Merge "Add parameter to API modules to apply change tags to log entries"

[lhc/web/wiklou.git] / includes / collation / CollationFa.php

diff --git a/includes/collation/CollationFa.php b/includes/collation/CollationFa.php
index b7e45cc..9cce087 100644 (file)
--- a/includes/collation/CollationFa.php
+++ b/includes/collation/CollationFa.php
@@ -19,9 +19,12 @@
  */
 
 /**
- * Temporary workaround for incorrect collation of Persian language ('fa') in ICU (bug T139110).
+ * Temporary workaround for incorrect collation of Persian language ('fa') in ICU 52 (bug T139110).
  *
- * 'ا' and 'و' should not be considered the same letter for the purposes of collation in Persian.
+ * All of the following will be considered separate letters for category headings in Persian:
+ *  - Characters 'و' 'ا' (often appear at the beginning of words)
+ *  - Characters 'ٲ' 'ٳ' (may appear at the beginning of words in loanwords)
+ *  - Characters 'ء' 'ئ' (don't appear at the beginning of words, but it's easier to implement)
  *
  * @since 1.29
  */
@@ -34,11 +37,14 @@ class CollationFa extends IcuCollation {
        }
 
        public function getPrimarySortKey( $string ) {
-               $firstLetter = mb_substr( $string, 0, 1 );
-               if ( $firstLetter === 'و' || $firstLetter === 'ا' ) {
+               $primary = parent::getPrimarySortKey( $string );
+               // We have to use a tertiary sortkey for everything with the primary sortkey of 2627.
+               // Otherwise, the "Remove duplicate prefixes" logic in IcuCollation would remove them.
+               // This matches sortkeys for the following characters: ء ئ ا و ٲ ٳ
+               if ( substr( $primary, 0, 2 ) === "\x26\x27" ) {
+                       wfDebug( "Using tertiary sortkey for '$string'\n" );
                        return $this->tertiaryCollator->getSortKey( $string );
                }
-
-               return parent::getPrimarySortKey( $string );
+               return $primary;
        }
 }