Changeset 9968

Timestamp:

06/25/2015 06:50:14 PM (9 years ago)

Author:

boonebgorges

Message:

The continuing adventures of bp_create_excerpt().

This changeset fixes a bug introduced in [9963] that caused excerpts to be
truncated too much in some cases. It also fixes some potential issues with
multibyte strings when html=true.

Merges [9967] to the 2.3 branch for 2.3.3.

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

@@ -867 +867 @@
     if ( empty( $r['exact'] ) ) {
         // 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 ] ) {
+        preg_match_all( '/<[a-z\!\/][^>]*>/', $truncate, $_truncate_tags, PREG_OFFSET_CAPTURE );
+
+        // Rekey tags by the string index of their last character.
+        $truncate_tags = array();
+        if ( ! empty( $_truncate_tags[0] ) ) {
+            foreach ( $_truncate_tags[0] as $_tt ) {
+                $_tt['start'] = $_tt[1];
+                $_tt['end']   = $_tt[1] + strlen( $_tt[0] );
+                $truncate_tags[ $_tt['end'] ] = $_tt;
+            }
+        }
+
+        $truncate_length = mb_strlen( $truncate );
+        $spacepos = $truncate_length + 1;
+        for ( $pos = $truncate_length - 1; $pos >= 0; $pos-- ) {
+            // Word boundaries are spaces and the close of HTML tags, when the tag is preceded by a space.
+            $is_word_boundary = ' ' === $truncate[ $pos ];
+            if ( ! $is_word_boundary && isset( $truncate_tags[ $pos - 1 ] ) ) {
+                $preceding_tag    = $truncate_tags[ $pos - 1 ];
+                if ( ' ' === $truncate[ $preceding_tag['start'] - 1 ] ) {
+                    $is_word_boundary = true;
+                    break;
+                }
+            }
+
+            if ( ! $is_word_boundary ) {
                 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] ) ) {
+            if ( empty( $truncate_tags ) ) {
                 $spacepos = $pos;
                 break;
@@ -885 +903 @@
 
             // 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;
+            $intag = false;
+            foreach ( $truncate_tags as $tt ) {
+                if ( $pos > $tt['start'] && $pos < $tt['end'] ) {
+                    $intag = true;
+                    break;
                 }
             }
-        }
-
-        if ( false !== $spacepos ) {
-            if ( $r['html'] ) {
-                $bits = mb_substr( $truncate, $spacepos );
-                preg_match_all( '/<\/([a-z]+)>/', $bits, $droppedTags, PREG_SET_ORDER );
-                if ( !empty( $droppedTags ) ) {
-                    foreach ( $droppedTags as $closingTag ) {
-                        if ( !in_array( $closingTag[1], $openTags ) ) {
-                            array_unshift( $openTags, $closingTag[1] );
-                        }
+
+            if ( ! $intag ) {
+                $spacepos = $pos;
+                break;
+            }
+        }
+
+        if ( $r['html'] ) {
+            $bits = mb_substr( $truncate, $spacepos );
+            preg_match_all( '/<\/([a-z]+)>/', $bits, $droppedTags, PREG_SET_ORDER );
+            if ( !empty( $droppedTags ) ) {
+                foreach ( $droppedTags as $closingTag ) {
+                    if ( !in_array( $closingTag[1], $openTags ) ) {
+                        array_unshift( $openTags, $closingTag[1] );
                     }
                 }
             }
-            $truncate = mb_substr( $truncate, 0, $spacepos );
-        }
+        }
+
+        $truncate = rtrim( mb_substr( $truncate, 0, $spacepos ) );
     }
     $truncate .= $ending;

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

@@ -52 +52 @@
 
     /**
+     * @ticket BP6517
+     */
+    public function test_exact_false_should_properly_account_for_accented_characters() {
+        $text = 'Toutes les connaissances que les hommes avaient mis sur Internet lui étaient accessible. Les grandes bibliothèques du monde entier n’avaient plus de secret pour lui. Il pouvait apprendre très vite, beaucoup plus vite que n’importe quel humain.
+Il avait appris toutes les connaissances du monde entier, visiter tout les pays. C’est lui qui avait fait en sorte qu’Internet se déploie ainsi.';
+        $expected = 'Toutes les connaissances que les hommes avaient mis sur Internet lui étaient accessible. Les';
+        $this->assertSame( $expected, bp_create_excerpt( $text, 98, array(
+            'ending' => '',
+            'exact' => false,
+        ) ) );
+    }
+
+    /**
      * @ticket BP6254
      */