nv-ingest-api 2025.4.24.dev20250424__tar.gz → 2025.4.25.dev20250425__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.

Potentially problematic release.

This version of nv-ingest-api might be problematic. Click here for more details.

Files changed (159) hide show
  1. {nv_ingest_api-2025.4.24.dev20250424/src/nv_ingest_api.egg-info → nv_ingest_api-2025.4.25.dev20250425}/PKG-INFO +1 -1
  2. {nv_ingest_api-2025.4.24.dev20250424 → nv_ingest_api-2025.4.25.dev20250425}/src/nv_ingest_api/internal/transform/embed_text.py +4 -1
  3. {nv_ingest_api-2025.4.24.dev20250424 → nv_ingest_api-2025.4.25.dev20250425}/src/nv_ingest_api/util/converters/type_mappings.py +1 -1
  4. {nv_ingest_api-2025.4.24.dev20250424 → nv_ingest_api-2025.4.25.dev20250425/src/nv_ingest_api.egg-info}/PKG-INFO +1 -1
  5. {nv_ingest_api-2025.4.24.dev20250424 → nv_ingest_api-2025.4.25.dev20250425}/LICENSE +0 -0
  6. {nv_ingest_api-2025.4.24.dev20250424 → nv_ingest_api-2025.4.25.dev20250425}/MANIFEST.in +0 -0
  7. {nv_ingest_api-2025.4.24.dev20250424 → nv_ingest_api-2025.4.25.dev20250425}/README.md +0 -0
  8. {nv_ingest_api-2025.4.24.dev20250424 → nv_ingest_api-2025.4.25.dev20250425}/pyproject.toml +0 -0
  9. {nv_ingest_api-2025.4.24.dev20250424 → nv_ingest_api-2025.4.25.dev20250425}/setup.cfg +0 -0
  10. {nv_ingest_api-2025.4.24.dev20250424 → nv_ingest_api-2025.4.25.dev20250425}/src/nv_ingest_api/__init__.py +0 -0
  11. {nv_ingest_api-2025.4.24.dev20250424 → nv_ingest_api-2025.4.25.dev20250425}/src/nv_ingest_api/interface/__init__.py +0 -0
  12. {nv_ingest_api-2025.4.24.dev20250424 → nv_ingest_api-2025.4.25.dev20250425}/src/nv_ingest_api/interface/extract.py +0 -0
  13. {nv_ingest_api-2025.4.24.dev20250424 → nv_ingest_api-2025.4.25.dev20250425}/src/nv_ingest_api/interface/mutate.py +0 -0
  14. {nv_ingest_api-2025.4.24.dev20250424 → nv_ingest_api-2025.4.25.dev20250425}/src/nv_ingest_api/interface/store.py +0 -0
  15. {nv_ingest_api-2025.4.24.dev20250424 → nv_ingest_api-2025.4.25.dev20250425}/src/nv_ingest_api/interface/transform.py +0 -0
  16. {nv_ingest_api-2025.4.24.dev20250424 → nv_ingest_api-2025.4.25.dev20250425}/src/nv_ingest_api/interface/utility.py +0 -0
  17. {nv_ingest_api-2025.4.24.dev20250424 → nv_ingest_api-2025.4.25.dev20250425}/src/nv_ingest_api/internal/__init__.py +0 -0
  18. {nv_ingest_api-2025.4.24.dev20250424 → nv_ingest_api-2025.4.25.dev20250425}/src/nv_ingest_api/internal/enums/__init__.py +0 -0
  19. {nv_ingest_api-2025.4.24.dev20250424 → nv_ingest_api-2025.4.25.dev20250425}/src/nv_ingest_api/internal/enums/common.py +0 -0
  20. {nv_ingest_api-2025.4.24.dev20250424 → nv_ingest_api-2025.4.25.dev20250425}/src/nv_ingest_api/internal/extract/__init__.py +0 -0
  21. {nv_ingest_api-2025.4.24.dev20250424 → nv_ingest_api-2025.4.25.dev20250425}/src/nv_ingest_api/internal/extract/audio/__init__.py +0 -0
  22. {nv_ingest_api-2025.4.24.dev20250424 → nv_ingest_api-2025.4.25.dev20250425}/src/nv_ingest_api/internal/extract/audio/audio_extraction.py +0 -0
  23. {nv_ingest_api-2025.4.24.dev20250424 → nv_ingest_api-2025.4.25.dev20250425}/src/nv_ingest_api/internal/extract/docx/__init__.py +0 -0
