What is CSL-JSON used for?

CSL-JSON (Citation Style Language JSON) is the native bibliography format for Pandoc, citeproc, and tools like Zotero's better-bibtex. Pandoc reads a .json bibliography with --bibliography to render citations in any CSL style.

How are authors split?

Authors joined by `and` are split into separate CSL name objects. Names in `Last, First` form map to family and given; a single token without a comma is treated as a family name.

How is the year handled?

The BibTeX year becomes a CSL issued date with a date-parts array, for example `[[2021]]`. This is the structure Pandoc and citeproc expect for sorting and rendering dates.

Which entry types are mapped?

Common types map to CSL types: article maps to article-journal, book to book, inproceedings to paper-conference, phdthesis to thesis, and techreport to report. Unknown types fall back to the generic document type.

Is anything uploaded?

No. The BibTeX is parsed and converted entirely in your browser with JavaScript. Your bibliography data never leaves your device.

What is the BibTeX to CSL-JSON Converter?

Produces CSL-JSON that Pandoc, Zotero, and scholarly markdown tools consume directly. Maps standard BibTeX fields to CSL variables, splits authors, and parses years into date-parts — all fully in your browser. It runs free in your browser on Gera Tools, with nothing uploaded.

BibTeX to CSL-JSON Converter

Name: BibTeX to CSL-JSON Converter
Creator: Gera Tools
License: https://creativecommons.org/licenses/by/4.0/

Get one useful tool a week

Like this tool? Enter your email and we'll send you one genuinely useful Gera tool a week — plus a link to come back to this one. No spam, one-click unsubscribe any time.

Convert BibTeX to CSL-JSON

CSL-JSON is the bibliography format that Pandoc, citeproc-js, and reference managers like Zotero use internally. If you write in Markdown and render with Pandoc, a .json bibliography is often cleaner to work with than .bib. This tool reads standard BibTeX entries and emits a CSL-JSON array you can drop into your Pandoc workflow.

How it works

The parser scans for @type{key, ...} blocks, tracking brace depth so nested braces inside a field value do not end the entry early. Each field = {value} or field = "value" pair is captured, and surrounding braces and quotes are stripped.

Fields are then mapped onto CSL variables: title→title, journal/booktitle→container-title, volume→volume, number→issue, pages→page (with -- normalised to -), doi→DOI, publisher→publisher. The author and editor fields are split on and into CSL name objects with family/given. The year becomes an issued object whose date-parts is [[year]]. The BibTeX entry type chooses the CSL type (@article→article-journal, @inproceedings→paper-conference, and so on).

When to use this converter

You will need CSL-JSON specifically when:

Passing a bibliography to Pandoc via --bibliography refs.json — Pandoc’s citeproc natively reads CSL-JSON without any extra filter.
Importing into Zotero using the Better BibTeX or CSL JSON import option, which preserves richer metadata than the BibTeX import path.
Using citation tools in Obsidian, Quarto, or R Markdown that call citeproc under the hood.
Building a citations API where JSON is far easier to parse downstream than raw .bib text.

Worked example

An @article entry like this:

@article{Smith2021,
  author  = {Smith, Alice and Jones, Bob},
  title   = {A study of things},
  journal = {Journal of Things},
  year    = {2021},
  volume  = {12},
  pages   = {100--110},
  doi     = {10.1234/jot.2021}
}

becomes a CSL-JSON item:

{
  "id": "Smith2021",
  "type": "article-journal",
  "title": "A study of things",
  "author": [
    {"family": "Smith", "given": "Alice"},
    {"family": "Jones", "given": "Bob"}
  ],
  "container-title": "Journal of Things",
  "volume": "12",
  "page": "100-110",
  "DOI": "10.1234/jot.2021",
  "issued": {"date-parts": [[2021]]}
}

Note that the page range -- is normalised to a single -, and the year becomes the issued date object that Pandoc sorts on.

Notes and edge cases

The output id of each CSL item is the BibTeX cite key, so in-text citations like [@Smith2021] keep working after conversion. Page ranges written with a single hyphen, an en-dash, or -- are all normalised to a single hyphen, which is what CSL expects.

Multiple BibTeX entries in one paste are all converted: the output is a single JSON array containing one object per entry, which is the format --bibliography expects. Fields not listed in the mapping are silently dropped rather than forwarded as unknown keys, keeping the JSON clean for citeproc.