo/discovery_engine/chunker_handler.py +0 -0
  94. {sunholo-0.100.2 → sunholo-0.101.0}/sunholo/discovery_engine/create_new.py +0 -0
  95. {sunholo-0.100.2 → sunholo-0.101.0}/sunholo/discovery_engine/discovery_engine_client.py +0 -0
  96. {sunholo-0.100.2 → sunholo-0.101.0}/sunholo/discovery_engine/get_ai_search_chunks.py +0 -0
  97. {sunholo-0.100.2 → sunholo-0.101.0}/sunholo/embedder/__init__.py +0 -0
  98. {sunholo-0.100.2 → sunholo-0.101.0}/sunholo/embedder/embed_chunk.py +0 -0
  99. {sunholo-0.100.2 → sunholo-0.101.0}/sunholo/excel/__init__.py +0 -0
  100. {sunholo-0.100.2 → sunholo-0.101.0}/sunholo/excel/plugin.py +0 -0
  101. {sunholo-0.100.2 → sunholo-0.101.0}/sunholo/gcs/__init__.py +0 -0
  102. {sunholo-0.100.2 → sunholo-0.101.0}/sunholo/gcs/download_folder.py +0 -0
  103. {sunholo-0.100.2 → sunholo-0.101.0}/sunholo/gcs/metadata.py +0 -0
  104. {sunholo-0.100.2 → sunholo-0.101.0}/sunholo/genai/__init__.py +0 -0
  105. {sunholo-0.100.2 → sunholo-0.101.0}/sunholo/genai/init.py +0 -0
  106. {sunholo-0.100.2 → sunholo-0.101.0}/sunholo/genai/safety.py +0 -0
  107. {sunholo-0.100.2 → sunholo-0.101.0}/sunholo/invoke/__init__.py +0 -0
  108. {sunholo-0.100.2 → sunholo-0.101.0}/sunholo/invoke/direct_vac_func.py +0 -0
  109. {sunholo-0.100.2 → sunholo-0.101.0}/sunholo/invoke/invoke_vac_utils.py +0 -0
  110. {sunholo-0.100.2 → sunholo-0.101.0}/sunholo/langfuse/__init__.py +0 -0
  111. {sunholo-0.100.2 → sunholo-0.101.0}/sunholo/langfuse/callback.py +0 -0
  112. {sunholo-0.100.2 → sunholo-0.101.0}/sunholo/langfuse/evals.py +0 -0
  113. {sunholo-0.100.2 → sunholo-0.101.0}/sunholo/llamaindex/__init__.py +0 -0
  114. {sunholo-0.100.2 → sunholo-0.101.0}/sunholo/llamaindex/get_files.py +0 -0
  115. {sunholo-0.100.2 → sunholo-0.101.0}/sunholo/llamaindex/import_files.py +0 -0
  116. {sunholo-0.100.2 → sunholo-0.101.0}/sunholo/llamaindex/llamaindex_class.py +0 -0
  117. {sunholo-0.100.2 → sunholo-0.101.0}/sunholo/llamaindex/user_history.py +0 -0
  118. {sunholo-0.100.2 → sunholo-0.101.0}/sunholo/lookup/__init__.py +0 -0
  119. {sunholo-0.100.2 → sunholo-0.101.0}/sunholo/lookup/model_lookup.yaml +0 -0
  120. {sunholo-0.100.2 → sunholo-0.101.0}/sunholo/patches/__init__.py +0 -0
  121. {sunholo-0.100.2 → sunholo-0.101.0}/sunholo/patches/langchain/__init__.py +0 -0
  122. {sunholo-0.100.2 → sunholo-0.101.0}/sunholo/patches/langchain/lancedb.py +0 -0
  123. {sunholo-0.100.2 → sunholo-0.101.0}/sunholo/patches/langchain/vertexai.py +0 -0
  124. {sunholo-0.100.2 → sunholo-0.101.0}/sunholo/pubsub/__init__.py +0 -0
  125. {sunholo-0.100.2 → sunholo-0.101.0}/sunholo/pubsub/process_pubsub.py +0 -0
  126. {sunholo-0.100.2 → sunholo-0.101.0}/sunholo/pubsub/pubsub_manager.py +0 -0
  127. {sunholo-0.100.2 → sunholo-0.101.0}/sunholo/qna/__init__.py +0 -0
  128. {sunholo-0.100.2 → sunholo-0.101.0}/sunholo/qna/parsers.py +0 -0
  129. {sunholo-0.100.2 → sunholo-0.101.0}/sunholo/qna/retry.py +0 -0
  130. {sunholo-0.100.2 → sunholo-0.101.0}/sunholo/streaming/__init__.py +0 -0
  131. {sunholo-0.100.2 → sunholo-0.101.0}/sunholo/streaming/langserve.py +0 -0
  132. {sunholo-0.100.2 → sunholo-0.101.0}/sunholo/streaming/stream_lookup.py +0 -0
