BioMed Article Converter

This is a simple package made to convert NCBI PubMed/PMC XML formats into more common and useable downstream formats such as HTML, MD, or plain text

This can be used as command line tool or as a library. For example to convert the XML from a pmc article to html

python -m biomed_article_converter input.xml output.html --source_type pmc

or via the library to convert to markdown

from biomed_article_converter import convert_format


with open("input.xml", "r") as fh:
    content = fh.read()

md = convert_format("xml", "md", content, source_type="pmc")

While the xml to html conversion tries to retain as much information as possible, the simpler md and txt formats make some assumptions on what you want to keep.

By default the md/txt formats drop the following

• author details
• references
• inline citations
• article metadata

as well as any sections with known non-informative titles defined by (OMIT_TITLES)

OMIT_TITLES = {
    "ethics statement",
    "conflict of interest",
    "conflicts of interest",
    "acknowledgments",
    "acknowledgment",
    "data deposition and access",
    "data availability",
    "code availability",
    "author contributions",
    "funding",
    "competing interest statement",
    "competing interests",
}

Table Normalization

For markdown or plain text, we cannot support the nested/complex table structures seen in html. Therefore the first step is simplifying the tables. For cells the span multiple columns or rows we simply create X new cells and repeat the content. For multi-row headers were concatenate each header level within a cell by row1: row2: cell content

XML inspection helpers

biomed_article_converter also provides small XML-structure helpers used when working with raw NCBI article responses.

Split PubMed article sets

PubMed efetch responses may contain multiple PubmedArticle or PubmedBookArticle records. Use split_articles_from_pubmed_xml() to split a response into individual XML documents keyed by PMID:

from biomed_article_converter import split_articles_from_pubmed_xml

articles = split_articles_from_pubmed_xml(xml)

article_xml = articles["32214442"]

The returned values are standalone XML strings for each article. Both regular PubMed articles and PubMed book articles are supported.

Detect abstract-only records

Use is_abstract_only() to determine whether an article XML record contains only abstract-level content:

from biomed_article_converter import is_abstract_only

if is_abstract_only(xml, source_type="pmc"):
    print("PMC did not provide substantive full text")

For PMC/JATS XML, a document is considered abstract-only when it has no substantive <body> content. PubMed XML records are abstract-level records by definition, so source_type="pubmed" returns True.

These helpers inspect article XML structure only; they do not perform format conversion.

Crossref JATS abstracts

Pass the Crossref API response's message["abstract"] string directly, without additional XML wrapping:

from biomed_article_converter import convert_format

abstract_xml = response.json()["message"].get("abstract")
if abstract_xml:
    md = convert_format("xml", "md", abstract_xml, source_type="crossref")

Crossref abstracts support HTML, Markdown, and plain-text output, including JATS section labels, paragraph boundaries, and inline formatting. Trailing AACR Citation Format: blocks are omitted. is_abstract_only() returns True for this source. The command line also accepts --source_type crossref. Conversion is local; it does not fetch Crossref records or look up PMIDs.