  24. {nv_ingest_api-2025.4.24.dev20250424 → nv_ingest_api-2025.4.25.dev20250425}/src/nv_ingest_api/internal/extract/docx/docx_extractor.py +0 -0
  25. {nv_ingest_api-2025.4.24.dev20250424 → nv_ingest_api-2025.4.25.dev20250425}/src/nv_ingest_api/internal/extract/docx/engines/__init__.py +0 -0
  26. {nv_ingest_api-2025.4.24.dev20250424 → nv_ingest_api-2025.4.25.dev20250425}/src/nv_ingest_api/internal/extract/docx/engines/docxreader_helpers/__init__.py +0 -0
  27. {nv_ingest_api-2025.4.24.dev20250424 → nv_ingest_api-2025.4.25.dev20250425}/src/nv_ingest_api/internal/extract/docx/engines/docxreader_helpers/docx_helper.py +0 -0
  28. {nv_ingest_api-2025.4.24.dev20250424 → nv_ingest_api-2025.4.25.dev20250425}/src/nv_ingest_api/internal/extract/docx/engines/docxreader_helpers/docxreader.py +0 -0
  29. {nv_ingest_api-2025.4.24.dev20250424 → nv_ingest_api-2025.4.25.dev20250425}/src/nv_ingest_api/internal/extract/image/__init__.py +0 -0
  30. {nv_ingest_api-2025.4.24.dev20250424 → nv_ingest_api-2025.4.25.dev20250425}/src/nv_ingest_api/internal/extract/image/chart_extractor.py +0 -0
  31. {nv_ingest_api-2025.4.24.dev20250424 → nv_ingest_api-2025.4.25.dev20250425}/src/nv_ingest_api/internal/extract/image/image_extractor.py +0 -0
  32. {nv_ingest_api-2025.4.24.dev20250424 → nv_ingest_api-2025.4.25.dev20250425}/src/nv_ingest_api/internal/extract/image/image_helpers/__init__.py +0 -0
  33. {nv_ingest_api-2025.4.24.dev20250424 → nv_ingest_api-2025.4.25.dev20250425}/src/nv_ingest_api/internal/extract/image/image_helpers/common.py +0 -0
  34. {nv_ingest_api-2025.4.24.dev20250424 → nv_ingest_api-2025.4.25.dev20250425}/src/nv_ingest_api/internal/extract/image/infographic_extractor.py +0 -0
  35. {nv_ingest_api-2025.4.24.dev20250424 → nv_ingest_api-2025.4.25.dev20250425}/src/nv_ingest_api/internal/extract/image/table_extractor.py +0 -0
  36. {nv_ingest_api-2025.4.24.dev20250424 → nv_ingest_api-2025.4.25.dev20250425}/src/nv_ingest_api/internal/extract/pdf/__init__.py +0 -0
  37. {nv_ingest_api-2025.4.24.dev20250424 → nv_ingest_api-2025.4.25.dev20250425}/src/nv_ingest_api/internal/extract/pdf/engines/__init__.py +0 -0
  38. {nv_ingest_api-2025.4.24.dev20250424 → nv_ingest_api-2025.4.25.dev20250425}/src/nv_ingest_api/internal/extract/pdf/engines/adobe.py +0 -0
  39. {nv_ingest_api-2025.4.24.dev20250424 → nv_ingest_api-2025.4.25.dev20250425}/src/nv_ingest_api/internal/extract/pdf/engines/llama.py +0 -0
  40. {nv_ingest_api-2025.4.24.dev20250424 → nv_ingest_api-2025.4.25.dev20250425}/src/nv_ingest_api/internal/extract/pdf/engines/nemoretriever.py +0 -0
  41. {nv_ingest_api-2025.4.24.dev20250424 → nv_ingest_api-2025.4.25.dev20250425}/src/nv_ingest_api/internal/extract/pdf/engines/pdf_helpers/__init__.py +0 -0
  42. {nv_ingest_api-2025.4.24.dev20250424 → nv_ingest_api-2025.4.25.dev20250425}/src/nv_ingest_api/internal/extract/pdf/engines/pdfium.py +0 -0
  43. {nv_ingest_api-2025.4.24.dev20250424 → nv_ingest_api-2025.4.25.dev20250425}/src/nv_ingest_api/internal/extract/pdf/engines/tika.py +0 -0
  44. {nv_ingest_api-2025.4.24.dev20250424 → nv_ingest_api-2025.4.25.dev20250425}/src/nv_ingest_api/internal/extract/pdf/engines/unstructured_io.py +0 -0
