fix(docling)!: store binary_hash as a string to avoid 64-bit overflow

[SyedShahmeerAli12]

Related Issues

• fixes DoclingConverter: binary_hash metadata can exceed signed 64-bit integer range #3604

Proposed Changes

MetaExtractor previously stored Docling's origin.binary_hash as an integer in document metadata. Since binary_hash is an unsigned 64-bit integer, values greater than 2^63 - 1 break document stores that map numeric metadata to signed 64-bit types (e.g. OpenSearch long), causing:

failed to parse field [dl_meta.origin.binary_hash] of type [long]
Numeric value (9768961288489567249) out of range of long

Following @julian-risch's guidance on #3604, binary_hash is now always stored as a string via a new _stringify_binary_hash() recursive helper applied in both extract_dl_doc_meta and extract_chunk_meta.

Reproduction

Before fix — binary_hash returned as int, overflows OpenSearch long:

# model_dump() returns: {binary_hash: 9768961288489567249}  � int, rejected by OpenSearch

After fix — stored as str, no overflow:

# _stringify_binary_hash() returns: {binary_hash: 9768961288489567249}  � safe keyword

Testing

• Added 3 new unit tests: oversized hash in doc meta, chunk meta, and nested structures
• All 8 TestMetaExtractor tests pass
• ruff check clean

Notes

• Breaking metadata-type change for callers that assumed binary_hash was int — aligns with maintainer suggestion for a major release
• Both small and large hashes are stringified for consistency

[github-actions]

Hi @SyedShahmeerAli12, thanks for your interest in contributing to Haystack! 🙏

⚠️ Issue #3604 is already being addressed by open pull request(s) #3607. Before opening a PR for an issue, please check whether a PR is already linked to it, and consider contributing to the existing PR instead. We may close duplicate PRs to keep the review queue manageable.

⚠️ You currently have 4 open pull requests in this repository (#3565, #3483, #3254 and this one). Our review capacity is limited, so please hold off opening more PRs until we've had a chance to review your first 2 open PRs. This helps us give each contribution the attention it deserves. Thank you!

This is an automated message to help us keep the review queue healthy.

[anakin87]

I just simplified it a bit.

Ready to go now. Thanks!