  133. {sunholo-0.100.2 → sunholo-0.101.0}/sunholo/streaming/streaming.py +0 -0
  134. {sunholo-0.100.2 → sunholo-0.101.0}/sunholo/summarise/__init__.py +0 -0
  135. {sunholo-0.100.2 → sunholo-0.101.0}/sunholo/summarise/summarise.py +0 -0
  136. {sunholo-0.100.2 → sunholo-0.101.0}/sunholo/terraform/__init__.py +0 -0
  137. {sunholo-0.100.2 → sunholo-0.101.0}/sunholo/terraform/tfvars_editor.py +0 -0
  138. {sunholo-0.100.2 → sunholo-0.101.0}/sunholo/tools/__init__.py +0 -0
  139. {sunholo-0.100.2 → sunholo-0.101.0}/sunholo/tools/web_browser.py +0 -0
  140. {sunholo-0.100.2 → sunholo-0.101.0}/sunholo/utils/__init__.py +0 -0
  141. {sunholo-0.100.2 → sunholo-0.101.0}/sunholo/utils/api_key.py +0 -0
  142. {sunholo-0.100.2 → sunholo-0.101.0}/sunholo/utils/big_context.py +0 -0
  143. {sunholo-0.100.2 → sunholo-0.101.0}/sunholo/utils/config.py +0 -0
  144. {sunholo-0.100.2 → sunholo-0.101.0}/sunholo/utils/config_class.py +0 -0
  145. {sunholo-0.100.2 → sunholo-0.101.0}/sunholo/utils/config_schema.py +0 -0
  146. {sunholo-0.100.2 → sunholo-0.101.0}/sunholo/utils/gcp.py +0 -0
  147. {sunholo-0.100.2 → sunholo-0.101.0}/sunholo/utils/gcp_project.py +0 -0
  148. {sunholo-0.100.2 → sunholo-0.101.0}/sunholo/utils/parsers.py +0 -0
  149. {sunholo-0.100.2 → sunholo-0.101.0}/sunholo/utils/timedelta.py +0 -0
  150. {sunholo-0.100.2 → sunholo-0.101.0}/sunholo/utils/user_ids.py +0 -0
  151. {sunholo-0.100.2 → sunholo-0.101.0}/sunholo/utils/version.py +0 -0
  152. {sunholo-0.100.2 → sunholo-0.101.0}/sunholo/vertex/__init__.py +0 -0
  153. {sunholo-0.100.2 → sunholo-0.101.0}/sunholo/vertex/extensions_call.py +0 -0
  154. {sunholo-0.100.2 → sunholo-0.101.0}/sunholo/vertex/extensions_class.py +0 -0
  155. {sunholo-0.100.2 → sunholo-0.101.0}/sunholo/vertex/genai_functions.py +0 -0
  156. {sunholo-0.100.2 → sunholo-0.101.0}/sunholo/vertex/init.py +0 -0
  157. {sunholo-0.100.2 → sunholo-0.101.0}/sunholo/vertex/memory_tools.py +0 -0
  158. {sunholo-0.100.2 → sunholo-0.101.0}/sunholo/vertex/safety.py +0 -0
  159. {sunholo-0.100.2 → sunholo-0.101.0}/sunholo/vertex/type_dict_to_json.py +0 -0
  160. {sunholo-0.100.2 → sunholo-0.101.0}/sunholo.egg-info/dependency_links.txt +0 -0
  161. {sunholo-0.100.2 → sunholo-0.101.0}/sunholo.egg-info/entry_points.txt +0 -0
  162. {sunholo-0.100.2 → sunholo-0.101.0}/sunholo.egg-info/requires.txt +0 -0
  163. {sunholo-0.100.2 → sunholo-0.101.0}/sunholo.egg-info/top_level.txt +0 -0
  164. {sunholo-0.100.2 → sunholo-0.101.0}/tests/test_chat_history.py +0 -0
  165. {sunholo-0.100.2 → sunholo-0.101.0}/tests/test_config.py +0 -0
@@ -1,9 +1,9 @@
1
1
  Metadata-Version: 2.1
2
2
  Name: sunholo
3
- Version: 0.100.2
3
+ Version: 0.101.0
4
4
  Summary: Large Language Model DevOps - a package to help deploy LLMs to the Cloud.
5
5
  Home-page: https://github.com/sunholo-data/sunholo-py
6
- Download-URL: https://github.com/sunholo-data/sunholo-py/archive/refs/tags/v0.100.2.tar.gz
6
+ Download-URL: https://github.com/sunholo-data/sunholo-py/archive/refs/tags/v0.101.0.tar.gz
7
7
  Author: Holosun ApS
8
8
  Author-email: multivac@sunholo.com
9
9
  License: Apache License, Version 2.0
@@ -1,6 +1,6 @@
1
1
  from setuptools import setup, find_packages
2
2
 
