From: Brion Vibber <brion@users.mediawiki.org>
Date: Thu, 8 May 2008 05:34:48 +0000 (+0000)
Subject: A bunch of Unicode constants got moved at some point out of the Unicode normalization... 
X-Git-Tag: 1.31.0-rc.0~47829
X-Git-Url: https://git.cyclocoop.org/%242?a=commitdiff_plain;h=4b4ae60c52b673f065bb4a46bfd5d2c0f71fee58;p=lhc%2Fweb%2Fwiklou.git

A bunch of Unicode constants got moved at some point out of the Unicode normalization library into core Defines.php, breaking the data file generation & testing.
Moved them out to a separate file in the library which can be cleanly included from both places for transparent happiness.
A fresh rebuild & test of UtfNormal data via 'make test' now works fine.
---

diff --git a/includes/Defines.php b/includes/Defines.php
index b0275ec234..3cc248ba81 100644
--- a/includes/Defines.php
+++ b/includes/Defines.php
@@ -184,57 +184,7 @@ define( 'LIST_OR', 4);
 /**
  * Unicode and normalisation related
  */
-define( 'UNICODE_HANGUL_FIRST', 0xac00 );
-define( 'UNICODE_HANGUL_LAST',  0xd7a3 );
-
-define( 'UNICODE_HANGUL_LBASE', 0x1100 );
-define( 'UNICODE_HANGUL_VBASE', 0x1161 );
-define( 'UNICODE_HANGUL_TBASE', 0x11a7 );
-
-define( 'UNICODE_HANGUL_LCOUNT', 19 );
-define( 'UNICODE_HANGUL_VCOUNT', 21 );
-define( 'UNICODE_HANGUL_TCOUNT', 28 );
-define( 'UNICODE_HANGUL_NCOUNT', UNICODE_HANGUL_VCOUNT * UNICODE_HANGUL_TCOUNT );
-
-define( 'UNICODE_HANGUL_LEND', UNICODE_HANGUL_LBASE + UNICODE_HANGUL_LCOUNT - 1 );
-define( 'UNICODE_HANGUL_VEND', UNICODE_HANGUL_VBASE + UNICODE_HANGUL_VCOUNT - 1 );
-define( 'UNICODE_HANGUL_TEND', UNICODE_HANGUL_TBASE + UNICODE_HANGUL_TCOUNT - 1 );
-
-define( 'UNICODE_SURROGATE_FIRST', 0xd800 );
-define( 'UNICODE_SURROGATE_LAST', 0xdfff );
-define( 'UNICODE_MAX', 0x10ffff );
-define( 'UNICODE_REPLACEMENT', 0xfffd );
-
-
-define( 'UTF8_HANGUL_FIRST', "\xea\xb0\x80" /*codepointToUtf8( UNICODE_HANGUL_FIRST )*/ );
-define( 'UTF8_HANGUL_LAST', "\xed\x9e\xa3" /*codepointToUtf8( UNICODE_HANGUL_LAST )*/ );
-
-define( 'UTF8_HANGUL_LBASE', "\xe1\x84\x80" /*codepointToUtf8( UNICODE_HANGUL_LBASE )*/ );
-define( 'UTF8_HANGUL_VBASE', "\xe1\x85\xa1" /*codepointToUtf8( UNICODE_HANGUL_VBASE )*/ );
-define( 'UTF8_HANGUL_TBASE', "\xe1\x86\xa7" /*codepointToUtf8( UNICODE_HANGUL_TBASE )*/ );
-
-define( 'UTF8_HANGUL_LEND', "\xe1\x84\x92" /*codepointToUtf8( UNICODE_HANGUL_LEND )*/ );
-define( 'UTF8_HANGUL_VEND', "\xe1\x85\xb5" /*codepointToUtf8( UNICODE_HANGUL_VEND )*/ );
-define( 'UTF8_HANGUL_TEND', "\xe1\x87\x82" /*codepointToUtf8( UNICODE_HANGUL_TEND )*/ );
-
-define( 'UTF8_SURROGATE_FIRST', "\xed\xa0\x80" /*codepointToUtf8( UNICODE_SURROGATE_FIRST )*/ );
-define( 'UTF8_SURROGATE_LAST', "\xed\xbf\xbf" /*codepointToUtf8( UNICODE_SURROGATE_LAST )*/ );
-define( 'UTF8_MAX', "\xf4\x8f\xbf\xbf" /*codepointToUtf8( UNICODE_MAX )*/ );
-define( 'UTF8_REPLACEMENT', "\xef\xbf\xbd" /*codepointToUtf8( UNICODE_REPLACEMENT )*/ );
-#define( 'UTF8_REPLACEMENT', '!' );
-
-define( 'UTF8_OVERLONG_A', "\xc1\xbf" );
-define( 'UTF8_OVERLONG_B', "\xe0\x9f\xbf" );
-define( 'UTF8_OVERLONG_C', "\xf0\x8f\xbf\xbf" );
-
-# These two ranges are illegal
-define( 'UTF8_FDD0', "\xef\xb7\x90" /*codepointToUtf8( 0xfdd0 )*/ );
-define( 'UTF8_FDEF', "\xef\xb7\xaf" /*codepointToUtf8( 0xfdef )*/ );
-define( 'UTF8_FFFE', "\xef\xbf\xbe" /*codepointToUtf8( 0xfffe )*/ );
-define( 'UTF8_FFFF', "\xef\xbf\xbf" /*codepointToUtf8( 0xffff )*/ );
-
-define( 'UTF8_HEAD', false );
-define( 'UTF8_TAIL', true );
+require_once dirname(__FILE__).'/normal/UtfNormalDefines.php';
 
