Sorts a BibTex file according to its cited order (used in conjunction with biber and parallel alphabetic style)

numeric_entry_sorter.py

#!/usr/bin/env python
from __future__ import print_function
import getopt, sys

bib_types = ("@misc", "@online")

def main(options):
    aux_file = options[0]
    bib_file = options[1]

    citations = extract_citations(aux_file)
    entries = extract_entries(bib_file)
    number_of_digits = add_running_indices(citations, entries)
    sortname_format = "%0" + str(number_of_digits) + "d"

    new_bib_file = []
    with open(bib_file) as f:
        for line in f:
            line = line.rstrip()
            l = line.lower().lstrip()

            if l.startswith("shorthand") or l.startswith("sortname"):
                pass
            else:
                new_bib_file.append(line)

            if l.startswith(bib_types):
                shorthand = entries.get(extract_citation_key(line), 0)
                if shorthand:
                    new_bib_file.append("  shorthand = {%s}," % shorthand)
                    new_bib_file.append(("  sortname = {" + sortname_format + "},") % shorthand)

    for line in new_bib_file:
        print(line)

def extract_citations(aux_file):
    with open(aux_file) as f:
        for line in f.readlines():
            if "abx@aux@cite" in line:
                yield line.split("{")[1].split("}")[0]

def extract_entries(bib_file):
    with open(bib_file) as f:
        entries = {}
        for line in f.readlines():
            l = line.lower().lstrip()
            if l.startswith(bib_types):
                citation_key = extract_citation_key(line)
                entries[citation_key] = 0
        return entries

def extract_citation_key(s):
    if s.lstrip().startswith("@"):
        return s.split("{")[1].split(",")[0].strip()
    else:
        return ""

def add_running_indices(citations, entries):
    running_index = 1
    for cite in citations:
        if cite in entries:
            entries[cite] = running_index
            running_index = running_index + 1
    number_of_digits = len(str(running_index))
    return number_of_digits

### Invocation
def help():
    print("Usage: %s AUX_FILE BIB_FILE" % sys.argv[0])
    print("""Sort BibTeX entries in BIB_FILE according to first cite in AUX_FILE

Outputs the annotated BibTeX entries to standard out.""")


def opterr(msg):
    print("parse error:", msg, file=sys.stderr)
    print("use `%s -h' to get help" % sys.argv[0], file=sys.stderr)
    sys.exit(1)

def parse_options():
    try:
        opts,args=getopt.getopt(sys.argv[1:],"h",\
                  ["help"])
    except getopt.GetoptError, message:
        opterr(message)
    for opt,arg in opts:
        if opt in ("-h","--help"):
            help()
            sys.exit(0)
    return args

def check_filetypes(args):
    if len(args) != 2:
        opterr("provide AUX_FILE and BIB_FILE")
    if not args[0].endswith(".aux"):
        opterr("First argument has to be the AUX_FILE")
    if not args[1].endswith(".bib"):
        opterr("Second argument has to be the BIB_FILE")

if __name__=="__main__":
    args=parse_options()
    check_filetypes(args)
    try:
        main(args)
    except KeyboardInterrupt:
        print("error :(", file=sys.stderr)