Add metadata support to Knowledge Source classes

Browse Source

- Implement _coerce_to_records function to handle both string and dict formats
- Update KnowledgeStorage.save() to accept list[str] | list[dict[str, Any]]
- Add metadata (filepath, chunk_index, source_type) to all Knowledge Source classes:
  - TextFileKnowledgeSource
  - PDFKnowledgeSource
  - CSVKnowledgeSource
  - JSONKnowledgeSource
  - ExcelKnowledgeSource (includes sheet_name for multi-sheet files)
  - StringKnowledgeSource
  - CrewDoclingSource
- Add comprehensive tests for metadata functionality
- Maintain backward compatibility with existing string-based chunks

Fixes #3812

Co-Authored-By: João <joao@crewai.com>

...

This commit is contained in:

Devin AI

2025-10-30 09:04:35 +00:00

parent 2e9eb8c32d

commit 82bb304147

10 changed files with 4626 additions and 4058 deletions

Show all changes Ignore whitespace when comparing lines Ignore changes in amount of whitespace Ignore changes in whitespace at EOL

Download Patch File Download Diff File Expand all files Collapse all files

8079

uv.lock generated

File diff suppressed because it is too large Load Diff