  45. {nv_ingest_api-2025.4.24.dev20250424 → nv_ingest_api-2025.4.25.dev20250425}/src/nv_ingest_api/internal/extract/pdf/pdf_extractor.py +0 -0
  46. {nv_ingest_api-2025.4.24.dev20250424 → nv_ingest_api-2025.4.25.dev20250425}/src/nv_ingest_api/internal/extract/pptx/__init__.py +0 -0
  47. {nv_ingest_api-2025.4.24.dev20250424 → nv_ingest_api-2025.4.25.dev20250425}/src/nv_ingest_api/internal/extract/pptx/engines/__init__.py +0 -0
  48. {nv_ingest_api-2025.4.24.dev20250424 → nv_ingest_api-2025.4.25.dev20250425}/src/nv_ingest_api/internal/extract/pptx/engines/pptx_helper.py +0 -0
  49. {nv_ingest_api-2025.4.24.dev20250424 → nv_ingest_api-2025.4.25.dev20250425}/src/nv_ingest_api/internal/extract/pptx/pptx_extractor.py +0 -0
  50. {nv_ingest_api-2025.4.24.dev20250424 → nv_ingest_api-2025.4.25.dev20250425}/src/nv_ingest_api/internal/mutate/__init__.py +0 -0
  51. {nv_ingest_api-2025.4.24.dev20250424 → nv_ingest_api-2025.4.25.dev20250425}/src/nv_ingest_api/internal/mutate/deduplicate.py +0 -0
  52. {nv_ingest_api-2025.4.24.dev20250424 → nv_ingest_api-2025.4.25.dev20250425}/src/nv_ingest_api/internal/mutate/filter.py +0 -0
  53. {nv_ingest_api-2025.4.24.dev20250424 → nv_ingest_api-2025.4.25.dev20250425}/src/nv_ingest_api/internal/primitives/__init__.py +0 -0
  54. {nv_ingest_api-2025.4.24.dev20250424 → nv_ingest_api-2025.4.25.dev20250425}/src/nv_ingest_api/internal/primitives/control_message_task.py +0 -0
  55. {nv_ingest_api-2025.4.24.dev20250424 → nv_ingest_api-2025.4.25.dev20250425}/src/nv_ingest_api/internal/primitives/ingest_control_message.py +0 -0
  56. {nv_ingest_api-2025.4.24.dev20250424 → nv_ingest_api-2025.4.25.dev20250425}/src/nv_ingest_api/internal/primitives/nim/__init__.py +0 -0
  57. {nv_ingest_api-2025.4.24.dev20250424 → nv_ingest_api-2025.4.25.dev20250425}/src/nv_ingest_api/internal/primitives/nim/default_values.py +0 -0
  58. {nv_ingest_api-2025.4.24.dev20250424 → nv_ingest_api-2025.4.25.dev20250425}/src/nv_ingest_api/internal/primitives/nim/model_interface/__init__.py +0 -0
  59. {nv_ingest_api-2025.4.24.dev20250424 → nv_ingest_api-2025.4.25.dev20250425}/src/nv_ingest_api/internal/primitives/nim/model_interface/cached.py +0 -0
  60. {nv_ingest_api-2025.4.24.dev20250424 → nv_ingest_api-2025.4.25.dev20250425}/src/nv_ingest_api/internal/primitives/nim/model_interface/decorators.py +0 -0
  61. {nv_ingest_api-2025.4.24.dev20250424 → nv_ingest_api-2025.4.25.dev20250425}/src/nv_ingest_api/internal/primitives/nim/model_interface/deplot.py +0 -0
  62. {nv_ingest_api-2025.4.24.dev20250424 → nv_ingest_api-2025.4.25.dev20250425}/src/nv_ingest_api/internal/primitives/nim/model_interface/helpers.py +0 -0
  63. {nv_ingest_api-2025.4.24.dev20250424 → nv_ingest_api-2025.4.25.dev20250425}/src/nv_ingest_api/internal/primitives/nim/model_interface/nemoretriever_parse.py +0 -0
  64. {nv_ingest_api-2025.4.24.dev20250424 → nv_ingest_api-2025.4.25.dev20250425}/src/nv_ingest_api/internal/primitives/nim/model_interface/paddle.py +0 -0
  65. {nv_ingest_api-2025.4.24.dev20250424 → nv_ingest_api-2025.4.25.dev20250425}/src/nv_ingest_api/internal/primitives/nim/model_interface/parakeet.py +0 -0
  66. {nv_ingest_api-2025.4.24.dev20250424 → nv_ingest_api-2025.4.25.dev20250425}/src/nv_ingest_api/internal/primitives/nim/model_interface/text_embedding.py +0 -0