 # Hook support constants
 define( 'MW_SUPPORTS_EDITFILTERMERGED', 1 );
diff --git a/includes/normal/Makefile b/includes/normal/Makefile
index 887f3ce651..c7df302ef9 100644
--- a/includes/normal/Makefile
+++ b/includes/normal/Makefile
@@ -6,7 +6,7 @@
 #BASE=http://www.unicode.org/Public/UNIDATA
 
 # Explicitly using Unicode 5.0
-BASE=http://www.unicode.org/Public/5.0.0/ucd/
+BASE=http://www.unicode.org/Public/5.0.0/ucd
 
 # Can override to php-cli or php5 or whatevah
 PHP=php
diff --git a/includes/normal/UtfNormalDefines.php b/includes/normal/UtfNormalDefines.php
new file mode 100644
index 0000000000..419f6f8c1f
--- /dev/null
+++ b/includes/normal/UtfNormalDefines.php
@@ -0,0 +1,53 @@
+<?php
+
+define( 'UNICODE_HANGUL_FIRST', 0xac00 );
+define( 'UNICODE_HANGUL_LAST',  0xd7a3 );
+
+define( 'UNICODE_HANGUL_LBASE', 0x1100 );
+define( 'UNICODE_HANGUL_VBASE', 0x1161 );
+define( 'UNICODE_HANGUL_TBASE', 0x11a7 );
+
+define( 'UNICODE_HANGUL_LCOUNT', 19 );
+define( 'UNICODE_HANGUL_VCOUNT', 21 );
+define( 'UNICODE_HANGUL_TCOUNT', 28 );
+define( 'UNICODE_HANGUL_NCOUNT', UNICODE_HANGUL_VCOUNT * UNICODE_HANGUL_TCOUNT );
+
+define( 'UNICODE_HANGUL_LEND', UNICODE_HANGUL_LBASE + UNICODE_HANGUL_LCOUNT - 1 );
+define( 'UNICODE_HANGUL_VEND', UNICODE_HANGUL_VBASE + UNICODE_HANGUL_VCOUNT - 1 );
+define( 'UNICODE_HANGUL_TEND', UNICODE_HANGUL_TBASE + UNICODE_HANGUL_TCOUNT - 1 );
+
+define( 'UNICODE_SURROGATE_FIRST', 0xd800 );
+define( 'UNICODE_SURROGATE_LAST', 0xdfff );
+define( 'UNICODE_MAX', 0x10ffff );
+define( 'UNICODE_REPLACEMENT', 0xfffd );
+
+
+define( 'UTF8_HANGUL_FIRST', "\xea\xb0\x80" /*codepointToUtf8( UNICODE_HANGUL_FIRST )*/ );
+define( 'UTF8_HANGUL_LAST', "\xed\x9e\xa3" /*codepointToUtf8( UNICODE_HANGUL_LAST )*/ );
+