3
- version = '0.100.2'
3
+ version = '0.101.0'
4
4
 
5
5
  setup(
6
6
  name='sunholo',
@@ -25,6 +25,40 @@ except ImportError:
25
25
  from ..custom_logging import log
26
26
  from ..utils import ConfigManager
27
27
 
28
+ def guess_image_type(file_path: str) -> str:
29
+ """
30
+ Guess the image type based on the file extension.
31
+
32
+ Args:
33
+ file_path (str): The path or URL of the image file.
34
+
35
+ Returns:
36
+ str: The guessed image type (e.g., "jpeg", "png", "gif", etc.)
37
+ or None if the extension is not recognized.
38
+ """
39
+ # Extract the file extension
40
+ _, ext = os.path.splitext(file_path)
41
+
42
+ # Normalize and remove the leading dot
43
+ ext = ext.lower().strip('.')
44
+
45
+ # Mapping of common file extensions to image types
46
+ extension_to_type = {
47
+ "jpg": "image/jpeg",
48
+ "jpeg": "image/jpeg",
49
+ "png": "image/png",
50
+ "gif": "image/gif",
51
+ "bmp": "image/bmp",
52
+ "tiff": "image/tiff",
53
+ "tif": "image/tiff",
54
+ "webp": "image/webp",
55
+ "ico": "image/ico",
56
+ "svg": "image/svg",
57
+ "pdf": "application/pdf",
58
+ }
59
+
60
+ return extension_to_type.get(ext.lower(), None)
61
+
28
62
 
29
63
  def handle_base64_image(base64_data: str, vector_name: str, extension: str):
30
64
  """
@@ -58,14 +92,7 @@ def handle_base64_image(base64_data: str, vector_name: str, extension: str):
58
92
  os.remove(filename) # Clean up the saved file
59
93
 
60
94
  # Determine MIME type based on extension
61
- mime_type = {
62
- ".jpg": "image/jpeg",
63
- ".jpeg": "image/jpeg",
64
- ".png": "image/png",
65
- ".gif": "image/gif",
66
- ".bmp": "image/bmp",
67
- ".tiff": "image/tiff"
68
- }.get(extension.lower(), "application/octet-stream") # Default MIME type if unknown
95
+ mime_type = guess_image_type(extension) or "application/octet-stream"
69
96
 
70
97
  return image_uri, mime_type
71
98
  except Exception as e:
@@ -176,6 +176,7 @@ def construct_download_link_simple(bucket_name:str, object_name:str) -> Tuple[st
176
176
  else:
177
177
  public_url = f"https://storage.cloud.google.com/{bucket_name}/{quote(object_name)}"
178
178
  filename = os.path.basename(object_name)
179
+ log.info(f"Created simple download link: {public_url}")
179
180
  return public_url, filename, False
180
181
 
