# Install dependencies
pip install requests

# Search by keywords
python scripts/biorxiv_client.py --terms "protein folding" --recent 30 --out results.json

# Search by author
python scripts/biorxiv_client.py --author "Chen" --recent 180

# Get specific paper by DOI
python scripts/biorxiv_client.py --doi "10.1101/2024.05.22.594321"

# Download PDF
python scripts/biorxiv_client.py --doi "10.1101/2024.05.22.594321" --fetch-pdf paper.pdf

from scripts.biorxiv_client import PreprintClient

client = PreprintClient(debug=True)

# Search by keywords
results = client.find_by_terms(
    terms=["enzyme engineering"],
    since="2024-01-01",
    until="2024-12-31",
    subject="biochemistry"
)

# Search by author
papers = client.find_by_author(name="Garcia", since="2023-01-01")

# Get paper by DOI
metadata = client.get_by_doi("10.1101/2024.05.22.594321")

# Download PDF
client.fetch_pdf(doi="10.1101/2024.05.22.594321", destination="paper.pdf")

# Normalize output
formatted = client.normalize(metadata, include_abstract=True)

{
  "query": {
    "terms": ["protein folding"],
    "since": "2024-03-01",
    "until": "2024-09-30",
    "subject": "biophysics"
  },
  "count": 87,
  "papers": [
    {
      "doi": "10.1101/2024.05.22.594321",
      "title": "Example Preprint Title",
      "authors": "Chen L, Patel R, Kim S",
      "corresponding_author": "Chen L",
      "institution": "Research Institute",
      "posted": "2024-05-22",
      "revision": "1",
      "category": "biophysics",
      "license": "cc_by",
      "paper_type": "new results",
      "abstract": "Abstract content here...",
      "pdf_link": "https://www.biorxiv.org/content/10.1101/2024.05.22.594321v1.full.pdf",
      "web_link": "https://www.biorxiv.org/content/10.1101/2024.05.22.594321v1",
      "journal_ref": ""
    }
  ]
}