+define( 'UTF8_HANGUL_LBASE', "\xe1\x84\x80" /*codepointToUtf8( UNICODE_HANGUL_LBASE )*/ );
+define( 'UTF8_HANGUL_VBASE', "\xe1\x85\xa1" /*codepointToUtf8( UNICODE_HANGUL_VBASE )*/ );
+define( 'UTF8_HANGUL_TBASE', "\xe1\x86\xa7" /*codepointToUtf8( UNICODE_HANGUL_TBASE )*/ );
+
+define( 'UTF8_HANGUL_LEND', "\xe1\x84\x92" /*codepointToUtf8( UNICODE_HANGUL_LEND )*/ );
+define( 'UTF8_HANGUL_VEND', "\xe1\x85\xb5" /*codepointToUtf8( UNICODE_HANGUL_VEND )*/ );
+define( 'UTF8_HANGUL_TEND', "\xe1\x87\x82" /*codepointToUtf8( UNICODE_HANGUL_TEND )*/ );
+
+define( 'UTF8_SURROGATE_FIRST', "\xed\xa0\x80" /*codepointToUtf8( UNICODE_SURROGATE_FIRST )*/ );
+define( 'UTF8_SURROGATE_LAST', "\xed\xbf\xbf" /*codepointToUtf8( UNICODE_SURROGATE_LAST )*/ );
+define( 'UTF8_MAX', "\xf4\x8f\xbf\xbf" /*codepointToUtf8( UNICODE_MAX )*/ );
+define( 'UTF8_REPLACEMENT', "\xef\xbf\xbd" /*codepointToUtf8( UNICODE_REPLACEMENT )*/ );
+#define( 'UTF8_REPLACEMENT', '!' );
+
+define( 'UTF8_OVERLONG_A', "\xc1\xbf" );
+define( 'UTF8_OVERLONG_B', "\xe0\x9f\xbf" );
+define( 'UTF8_OVERLONG_C', "\xf0\x8f\xbf\xbf" );
+
+# These two ranges are illegal
+define( 'UTF8_FDD0', "\xef\xb7\x90" /*codepointToUtf8( 0xfdd0 )*/ );
+define( 'UTF8_FDEF', "\xef\xb7\xaf" /*codepointToUtf8( 0xfdef )*/ );
+define( 'UTF8_FFFE', "\xef\xbf\xbe" /*codepointToUtf8( 0xfffe )*/ );
+define( 'UTF8_FFFF', "\xef\xbf\xbf" /*codepointToUtf8( 0xffff )*/ );
+
+define( 'UTF8_HEAD', false );
+define( 'UTF8_TAIL', true );
diff --git a/includes/normal/UtfNormalUtil.php b/includes/normal/UtfNormalUtil.php
index 991e81cee2..39acc6d867 100644
--- a/includes/normal/UtfNormalUtil.php
+++ b/includes/normal/UtfNormalUtil.php
@@ -26,6 +26,7 @@
  */
 
 /** */
+require_once dirname(__FILE__).'/UtfNormalDefines.php';
 
 /**
  * Return UTF-8 sequence for a given Unicode code point.