  67. {nv_ingest_api-2025.4.24.dev20250424 → nv_ingest_api-2025.4.25.dev20250425}/src/nv_ingest_api/internal/primitives/nim/model_interface/vlm.py +0 -0
  68. {nv_ingest_api-2025.4.24.dev20250424 → nv_ingest_api-2025.4.25.dev20250425}/src/nv_ingest_api/internal/primitives/nim/model_interface/yolox.py +0 -0
  69. {nv_ingest_api-2025.4.24.dev20250424 → nv_ingest_api-2025.4.25.dev20250425}/src/nv_ingest_api/internal/primitives/nim/nim_client.py +0 -0
  70. {nv_ingest_api-2025.4.24.dev20250424 → nv_ingest_api-2025.4.25.dev20250425}/src/nv_ingest_api/internal/primitives/nim/nim_model_interface.py +0 -0
  71. {nv_ingest_api-2025.4.24.dev20250424 → nv_ingest_api-2025.4.25.dev20250425}/src/nv_ingest_api/internal/primitives/tracing/__init__.py +0 -0
  72. {nv_ingest_api-2025.4.24.dev20250424 → nv_ingest_api-2025.4.25.dev20250425}/src/nv_ingest_api/internal/primitives/tracing/latency.py +0 -0
  73. {nv_ingest_api-2025.4.24.dev20250424 → nv_ingest_api-2025.4.25.dev20250425}/src/nv_ingest_api/internal/primitives/tracing/logging.py +0 -0
  74. {nv_ingest_api-2025.4.24.dev20250424 → nv_ingest_api-2025.4.25.dev20250425}/src/nv_ingest_api/internal/primitives/tracing/tagging.py +0 -0
  75. {nv_ingest_api-2025.4.24.dev20250424 → nv_ingest_api-2025.4.25.dev20250425}/src/nv_ingest_api/internal/schemas/__init__.py +0 -0
  76. {nv_ingest_api-2025.4.24.dev20250424 → nv_ingest_api-2025.4.25.dev20250425}/src/nv_ingest_api/internal/schemas/extract/__init__.py +0 -0
  77. {nv_ingest_api-2025.4.24.dev20250424 → nv_ingest_api-2025.4.25.dev20250425}/src/nv_ingest_api/internal/schemas/extract/extract_audio_schema.py +0 -0
  78. {nv_ingest_api-2025.4.24.dev20250424 → nv_ingest_api-2025.4.25.dev20250425}/src/nv_ingest_api/internal/schemas/extract/extract_chart_schema.py +0 -0
  79. {nv_ingest_api-2025.4.24.dev20250424 → nv_ingest_api-2025.4.25.dev20250425}/src/nv_ingest_api/internal/schemas/extract/extract_docx_schema.py +0 -0
  80. {nv_ingest_api-2025.4.24.dev20250424 → nv_ingest_api-2025.4.25.dev20250425}/src/nv_ingest_api/internal/schemas/extract/extract_image_schema.py +0 -0
  81. {nv_ingest_api-2025.4.24.dev20250424 → nv_ingest_api-2025.4.25.dev20250425}/src/nv_ingest_api/internal/schemas/extract/extract_infographic_schema.py +0 -0
  82. {nv_ingest_api-2025.4.24.dev20250424 → nv_ingest_api-2025.4.25.dev20250425}/src/nv_ingest_api/internal/schemas/extract/extract_pdf_schema.py +0 -0
  83. {nv_ingest_api-2025.4.24.dev20250424 → nv_ingest_api-2025.4.25.dev20250425}/src/nv_ingest_api/internal/schemas/extract/extract_pptx_schema.py +0 -0
  84. {nv_ingest_api-2025.4.24.dev20250424 → nv_ingest_api-2025.4.25.dev20250425}/src/nv_ingest_api/internal/schemas/extract/extract_table_schema.py +0 -0
  85. {nv_ingest_api-2025.4.24.dev20250424 → nv_ingest_api-2025.4.25.dev20250425}/src/nv_ingest_api/internal/schemas/message_brokers/__init__.py +0 -0
  86. {nv_ingest_api-2025.4.24.dev20250424 → nv_ingest_api-2025.4.25.dev20250425}/src/nv_ingest_api/internal/schemas/message_brokers/message_broker_client_schema.py +0 -0
  87. {nv_ingest_api-2025.4.24.dev20250424 → nv_ingest_api-2025.4.25.dev20250425}/src/nv_ingest_api/internal/schemas/message_brokers/request_schema.py +0 -0
