Perl正则表达式在线PCRE测试员工作但不在perl命令中

Question

I've written the following PCRE regex to strip scripts from HTML pages: <script.*?>[\\s\\S]*?< *?\\/ *?script *?>

It works on many online PCRE regex testers:

https://regex101.com/r/lsxyI6/1

https://www.regextester.com/?fam=102647

It does NOT work when I run the following perl substitution command in a bash terminal: cat tmp.html | perl -pe 's/<script.*?>[\\s\\S]*?< *?\\/ *?script *?>//g' cat tmp.html | perl -pe 's/<script.*?>[\\s\\S]*?< *?\\/ *?script *?>//g'

I am using the following test data:

<script>
                       $(document).ready(function() {
                           var url = window.location.href;
                           var element = $('ul.nav a').filter(function() {
                               if (url.charAt(url.length - 1) == '/') {
                                   url = url.substring(0, url.length - 1);
                               }

                               return this.href == url;
                           }).parent();

                           if (element.is('li')) {
                               element.addClass('active');
                           }
                       });
                   </script>

PS I am using regex to parse HTML because the HTML parser I am forced to use (xmlpath) breaks when there are complex scripts on the page. I am using this regex to remove scripts from the page before passing it to the parser.

Answer 1

You need to tell perl not to break up each line of the file into its own separate record with -0 .

 perl -0 -pe 's/<script.*?>[\s\S]*?< *?\/ *?script *?>//g' tmp.html

This actually tells perl to break up records on '\\0' . perl -0777 will very explicitly slurp the whole file.

Answer 2

By the way, because I find slurping whole files distasteful, and because I don't care what html has to say about line breaks...a quicker, cleaner, more correct way to do this IF you can guarantee there is no important content on <script> tag lines is:

perl -ne 'print if !(/<script>/../<\/script>/)' tmp.html

(modifying the two regexes to your fancy, of course) .. is a stateful operator that is flipped on by the expression before it being true and off by the one after being true.

~/test£ cat example.html
<important1/>
<edgecase1/><script></script><edgecase2/>
<important2/>
<script></script>
<important3/>
<script>
<notimportant/>
</script>

~/test£ perl -ne 'print if !(/<script>/../<\/script>/)' example.html
<important1/>
<important2/>
<important3/>

And to (mostly) address content on script tag lines but outside tags:

~/test£ perl -ne 'print if !(/<script>/../<\/script>/);print "$1\n" if /(.+)<script>/;print "$1\n" if /<\/script>(.+)/;' example.html
<important1/>
<edgecase1/>
<edgecase2/>
<important2/>
<important3/>