#!/usr/bin/env python

# Copyright (C) 2008 Richard Hacker
# License: GPLv2

import mmap
import re
import sys

class Document:
    # Regex to find a reference
    ref_regex = re.compile(r"\[[0-9]+\]")

    # Offset in file pointing to "@footnote:\n"
    footnote_pos = 0

    # Original footnote reference number appearance
    orig_footnote = []

    # New footnote sorting with original numbers
    new_footnote = []

    # Mapping {original_number} -> [new_number, offset of footnote text]
    footnote = {}

    # Current reference index
    refidx = 1

    def __init__(self,file):
        f = open(file)
        self.fmap = mmap.mmap(f.fileno(), 0, mmap.MAP_SHARED, mmap.PROT_READ)

        # Find footnote separator
        marker = "\n@footnote:\n"
        self.footnote_pos = self.fmap.find(marker) + 1

        self.__read_footnotes()

    # Go through the footnotes in original order and renumber them
    def sort_by_footnote(self):
        for i in xrange(len(self.orig_footnote)):
            self.__new_reference(self.orig_footnote[i])

    # Print the file body and footnote
    def output(self):
        self.__print_body()
        self.__print_footnote()

    # Fill in the list self.orig_footnote and the map self.footnote
    def __read_footnotes(self):
        self.fmap.seek(self.footnote_pos)
        while self.fmap.tell() != self.fmap.size():
            pos = self.fmap.tell()
            line = self.fmap.readline()
            
            # All lines proceeding the footnote marker have to begin with
            # [0-9], otherwise they are discarded. Use Python's match() method
            # to force a match at the line start
            m = self.ref_regex.match(line)
            if m != None:
                refnumber = int(m.group(0)[1:-1])
                self.orig_footnote.append(refnumber)
                self.footnote[refnumber] = [0, pos+m.end(), 9]

        self.new_footnote = [0] * len(self.orig_footnote)

    # Return the new reference number as a string "[new-number]"
    def __new_reference_str(self, pat):
        ref = self.__new_reference(int(pat.group(0)[1:-1]))
        if ref:
            return "[%i]" % ref
        else:
            return "[??unknown??]"

    # Allocate and return a reference a new number. If the original number does
    # not appear in the footnotes section, return 0
    def __new_reference(self, oldref):
        if not self.footnote.has_key(oldref):
            return 0

        if not self.footnote[oldref][0]:
            # Assign new reference number
            self.footnote[oldref][0] = self.refidx
            self.new_footnote[self.refidx-1] = oldref
            self.refidx = self.refidx + 1

        return self.footnote[oldref][0]

    def __print_body(self):
        self.fmap.seek(0)
        while self.fmap.tell() < self.footnote_pos:
            line = self.fmap.readline()

            # Substitute all references with their new values
            print self.ref_regex.sub(self.__new_reference_str, line),

    def __print_footnote(self):
        self.fmap.seek(self.footnote_pos)
        print self.fmap.readline(),
        for ref in self.new_footnote:
            self.fmap.seek(self.footnote[ref][1])
            print "[%i] %s" % (self.footnote[ref][0],  self.fmap.readline()),

doc = Document(sys.argv[-1])

# Calling this program with the -a switch to preserve the original footnote
# ordering
if "-a" in sys.argv:
    doc.sort_by_footnote()

doc.output()