  88. {nv_ingest_api-2025.4.24.dev20250424 → nv_ingest_api-2025.4.25.dev20250425}/src/nv_ingest_api/internal/schemas/message_brokers/response_schema.py +0 -0
  89. {nv_ingest_api-2025.4.24.dev20250424 → nv_ingest_api-2025.4.25.dev20250425}/src/nv_ingest_api/internal/schemas/meta/__init__.py +0 -0
  90. {nv_ingest_api-2025.4.24.dev20250424 → nv_ingest_api-2025.4.25.dev20250425}/src/nv_ingest_api/internal/schemas/meta/base_model_noext.py +0 -0
  91. {nv_ingest_api-2025.4.24.dev20250424 → nv_ingest_api-2025.4.25.dev20250425}/src/nv_ingest_api/internal/schemas/meta/ingest_job_schema.py +0 -0
  92. {nv_ingest_api-2025.4.24.dev20250424 → nv_ingest_api-2025.4.25.dev20250425}/src/nv_ingest_api/internal/schemas/meta/metadata_schema.py +0 -0
  93. {nv_ingest_api-2025.4.24.dev20250424 → nv_ingest_api-2025.4.25.dev20250425}/src/nv_ingest_api/internal/schemas/mutate/__init__.py +0 -0
  94. {nv_ingest_api-2025.4.24.dev20250424 → nv_ingest_api-2025.4.25.dev20250425}/src/nv_ingest_api/internal/schemas/mutate/mutate_image_dedup_schema.py +0 -0
  95. {nv_ingest_api-2025.4.24.dev20250424 → nv_ingest_api-2025.4.25.dev20250425}/src/nv_ingest_api/internal/schemas/store/__init__.py +0 -0
  96. {nv_ingest_api-2025.4.24.dev20250424 → nv_ingest_api-2025.4.25.dev20250425}/src/nv_ingest_api/internal/schemas/store/store_embedding_schema.py +0 -0
  97. {nv_ingest_api-2025.4.24.dev20250424 → nv_ingest_api-2025.4.25.dev20250425}/src/nv_ingest_api/internal/schemas/store/store_image_schema.py +0 -0
  98. {nv_ingest_api-2025.4.24.dev20250424 → nv_ingest_api-2025.4.25.dev20250425}/src/nv_ingest_api/internal/schemas/transform/__init__.py +0 -0
  99. {nv_ingest_api-2025.4.24.dev20250424 → nv_ingest_api-2025.4.25.dev20250425}/src/nv_ingest_api/internal/schemas/transform/transform_image_caption_schema.py +0 -0
  100. {nv_ingest_api-2025.4.24.dev20250424 → nv_ingest_api-2025.4.25.dev20250425}/src/nv_ingest_api/internal/schemas/transform/transform_image_filter_schema.py +0 -0
  101. {nv_ingest_api-2025.4.24.dev20250424 → nv_ingest_api-2025.4.25.dev20250425}/src/nv_ingest_api/internal/schemas/transform/transform_text_embedding_schema.py +0 -0
  102. {nv_ingest_api-2025.4.24.dev20250424 → nv_ingest_api-2025.4.25.dev20250425}/src/nv_ingest_api/internal/schemas/transform/transform_text_splitter_schema.py +0 -0
  103. {nv_ingest_api-2025.4.24.dev20250424 → nv_ingest_api-2025.4.25.dev20250425}/src/nv_ingest_api/internal/store/__init__.py +0 -0
  104. {nv_ingest_api-2025.4.24.dev20250424 → nv_ingest_api-2025.4.25.dev20250425}/src/nv_ingest_api/internal/store/embed_text_upload.py +0 -0
  105. {nv_ingest_api-2025.4.24.dev20250424 → nv_ingest_api-2025.4.25.dev20250425}/src/nv_ingest_api/internal/store/image_upload.py +0 -0
  106. {nv_ingest_api-2025.4.24.dev20250424 → nv_ingest_api-2025.4.25.dev20250425}/src/nv_ingest_api/internal/transform/__init__.py +0 -0
  107. {nv_ingest_api-2025.4.24.dev20250424 → nv_ingest_api-2025.4.25.dev20250425}/src/nv_ingest_api/internal/transform/caption_image.py +0 -0
  108. {nv_ingest_api-2025.4.24.dev20250424 → nv_ingest_api-2025.4.25.dev20250425}/src/nv_ingest_api/internal/transform/split_text.py +0 -0
