Merge "Fix 'Tags' padding to keep it farther from the edge and document the source...

[lhc/web/wiklou.git] / languages / data / CrhExceptions.php

diff --git a/languages/data/CrhExceptions.php b/languages/data/CrhExceptions.php
index c759220..fcba6dc 100644 (file)
--- a/languages/data/CrhExceptions.php
+++ b/languages/data/CrhExceptions.php
@@ -126,7 +126,6 @@ class CrhExceptions {
                'beyude' => 'бейуде', 'beyüde' => 'бейуде',
                'curat' => 'джурьат', 'cürat' => 'джурьат',
                'mesul' => 'месуль', 'mesül' => 'месуль',
-               'yetsin' => 'етсин', 'etsin' => 'етсин',
        ];
 
        # map Cyrillic to Latin and back, simple string match only (no regex)
@@ -367,7 +366,7 @@ class CrhExceptions {
                'козь' => 'köz', '-юнджи' => '-ünci', '-юнджиде' => '-üncide', '-юнджиден' => '-ünciden',
 
                # originally L2C, here swapped
-               'еÑ\82Ñ\81ин' => 'etsin', 'лÑ\8cнаÑ\8f' => 'lnaya', 'лÑ\8cное' => 'lnoye', 'лÑ\8cнÑ\8bй' => 'lnıy', 'лÑ\8cний' => 'lniy',
+               'льная' => 'lnaya', 'льное' => 'lnoye', 'льный' => 'lnıy', 'льний' => 'lniy',
                'льская' => 'lskaya', 'льский' => 'lskiy', 'льское' => 'lskoye', 'ополь' => 'opol',
                'щее' => 'şçeye', 'щий' => 'şçiy', 'щая' => 'şçaya', 'цепс' => 'tseps',
 
@@ -389,8 +388,8 @@ class CrhExceptions {
                'му([иэИЭ])' => 'mü$1',
 
                # originally L2C, here swapped
-               'роль$1' => 'rol([^ü])',
-               'усть$1' => 'üst([^ü])',
+               'роль$1' => 'rol([^ü]|'.self::WB.')',
+               'усть$1' => 'üst([^ü]|'.self::WB.')',
 
                # more prefixes
                'ком-кок' => 'köm-kök',
@@ -460,6 +459,10 @@ class CrhExceptions {
                        '/'.self::WB.'Джонкю'.self::WB.'/u' => 'Cönkü',
                        '/'.self::WB.'ДЖОНКЮ'.self::WB.'/u' => 'CÖNKÜ',
 
+                       '/'.self::WB.'куркчи/u' => 'kürkçi',
+                       '/'.self::WB.'Куркчи/u' => 'Kürkçi',
+                       '/'.self::WB.'КУРКЧИ/u' => 'KÜRKÇI',
+
                        '/'.self::WB.'устке'.self::WB.'/u' => 'üstke',
                        '/'.self::WB.'Устке'.self::WB.'/u' => 'Üstke',
                        '/'.self::WB.'УСТКЕ'.self::WB.'/u' => 'ÜSTKE',
@@ -615,13 +618,21 @@ class CrhExceptions {
                        '/'.self::WB.'Mer'.self::WB.'/u' => 'Мэр',
                        '/'.self::WB.'MER'.self::WB.'/u' => 'МЭР',
 
-                       '/'.self::WB.'джонк/u' => 'cönk',
-                       '/'.self::WB.'Джонк/u' => 'Cönk',
-                       '/'.self::WB.'ДЖОНК/u' => 'CÖNK',
+                       '/'.self::WB.'cönk/u' => 'джонк',
+                       '/'.self::WB.'Cönk/u' => 'Джонк',
+                       '/'.self::WB.'CÖNK/u' => 'ДЖОНК',
 
-                       '/'.self::WB.'куркчи/u' => 'kürkçi',
-                       '/'.self::WB.'Куркчи/u' => 'Kürkçi',
-                       '/'.self::WB.'КУРКЧИ/u' => 'KÜRKÇI',
+                       # (y)etsin -> етсин/этсин
+                       # note that target starts with CYRILLIC е/Е!
+                       '/yetsin/u' => 'етсин',
+                       '/Yetsin/u' => 'Етсин',
+                       '/YETSİN/u' => 'ЕТСИН',
+
+                       # note that target starts with LATIN e/E!
+                       # (other transformations will determine CYRILLIC е/э as needed)
+                       '/etsin/u' => 'eтсин',
+                       '/Etsin/u' => 'Eтсин',
+                       '/ETSİN/u' => 'EТСИН',
 
                        # буква Ё - первый заход
                        # расставляем Ь после согласных
@@ -666,10 +677,6 @@ class CrhExceptions {
                        '/(['.Crh::L_F.'])l(['.Crh::L_CONS_LC.']|'.self::WB.')/u' => '$1ль$2',
                        '/(['.Crh::L_F_UC.'])L(['.Crh::L_CONS.']|'.self::WB.')/u' => '$1ЛЬ$2',
 
-                       '/etsin'.self::WB.'/u' => 'етсин',
-                       '/Etsin'.self::WB.'/u' => 'Етсин',
-                       '/ETSİN'.self::WB.'/u' => 'ЕТСИН',
-
                        # относятся к началу слова
                        '/'.self::WB.'ts/u' => 'ц',
                        '/'.self::WB.'T[sS]/u' => 'Ц',