Dify Extractor

Dify Extractor converts uploaded documents into Markdown plus structured document records for RAG pipelines and workflows.

Supported formats

• PDF (.pdf)
• Microsoft Word (.docx)
• Microsoft PowerPoint (.pptx)
• Microsoft Excel (.xls, .xlsx)
• Markdown (.md, .markdown, .mdx)
• HTML (.htm, .html)
• CSV (.csv)
• JSON (.json)
• YAML (.yaml, .yml)
• Plain-text formats such as .txt, .log, .rst, .ini, .cfg, .conf, and .xml

Text MIME types are also accepted when a file has no recognized extension. Unsupported binary formats return a clear error instead of being decoded as text.

Outputs

• The text output contains the complete extracted Markdown.
• documents contains format-appropriate records: rows for CSV/Excel, pages for PDF, sections for Markdown, and a whole-file record for the other formats.
• images is emitted when embedded or referenced images are successfully uploaded to Dify.

Malformed, empty, undecodable, or resource-limit-breaking files return one error message and do not emit partial document variables.

Image and archive safeguards

Remote Markdown and DOCX images are imported on a best-effort basis. Only HTTP(S) URLs are used, with a 30-second request timeout, a maximum of 20 images, a 15 MiB per-image limit, and a 100 MiB cumulative limit. A failed image download or upload does not discard otherwise extractable text.

ZIP-based Office files are rejected before parsing if they contain more than 10,000 entries, expand beyond 200 MiB, or exceed a 1,000:1 compression ratio.

Development

The plugin targets Python 3.12 and uses uv for dependency management.

uv sync --all-groups --frozen --python 3.12
uv run pytest
uv run ruff check .
uv run ruff format --check .