Parse markdown-style citations in .docx using pandocfilters

citefilter.py

#!/usr/bin/env python

"""
Pandoc filter to convert text of form [@foo,@bar] to citations.
Handy for e.g. word documents...
"""

import re
from pandocfilters import toJSONFilter, Str

reg = re.compile("\w+")


def Cite(
    citationIds,
    citationPrefix=[],
    citationSuffix=[],
    citationMode="NormalCitation",
    citationNoteNum=1,
    citationHash=0,
):
    return {
        "t": "Cite",
        "c": [
            [
                {
                    "citationId": citationId,
                    "citationPrefix": citationPrefix,
                    "citationSuffix": citationSuffix,
                    "citationMode": {
                        "t": citationMode,
                    },
                    "citationNoteNum": citationNoteNum,
                    "citationHash": citationHash,
                }
                for citationId in citationIds
            ],
            [{"t": "Str", "c": citationId} for citationId in citationIds],
        ],
    }


def mkcite(key, value, format, meta):
    if key == "Str":
        # match stuff of the form [@foo,@bar].
        match = re.match("(\[\@[\@\w+,]+\])([\.\w]*)", value)
        if match:
            citekeys_raw, remainder = match.groups()
            citekeys = reg.findall(citekeys_raw)
            cites = Cite(citekeys)
            return [cites, Str(remainder)]


if __name__ == "__main__":
    toJSONFilter(mkcite)