http://qs1969.pair.com?node_id=11130401


in reply to extracting sub elements from DOM by class

I must admit: I dont understand the DOM and scraping websites is a pain if you dont know it.

The basics of DOM are actually not all too difficult - it's basically a tree structure with nodes of different types. They're often represented as objects with a base "node" class that supports methods like "what are the children of this node", and the different node types are implemented as subclasses of this node (XML::LibXML works this way; Mojo::DOM AFAIK doesn't, but these are just implementation details). The two most common are "element" nodes, that represent <elements>s (including their attributes), and text nodes, that represent any text in between elements. There's also "comment" nodes that represent <!-- comments -->, etc.

In my experience, probably one of the most common things to confuse people is that this structure is very formal and rigid, asking a question like "what is the text content of <p>Hello, <b>cool</b> World!</p>?" is not as obvious as one might think. This <p> element has three children: the text "Hello, ", the element <b>, and the text " World!". To get all the text content means to walk down the tree and include the text child node "cool" of the <b> element too. Most libraries have functions that do this for you though.

Anyway, one nice thing about Mojo::DOM is that it supports CSS selectors. This is related to the DOM of course, but actually simplifies finding things in the DOM a lot. They're a little bit like a more flexible XPath. See Mojo::DOM::CSS: ids can be selected via #idname and classes can be selected via .classname, with automatic handling of multiple classes, e.g. your class="fa fa fa-mobile-phone" can be selected via e.g. .fa-mobile-phone or perhaps .fa.fa-mobile-phone, though interestingly I don't see a mention of the latter in the docs (it's in the W3C specs though).

Your HTML appears to be structured as a class="item-list" with <div class="item">s containing the data, so that's what I'd start with. What I think is quite strange is <span><i class="fa fa fa-phone"></i>011111111</span>, it's unclear to me why the class="fa fa fa-phone" isn't on the <span> that actually contains the data but is instead on the empty <i> in front of it. But oh well, we can deal with that too. (Update: Oh, they're Font Awesome icons.)

use Mojo::Base -strict, -signatures; use Mojo::DOM; use Mojo::Util qw/trim dumper/; my $dom = Mojo::DOM->new( do { local $/; <DATA> } ); my %members; $dom->find('#members-list .item')->map(sub { # assume only one .item-title (use ->find instead of ->at otherwis +e) my $name = trim( $_->at('.item-title')->all_text ); $_->find('.woffice-xprofile-list .fa')->map(sub { my $class = $_->attr('class'); # go up one node from the <i> to the <span> my $content = $_->parent->all_text; # assume no duplicates $members{$name}{$class} = $content; }); }); print dumper(\%members);