Document loader integrations
Document loaders provide a standard interface for reading data from different sources (such as Slack, Notion, or Google Drive) into LangChain’s Document format. This ensures that data can be handled consistently regardless of the source.
All document loaders implement the BaseLoader interface.
Integration components
- Typical retrieval flow: Document loaders (this page) → Text splitters → Embedding models → Vector stores → Retrievers → Chat models
- Agent capabilities: Tools, Middleware
- Execution and state: Sandboxes, Backends, Checkpointers, Stores
Warning
Community document loaders are user-contributed and unverified. LangChain does not review or endorse these integrations; use them at your own risk.
Interface
Each document loader may define its own parameters, but they share a common API:
load(): Loads all documents at once.lazy_load(): Streams documents lazily, useful for large datasets.
from langchain_docling.loader import DoclingLoader
FILE_PATH = "https://arxiv.org/pdf/2408.09869"
loader = DoclingLoader(file_path=FILE_PATH)
# Load all documents
documents = loader.load()
# For large datasets, lazily load documents
for document in loader.lazy_load():
print(document)By category
Productivity tools
The below document loaders allow you to load data from commonly used productivity tools.
| Document Loader | API reference |
|---|---|
| AgentMail | AgentMailLoader |
| Google Classroom | GoogleClassroomLoader |
Webpages
The below document loaders allow you to load webpages.
| Document Loader | Description | Package/API |
|---|---|---|
| Unstructured | Uses Unstructured to load and parse web pages | Package |
| Apify Dataset | Load documents from Apify datasets | API |
| Apify Crawl | Crawl a website on demand and load each page as a Document | API |
| Diffbot Crawl Loader | Crawl a website and load each page as ontologically consistent JSON using the Diffbot Crawl API | API |
| Diffbot Extract Loader | Load web pages as ontologically consistent JSON using the Diffbot Extract API | API |
| Docling | Uses Docling to load and parse web pages | Package |
| DomPruner | DOM AST pruning: loads web pages as compact Markdown with 97%+ token reduction, no API key | Package |
| Firecrawl | Turns websites into clean, LLM-ready data via scrape/crawl/map/extract/search | API |
| Hyperbrowser | Platform for running and scaling headless browsers, can be used to scrape/crawl any site | API |
| OpeddFeedLoader | Load a licensed Opedd content catalog as Documents with licensing provenance | API |
| ProxyHatLoader | Load web pages through ProxyHat residential proxies as Documents | API |
| AgentQL | Web interaction and structured data extraction from any web page using an AgentQL query or a Natural Language prompt | API |
| CRW | Open-source Firecrawl-compatible web scraper via local binary or fastcrw.com cloud | Package |
| Plasmate | Agent-native headless browser with Set of Mark (SOM) structured UI extraction | Package |
| Spidra | AI-powered web scraper with real browsers, CAPTCHA solving, and structured data extraction | API |
PDFs
Partner and community loaders for PDF files. Hosted partner guides are listed first. Classic loaders from langchain-community and other community packages follow. For the full download-sorted catalog, see All document loaders.
| Document Loader | Description | Package/API |
|---|---|---|
| Unstructured | Uses Unstructured's open source library to load PDFs | Package |
| Upstage Document Parse Loader | Load PDF files using UpstageDocumentParseLoader | Package |
| Docling | Load PDF files using Docling | Package |
PyPDFLoader | Load PDF files with pypdf (langchain-community) | Community |
PyMuPDFLoader | Load PDF files with PyMuPDF (langchain-community) | Community |
PDFMinerLoader | Load PDF files with PDFMiner (langchain-community) | Community |
PDFPlumberLoader | Load PDF files with pdfplumber (langchain-community) | Community |
| MinerU | Load PDF and other documents using MinerU | Package |
| UnDatasIO | Load PDF files using UnDatasIO | Package |
| OpenDataLoader PDF | Load PDF files using OpenDataLoader PDF | Package |
| CVFileLoader | Load .cv PDF/A-3u files with embedded Markdown, HTML, and JSON Resume payloads | Package |
| pdfmuse | Load PDF and DOCX files deterministically, with exact coordinates, tables and per-block section metadata for RAG | Package |
| oxidize-pdf | Load PDF files using a Rust engine with element-disjoint RAG chunking | Package |
| pdf-inspector | Load PDF files using pdf-inspector | Package |
Cloud providers
The below document loaders allow you to load documents from your favorite cloud providers.
| Document Loader | Description | Partner Package | API reference / Source |
|---|---|---|---|
| Azure Blob Storage | Load documents from Azure Blob Storage containers or individual blobs | ✅ | langchain-ai/langchain-azure |
| Google Cloud Storage Directory | Load documents from GCS bucket | ✅ | GCSDirectoryLoader |
| Google Cloud Storage File | Load documents from GCS file object | ✅ | GCSFileLoader |
| Google Drive | Load documents from Google Drive (Google Docs only) | ✅ | GoogleDriveLoader |
Common file types
The below document loaders allow you to load data from common data formats.
| Document Loader | Data Type |
|---|---|
Unstructured | Many file types (see https://docs.unstructured.io/platform/supported-file-types) |
HwpHwpxLoader | HWP/HWPX files |
DoclingLoader | Various file types (see https://docling-project.github.io/docling/) |
PolarisAIDataInsightLoader | Various file types (see https://datainsight.polarisoffice.com/documentation?docType=doc_extract) |