  109. {nv_ingest_api-2025.4.24.dev20250424 → nv_ingest_api-2025.4.25.dev20250425}/src/nv_ingest_api/util/__init__.py +0 -0
  110. {nv_ingest_api-2025.4.24.dev20250424 → nv_ingest_api-2025.4.25.dev20250425}/src/nv_ingest_api/util/control_message/__init__.py +0 -0
  111. {nv_ingest_api-2025.4.24.dev20250424 → nv_ingest_api-2025.4.25.dev20250425}/src/nv_ingest_api/util/control_message/validators.py +0 -0
  112. {nv_ingest_api-2025.4.24.dev20250424 → nv_ingest_api-2025.4.25.dev20250425}/src/nv_ingest_api/util/converters/__init__.py +0 -0
  113. {nv_ingest_api-2025.4.24.dev20250424 → nv_ingest_api-2025.4.25.dev20250425}/src/nv_ingest_api/util/converters/bytetools.py +0 -0
  114. {nv_ingest_api-2025.4.24.dev20250424 → nv_ingest_api-2025.4.25.dev20250425}/src/nv_ingest_api/util/converters/containers.py +0 -0
  115. {nv_ingest_api-2025.4.24.dev20250424 → nv_ingest_api-2025.4.25.dev20250425}/src/nv_ingest_api/util/converters/datetools.py +0 -0
  116. {nv_ingest_api-2025.4.24.dev20250424 → nv_ingest_api-2025.4.25.dev20250425}/src/nv_ingest_api/util/converters/dftools.py +0 -0
  117. {nv_ingest_api-2025.4.24.dev20250424 → nv_ingest_api-2025.4.25.dev20250425}/src/nv_ingest_api/util/converters/formats.py +0 -0
  118. {nv_ingest_api-2025.4.24.dev20250424 → nv_ingest_api-2025.4.25.dev20250425}/src/nv_ingest_api/util/detectors/__init__.py +0 -0
  119. {nv_ingest_api-2025.4.24.dev20250424 → nv_ingest_api-2025.4.25.dev20250425}/src/nv_ingest_api/util/detectors/language.py +0 -0
  120. {nv_ingest_api-2025.4.24.dev20250424 → nv_ingest_api-2025.4.25.dev20250425}/src/nv_ingest_api/util/exception_handlers/__init__.py +0 -0
  121. {nv_ingest_api-2025.4.24.dev20250424 → nv_ingest_api-2025.4.25.dev20250425}/src/nv_ingest_api/util/exception_handlers/converters.py +0 -0
  122. {nv_ingest_api-2025.4.24.dev20250424 → nv_ingest_api-2025.4.25.dev20250425}/src/nv_ingest_api/util/exception_handlers/decorators.py +0 -0
  123. {nv_ingest_api-2025.4.24.dev20250424 → nv_ingest_api-2025.4.25.dev20250425}/src/nv_ingest_api/util/exception_handlers/detectors.py +0 -0
  124. {nv_ingest_api-2025.4.24.dev20250424 → nv_ingest_api-2025.4.25.dev20250425}/src/nv_ingest_api/util/exception_handlers/pdf.py +0 -0
  125. {nv_ingest_api-2025.4.24.dev20250424 → nv_ingest_api-2025.4.25.dev20250425}/src/nv_ingest_api/util/exception_handlers/schemas.py +0 -0
  126. {nv_ingest_api-2025.4.24.dev20250424 → nv_ingest_api-2025.4.25.dev20250425}/src/nv_ingest_api/util/image_processing/__init__.py +0 -0
  127. {nv_ingest_api-2025.4.24.dev20250424 → nv_ingest_api-2025.4.25.dev20250425}/src/nv_ingest_api/util/image_processing/clustering.py +0 -0
  128. {nv_ingest_api-2025.4.24.dev20250424 → nv_ingest_api-2025.4.25.dev20250425}/src/nv_ingest_api/util/image_processing/processing.py +0 -0
  129. {nv_ingest_api-2025.4.24.dev20250424 → nv_ingest_api-2025.4.25.dev20250425}/src/nv_ingest_api/util/image_processing/table_and_chart.py +0 -0
  130. {nv_ingest_api-2025.4.24.dev20250424 → nv_ingest_api-2025.4.25.dev20250425}/src/nv_ingest_api/util/image_processing/transforms.py +0 -0
  131. {nv_ingest_api-2025.4.24.dev20250424 → nv_ingest_api-2025.4.25.dev20250425}/src/nv_ingest_api/util/logging/__init__.py +0 -0
