addlongdesc

#!/usr/local/bin/perl
# Add d-links from images to long descriptions.
# Two cases:
#   Image is part of a figure with a caption (<P class="caption">):
#      Insert d-link at beginning of caption.
#   Not part of a figure or no caption.
#      Add d-link just after image.
#
# Ian Jacobs ij@w3.org
# $Id: addlongdesc,v 1.5 2006-10-09 18:55:52 ihickson Exp $

use English;
use DB_File;
use lib 'bin';
use utils;

$PROG = substr($0, rindex($0, "/") + 1);
$USAGE = "Usage: $PROG input [output]\n";
($file = $ARGV[0]) || die $USAGE;
shift;
($output = $ARGV[0]) || ($output = "-");
shift;

sub image_name {
    my $f = `basename $_[0] '.gif'`;
    $f = `basename $_[0] '.png'`;
    chop $f;
    return($f);
}

sub desc_title {
    my $f = "$_[0]";
    my $title = "";
    if (open(DLINK, $f)) {
	local $/ = undef;
	my $b = <DLINK>;
	close(DLINK);
	$b =~ /<TITLE>(.*?)<\/TITLE>/iso;
	# Probably need to strip title of markup here.
	$title = $1;
    }
    return($title);
}


open(FILE, $file) || 
    die "$PROGNAME Error: Cannot open $file\n";

open(OUTPUT, "> $output") || 
    die "$PROGNAME Error: Cannot output to $output\n";

$INPUT_RECORD_SEPARATOR="<";
$buf = "";
while ($buf = <FILE>) {
    print OUTPUT $buf;
    $INPUT_RECORD_SEPARATOR=">";
    unless (eof(FILE)) {
	$anchor=<FILE>;
	# Remove final ">"
	chop($anchor);
	$INPUT_RECORD_SEPARATOR="<";
	$dlink = "";
	if ($anchor =~ /^IMG.*?src\s*=\s*(?:([^\s>]+)|\"([^\"]+)\").*/is){
	    my $imagename = image_name("$1$2");
	    my $descfile = "images/longdesc/$imagename-desc.html";
	    if (-f "$descfile") {
		$title = desc_title($descfile);
		if ($title) {
		    $dlink = "<SPAN class=\"dlink\">&nbsp;&nbsp;&nbsp;<A name=\"img-$imagename\" href=\"$descfile\" title=\"$title\">[D]</A></SPAN>";
		}
	    }
	} else {
	    # No change to the anchor.
	}
	# Put back chopped ">"
	print OUTPUT "$anchor>$dlink";
    }
}
close(FILE);
close(OUTPUT);