This is the blog of Adam Kalsey. Unusual depth and complexity. Rich, full body with a hint of nutty earthiness.
Freshness Warning
This blog post is over 19 years old. It's possible that the information you read below isn't current and the links no longer work.
24 Sep 2003
I’m working on a site and realized that I had left off the alt text from some of my images. Going through several hundred pages looking for images that didn’t contain the alt text didn’t sound like much fun, so I whipped up a quick Perl script to add the text for me.
The program walks through a directory and opens every file ending with .html or .htm. Then it looks for image tags without alt attributes and adds them in. The alt text that it adds comes from a pipe-delimited data file containing image file names followed by the alt text. If the program comes across an image that doesn’t have an alt attribute and it isn’t listed in the data file, an empty alt attribute is inserted. Image tags that already have alt attributes are ignored.
To use the program, create a text file that looks like this:
someimage.jpg|Some Imagelogo.gif|Kalsey Consulting Groupyomama.jpg|Your Mother
Then run the Perl app like so…
ALTer.pl /path/to/data_file.txt /path/to/search/inside/
Please note: This is provided with no warranty or tech support whatsoever. Make backups. If you come across a problem, I probably won’t be able to help you. The code is licensed under the MIT License, copyright 2003 Kalsey Consulting Group.
Here’s the contents of ALTer.pl
#!/usr/bin/perluse File::Find;use strict;my $directory = './';my $data_file;my @valid_extensions = ('\.html?');my %texts;$directory = $ARGV[1] if $ARGV[1];$data_file = $ARGV[0];open (DATA, $data_file) or die "Can’t open $data_file: $!";while (<DATA>) { chomp; my ($src, $alt) = split /\|/; $texts{$src} = $alt;}chomp($directory);$directory =~ s/\\/\//g;find(\&add_alt, $directory);sub add_alt() { my $file = $File::Find::name; my @html; my $blnValidType = 0; # skip directories return if(-d $file); # only insert alt text if this is an HTML file foreach(@valid_extensions) { if($file =~ m/$_$/) { $blnValidType = 1; last; } } return if($blnValidType == 0); chomp($file); open(IN, $file) || die "Cannot open $file for reading: $!"; @html = <IN>; close(IN); my $line = join("\n", @html); $line =~ s+<img([^>]*)?>+ my $orgtxt = $1; if($1 !~ /alt\=\"/) { my $alttxt = ''; if ($1 =~ m/src="([^"]*)/) { local $_ = $1; s/^.*(\\|\/)//; $alttxt = $texts{$_}; } "<img alt=\"$alttxt\" $orgtxt>"; } else { "<img$orgtxt>"; } +egs; open (OUT, ">$file"); print OUT $line; close OUT;}
That looks to be a handy script. You can make your regex handle uppercase tags by using the 'i' modifier (which makes the regex case insensitive). In the img regex, s+]*)?>+ would become s+]*)?>+i. (the period is a full stop and not part of the regex) If you wanted to write some messy regex (is there any other kind :-) ), you could find unquoted attributes by looking for anything but spaces and single/double quote tags after the equals sign, ala m/src=([^"/s']*)/ . I'm sure the regex I just gave fails in some instances. YMMV.
You'd want \s for a space character, not /s. I'm aware of the i modifier, and you'd actually have to use it three times. Once after /alt\=\"/ once after m/src="([^"]*)/ and once after +egs Since the script was written primarily for my use, I'm not too concerned with different HTML styles. But if people want to suggest changes that will make things a bit more robust, please do. It could be handy.
I thought I would post this for you since I got so much milage from your Forms Errors Simplified article. Just for info here's how you could do it with HTML::TokeParser::Simple which will be more accepting about the nature of the HTML that's put in. -- replace lines 45 - 65 in your script use HTML::TokeParser::Simple; my $parser = HTML::TokeParser::Simple->new( $file ) || die "Can't open $file for reading: $!"; my $html; while (my $token = $parser->get_token) { # find image files without alt if ( $token->is_start_tag('img') ) { my $tag = $token->return_tag; my $attr = $token->return_attr; # If we have an alt entry already we'll skip it if ( defined $attr->{alt} ){ # we could change the above test to look in %texts for new alt contents # for tags that exist but are empty # if ( defined $attr->{alt} and $attr->{alt} ne '' ) next; } # Add an alt attribute my $alt = exists $texts{ $attr->{src} } ? $texts{ $attr->{src} } : ''; $token->set_attr( alt => $alt ); } $html .= $token->as_is; } open (OUT, ">$file"); print OUT $html; -- above code is mostly untested unless otherwise stated Also here's a link to an example of a script I wrote to extract inline css to external style sheets. Extract inline styles to an external style sheet http://www.perlmonks.org/index.pl?node_id=292225 Clayton PS the code would look nicer if the comment system would let me put code tags around it. :)
That's a good idea. I hadn't thought about using an actual HTML parser. I wonder how fast this method would be though. One advantage of the big ugly regex is that the script takes under a second to process 150 3kb HTML files. On the other hand, since this is a utility script that is rarely run, speed probably isn't much of an issue as long as it doesn't take more than a few minutes to process.
Since it's a rarely used utility script I'm more worried about maintainability and robustness than speed. I'd use a set of similar scripts for all kinds of HTML transformations when refactoring a new customers old website.
It would be nice to set up a find and replace script for this in BBEdit using grep. I would if I knew how.
Looking for a regex to find and tags that have file:/// in them... any ideas?
use while ( my $line = ) {} instead of @lines = ; for those really big files, you will save a lot of memory this way. thx for the script
What a "perler" of an idea (sorry about that) but seriously this is very clever. As a bit of a novice I use dreamweaver to do all my design work and use the find & replace all the time to try and save work, it would be nice to just whip up a script sometimes!
This discussion has been closed.
Adam Kalsey
September 24, 2003 1:13 PM
Some quick notes... I'm not doing very robust HTML parsing here, so if you don't write your HTML like me then things probably won't work. The code won't handle uppercase tag or attribute names, spaces around the = signs, or attributes that aren't double-quoted. It will work with attrbutes that are in odd orders, though. If your image tag ends with an XHTML slash at the end, that slash will still be there after ALTer does its thing. If the tag doesn't end in a slash, ALTer won't add it.