improved whitespace trimming in Mojo::DOM

mojolicious · Jun 17, 2011 · 0d69830 · 0d69830
1 parent 314fec1
commit 0d69830
Show file tree

Hide file tree

Showing 3 changed files with 20 additions and 2 deletions.
diff --git a/Changes b/Changes
@@ -4,11 +4,14 @@ This file documents the revision history for Perl extension Mojolicious.
         - Added EXPERIMENTAL self-restarting Morbo development web server and
           removed old "--reload" support since there have been too many
           negative side effects.
+          This also improves overall performance by about 5-10% and reduces
+          memory usage by about 10%.
         - Added EXPERIMENTAL application mount plugin.
         - Updated prettify.js to version 1-Jun-2011.
         - Updated WebSocket diagnostics test in Mojo::HelloWorld for latest
           Firefox Aurora.
         - Improved inline template and static file performance.
+        - Improved whitespace trimming in Mojo::DOM. (sri, DaTa)
         - Improved documentation.
         - Improved tests.
         - Fixed Morbo file discovery bug. (crab)

diff --git a/lib/Mojo/DOM.pm b/lib/Mojo/DOM.pm
@@ -1280,7 +1280,7 @@ sub _trim {
   $text =~ s/\s*\n+\s*/\ /g;
 
   # Add leading whitespace
-  $text = " $text" if $ws;
+  $text = " $text" if $ws && $text =~ /^[^\,\.\;]/;
 
   return $text;
 }

diff --git a/t/mojo/dom.t b/t/mojo/dom.t
@@ -5,7 +5,7 @@ use warnings;
 
 use utf8;
 
-use Test::More tests => 591;
+use Test::More tests => 595;
 
 # "Homer gave me a kidney: it wasn't his, I didn't need it,
 #  and it came postage due- but I appreciated the gesture!"
@@ -1309,6 +1309,21 @@ is $dom->find('body > ul > li > p')->[2]->text, '',           'no text';
 is $dom->find('body > ul > li')->[2]->all_text, 'Test 3 2 1', 'right text';
 is $dom->find('body > ul > li > p')->[2]->all_text, '', 'no text';
 
+# Advanced whitespace trimming
+$dom = Mojo::DOM->new->parse(<<EOF);
+<html>
+  <head>
+    <title>Real World!</title>
+  <body>
+    <div>foo <strong>bar</strong>.</div>
+    <div>foo<strong>, bar</strong>baz<strong>; yada</strong>.</div>
+EOF
+is $dom->find('html > head > title')->[0]->text, 'Real World!', 'right text';
+is $dom->find('body > div')->[0]->all_text,      'foo bar.',    'right text';
+is $dom->find('body > div')->[1]->all_text, 'foo, bar baz; yada.',
+  'right text';
+is $dom->find('body > div')->[1]->text, 'foo baz.', 'right text';
+
 # Real world JavaScript and CSS
 $dom = Mojo::DOM->new->parse(<<EOF);
 <html>