Changeset 9963

Timestamp:

06/23/2015 01:55:40 PM (9 years ago)

Author:

boonebgorges

Message:

Improve the HTML-handling logic of bp_create_excerpt().

• The regular expression that detects tags should be less generous, to avoid false matches.
• Word boundary detection, as used when exact=false, should never return a space character inside of an HTML tag (such as the space in <a href="http://example.com">).

These changes fix some tag-parsing issues introduced in [9523].

Merges [9962] to the 2.3 branch for 2.2.3.

Fixes #6517.

Location:

branches/2.3

Files:

• . (modified) (1 prop)
• src/bp-core/bp-core-template.php (modified) (2 diffs)
• tests/phpunit/testcases/core/template/bpCreateExcerpt.php (modified) (1 diff)

branches/2.3/src/bp-core/bp-core-template.php

@@ -814 +814 @@
 
         // Find all the tags and HTML comments and put them in a stack for later use
-        preg_match_all( '/(<\/?([\w!].+?)[^>]*>)?([^<>]*)/', $text, $tags, PREG_SET_ORDER );
+        preg_match_all( '/(<\/?([\w+!]+)[^>]*>)?([^<>]*)/', $text, $tags, PREG_SET_ORDER );
 
         foreach ( $tags as $tag ) {
@@ -866 +866 @@
     // If $exact is false, we can't break on words
     if ( empty( $r['exact'] ) ) {
-        $spacepos = mb_strrpos( $truncate, ' ' );
+        // Find the position of the last space character not part of a tag.
+        preg_match_all( '/<[a-z\!\/][^>]*>/', $truncate, $truncate_tags, PREG_OFFSET_CAPTURE );
+        $rtruncate = strrev( $truncate );
+        $spacepos = false;
+        for ( $i = strlen( $rtruncate ) - 1; $i >= 0; $i-- ) {
+            if ( ' ' !== $rtruncate[ $i ] ) {
+                continue;
+            }
+
+            // Convert rpos to negative offset on forward-facing string.
+            $pos = -1 - $i;
+
+            // If there are no tags in the string, the first space found is the right one.
+            if ( empty( $truncate_tags[0] ) ) {
+                $spacepos = $pos;
+                break;
+            }
+
+            // Look at each tag to see if the space is inside of it.
+            foreach ( $truncate_tags[0] as $truncate_tag ) {
+                $start = $truncate_tag[1];
+                $end   = $start + strlen( $truncate_tag[0] );
+                if ( $pos > $start && $pos < $end ) {
+                    $spacepos = $pos;
+                    break 2;
+                }
+            }
+        }
+
         if ( false !== $spacepos ) {
             if ( $r['html'] ) {

branches/2.3/tests/phpunit/testcases/core/template/bpCreateExcerpt.php

@@ -62 +62 @@
         ) ) );
     }
+
+    /**
+     * @ticket BP6517
+     */
+    public function test_string_should_not_be_cut_mid_tag_when_exact_is_false() {
+        $text = '<p><span>Foo</span> <a href="http://example.com">Bar</a> Baz.</p><p>Foo Bar Baz</p>';
+        $actual = bp_create_excerpt( $text, 7, array(
+            'html' => true,
+            'ending' => '',
+            'exact' => false,
+        ) );
+        $this->assertSame( '<p><span>Foo</span> <a href="http://example.com">Bar</a></p>', $actual );
+    }
 }