181
182
  def parse_gs_uri(gs_uri: str) -> Tuple[str, str]:
@@ -0,0 +1,41 @@
1
+ import re
2
+ import asyncio
3
+
4
+ from .download_url import construct_download_link
5
+ from ..utils.mime import guess_mime_type
6
+ from ..custom_logging import log
7
+
8
+ async def extract_gs_uris_and_sign(content, pattern=r'gs://[^\n]+\.(?:png|jpg|jpeg|pdf|txt|md)'):
9
+
10
+ gs_matches = re.findall(pattern, content)
11
+ unique_gs_matches = set(gs_matches)
12
+ image_signed_urls = []
13
+ if unique_gs_matches:
14
+ log.info(f"Got gs matches: {unique_gs_matches}")
15
+
16
+ async def process_link(gs_url):
17
+ log.info(f"Processing {gs_url}")
18
+ link, encoded_filename, signed = await asyncio.to_thread(construct_download_link, gs_url)
19
+ if signed:
20
+ try:
21
+ mime_type = guess_mime_type(gs_url)
22
+ except Exception as err:
23
+ log.error(f"Could not find mime_type for {link} - {str(err)}")
24
+ mime_type = "application/octet-stream"
25
+
26
+ return {
27
+ "original": gs_url,
28
+ "link": link,
29
+ "name": encoded_filename,
30
+ "mime": mime_type,
31
+ "signed": signed
32
+ }
33
+ else:
34
+ log.info(f"Could not sign this GS_URI: {gs_url} - skipping")
35
+ return None
36
+
37
+ # Gather all tasks and run them concurrently
38
+ image_signed_urls = await asyncio.gather(*(process_link(gs_url) for gs_url in unique_gs_matches))
39
+
40
+ log.info(f"found files to msg: {image_signed_urls}")
41
+ return image_signed_urls
@@ -0,0 +1,38 @@
1
+ import re
2
+ from ..utils.mime import guess_mime_type
3
+ from ..gcs import get_bytes_from_gcs
4
+ from ..custom_logging import log
5
+ import io
6
+ import os
7
+ try:
8
+ import google.generativeai as genai
9
+ except ImportError:
10
+ genai = None
11
+
12
+ def extract_gs_images_and_genai_upload(content:str):
13
+ # Regular expression to find gs:// URLs
14
+ pattern = r'gs://[^ ]+\.(?:png|jpg|jpeg|pdf)'
15
+
16
+ gs_matches = re.findall(pattern, content)
17
+ unique_gs_matches = set(gs_matches)
18
+ output_gs_images = []
19
+ for gs_uri in unique_gs_matches:
20
+ mime_type = guess_mime_type(gs_uri)
21
+ if mime_type is None:
22
+ continue
23
+
24
+ log.info(f"Getting bytes from GCS: {gs_uri}")
25
+ image_bytes = get_bytes_from_gcs(gs_uri)
26
+ if image_bytes is None:
27
+ continue
28
+ image_file = io.BytesIO(image_bytes)
29
+ image_file.name = os.path.basename(gs_uri) # Assign a name, as some APIs require it
30
+
31
+ try:
32
+ uploaded_file = genai.upload_file(image_file)
33
+ output_gs_images.append(uploaded_file)
34
+
35
+ except Exception as e:
36
+ log.error(f"Error adding {gs_uri} to base64: {str(e)}")
37
+
38
+ return output_gs_images
@@ -17,9 +17,13 @@ try:
17
17
  except ImportError:
18
18
  genai = None
19
19
 
20
+ from .images import extract_gs_images_and_genai_upload
21
+
20
22
  if TYPE_CHECKING:
21
23
  from google.generativeai.protos import Part
22
24
 
25
+
26
+
23
27
  class GenAIFunctionProcessor:
24
28
  """
25
29
  A generic class for processing function calls from google.generativeai function calling models.
@@ -366,20 +370,22 @@ class GenAIFunctionProcessor:
366
370
 
367
371
  return clean
368
372
 
369
- def run_agent_loop(self, chat, content, callback, guardrail_max=10, loop_return=3):
373
+ def run_agent_loop(self, chat, content, callback=None, guardrail_max=10, loop_return=3):
370
374
  """
371
375
  Runs the agent loop, sending messages to the orchestrator, processing responses, and executing functions.
372
376
 
373
377
  Args:
374
378
  chat: The chat object for interaction with the orchestrator.
375
379
  content: The initial content to send to the agent.
376
- callback: The callback object for handling intermediate responses.
380
+ callback: The callback object for handling intermediate responses. If not supplied will use self.IOCallback()
377
381
  guardrail_max (int): The maximum number of iterations for the loop.
378
382
  loop_return (int): The number of last loop iterations to return. Default 3 will return last 3 iterations. If loop_return > guardrail_max then all iterations are returned.
379
383
 
380
384
  Returns:
381
385
  tuple: (big_text, usage_metadata) from the loop execution.
382
386
  """
387
+ if not callback:
388
+ callback = self.IOCallback()
383
389
  guardrail = 0
384
390
  big_result = []
385
391
  usage_metadata = {
@@ -535,6 +541,13 @@ class GenAIFunctionProcessor:
535
541
 
536
542
  if this_text:
537
543
  content.append(f"Agent: {this_text}")
544
+ # if text includes gs:// try to download it
545
+ image_uploads = extract_gs_images_and_genai_upload(this_text)
546
+ if image_uploads:
547
+ for img in image_uploads:
548
+ log.info(f"Adding {img=}")
549
+ content.append(img)
550
+ content.append(f"{img.name} was created by agent and added")
538
551
  log.info(f"[{guardrail}] Updated content:\n{this_text}")
539
552
  big_result.append(this_text)
540
553
  else:
@@ -567,6 +580,15 @@ class GenAIFunctionProcessor:
567
580
 