  132. {nv_ingest_api-2025.4.24.dev20250424 → nv_ingest_api-2025.4.25.dev20250425}/src/nv_ingest_api/util/logging/configuration.py +0 -0
  133. {nv_ingest_api-2025.4.24.dev20250424 → nv_ingest_api-2025.4.25.dev20250425}/src/nv_ingest_api/util/message_brokers/__init__.py +0 -0
  134. {nv_ingest_api-2025.4.24.dev20250424 → nv_ingest_api-2025.4.25.dev20250425}/src/nv_ingest_api/util/message_brokers/simple_message_broker/__init__.py +0 -0
  135. {nv_ingest_api-2025.4.24.dev20250424 → nv_ingest_api-2025.4.25.dev20250425}/src/nv_ingest_api/util/message_brokers/simple_message_broker/broker.py +0 -0
  136. {nv_ingest_api-2025.4.24.dev20250424 → nv_ingest_api-2025.4.25.dev20250425}/src/nv_ingest_api/util/message_brokers/simple_message_broker/ordered_message_queue.py +0 -0
  137. {nv_ingest_api-2025.4.24.dev20250424 → nv_ingest_api-2025.4.25.dev20250425}/src/nv_ingest_api/util/message_brokers/simple_message_broker/simple_client.py +0 -0
  138. {nv_ingest_api-2025.4.24.dev20250424 → nv_ingest_api-2025.4.25.dev20250425}/src/nv_ingest_api/util/metadata/__init__.py +0 -0
  139. {nv_ingest_api-2025.4.24.dev20250424 → nv_ingest_api-2025.4.25.dev20250425}/src/nv_ingest_api/util/metadata/aggregators.py +0 -0
  140. {nv_ingest_api-2025.4.24.dev20250424 → nv_ingest_api-2025.4.25.dev20250425}/src/nv_ingest_api/util/multi_processing/__init__.py +0 -0
  141. {nv_ingest_api-2025.4.24.dev20250424 → nv_ingest_api-2025.4.25.dev20250425}/src/nv_ingest_api/util/multi_processing/mp_pool_singleton.py +0 -0
  142. {nv_ingest_api-2025.4.24.dev20250424 → nv_ingest_api-2025.4.25.dev20250425}/src/nv_ingest_api/util/nim/__init__.py +0 -0
  143. {nv_ingest_api-2025.4.24.dev20250424 → nv_ingest_api-2025.4.25.dev20250425}/src/nv_ingest_api/util/pdf/__init__.py +0 -0
  144. {nv_ingest_api-2025.4.24.dev20250424 → nv_ingest_api-2025.4.25.dev20250425}/src/nv_ingest_api/util/pdf/pdfium.py +0 -0
  145. {nv_ingest_api-2025.4.24.dev20250424 → nv_ingest_api-2025.4.25.dev20250425}/src/nv_ingest_api/util/schema/__init__.py +0 -0
  146. {nv_ingest_api-2025.4.24.dev20250424 → nv_ingest_api-2025.4.25.dev20250425}/src/nv_ingest_api/util/schema/schema_validator.py +0 -0
  147. {nv_ingest_api-2025.4.24.dev20250424 → nv_ingest_api-2025.4.25.dev20250425}/src/nv_ingest_api/util/service_clients/__init__.py +0 -0
  148. {nv_ingest_api-2025.4.24.dev20250424 → nv_ingest_api-2025.4.25.dev20250425}/src/nv_ingest_api/util/service_clients/client_base.py +0 -0
  149. {nv_ingest_api-2025.4.24.dev20250424 → nv_ingest_api-2025.4.25.dev20250425}/src/nv_ingest_api/util/service_clients/kafka/__init__.py +0 -0
  150. {nv_ingest_api-2025.4.24.dev20250424 → nv_ingest_api-2025.4.25.dev20250425}/src/nv_ingest_api/util/service_clients/redis/__init__.py +0 -0
  151. {nv_ingest_api-2025.4.24.dev20250424 → nv_ingest_api-2025.4.25.dev20250425}/src/nv_ingest_api/util/service_clients/redis/redis_client.py +0 -0
  152. {nv_ingest_api-2025.4.24.dev20250424 → nv_ingest_api-2025.4.25.dev20250425}/src/nv_ingest_api/util/service_clients/rest/__init__.py +0 -0
  153. {nv_ingest_api-2025.4.24.dev20250424 → nv_ingest_api-2025.4.25.dev20250425}/src/nv_ingest_api/util/service_clients/rest/rest_client.py +0 -0
