#!/usr/local/bin/perl -w

# by Jamie (jamie@westlake.com)
# asks for starting dir and HTML document root, then uses
# Image::Size to set HEIGHT and WIDTH attrs of all <IMG> tags

require 5.0;
use Image::Size;

# configure:
$debug = 0;

#--------------------

$examined_count =0;
$changed_count =0;
$total_errors =0;
$longest_line =0;

$| = 1; # don't buffer output - helps filename status display look right

$here = `pwd`;
chomp($here);

print "Starting directory (return for $here)?";
$startdir = <STDIN>;
chomp($startdir);
if ($startdir)
{
    chdir($startdir) || die "Can't chdir to $startdir.\n";
}
else
{
    $startdir=$here;
}

print "Site's HTML document root (return for $here)?";
$docroot = <STDIN>;
chomp($docroot);
if(!$docroot) {$docroot = $here;}
if(! -d $docroot) {die "Can't see $docroot. Sorry.\n";}
if($docroot =~ /\/$/) {chop($docroot);} # get rid of trailing slash if it's there

# get a list of all the .html files
@htmlfiles = grep { m/\.(\/.*\.s?html)\s*$/; $_ = $1 ? $startdir.$1 : undef; } `du -a`;

while ($current_html_doc = shift(@htmlfiles))
{
    chomp($current_html_doc);
    show_file("$current_html_doc");

    &setImgTagSize($current_html_doc);

    $total_errors += $errs_in_file;
    $examined_count++;
}
show("$examined_count files examined, $changed_count files changed, $total_errors errors.\n\n");
exit;

sub setImgTagSize
{
    ($html_filename) = @_;
    my($html_file, $temp_filename);

    if (open (HTMLDOC, $html_filename))
    {
	$html_file = join('',(<HTMLDOC>));
	close(HTMLDOC);
    }
    else
    {
	show("Can't read $html_filename\n");
	return;
    }

    $htmlfile_dir = $html_filename;
    while (chop($htmlfile_dir) ne "/") {}
    $temp_filename = "$htmlfile_dir/imagesizer_tmp$$";

    $changed_file = &substImgSizes($html_file);

    if ($changed_file eq $html_file)
    {
	return;
    }
    
    if (open (TEMPFILE, ">$temp_filename")) # for writing!
    {
	print TEMPFILE $changed_file; 
	close(TEMPFILE);

 	### delete original file
	unlink $html_filename; 
	### and rename tempfile to original filename
	link $temp_filename, $html_filename; 
	unlink $temp_filename;

	print "\rChanged file $html_filename" . " " x 20 . "\n";
	$changed_count++;
    }
    else
    {
	show("Can't create temp file in $htmlfile_dir\n");
	return;
    }
}



sub substImgSizes
{
    my($file)=@_;

    # grab ... from <IMG ...> or <img ...> and call substOneImgSize on it
    $file =~ s!\<IMG\s*([^\>]*)\>!"<IMG ".&substOneImgSize($1).">"!geis;

    return $file;
}



sub substOneImgSize
{
    my($imgtagattrs)=@_;
    my($imgfile, $imgpath);

    if ($debug) {print "\nGot IMG tag attrs: [$imgtagattrs]\n";}

    # getting something like:
    #  SRC="blah" ISMAP ALIGN="right"
    # or maybe
    #  SRC="blah" ALIGN="right" HEIGHT="20" width=100
    # set height and width if not present; if one/both is present, don't touch it
    
    if ($imgtagattrs =~ /height/i || $imgtagattrs =~ /width/i)
    {
	return $imgtagattrs; # already specifying height, width, or both, don't touch
    }
    
    # get image filename from src="..." or src = ...
    $imgtagattrs =~ m/src\s*=\s*([\"\']?)([^\"\' ]*)\1/i;

    $imgfile = $2;
    if ($debug) {print "\nImage file found in SRC attr: $imgfile\n";}

    if (!$imgfile)
    {
	show_err("Required SRC attribute not found in IMG tag!");
	return $imgtagattrs;
    }

    # make sure we are in the right directory - the image file could be url-relative or absolute
    if ($imgfile =~ /^\//)
    {
        # leading slash = URL-absolute; prepend document root to get pathname
	$imgpath = $docroot . $imgfile;
    }
    else
    {
        # no leading slash = URL-relative; chdir to HTML document's dir
	$imgpath=$imgfile;
	chdir($htmlfile_dir);
    }

    if (! -f $imgpath)
    {
	# file doesn't exist in filesystem
	if ($imgfile =~ m!^http://!)
	{
	    show_err(" Skipping $imgfile")
	}
    }
    else
    {
	# get image size via Image::Size module
	my($img_width, $img_height) = imgsize($imgpath);
	
	if ($img_width && $img_height)
	{
	    if ($debug) {print "\nGot width $img_width height $img_height\n";}
	    $imgtagattrs .= " HEIGHT=\"$img_height\" WIDTH=\"$img_width\""; 
	}
	else
	{
	    show_err(" $imgfile exists but I can't get its size.\n");
	}
    }

    return $imgtagattrs;
}


sub show
{
    # this cool fn displays messages with CR not LF so they all show on the same line
    my($mess) = @_;
    
    $mess_len = length($mess);
    $longest_line = ($longest_line >= $mess_len) ? $longest_line : $mess_len;

    print "\r$mess" . (" " x ($longest_line - $mess_len));
    
    if ($debug) {print "\n";}
}


sub show_err
{
    my($mesg)=@_;
    if ($errs_in_file == 0) {$mesg="\n".$mesg;}
    show($mesg);
    $errs_in_file++;
}

sub show_file
{
    my($mesg)=@_;
    show("Examining file: $mesg");
    $errs_in_file=0;
    return;
}