Fix issue with HtmlDiff matching skipping some single word matches

iluuu1994 · Feb 2, 2016 · 4fb7b52 · 4fb7b52
1 parent 36467f2
commit 4fb7b52
Show file tree

Hide file tree

Showing 2 changed files with 16 additions and 6 deletions.
diff --git a/lib/Caxy/HtmlDiff/AbstractDiff.php b/lib/Caxy/HtmlDiff/AbstractDiff.php
@@ -224,11 +224,11 @@ protected function convertHtmlToListOfWords($characterString)
                     }
                     $current_word = "<";
                     $mode = 'tag';
-                } elseif ( preg_match( "[^\s]", $character ) > 0 ) {
-                    if ($current_word != '') {
+                } elseif (preg_match("/\s/", $character)) {
+                    if ($current_word !== '') {
                         $words[] = $current_word;
                     }
-                    $current_word = $character;
+                    $current_word = preg_replace('/\s+/S', ' ', $character);
                     $mode = 'whitespace';
                 } else {
                     if (
@@ -259,13 +259,14 @@ protected function convertHtmlToListOfWords($characterString)
                 break;
                 case 'whitespace':
                 if ( $this->isStartOfTag( $character ) ) {
-                    if ($current_word != '') {
+                    if ($current_word !== '') {
                         $words[] = $current_word;
                     }
                     $current_word = "<";
                     $mode = 'tag';
-                } elseif ( preg_match( "[^\s]", $character ) ) {
+                } elseif ( preg_match( "/\s/", $character ) ) {
                     $current_word .= $character;
+                    $current_word = preg_replace('/\s+/S', ' ', $current_word);
                 } else {
                     if ($current_word != '') {
                         $words[] = $current_word;

diff --git a/lib/Caxy/HtmlDiff/HtmlDiff.php b/lib/Caxy/HtmlDiff/HtmlDiff.php
@@ -468,7 +468,16 @@ protected function findMatch($startInOld, $endInOld, $startInNew, $endInNew)
                 }
                 $newMatchLength = ( isset( $matchLengthAt[ $indexInNew - 1 ] ) ? $matchLengthAt[ $indexInNew - 1 ] : 0 ) + 1;
                 $newMatchLengthAt[ $indexInNew ] = $newMatchLength;
-                if ($newMatchLength > $bestMatchSize) {
+                if ($newMatchLength > $bestMatchSize ||
+                    (
+                        $this->isGroupDiffs() &&
+                        $bestMatchSize > 0 &&
+                        preg_match(
+                            '/^\s+$/',
+                            implode('', array_slice($this->oldWords, $bestMatchInOld, $bestMatchSize))
+                        )
+                    )
+                ) {
                     $bestMatchInOld = $indexInOld - $newMatchLength + 1;
                     $bestMatchInNew = $indexInNew - $newMatchLength + 1;
                     $bestMatchSize = $newMatchLength;