568
581
  return big_text, usage_metadata
569
582
 
583
+ class IOCallback:
584
+ """
585
+ This is a default callback that will print to console any tokens it recieves.
586
+ """
587
+ def on_llm_new_token(self, token:str):
588
+ print(token)
589
+ def on_llm_end(self, response):
590
+ print(f"\nFull response: \n{response}")
591
+
570
592
  @staticmethod
571
593
  def decide_to_go_on(go_on: bool, chat_summary: str) -> dict:
572
594
  """
@@ -0,0 +1,178 @@
1
+ import asyncio
2
+ from typing import Callable, Any, AsyncGenerator, Dict
3
+ import time
4
+ import traceback
5
+ from ..custom_logging import setup_logging
6
+ from tenacity import AsyncRetrying, retry_if_exception_type, wait_random_exponential, stop_after_attempt
7
+
8
+ log = setup_logging("sunholo_AsyncTaskRunner")
9
+
10
+ class AsyncTaskRunner:
11
+ def __init__(self, retry_enabled=False, retry_kwargs=None):
12
+ self.tasks = []
13
+ self.retry_enabled = retry_enabled
14
+ self.retry_kwargs = retry_kwargs or {}
15
+
16
+ def add_task(self, func: Callable[..., Any], *args: Any):
17
+ """
18
+ Adds a task to the list of tasks to be executed.
19
+ """
20
+ log.info(f"Adding task: {func.__name__} with args: {args}")
21
+ self.tasks.append((func.__name__, func, args))
22
+
23
+ async def run_async_as_completed(self) -> AsyncGenerator[Dict[str, Any], None]:
24
+ """
25
+ Runs all tasks concurrently and yields results as they complete,
26
+ while periodically sending heartbeat messages.
27
+ """
28
+ log.info("Running tasks asynchronously and yielding results as they complete")
29
+
30
+ # Create a queue for inter-coroutine communication
31
+ queue = asyncio.Queue()
32
+
33
+ # List to keep track of all running tasks and their heartbeats
34
+ task_infos = []
35
+
36
+ # Start all tasks and their corresponding heartbeats
37
+ for name, func, args in self.tasks:
38
+ # Create an event to signal task completion to the heartbeat
39
+ completion_event = asyncio.Event()
40
+
41
+ # Start the main task with retries
42
+ task_coro = self._run_with_retries(name, func, *args, queue=queue, completion_event=completion_event)
43
+ task = asyncio.create_task(task_coro)
44
+
45
+ # Start the heartbeat coroutine
46
+ heartbeat_coro = self._send_heartbeat(name, completion_event, queue)
47
+ heartbeat_task = asyncio.create_task(heartbeat_coro)
48
+
49
+ # Store task information for management
50
+ task_infos.append({
51
+ 'name': name,
52
+ 'task': task,
53
+ 'heartbeat_task': heartbeat_task,
54
+ 'completion_event': completion_event
55
+ })
56
+
57
+ log.info(f"Started task '{name}' and its heartbeat")
58
+
59
+ log.info(f"Started async run with {len(self.tasks)} tasks and heartbeats")
60
+
61
+ # Create a monitor task to detect when all tasks and heartbeats are done
62
+ monitor = asyncio.create_task(self._monitor_tasks(task_infos, queue))
63
+
64
+ # Continuously yield messages from the queue until sentinel is received
65
+ while True:
66
+ message = await queue.get()
67
+ if message is None:
68
+ log.info("Received sentinel. Exiting message loop.")
69
+ break # Sentinel received, all tasks and heartbeats are done
70
+ log.info(f"Received message from queue: {message}")
71
+ yield message
72
+
73
+ # Wait for the monitor to finish
74
+ await monitor
75
+
76
+ log.info("All tasks and heartbeats have completed")
77
+
78
+ async def _monitor_tasks(self, task_infos, queue):
79
+ """
80
+ Monitors the tasks and heartbeats, and sends a sentinel to the queue when done.
81
+ """
82
+ # Wait for all main tasks to complete
83
+ main_tasks = [info['task'] for info in task_infos]
84
+ log.info("Monitor: Waiting for all main tasks to complete")
85
+ await asyncio.gather(*main_tasks, return_exceptions=True)
86
+ log.info("Monitor: All main tasks have completed")
87
+
88
+ # Cancel all heartbeat tasks
89
+ for info in task_infos:
90
+ info['heartbeat_task'].cancel()
91
+ try:
92
+ await info['heartbeat_task']
93
+ except asyncio.CancelledError:
94
+ pass
95
+ log.info(f"Monitor: Heartbeat for task '{info['name']}' has been canceled")
96
+
97
+ # Send a sentinel to indicate completion
98
+ await queue.put(None)
99
+ log.info("Monitor: Sent sentinel to queue")
100
+
101
+ async def _run_with_retries(self, name: str, func: Callable[..., Any], *args: Any, queue: asyncio.Queue, completion_event: asyncio.Event) -> None:
102
+ """
103
+ Executes a task with optional retries and sends completion or error messages to the queue.
104
+ """
105
+ try:
106
+ if self.retry_enabled:
107
+ retry_kwargs = {
108
+ 'wait': wait_random_exponential(multiplier=1, max=60),
109
+ 'stop': stop_after_attempt(5),
110
+ 'retry': retry_if_exception_type(Exception),
111
+ **self.retry_kwargs
112
+ }
113
+ async for attempt in AsyncRetrying(**retry_kwargs):
114
+ with attempt:
115
+ log.info(f"Starting task '{name}' with retry")
116
+ result = await self._execute_task(func, *args)
117
+ await queue.put({'type': 'task_complete', 'func_name': name, 'result': result})
118
+ log.info(f"Sent 'task_complete' message for task '{name}'")
119
+ return
120
+ else:
121
+ log.info(f"Starting task '{name}' with no retry")
122
+ result = await self._execute_task(func, *args)
123
+ await queue.put({'type': 'task_complete', 'func_name': name, 'result': result})
124
+ log.info(f"Sent 'task_complete' message for task '{name}'")
125
+ except Exception as e:
126
+ log.error(f"Error in task '{name}': {e}\n{traceback.format_exc()}")
127
+ await queue.put({'type': 'task_error', 'func_name': name, 'error': f'{e}\n{traceback.format_exc()}'})
128
+ log.info(f"Sent 'task_error' message for task '{name}'")
129
+ finally:
130
+ log.info(f"Task '{name}' completed.")
131
+ # Set the completion event after sending the message
132
+ completion_event.set()
133
+
134
+ async def _execute_task(self, func: Callable[..., Any], *args: Any) -> Any:
135
+ """
136
+ Executes the given task function and returns its result.
137
+
138
+ Args:
139
+ func (Callable): The callable to execute.
140
+ *args: Arguments to pass to the callable.
141
+
142
+ Returns:
143
+ Any: The result of the task.
144
+ """
145
+ if asyncio.iscoroutinefunction(func):
146
+ return await func(*args)
147
+ else:
148
+ return await asyncio.to_thread(func, *args)
149
+
150
+ async def _send_heartbeat(self, func_name: str, completion_event: asyncio.Event, queue: asyncio.Queue, interval: int = 2):
151
+ """
152
+ Sends periodic heartbeat updates to indicate the task is still in progress.
153
+
154
+ Args:
155
+ func_name (str): The name of the task function.
156
+ completion_event (asyncio.Event): Event to signal when the task is completed.
157
+ queue (asyncio.Queue): The queue to send heartbeat messages to.
158
+ interval (int): How frequently to send heartbeat messages (in seconds).
159
+ """
160
+ start_time = time.time()
161
+ log.info(f"Starting heartbeat for task '{func_name}' with interval {interval} seconds")
162
+ try:
163
+ while not completion_event.is_set():
164
+ await asyncio.sleep(interval)
165
+ elapsed_time = int(time.time() - start_time)
166
+ heartbeat_message = {
167
+ 'type': 'heartbeat',
168
+ 'name': func_name,
169
+ 'interval': interval,
170
+ 'elapsed_time': elapsed_time
171
+ }
172
+ log.info(f"Sending heartbeat for task '{func_name}', running for {elapsed_time} seconds")
173
+ await queue.put(heartbeat_message)
174
+ except asyncio.CancelledError:
175
+ log.info(f"Heartbeat for task '{func_name}' has been canceled")
176
+ finally:
177
+ log.info(f"Heartbeat for task '{func_name}' stopped")
178
+
@@ -2,13 +2,14 @@ from ..custom_logging import log
2
2
  from ..utils import ConfigManager
3
3
 
