Installation

Install Structure-D via pip with optional extras, or build the Rust CLI from source.

Requirements

Python 3.10 or higher
pip 23+ (for editable installs with pyproject.toml)
Rust 1.80+ (optional — only if building the CLI)

Python package

Install from PyPI (or directly from the repo) with only the extras you need:

bash

# Minimal — pipeline core only (no parsers, no LLM SDK)
pip install structure-d

# Recommended for most users
pip install "structure-d[ingestion,api,llm]"

# Everything
pip install "structure-d[all]"

Extras reference

Extra	Contents	When to use
`ingestion`	pymupdf, pdfplumber, pytesseract, python-docx, openpyxl, beautifulsoup4	Processing PDFs, images, Office files, HTML
`llm`	openai ≥1.30, anthropic ≥0.28, google-generativeai ≥0.8	Using cloud LLM providers
`inference`	vllm ≥0.6, transformers, torch	Self-hosted vLLM inference
`retrieval`	sentence-transformers, chromadb, langchain, llama-index	RAG pipelines and vector search
`storage`	sqlalchemy, asyncpg, pgvector, asyncmy	Writing to databases
`connectors`	boto3, google-cloud-storage, azure-storage-blob, paramiko	Reading from S3, GCS, Azure, SFTP
`destinations`	snowflake-connector-python, google-cloud-bigquery	Writing to Snowflake, BigQuery
`api`	fastapi, uvicorn, python-multipart	Running the HTTP service
`monitoring`	prometheus-client, opentelemetry	Metrics and tracing
`dev`	pytest, pytest-asyncio, ruff, mypy, pre-commit	Development and testing
`all`	All of the above	Full feature set

Rust CLI

The structure-d CLI is a native Rust binary for high-throughput batch extraction. It ships the same pipeline logic without the Python overhead.

bash

# Build with Cargo (requires Rust 1.80+)
cargo install --path cli/

# Or use the Makefile shortcut
make install

Verify the CLI is available:

bash

structure-d --version
# structure-d 0.2.0

structure-d formats   # list supported file formats
structure-d models    # list registered models

From source

bash

git clone https://github.com/jagguvarma15/Structure-D.git
cd Structure-D

# Set up virtual env
python -m venv .venv
source .venv/bin/activate

# Install in editable mode with all extras
pip install -e ".[all]"

# Build Rust CLI
make install

Verify the install

bash

python -c "import structure_d; print(structure_d.__version__)"
# 0.2.0

# Run the test suite
pytest tests/ -v -m "not slow and not integration"

ℹ

OCR dependencies

The tesseract_image and ocr_pdf parsers require Tesseract to be installed on your system: brew install tesseract on macOS or apt install tesseract-ocr on Linux.

Edit this page on GitHub