Trying to get perl regex to find multi-line AND single-line HTML comments

Question

I'm trying to find both single and multi-line comments in an HTML file. I've stripped it down to just a few examples, and some other content just to have something there.

I've read a lot of the entries here but can't get a definitive answer to this. I'm reading in the HTML file in "slurp" mode, and doing a match of my pattern. This code runs now and prints only the first match.

#!C:\Perl\bin\perl.exe 

BEGIN {  unshift @INC, 'C:\rmhperl'; } 

use warnings;
no warnings 'uninitialized';

chdir 'c:\watts\html'; 

open FILE, "test.html" or print 'error opening file "test.html" ';
my $text = do { local $/; <FILE> };
close(FILE);

if ($text =~ m/(?s)(<!--.*?)(-->\n)/sg) {
    print "1 = $1  2= $2\n";
}

exit;

I've set up single and multi-line comments in the HTML file. I can get one or the other printed but not both (at least in "slurp" mode).

I'm told I should be able to accomplish this with a single regex, so the objective is "find all HTML comments, regardless of their being single/multi-line comments" .

I built the regex to find both, but finds only the first match -- a multi-line comment.

I'm trying to find a way to find every match, whether it occurs on one line or multiple lines. I can find one or the other, but I can't get them to work with one regex.

I can do non-slurp mode, and find the  tag, but wanted to see if I can get it to work with a single regex.

I've been reading about this, and trying to find relevant examples. can't see what I'm missing. Here's the HTML file snippet I have been using for the regex:

HTML file

<!DOCTYPE html>
 
<script type="text/javascript" src="fadeslideshow.js"></script>
<style>

.divTable {
    display: block;
    width: 100%;
}

.divTableBody, .divTableRow{ clear: both; }

.divTableCell {
    border: 1px solid #999999;
    float: left;
    overflow: hide;
    padding: 2%;
    width: 45%; }

.divTable:after {
    display: block;
    font-size: 0;
    content: " ";
    clear: both;
    height: 100px; }
</style>
<style type="text/css">
<!--
a:link {color: #0000ff;}
 a:visited {color: #3563a8;} 
 a:active {color: #000000;}
 a:hover {background-color: #000000;}
 a {text-decoration: none;}
 -->
 </style> 
</head>
    <body class="home">

 <div id="white_back">
<div style="text-align: center">
</div>
<div class="chromestyle" id="chromemenu">
<ul>
<!-- <li><a href="xyz.com">Home</a></li>
 -->
 <li><a href="#" rel="dropmenu0">About Us</a></li>
<li><a href="#" rel="dropmenu5">Publications</a></li>   
</ul>
</div>

<!--1st drop down menu
-->                                                   
<div id="dropmenu0" class="dropmenudiv">
</div>

<!--2nd drop down menu -->
<div id="dropmenu1" class="dropmenudiv">
</div>

Answer 1

I presume this is production code, in which case your manager is a scary man as this sort of practice can result in hard-to-find bugs. That's acceptable if the code is only for yourself, but inflicting that on others is unfair

Some notes on your code

The shebang line #! is unnecessary on Windows systems, and in fact does nothing unless you specify command-line options there. It's best to drop it altogether
Always use strict and use warnings 'all' , and fix the bugs rather than disabling warnings with no warnings 'uninitialized'
BEGIN { unshift @INC, 'C:\\rmhperl' } is best written use lib 'C:\\rmhperl' but you're not using libraries in this case so it will have no effect
You should use lexical file handles with the three-parameter form of open
There is no need for (?s) in the regex pattern as well as the /s modifier. Unless you are doing something fancy like enabling options for only part of the pattern (which you're not) then people will understand you better if you use the modifier /s

The reason you're only finding one comment is that you're only asking for one. In scalar context a global regex pattern match will iterate through all the matches in the target string one at a time. You only call it once so it finds only the first. You can fix that by using a while in place of if

I've improved your regex pattern somewhat by making sure that the opening <-- isn't followed by > or by -> which would form an illegal HTML comment. There may also be optional space after the closing -- and the > so I've allowed for that. And you are insisting on a newline after the end of the comment which may not be there, so I've removed that

This code seems to work with your data

use strict;
use warnings 'all';

my $text = do {
    open my $fh, '<', 'test.html' or print qq{Unable to open file "test.html" for input: $!};
    local $/;
    <$fh>;
};

while ( $text =~ /(<!--(?!-?>).*?--\s*>)/sg ) {
    my $comment = $1;
    print $comment, "\n";
}

output

<!--
a:link {color: #0000ff;}
 a:visited {color: #3563a8;} 
 a:active {color: #000000;}
 a:hover {background-color: #000000;}
 a {text-decoration: none;}
 -->
<!-- <li><a href="xyz.com">Home</a></li>
 -->
<!--1st drop down menu
-->
<!--2nd drop down menu -->

Trying to get perl regex to find multi-line AND single-line HTML comments

Question

HTML file

1 answers

solution1
2 2016-03-26 21:30:09

output

Trying to get perl regex to find multi-line AND single-line HTML comments

Question

HTML file

1 answers

solution1 2 2016-03-26 21:30:09

output

solution1
2 2016-03-26 21:30:09