Merge branch 'master' into use-trusty

Author: danielbachhuber

features/db-search.feature

@@ -613,6 +613,8 @@ Feature: Search through the database
   Scenario: Search with multibyte strings
     Given a WP install
     And I run `wp option update multibytetest 'あいうえおかきくけこさしすせとたちつてと'`
+    # Note ö is o with combining umlaut.
+    And I run `wp option update plaintst_combining 'lllllムnöppppp'`
 
     When I run `wp db search "かきくけこ" --before_context=0 --after_context=0`
     Then STDOUT should contain:
@@ -647,6 +649,38 @@ Feature: Search through the database
       :あいうえおかきくけこさしすせとたちつてと
       """
 
+    When I run `wp db search 'ppppp' --before_context=3 --after_context=4`
+    Then STDOUT should contain:
+      """
+      :ムnöppppp
+      """
+
+    When I run `wp db search 'ppppp' --before_context=1 --after_context=1`
+    Then STDOUT should contain:
+      """
+      :öppppp
+      """
+
+    When I run `wp db search 'ムn' --before_context=2 --after_context=1`
+    Then STDOUT should contain:
+      """
+      :llムnö
+      """
+    And STDOUT should not contain:
+      """
+      :llムnöp
+      """
+
+    When I run `wp db search 'ムn' --before_context=2 --after_context=2`
+    Then STDOUT should contain:
+      """
+      :llムnöp
+      """
+    And STDOUT should not contain:
+      """
+      :llムnöpp
+      """
+
   Scenario: Search with regular expressions
     Given a WP install
     And I run `wp option update regextst '12345é789あhttps://regextst.com1234567890123456789éhttps://regextst.com12345678901234567890regextst.com34567890t.com67890'`
@@ -714,15 +748,19 @@ Feature: Search through the database
       https://r
       """
 
-    # Bug: context counts combining characters as 2 characters.
     When I run `wp db search 'ppppp' --regex --before_context=3 --after_context=4`
     Then STDOUT should contain:
       """
-      :nöppppp
+      :ムnöppppp
       """
 
-    # Bug: context counts combining characters as 2 characters.
-    When I run `wp db search 'ムn' --regex --before_context=2 --after_context=2`
+    When I run `wp db search 'ppppp' --regex --before_context=1 --after_context=1`
+    Then STDOUT should contain:
+      """
+      :öppppp
+      """
+
+    When I run `wp db search 'ムn' --before_context=2 --after_context=1`
     Then STDOUT should contain:
       """
       :llムnö
@@ -732,6 +770,16 @@ Feature: Search through the database
       :llムnöp
       """
 
+    When I run `wp db search 'ムn' --regex --before_context=2 --after_context=2`
+    Then STDOUT should contain:
+      """
+      :llムnöp
+      """
+    And STDOUT should not contain:
+      """
+      :llムnöpp
+      """
+
     When I run `wp db search 't\.c' --regex --before_context=1 --after_context=1`
     Then STDOUT should contain:
       """

src/DB_Command.php

@@ -844,9 +844,11 @@ public function search( $args, $assoc_args ) {
 			}
 		} else {
 			$safe_search = preg_quote( $search, '#' );
-			$search_regex = '#(.{0,' . $before_context . '})(' . $safe_search .')(.{0,' . $after_context . '})#i';
 			if ( 0 === strpos( $wpdb->charset, 'utf8' ) ) {
-				$search_regex .= 'u';
+				$context_re = \cli\can_use_pcre_x() ? '\X' : '.';
+				$search_regex = '#(' . $context_re . '{0,' . $before_context . '})(' . $safe_search .')(' . $context_re . '{0,' . $after_context . '})#iu';
+			} else {
+				$search_regex = '#(.{0,' . $before_context . '})(' . $safe_search .')(.{0,' . $after_context . '})#i';
 			}
 			$esc_like_search = '%' . self::esc_like( $search ) . '%';
 		}
@@ -910,14 +912,13 @@ public function search( $args, $assoc_args ) {
 								foreach ( $matches[0] as $match_arr ) {
 									$match = $match_arr[0];
 									$offset = $match_arr[1];
-									// Offsets are in bytes, so need to use `strlen()` and  `substr()` before using `safe_substr()`. Note: not catering for combining chars.
+									// Offsets are in bytes, so need to use `strlen()` and  `substr()` before using `safe_substr()`.
 									$before = $before_context && $offset ? \cli\safe_substr( substr( $col_val, 0, $offset ), -$before_context, null /*length*/, false /*is_width*/, $encoding ) : '';
 									$after = $after_context ? \cli\safe_substr( substr( $col_val, $offset + strlen( $match ) ), 0, $after_context, false /*is_width*/, $encoding ) : '';
 									$bits[] = $before . $colors['match'][0] . $match . $colors['match'][1] . $after;
 								}
 							} else {
 								foreach ( $matches[0] as $key => $value ) {
-									// Note: not catering for combining chars.
 									$bits[] = $matches[1][ $key ] . $colors['match'][0] . $matches[2][ $key ] . $colors['match'][1] . $matches[3][ $key ];
 								}
 							}