4
4
  # Load the YAML file
5
- def load_prompt_from_yaml(key, prefix="sunholo", load_from_file=False):
5
+ def load_prompt_from_yaml(key, prefix="sunholo", load_from_file=False, f_string=True):
6
6
  """
7
- Returns a string you can use with Langfuse PromptTemplate.from_template()
7
+ Returns a string you can use with prompts.
8
8
 
9
9
  Will first try to load from the Langfuse prompt library, if unavailable will look in promptConfig type file.
10
10
 
11
- Langfuse prompts have {{ two braces }}, Langchain prompts have { one brace }.
11
+ If f_string is True will be in a Langchain style prompt e.g. { one brace }
12
+ If f_string is False will be Langfuse style prompt e.g. {{ two braces }}
12
13
 
13
14
  Example:
14
15
 
@@ -36,7 +37,10 @@ def load_prompt_from_yaml(key, prefix="sunholo", load_from_file=False):
36
37
 
37
38
  langfuse_prompt = langfuse.get_prompt(langfuse_template, cache_ttl_seconds=300)
38
39
 
39
- return langfuse_prompt.get_langchain_prompt()
40
+ if f_string:
41
+ return langfuse_prompt.get_langchain_prompt()
42
+
43
+ return langfuse_prompt
40
44
 
41
45
  except Exception as err:
42
46
  log.warning(f"Could not find langfuse template: {langfuse_template} - {str(err)} - attempting to load from promptConfig")
@@ -106,7 +106,6 @@ class ContentBuffer:
106
106
  """
107
107
  async with self.lock:
108
108
  self.content = ""
109
- log.debug("Content buffer cleared")
110
109
  self.content_available.clear()
111
110
 
