http://qs1969.pair.com?node_id=11140633


in reply to Replace empty alt tag on <img> tag

<image href = "images/cron1.png"><<alt>></alt></image>

More than a single example would be better. Note that <image> is "... an ancient and poorly supported precursor to the <img> element. It should not be used." whereas <img> is an empty element, meaning it can't have content. <img hrefsrc="images/cron1.png" alt="" /> would be a valid HTML example.

But assuming your example is accurate, i.e. it is not actually HTML (and ignoring the likely erroneous <<alt>>), you can use Mojo::DOM in XML mode to parse your data. I've made a guess that by "empty" you mean that it does not contain other tags or non-whitespace text, but please clarify this as well.

use warnings; use strict; use Mojo::DOM; my $html = <<'END_HTML'; <image href = "images/cron1.png"></image> <image href = "images/cron2.png"> </image> <image href = "images/cron3.png">abc</image> <image href = "images/cron4.png"><alt></alt></image> <image href = "images/cron5.png"><alt><br/></alt></image> <image href = "images/cron6.png"><i><alt> </alt></i></image> <image href = "images/cron7.png"><alt>xyz</alt></image> <image href = "images/cron8.png"><alt><b>xyz</b>ijk</alt></image> <image href = "images/cron9.png">abc <alt><!-- foo --></alt> def</imag +e> END_HTML my $dom = Mojo::DOM->new->xml(1)->parse($html); $dom->find('image alt')->grep(sub { # find all <alt> tags inside <imag +e> not $_->child_nodes->grep(sub { # check whether they have content $_->type eq 'text' || $_->type eq 'cdata' ? $_->content=~/\S/ : $_->type ne 'comment' })->size; })->map('remove'); print $dom; __END__ <image href="images/cron1.png" /> <image href="images/cron2.png"> </image> <image href="images/cron3.png">abc</image> <image href="images/cron4.png" /> <image href="images/cron5.png"><alt><br /></alt></image> <image href="images/cron6.png"><i /></image> <image href="images/cron7.png"><alt>xyz</alt></image> <image href="images/cron8.png"><alt><b>xyz</b>ijk</alt></image> <image href="images/cron9.png">abc def</image>

Minor edits; changed selector from 'image > alt' to 'image alt' to find all descendants. Also: Comments are now not considered as content.