pubmed-markdown 0.2.2

Convert PubMed articles (PMIDs or PMCIDs) to clean, structured markdown with full text, abstracts, and supplementary materials

PubMed Downloader

Convert PubMed articles to clean, structured markdown. Handles the full pipeline: PMID resolution, full-text extraction via PubMed Central, HTML-to-markdown conversion, and supplementary material retrieval.

Articles without open-access full text automatically fall back to abstract-only download.

Installation

pip install pubmed-markdown

Setup

Set your email for NCBI API identification (optional but recommended):

export NCBI_EMAIL=[e-mail]

Or create a .env file in your working directory:

NCBI_EMAIL=[e-mail]

Usage

Python API

Get markdown strings (single or batch, no files created):

from pubmed_markdown import PubMedMarkdown

downloader = PubMedMarkdown()

# From PMID — accepts a single string or a list
markdown = downloader.pmid_to_markdown("12895196")
markdowns = downloader.pmid_to_markdown(["12895196", "17872605"])

# From PMCID directly — also accepts a single string or a list
markdown = downloader.pmcid_to_markdown("PMC1884285")
markdowns = downloader.pmcid_to_markdown(["PMC1884285", "PMC6435416"])

Save markdown files to disk (single or batch):

from pubmed_markdown import PubMedMarkdown

downloader = PubMedMarkdown()
downloader.pmids_to_markdown_files(["12895196", "17872605"], save_dir="data")

# Also works with a single PMID
downloader.pmids_to_markdown_files("25051018", save_dir="data")

This creates:

data/
├── html/          # Raw HTML from PMC
└── markdown/      # Converted markdown files

Individual utility functions:

from pubmed_markdown import (
    get_pmcid_from_pmid,
    get_html_from_pmcid,
    get_abstract_markdown_from_pmid,
    fetch_bioc_supplement,
)

# Resolve PMIDs to PMCIDs
mapping = get_pmcid_from_pmid(["12895196", "17872605"])

# Fetch raw HTML from PMC
html = get_html_from_pmcid("PMC1884285")

# Get abstract for non-open-access articles
abstract_md = get_abstract_markdown_from_pmid("12345678")

# Get supplementary material text
supplement = fetch_bioc_supplement("PMC6435416")

Command Line

# Convert PMIDs from a file (one PMID per line)
pubmed-download --file_path=pmids.txt --save_dir=data

API Reference

Method	Creates Files	Returns	Use Case
pmid_to_markdown()	No	Markdown string(s)	Single or batch, programmatic use
pmcid_to_markdown()	No	Markdown string(s)	Direct PMCID conversion
pmids_to_markdown_files()	Yes	None	Batch processing, building datasets
local_html_to_markdown()	Yes	None	Re-convert existing HTML files

All methods accepting IDs take either a single string or a list of strings.

How It Works

1. PMID to PMCID -- Uses NCBI's ID Converter API with batching and rate limiting
2. HTML extraction -- Fetches full article HTML from PubMed Central
3. Markdown conversion -- Converts HTML to structured markdown preserving tables, figures, citations, and references
4. Supplementary materials -- Fetches pre-processed supplement text via NCBI's BioC API
5. Abstract fallback -- Articles not in PMC Open Access get abstract + metadata via NCBI E-Fetch

Configuration

Environment Variable	Default	Description
NCBI_EMAIL	None	Email for NCBI API identification

License

MIT