#!/usr/bin/perl
# Author: Dimitri Antoniou <dimitri@risc2.aecom.yu.edu>
# usage: ltags filename
# handles: \label and \cite{ } with one or more arguments
#          fails if arguments of cite spread over more than one line
# also searches in files that are \include or \input in the main file

# get main LaTeX source file from command line:
$mainfile = shift;

# get names of included files and store them in an array
open MAIN, $mainfile or die "$!" ;
@mainfile=<MAIN>;
@allsrcfiles = map{ /^\\(?:input|include){(.*?)}/ } @mainfile;
unshift @allsrcfiles, $mainfile;

# loop over all source files
for $srcfile (@allsrcfiles) {
    # if \input{fname} append .tex to fname
    unless ( $srcfile =~ m/\.tex/ ) { $srcfile = $srcfile . "\.tex" }
    open SRC, $srcfile or die "$!" ;
    # store contents of source file in array @texfile
    @texfile=<SRC>;

    # store lines with \label and \cite (or \citeonline) in arrays
    @labelList  = grep{ /\\label{/ } @texfile;
    @citeList = grep{ /\\(cite|citeonline){/ } @texfile;

    # see if we use an external database; if yes, store its name in $bibfile
    ($dbase) = grep{ /^\\bibliography{/ } @texfile;
    if ($dbase) {
       $dbase =~ m/\\bibliography{(.*?)}/;
       $bibfile = $1;
    }

    # write \bibitem in tags file 
    @mrefs=();
    @refs=();
    @multirefs=();
    foreach (@citeList) {
        while ( m/\\(?:cite|citeonline){(.*?)}/g ) {
            $refs = $1;
            # if \cite has more than one argument, split them:
            if ($refs =~ /,/) {
                @mrefs = split /,/, $refs;
                # there might be more than one \cite in a line:
                push (@multirefs, @mrefs);
            }
            else {
                @refs = ($refs);
                push (@multirefs, @refs);
            }
        }
        # in BibTeX, format is @ARTICLE{Name, }; in source file, \bibitem{Name}
        for $ref (@multirefs) {
           if ( $dbase ) {
               push @unsorttag, "$ref\t$bibfile\t/{$ref,/\n"
           }
           else {
               push @unsorttag, "$ref\t$srcfile\t/bibitem{$ref}/\n"
           }
        }
    }

    # write \label in tag file
    foreach (@labelList) {
         m/\\label{(.*?)}/;
         push @unsorttag, "$1\t$srcfile\t/label{$1}/\n";
    }
}

# sort tag file; then, eliminate duplicates
@sortedtag = sort @unsorttag;
%seen = ();
@uniqtag = grep { ! $seen{$_} ++ } @sortedtag;

open(TAGS, "> tags");
print TAGS @uniqtag;