langchain-xberg
A LangChain document loader backed by Xberg. XbergLoader extracts text and metadata from 107 formats — running OCR where needed — and returns LangChain Document objects. Extraction is async at the core; multiple sources go through Xberg's extract_batch in a single native call, so concurrency happens Rust-side.
Install
pip install langchain-xberg
Requires Python 3.10+.
Load
Pass a path, a list of paths, a directory, or raw bytes. One source becomes one Document.
from langchain_xberg import XbergLoader
# Single file
docs = XbergLoader(file_path="report.pdf").load()
print(docs[0].page_content) # extracted markdown
print(docs[0].metadata["title"]) # source, mime_type, title, authors, detected_languages, page_count, ...
# Multiple files — one batched extraction
docs = XbergLoader(file_path=["report.pdf", "notes.docx"]).load()
# A directory with a glob
docs = XbergLoader(file_path="./corpus/", glob="**/*.pdf").load()
# Raw bytes (mime_type required)
docs = XbergLoader(data=raw_bytes, mime_type="application/pdf").load()
Chunk for retrieval
Enable Xberg's native chunking to emit one Document per chunk, sized for embedding. Each chunk carries chunk_index, total_chunks, heading_path, page, and token_count in its metadata.
from langchain_xberg import XbergLoader
from xberg import ChunkingConfig, ExtractionConfig
config = ExtractionConfig(chunking=ChunkingConfig(max_characters=1000, overlap=200))
docs = XbergLoader(file_path="report.pdf", config=config).load() # one Document per chunk
To split by page instead, pass pages=PageConfig(extract_pages=True); each Document then gets a 0-indexed page.
Configure extraction
Pass any Xberg ExtractionConfig to control OCR, output format, and batch concurrency.
from xberg import ExtractionConfig, OcrConfig
config = ExtractionConfig(
output_format="markdown",
ocr=OcrConfig(backend="tesseract"),
force_ocr=True,
max_concurrent_extractions=8,
)
docs = XbergLoader(file_path="./corpus/", config=config).load()
Async
Inside an event loop, use the async API — await loader.aload() or async for doc in loader.alazy_load(). These use Xberg's native async extraction end to end. The synchronous load / lazy_load bridge to it and must not run inside a running loop.
loader = XbergLoader(file_path="report.pdf")
docs = await loader.aload()
Errors
A per-input failure raises xberg.XbergError with the offending source and message. In batch mode the failure comes from ExtractionResult.errors; a single load surfaces the raised exception directly.
For the full API, see the Xberg documentation.
2 maintainers
Metadata
2 maintainers
Release files for langchain-xberg 1.3.0
For a detailed explanation of source distributions (sdists) and built distributions (wheels), please see the package formats documentation.
Source distribution (sdist)
| File | Size | Uploaded | |
|---|---|---|---|
| langchain_xberg-1.3.0.tar.gz | 12.6 kB | Details |
Built distribution (wheel)
| File | Interpreter | ABI | Platform | Reset |
|---|---|---|---|---|
| langchain_xberg-1.3.0-py3-none-any.whl | Python 3 | none | any | Details |
Total release size: 21.8 kB
Release files / langchain_xberg-1.3.0.tar.gz
| Download URL | langchain_xberg-1.3.0.tar.gz |
|---|---|
| Size | 12.6 kB |
| Tags | Source |
|
SHA-256 checksum How to use checksums |
dd9c8968fe31251ebdfd97046f411c51fcff30ea4b62fe70651b903be2af7bc8
|
|
BLAKE2b-256 checksum How to use checksums |
2752e807f11f26393a9f1662aa6a44d8334af6e2f39427a81a02d79070db6292
|
| Upload date | |
|
Uploaded using Trusted Publishing? What is trusted publishing? |
Yes |
| Uploaded via |
uv/0.12.20 {"installer":{"name":"uv","version":"0.12.20","subcommand":["publish"]},"python":null,"implementation":{"name":null,"version":null},"distro":{"name":"Ubuntu","version":"24.04","id":"noble","libc":null},"system":{"name":null,"release":null},"cpu":null,"openssl_version":null,"setuptools_version":null,"rustc_version":null,"ci":true}
|
Release files / langchain_xberg-1.3.0-py3-none-any.whl
| Download URL | langchain_xberg-1.3.0-py3-none-any.whl |
|---|---|
| Size | 9.2 kB |
| Tags | Python 3 |
|
SHA-256 checksum How to use checksums |
08b59a16fe746f3512682761c7654d727c1ad1ad8da467c50b820304b1781508
|
|
BLAKE2b-256 checksum How to use checksums |
3b201b48289a2c986cc5c38bb06542244f2ee4dc65dcc47f65f2fb3c5740ae20
|
| Upload date | |
|
Uploaded using Trusted Publishing? What is trusted publishing? |
Yes |
| Uploaded via |
uv/0.12.20 {"installer":{"name":"uv","version":"0.12.20","subcommand":["publish"]},"python":null,"implementation":{"name":null,"version":null},"distro":{"name":"Ubuntu","version":"24.04","id":"noble","libc":null},"system":{"name":null,"release":null},"cpu":null,"openssl_version":null,"setuptools_version":null,"rustc_version":null,"ci":true}
|
