dlsci

Download scientific paper PDFs by DOI.

Tries OA sources first (Unpaywall, Semantic Scholar, arXiv, PMC, bioRxiv, Publisher Direct), then auto-fallback to Sci-Hub.

Features

• 6 OA sources priority: Unpaywall → Semantic Scholar → arXiv → PMC → bioRxiv → Publisher Direct
• Auto Sci-Hub fallback when OA sources fail
• Batch download support
• Structured JSON output for agents

Installation

git clone https://github.com/cypgg/dlsci.git
cd dlsci
pip install requests beautifulsoup4

Usage

# Download a paper (OA + Sci-Hub auto fallback)
python scripts/dlsci.py "10.1038/nature12385"

# Download to specific directory
python scripts/dlsci.py "DOI" --out ./papers

# Disable Sci-Hub (OA sources only)
python scripts/dlsci.py "DOI" --no-scihub

# Batch download
python scripts/dlsci.py --batch dois.txt

# JSON output (for agents)
python scripts/dlsci.py "DOI" --json

Environment Variables

Variable	Description
UNPAYWALL_EMAIL	Email for Unpaywall API (higher OA hit rate)
SEMANTIC_SCHOLAR_API_KEY	Semantic Scholar API key
NCBI_API_KEY	NCBI/PMC API key

Sources

1. Unpaywall - Legal OA via DOI lookup
2. Semantic Scholar - openAccessPdf field
3. arXiv - Preprint server
4. PMC - PubMed Central
5. bioRxiv/medRxiv - Preprints
6. Publisher Direct - IOP, MDPI, Frontiers, PLOS
7. Sci-Hub - Fallback (auto-enabled)

License

MIT