112
111
 
@@ -0,0 +1,63 @@
1
+ import os
2
+
3
+ def guess_mime_type(file_path: str) -> str:
4
+ """
5
+ Guess the mime type based on the file extension.
6
+
7
+ Args:
8
+ file_path (str): The path or URL of the image file.
9
+
10
+ Returns:
11
+ str: The guessed image type (e.g., "jpeg", "png", "gif", etc.)
12
+ or None if the extension is not recognized.
13
+ """
14
+ # Extract the file extension
15
+ _, ext = os.path.splitext(file_path)
16
+
17
+ # Normalize and remove the leading dot
18
+ ext = ext.lower().strip('.')
19
+
20
+ # Mapping of common file extensions to file types
21
+ extension_to_type = {
22
+ "jpg": "image/jpeg",
23
+ "jpeg": "image/jpeg",
24
+ "png": "image/png",
25
+ "gif": "image/gif",
26
+ "bmp": "image/bmp",
27
+ "tiff": "image/tiff",
28
+ "tif": "image/tiff",
29
+ "webp": "image/webp",
30
+ "ico": "image/vnd.microsoft.icon",
31
+ "svg": "image/svg+xml",
32
+ "pdf": "application/pdf",
33
+ "txt": "text/plain",
34
+ "md": "text/markdown",
35
+ "html": "text/html",
36
+ "css": "text/css",
37
+ "js": "application/javascript",
38
+ "json": "application/json",
39
+ "xml": "application/xml",
40
+ "csv": "text/csv",
41
+ "py": "text/x-python",
42
+ "java": "text/x-java-source",
43
+ "c": "text/x-c",
44
+ "cpp": "text/x-c++",
45
+ "h": "text/x-c",
46
+ "hpp": "text/x-c++",
47
+ "sh": "application/x-sh",
48
+ "bat": "application/x-msdos-program",
49
+ "php": "application/x-httpd-php",
50
+ "rb": "application/x-ruby",
51
+ "pl": "application/x-perl",
52
+ "swift": "application/x-swift",
53
+ "r": "text/x-r",
54
+ "go": "text/x-go",
55
+ "sql": "application/sql",
56
+ "yaml": "text/yaml",
57
+ "yml": "text/yaml",
58
+ "ts": "application/typescript",
59
+ "tsx": "text/tsx",
60
+ "jsx": "text/jsx",
61
+ }
62
+
63
+ return extension_to_type.get(ext, "")
@@ -1,9 +1,9 @@
1
1
  Metadata-Version: 2.1
2
2
  Name: sunholo
3
- Version: 0.100.2
3
+ Version: 0.101.0
4
4
  Summary: Large Language Model DevOps - a package to help deploy LLMs to the Cloud.
5
5
  Home-page: https://github.com/sunholo-data/sunholo-py
6
- Download-URL: https://github.com/sunholo-data/sunholo-py/archive/refs/tags/v0.100.2.tar.gz
6
+ Download-URL: https://github.com/sunholo-data/sunholo-py/archive/refs/tags/v0.101.0.tar.gz
7
7
  Author: Holosun ApS
8
8
  Author-email: multivac@sunholo.com
9
9
  License: Apache License, Version 2.0
@@ -94,8 +94,10 @@ sunholo/gcs/__init__.py
94
94
  sunholo/gcs/add_file.py
95
95
  sunholo/gcs/download_folder.py
96
96
  sunholo/gcs/download_url.py
97
+ sunholo/gcs/extract_and_sign.py
97
98
  sunholo/gcs/metadata.py
98
99
  sunholo/genai/__init__.py
100
+ sunholo/genai/images.py
99
101
  sunholo/genai/init.py
100
102
  sunholo/genai/process_funcs_cls.py
101
103
  sunholo/genai/safety.py
@@ -143,6 +145,7 @@ sunholo/utils/config_class.py
143
145
  sunholo/utils/config_schema.py
144
146
  sunholo/utils/gcp.py
145
147
  sunholo/utils/gcp_project.py
148
+ sunholo/utils/mime.py
146
149
  sunholo/utils/parsers.py
147
150
  sunholo/utils/timedelta.py
148
151
  sunholo/utils/user_ids.py
@@ -155,5 +158,6 @@ sunholo/vertex/init.py
155
158
  sunholo/vertex/memory_tools.py
156
159
  sunholo/vertex/safety.py
157
160
  sunholo/vertex/type_dict_to_json.py
161
+ tests/test_async.py
158
162
  tests/test_chat_history.py
159
163
  tests/test_config.py
@@ -0,0 +1,93 @@
1
+ import asyncio
2
+ from sunholo.invoke import AsyncTaskRunner
3
+ # Mock logger for demonstration
4
+ import logging
5
+ logging.basicConfig(level=logging.INFO)
6
+ log = logging.getLogger(__name__)
7
+
8
+
9
