#!/usr/bin/env python

# veraenderte Python-Loesung von Uwe Hoffmann
# eine Anpassung der Funktion nodups() sorgt bei grossen Datein
# fuer befriedigende Performance

"""USAGE: footnotes.py [-a] [--test] [< input.txt]

    --test      Self-test using text SAMPLE in script source
    -a          Number footnotes in order of appearance in text
"""

SAMPLE = '''A great brown fox [13] jumped of a pile of lorem ipsum [4], [7]. 
He
met with a silver penguin, browsing the Linux Kernel Mailinglist [3]. They
debated other the question whether to start a C-program with "main (int argc,
char **argv)" or with "main (int argc, char *argv[])". Square brackets annoyed
them [9999]. Multiple references may exist to same targets [4].

@footnote:

[13] Al Fabetus: "On characters and animals", 1888, self published.
[4] Lorem Ipsum, <a href="http://en.wikipedia.org/wiki/Lorem_ipsum">Web 
Link</a>
[9999] Annoying Link.
[7] B. Fox: "More on Blind Text".
[3] Linux Kernel Maintainers: LKML
'''
import sys, re

def nodups(lst):
    new = set()
    lst.reverse() # es soll aus Originalliste gelöscht werden
    for ind in xrange(len(lst)-1,0,-1): 
        x=lst[ind]
        if x not in new:
            new.add(x)
        else:
            del lst[ind] # das Duplikat wird aus der Originalliste entfernt
    lst.reverse() # und wieder zurück zur originalen Reihenfolge
    return lst
   
def mkMap(lst):
    return dict([(ref, "[%d]" % (n+1)) for n, ref in enumerate(lst)])

def asNum(line):
    try:
        key, _ = line[1:].split(']', 1)
        key = int(key)
    except:
        key = line or None    # If line looks wrong, just return the raw line
    return key

def footnote(s, usebody=False):
    FOOTSPLIT = '\n@footnote:\n'
    body, foots = s.split(FOOTSPLIT)
    links = nodups(re.findall(r'\[\d+\]', body))
    targets = nodups(re.findall(r'\[\d+\]', foots))
    mapping = mkMap(links if usebody else targets) # Ref order of body xor foots
    # If Python < 2.5, could use:
    # mapping = mkMap((targets, links)[usebody])

    # Replace source->target footnote numbers
    def numsub(m):
        return mapping[m.group(1)]
    body = re.sub(r'(\[\d+\])', numsub , body)
    foots = re.sub(r'(\[\d+\])', numsub, foots)

    # May need to reorder target lines in foots
    if usebody:
        foots = '\n'.join(sorted(foots.splitlines(), key=asNum))

    return body + FOOTSPLIT + "\n" + foots.strip()

if __name__ == '__main__':
    if '--test' in sys.argv[1:]:
        print footnote(SAMPLE, usebody='-a' in sys.argv[1:])
    else:
        print footnote(sys.stdin.read(), usebody='-a' in sys.argv[1:])