  154. {nv_ingest_api-2025.4.24.dev20250424 → nv_ingest_api-2025.4.25.dev20250425}/src/nv_ingest_api/util/string_processing/__init__.py +0 -0
  155. {nv_ingest_api-2025.4.24.dev20250424 → nv_ingest_api-2025.4.25.dev20250425}/src/nv_ingest_api.egg-info/SOURCES.txt +0 -0
  156. {nv_ingest_api-2025.4.24.dev20250424 → nv_ingest_api-2025.4.25.dev20250425}/src/nv_ingest_api.egg-info/dependency_links.txt +0 -0
  157. {nv_ingest_api-2025.4.24.dev20250424 → nv_ingest_api-2025.4.25.dev20250425}/src/nv_ingest_api.egg-info/requires.txt +0 -0
  158. {nv_ingest_api-2025.4.24.dev20250424 → nv_ingest_api-2025.4.25.dev20250425}/src/nv_ingest_api.egg-info/top_level.txt +0 -0
  159. {nv_ingest_api-2025.4.24.dev20250424 → nv_ingest_api-2025.4.25.dev20250425}/src/version.py +0 -0
{nv_ingest_api-2025.4.24.dev20250424/src/nv_ingest_api.egg-info → nv_ingest_api-2025.4.25.dev20250425}/PKG-INFO RENAMED
@@ -1,6 +1,6 @@
1
1
  Metadata-Version: 2.4
2
2
  Name: nv-ingest-api
3
- Version: 2025.4.24.dev20250424
3
+ Version: 2025.4.25.dev20250425
4
4
  Summary: Python module with core document ingestion functions.
5
5
  Author-email: Jeremy Dyer <jdyer@nvidia.com>
6
6
  License: Apache License
{nv_ingest_api-2025.4.24.dev20250424 → nv_ingest_api-2025.4.25.dev20250425}/src/nv_ingest_api/internal/transform/embed_text.py RENAMED
@@ -453,13 +453,16 @@ def transform_create_text_embeddings_internal(
453
453
 
454
454
  logger.debug("Generating text embeddings for supported content types: TEXT, STRUCTURED, IMAGE.")
455
455
 
456
+ def _content_type_getter(row):
457
+ return row["content_metadata"]["type"]
458
+
456
459
  # Process each supported content type.
457
460
  for content_type, content_getter in pandas_content_extractor.items():
458
461
  if not content_getter:
459
462
  logger.debug(f"Skipping unsupported content type: {content_type}")
460
463
  continue
461
464
 
462
- content_mask = df_transform_ledger["document_type"] == content_type.value
465
+ content_mask = df_transform_ledger["metadata"].apply(_content_type_getter) == content_type.value
463
466
  if not content_mask.any():
464
467
  continue
465
468
 
{nv_ingest_api-2025.4.24.dev20250424 → nv_ingest_api-2025.4.25.dev20250425}/src/nv_ingest_api/util/converters/type_mappings.py RENAMED
@@ -7,7 +7,7 @@ from nv_ingest_api.internal.enums.common import ContentTypeEnum
7
7
  DOC_TO_CONTENT_MAP = {
8
8
  DocumentTypeEnum.BMP: ContentTypeEnum.IMAGE,
9
9
  DocumentTypeEnum.DOCX: ContentTypeEnum.STRUCTURED,
10
- DocumentTypeEnum.HTML: ContentTypeEnum.STRUCTURED,
10
+ DocumentTypeEnum.HTML: ContentTypeEnum.TEXT,
11
11
  DocumentTypeEnum.JPEG: ContentTypeEnum.IMAGE,
12
12
  DocumentTypeEnum.MP3: ContentTypeEnum.AUDIO,
13
13
  DocumentTypeEnum.PDF: ContentTypeEnum.STRUCTURED,
{nv_ingest_api-2025.4.24.dev20250424 → nv_ingest_api-2025.4.25.dev20250425/src/nv_ingest_api.egg-info}/PKG-INFO RENAMED
@@ -1,6 +1,6 @@
1
1
  Metadata-Version: 2.4
2
2
  Name: nv-ingest-api
3
- Version: 2025.4.24.dev20250424
3
+ Version: 2025.4.25.dev20250425
4
4
  Summary: Python module with core document ingestion functions.
5
5
  Author-email: Jeremy Dyer <jdyer@nvidia.com>
6
6
  License: Apache License