+ # Dummy task functions for testing
10
+ async def google_search(query, config_manager):
11
+ await asyncio.sleep(5) # Simulate a long-running task
12
+ return f"Results for '{query}'"
13
+
14
+ async def use_pdfs(query, arg2, arg3):
15
+ await asyncio.sleep(2) # Simulate a medium-running task
16
+ return "No image_uri specified\n"
17
+
18
+ async def process_urls(url):
19
+ await asyncio.sleep(1) # Simulate a short-running task
20
+ return "No URLs were found"
21
+
22
+ # Mock callback for demonstration
23
+ class MockCallback:
24
+ async def async_on_llm_new_token(self, token):
25
+ log.info(f"Callback received token: {token}")
26
+
27
+ # Mock functions for demonstration
28
+ def format_token_output(func_name, tokens, result=None, error=None):
29
+ if error:
30
+ return f"Error in {func_name}: {error}"
31
+ return f"{func_name} completed with {tokens} tokens. Result: {result}"
32
+
33
+ async def count_tokens(result):
34
+ return len(result.split())
35
+
36
+ # Example usage
37
+ async def main():
38
+ runner = AsyncTaskRunner(retry_enabled=True)
39
+
40
+ # Add your tasks
41
+ runner.add_task(google_search, "<original user question>...", "<config>")
42
+ runner.add_task(use_pdfs, "<original user question>...", None, None)
43
+ runner.add_task(process_urls, "please give me a forecast of ppas for spain using wind energy")
44
+
45
+ callback = MockCallback()
46
+ answers = {}
47
+ total_context_tokens = 0
48
+
49
+ async for message in runner.run_async_as_completed():
50
+ log.info(f"Runner message={message}")
51
+ if message['type'] == 'heartbeat':
52
+ func_name = message['name']
53
+ elapsed_time = message['elapsed_time']
54
+ log.info(f"Runner Heartbeat for {func_name}, elapsed_time={elapsed_time}")
55
+ # Send heartbeat to callback
56
+ update_html = (
57
+ f'<div style="display: none;" data-update-id="{func_name}-spinner">'
58
+ f'<span class="elapsed-time">{elapsed_time} seconds elapsed</span>'
59
+ f'</div>'
60
+ )
61
+ await callback.async_on_llm_new_token(token=f"[[HEARTBEAT]]{update_html}[[/HEARTBEAT]]")
62
+ elif message['type'] == 'task_complete':
63
+ func_name = message['func_name']
64
+ result = message['result']
65
+ log.info(f"Runner completed task: {func_name}")
66
+ # Process result
67
+ if isinstance(result, Exception):
68
+ log.info(f"Error Exception for {func_name}: {str(result)}")
69
+ formatted_output = format_token_output(func_name, 0, error="No results")
70
+ await callback.async_on_llm_new_token(token=formatted_output)
71
+ else:
72
+ # Stream the result to the callback
73
+ tokens = await count_tokens(result)
74
+ total_context_tokens += tokens
75
+ log.info(f"Got task {func_name} to stream result length [{len(result)}] [{tokens} tokens]")
76
+ formatted_output = format_token_output(func_name, tokens, result=result)
77
+ await callback.async_on_llm_new_token(token=formatted_output)
78
+ answers[func_name] = result
79
+ elif message['type'] == 'task_error':
80
+ func_name = message['func_name']
81
+ error = message['error']
82
+ log.info(f"Error Exception for {func_name}: {str(error)}")
83
+ formatted_output = format_token_output(func_name, 0, error="No results")
84
+ await callback.async_on_llm_new_token(token=formatted_output)
85
+
86
+ await callback.async_on_llm_new_token(
87
+ token=f'<div style="margin-top: 20px; color: #333;"><strong>-- Using [{total_context_tokens}] tokens in context for final answer --</strong></div>'
88
+ )
89
+ log.info("All tasks have been processed.")
90
+
91
+ # Run the main coroutine
92
+ if __name__ == "__main__":
93
+ asyncio.run(main())