bibtex-filter

#! /usr/bin/env python

"""
Filter a .bib file with keys referred to in an .aux file. Print result to STDOUT.
"""

import argparse
import bibtexparser
import re
import itertools
import toolz

def determine_citation_keys_and_source(aux):
    """Determine the citation keys that we need to keep. Returns a list of citation keys"""

    with open(aux, 'r') as f:
        text = f.read()

    regex = re.compile('\citation{(.*)}')
    keys = regex.findall(text)
    # The argument to \citation{} may contain multiple, comma-separated values.
    keys = sorted(list(set(list(itertools.chain.from_iterable(map(lambda x: x.split(','), keys))))))

    source = re.compile('bibdata{(.*)}').findall(text)[0]

    return keys, source


def _get_entry(entries, key):
    for entry in entries:
        if entry.ID == key:
            return entry

def main():

  parser = argparse.ArgumentParser()
  parser.add_argument('aux', type=str)

  args = parser.parse_args()

  keys, source = determine_citation_keys_and_source(args.aux)

  with open(source, 'r') as f:
    db = bibtexparser.load(f)

  entries = db.get_entry_dict()
  selection = [entries[key] for key in keys]
  # Update db with selected entries
  db.entries = selection
  db._entries_dict = {} # Wipe this ugly piece of state

  print(bibtexparser.dumps(db))

  #with open(args.dest, 'w') as f:
    #bibtexparser.dump(db, f)


if __name__ == '__main__':
  main()