feat: Added "document parser" OCR strategy

[dlew]

Pull Request Template

Summary

The document parser uses libraries to parse the text out of known document types. This lets LibreChat handle some complex document types without having to use a secondary service (like Mistral or standing up a RAG API server).

To enable the document parser, set the ocr strategy to "document_parser" in librechat.yaml.

We now support:

• PDFs using pdfjs
• DOCX using mammoth
• XLS/XLSX using SheetJS

(The associated packages were also added to the project.)

Here's a documentation update PR as well.

Change Type

Please delete any irrelevant options.

• New feature (non-breaking change which adds functionality)
• This change requires a documentation update (here)

Testing

I have added automated tests for most cases (the exception being PDFs, as getting Jest to work with ECMAScript modules would be a big lift just for this one small PR).

I also manually tested uploading PDFs, Word documents, and Excel sheets to LibreChat as text, to make sure they are parsed out.

Test Configuration:

• Enable ocr agent capability in librechat.yaml:

  agents:
   capabilities:
     - "ocr"
  

• Set ocr strategy to document_parser:

  ocr:
    strategy: "document_parser"
  

Checklist

Please delete any irrelevant options.

• My code adheres to this project's style guidelines
• I have performed a self-review of my own code
• I have commented in any complex areas of my code
• I have made pertinent documentation changes
• My changes do not introduce new warnings
• I have written tests demonstrating that my changes are effective or that my feature works
• Local unit tests pass with my changes
• A pull request for updating the documentation has been submitted (here).

[Copilot]

Pull request overview

Adds a new "document_parser" OCR strategy that extracts text directly from common document formats (PDF/DOCX/XLS/XLSX) using local parsing libraries, avoiding external OCR/RAG services for these cases.

Changes:

• Introduces a new document_parser OCR strategy and wires it into the file strategy selector.
• Implements document parsing for PDF (pdfjs), DOCX (mammoth), and XLS/XLSX (SheetJS/xlsx).
• Adds dependencies plus Jest coverage for DOCX/XLSX parsing (with fixtures).

Reviewed changes

Copilot reviewed 6 out of 10 changed files in this pull request and generated 6 comments.

Show a summary per file

File	Description
packages/data-provider/src/types/files.ts	Adds document_parser to FileSources so it can be referenced across packages.
packages/data-provider/src/config.ts	Adds DOCUMENT_PARSER to OCRStrategy and allows it via the config schema.
api/server/services/Files/strategies.js	Registers the new document_parser strategy and routes it to parseDocument.
api/server/services/Files/Documents/crud.js	Implements document parsing logic for PDF/DOCX/XLS/XLSX and returns extracted text payload.
api/server/services/Files/Documents/tests/documents.spec.js	Adds unit tests for DOCX/XLSX parsing and error cases.
api/server/services/Files/Documents/tests/sample.xlsx	XLSX fixture for tests.
api/server/services/Files/Documents/tests/sample.docx	DOCX fixture for tests.
api/server/services/Files/Documents/tests/empty.docx	Empty DOCX fixture for tests.
api/package.json	Adds new runtime deps: mammoth, pdfjs-dist, xlsx.
package-lock.json	Locks newly added dependencies and transitive tree.

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

[Copilot]

The PDF branch (case 'application/pdf') introduces new behavior but isn't covered by tests (only docx/xlsx are covered). Even a minimal test (e.g., a tiny fixture PDF or mocking pdfjs) would help prevent regressions in the PDF path.

[dlew]

I tried hard to make this happen, however since pdfjs uses ESM modules, we'd have to do a larger refactor to allow for ESM modules in LibreChat's Jest tests. Such a refactor didn't seem worth it for now.

A minimal test would only check that we call pdfToText() with the right mimetype, which didn't seem worth the effort of writing (it would only give minimal benefit & provide the illusion that this path is being tested).

[dlew]

I've addressed & responded to all the Copilot feedback.

[danny-avila]

I wanted a few more changes before merging, continued here: #11900

I don't have access to push changes to your fork so opened a new PR with your original commits.

[dlew]

Sounds good - do you also want to merge the docs update? LibreChat-AI/librechat.ai#492