Fix CITE-KEY regex

cite_key.py

import os
import re
import sys


CITE_KEY_RE = re.compile(ur"\s*:CITE-KEY:(?:\s*(\w+)(-\d+)?)?", re.IGNORECASE)
PROPERTIES_RE = re.compile(ur"(\s*):PROPERTIES:\s*", re.IGNORECASE)
END_RE = re.compile(ur"\s*:END:\s*", re.IGNORECASE)
META_GENRE_RE = re.compile(ur"\s*:GENRE:\s*(note|preface|section)\b", re.IGNORECASE)

citeKeys = {}


def skip_until_re(regex, lines, i, output):
    numLines = len(lines)

    while i < numLines and not regex.match(lines[i]):
        output.append(lines[i])
        i += 1

    return i

def parse(lines):
    output = []
    numLines = len(lines)
    i = 0

    # Scan until we find the first CITE-KEY
    i = skip_until_re(CITE_KEY_RE, lines, i, output)

    if i >= numLines:
        print("ERROR: missing initial CITE-KEY")
        return output

    citeKey = CITE_KEY_RE.match(lines[i]).group(1)

    if citeKey not in citeKeys:
        citeKeys[citeKey] = 0

    output.append(lines[i])
    i += 1

    # Now scan until we find :END:
    while i < numLines:
        i = skip_until_re(PROPERTIES_RE, lines, i, output)

        if i >= numLines:
            return output

        indent = PROPERTIES_RE.match(lines[i]).group(1)
        output.append(lines[i])
        i += 1
        haveMeta = False
        haveEnd = False

        while i <= numLines:
            line = lines[i]
            m = CITE_KEY_RE.match(line)

            if m:
                i += 1
                continue
            else:
                m = META_GENRE_RE.match(line)

                if m:
                    haveMeta = True
                else:
                    m = END_RE.match(line)

                    if m:
                        # Insert next CITE-KEY
                        if haveMeta:
                            output.append("%s:CITE-KEY: %s:%dn\n" % (indent, citeKey, citeKeys[citeKey] + 1))
                        else:
                            citeKeys[citeKey] += 1
                            output.append("%s:CITE-KEY: %s:%d\n" % (indent, citeKey, citeKeys[citeKey]))

                        haveEnd = True

            output.append(line)
            i += 1

            if haveEnd:
                break

    return output



if __name__ == "__main__":

    rootPath = os.path.realpath(os.path.expanduser(sys.argv[1]))

    for root, dirs, files in os.walk(rootPath):
        for filename in [filename for filename in files if filename.endswith(".txt")]:
            print filename
            filePath = os.path.join(root, filename)

            with open(filePath) as f:
                lines = f.readlines()

            output = parse(lines)

            with open(filePath, "w") as f:
                f.write("".join(output))

    print "\ncite keys:\n"

    for key in sorted(citeKeys.keys(), cmp=lambda x,y: cmp(x.lower(), y.lower())):
        print "%s:%d" % (key, citeKeys[key])