langchain-oxidize-pdf

LangChain document loader backed by oxidize-pdf, a fast Rust-powered PDF engine with first-class RAG chunking.

0.1.0 (2026-04-24) — Requires oxidize-pdf>=0.4.3 (oxidize-pdf-core 2.5.5). First release. The sibling llama-index-readers-oxidize-pdf 0.1.0 shipped with shape-only tests that missed a quadratic accumulation bug in the underlying chunker; this loader ships from day one with the semantic regression suite (test_loader_disjoint.py) that guarantees the disjointness contract end-to-end.

Install

pip install langchain-oxidize-pdf

Usage

LangChain convention binds the file path to the loader instance and uses lazy_load() as the primary entry point; load() is inherited from BaseLoader as a convenience that materializes the iterator.

RAG chunks (default)

from langchain_oxidize_pdf import OxidizePdfLoader

loader = OxidizePdfLoader("paper.pdf")  # mode="rag" by default
documents = loader.load()

for doc in documents:
    print(doc.metadata["chunk_index"], doc.metadata["heading_context"])
    print(doc.page_content[:200])

Each Document carries:

Field Description
chunk_index 0-based index within the document
page_numbers list of 1-indexed pages covered by the chunk
element_types list of semantic types detected (e.g. title, paragraph)
heading_context nearest surrounding heading, or None
token_estimate rough token count for budget planning
file_path / file_name / total_pages / pdf_version source metadata

One document per page

loader = OxidizePdfLoader("paper.pdf", mode="pages")
for doc in loader.lazy_load():
    print(doc.metadata["page_number"], len(doc.page_content))

Whole PDF as markdown

loader = OxidizePdfLoader("paper.pdf", mode="markdown")
[doc] = loader.load()
print(doc.page_content)

Adding caller metadata

loader = OxidizePdfLoader(
    "paper.pdf",
    extra_info={"source": "arxiv:2501.12345", "collection": "benchmarks"},
)

Keys in extra_info override base metadata (file_path, file_name, total_pages, pdf_version) if they collide — explicit caller intent.

Why oxidize-pdf

  • Rust parser: fast on large PDFs, low memory footprint.
  • Native RAG primitives: element-disjoint semantic chunking, element partitioning, heading-aware context — no post-processing needed. The disjointness contract (no chunk's text is a substring of another's; each source element appears in exactly one chunk) is enforced by regression tests in both this loader and the underlying bridge.
  • CJK friendly: compact output for multibyte documents (see oxidize-pdf 2.5.4 subsetter fixes).
  • Pure Python install: ships as a wheel for Linux/macOS/Windows via the oxidize-pdf package; no system dependencies.
  • Real lazy loading: lazy_load() returns a generator, so large PDFs don't force every Document into memory upfront.

Source

Part of oxidize-pdf-integrations, the ecosystem of integrations around oxidize-pdf. The Rust core and Python bridge live in oxidize-python.

License

MIT

Project links

Data verified by PyPI on May 19, 2026
Data provided by the project maintainers, verified at the time the release was uploaded to PyPI.

Key dates

PyPI data
Data sourced directly from PyPI's database.
  • Released:
Latest release

1 maintainer

PyPI data
Data sourced directly from PyPI's database.
Avatar for bytaro from gravatar.com bytaro

Credits

Author: Santiago Fernández Muñoz

GitHub Statistics

Data verified by PyPI on May 19, 2026
The GitHub source repository was provided by the project maintainers and verified by PyPI at the time of upload. Stars, forks, and open issues/PRs are derived from that repository and have not been independently verified.

License

MIT License (MIT)

Requires

Python <4.0, >=3.10

Tags

document-loader langchain oxidize-pdf pdf rag rust

Classifiers

Report project as malware

Download files

Download the file for your platform. If you're not sure which to choose, learn more about installing packages.

Source Distribution

langchain_oxidize_pdf-0.1.0.tar.gz (4.0 kB view details)

Uploaded Source

Built Distribution

If you're not sure about the file name format, learn more about wheel file names.

langchain_oxidize_pdf-0.1.0-py3-none-any.whl (4.6 kB view details)

Uploaded Python 3

File details

Details for the file langchain_oxidize_pdf-0.1.0.tar.gz.

File metadata

  • Download URL: langchain_oxidize_pdf-0.1.0.tar.gz
  • Upload date:
  • Size: 4.0 kB
  • Tags: Source
  • Uploaded using Trusted Publishing? Yes
  • Uploaded via: twine/6.1.0 CPython/3.13.12

File hashes

Hashes for langchain_oxidize_pdf-0.1.0.tar.gz
Algorithm Hash digest
SHA256 67c7e834e4b109714f5e3e807120733643532f75588b58f616d5ef431588b20b
MD5 71b34c5c9d9c7ce1afebf9dbed4ec5e4
BLAKE2b-256 b154c1acd610b9b0e77d36dd61f58d15e32c6d558d2abdd90d9790e0f34ca531

See more details on using hashes here.

Provenance

The following attestation bundles were made for langchain_oxidize_pdf-0.1.0.tar.gz:

Publisher: release-langchain.yml on bzsanti/oxidize-pdf-integrations

Attestations: Values shown here reflect the state when the release was signed and may no longer be current.

File details

Details for the file langchain_oxidize_pdf-0.1.0-py3-none-any.whl.

File metadata

File hashes

Hashes for langchain_oxidize_pdf-0.1.0-py3-none-any.whl
Algorithm Hash digest
SHA256 520c8e7df95690acaa242433ed04e50c20873e221b1962d938a6c11603567125
MD5 f7e2519d5bddf7732b2db6c332a4f6b4
BLAKE2b-256 309e115cadf2ea9dd89268e7f54fc3b647b195ecec128c18385ef45820d61b9e

See more details on using hashes here.

Provenance

The following attestation bundles were made for langchain_oxidize_pdf-0.1.0-py3-none-any.whl:

Publisher: release-langchain.yml on bzsanti/oxidize-pdf-integrations

Attestations: Values shown here reflect the state when the release was signed and may no longer be current.

Release history Release notifications | RSS feed

This release

0.1.0 This release

2 files

Supported by

AWS Cloud computing and Security Sponsor Datadog Monitoring Depot Continuous Integration Fastly CDN Google Download Analytics Sentry Error logging StatusPage Status page