#!/usr/bin/perl
use strict;  use warnings;


sub help {
  my $cmdname = $0;
  $cmdname =~ s:.*/::;
  print "$cmdname - list bib keys cited in LaTeX files and not defined in bib files.\n";
  print "Input is read from stdin or from named files.\n";
  print "The keys are those found in \\cite commands.\n";
  print "Each \\cite command can contain several keys, and can span lines.\n";
  print "The keys are presented sorted lexicographically, with each key listed once\n";
  print "even if it appears more than once.\n";
  print "If any .bib files are named on the command line, these are scanned for\n";
  print "definitions of keys.  Then only cited keys that are not defined are listed.\n";
  print "  -help  Show this help.\n";
  print "  -1     Display keys one per line.\n";
  print "  -,     Display keys in a comma-separated list.\n";
  print "  -v     Verbose.\n";
  print "\n";
  print "Here is an example use of $cmdname with bibfile to collect all the bib\n";
  print "entries needed for a group of .tex files:\n";
  print "bibfile -keep=`$cmdname *.tex` -expandCrossrefs";
  print " -keepStandardFields -showEntry ~/alspaugh.bib >cited.bib\n";
}

my $arg;
my $numFiles = 0;
my $sep = ",";
my $v = "";
my $k = "";
my %keys = ();
my %keysDefined = ();
my $count = 0;
foreach $arg (@ARGV) {
  if ($arg =~ /^-/) {
    if    ($arg =~ /^-+help$/) {
      &help;
      exit 0;
    }
    elsif ($arg =~ /^-+1$/) {
      $sep = "\n";
    }
    elsif ($arg =~ /^-+,$/) {
      $sep = ",";
    }
    elsif ($arg =~ /^-+v$/) {
      $v = "y";
    }
    else {
      print "Unexpected argument \"$arg\".\n";
      exit 0;
    }
  }
  else {
    if (open(HIN, $arg)) {
      if ($arg =~ /\.bib$/) {
        &bibFile(\*HIN, $arg);
      }
      else {
        &perFile(\*HIN, $arg);
      }
      close HIN;
      ++$numFiles;
    }
    else {
      print STDERR "Can't open input file \"$arg\".\n";
      exit -1;
    }
  }
}
if (0 == $numFiles) {
  &perFile(\*STDIN);
}
if (scalar %keys) {
  foreach $k (sort keys %keys) {
    if (defined $keysDefined{$k}) {
      # do nothing
    }
    else {
      if ($count++) {
        print "$sep";
      }
      print "$k";
    }
  }
  print "\n";
}

sub bibFile {
  my $IN = $_[0];
  my $fname = $_[1];
  my $lineno = 0;
  while (<$IN>) {
    ++$lineno;
    chomp;
    if (/^@[a-zA-Z]+[({]([-0-9a-zA-Z+]+),$/) {
       s/^@[a-zA-Z]+[({]([-0-9a-zA-Z+]+),$/$1/;
      #print "$_\n";
      my $key = $_;
      if (defined $keysDefined{$key}) {
        print STDERR "($fname:$lineno)  '$key' multiply defined ($keys{$key}).\n";
      }
      $keysDefined{$_} = "$fname:$lineno";
    }
  }
}

sub perFile {
  my $IN = $_[0];
  my $fname = $_[1];
  my $inCite = "";
  my $line = "";
  my $lineno = 0;
  while (<$IN>) {
    ++$lineno;
    chomp;
    s/^%.*$//;
    s/([^\\])%.*/$1/g;
    if ($inCite) {
      s/^[ \t]+//;
      $line = "${inCite}$_";
      $inCite = "";
      if ($v) {  print STDERR "Joining lines:  `$line'\n";  }
    }
    else {
      $line = $_;
    }
    while ($line =~  /^.*?(\\cite{) *([^,}]+),?([^}]*}.*)/) {  $k = $line;
      $k         =~ s/^.*?(\\cite{) *([^,}]+),?([^}]*}.*)/$2/;
      $line      =~ s/^.*?(\\cite{) *([^,}]+),?([^}]*}.*)/$1$3/;
      $keys{$k} = "$fname:$lineno";
      if ($v) {  print STDERR "Cited `$k'\n";  }
    }
    if ($line =~  /.*(\\cite{[^}]*$)/) {  $inCite = $line;
      $inCite =~ s/.*(\\cite{[^}]*$)/$1/;
    }
  }
}