dwrite: Support for Greek and Cyrillic ranges

[Nikolay Sivov]

Support for Greek and Cyrillic ranges
-------------- next part --------------
>From 10c849f7b81e63c36ea247681b3edfd8dfc6939b Mon Sep 17 00:00:00 2001
From: Nikolay Sivov <nsivov at codeweavers.com>
Date: Tue, 30 Oct 2012 11:53:19 -0400
Subject: [PATCH 1/1] Support for Greek and Cyrillic ranges

---
 dlls/dwrite/analyzer.c       |   30 +++++++++++++++++++++++-------
 dlls/dwrite/tests/analyzer.c |   29 +++++++++++++++++++++++++----
 2 files changed, 48 insertions(+), 11 deletions(-)

diff --git a/dlls/dwrite/analyzer.c b/dlls/dwrite/analyzer.c
index 3e22f84..541914b 100644
--- a/dlls/dwrite/analyzer.c
+++ b/dlls/dwrite/analyzer.c
@@ -30,8 +30,11 @@ WINE_DEFAULT_DEBUG_CHANNEL(dwrite);
 enum scriptcode {
     Script_Arabic = 0,
     Script_C1Controls = 12,
+    Script_Coptic = 13,
+    Script_Cyrillic = 16,
+    Script_Greek = 23,
     Script_Latin  = 38,
-    Script_Latin_Symb = 77,
+    Script_Symbol = 77,
     Script_Unknown = (UINT16)-1
 };
 
@@ -46,15 +49,15 @@ static const struct script_range script_ranges[] = {
     /* ASCII punctuation and symbols: U+0020???U+002F */
     /* ASCII digits: U+0030???U+0039 */
     /* ASCII punctuation and symbols: U+003A???U+0040 */
-    { Script_Latin_Symb, 0x00, 0x040 },
+    { Script_Symbol, 0x00, 0x040 },
     /* Latin uppercase: U+0041???U+005A */
     { Script_Latin, 0x41, 0x5a },
     /* ASCII punctuation and symbols: U+005B???U+0060 */
-    { Script_Latin_Symb, 0x5b, 0x060 },
+    { Script_Symbol, 0x5b, 0x060 },
     /* Latin lowercase: U+0061???U+007A */
     { Script_Latin, 0x61, 0x7a },
     /* ASCII punctuation and symbols, control char DEL: U+007B???U+007F */
-    { Script_Latin_Symb, 0x7b, 0x7f },
+    { Script_Symbol, 0x7b, 0x7f },
     /* C1 Controls: U+0080???U+009F */
     { Script_C1Controls, 0x80, 0x9f },
     /* Latin-1 Supplement: U+00A0???U+00FF */
@@ -63,9 +66,22 @@ static const struct script_range script_ranges[] = {
     /* IPA Extensions: U+0250???U+02AF */
     /* Spacing Modifier Letters: U+02B0???U+02FF */
     { Script_Latin, 0xa0, 0x2ff },
+    /* Combining Diacritical Marks: U+0300???U+036F */
+    { Script_Symbol, 0x300, 0x36f },
+    /* Greek: U+0370???U+03E1 */
+    { Script_Greek, 0x370, 0x3e1 },
+    /* Coptic: U+03E2???U+03Ef */
+    { Script_Coptic, 0x3e2, 0x3ef },
+    /* Greek: U+03F0???U+03FF */
+    { Script_Greek, 0x3f0, 0x3ff },
+    /* Cyrillic: U+0400???U+04FF */
+    /* Cyrillic Supplement: U+0500???U+052F */
+    /* Cyrillic Supplement range is incomplete cause it's based on Unicode 5.2
+       that doesn't define some Abkhaz and Azerbaijani letters, we support Unicode 6.0 range here */
+    { Script_Cyrillic, 0x400, 0x52f },
     /* Arabic: U+0600???U+06FF */
     { Script_Arabic, 0x600, 0x6ef },
-    /* unsuppoted range */
+    /* unsupported range */
     { Script_Unknown }
 };
 
@@ -102,8 +118,8 @@ static HRESULT analyze_script(const WCHAR *text, UINT32 len, IDWriteTextAnalysis
         UINT16 script = get_char_script(text[i]);
 
         /* Script_Latin_Symb script type is ignored when preceded or followed by another script */
-        if (sa.script == Script_Latin_Symb) sa.script = script;
-        if (script    == Script_Latin_Symb) script = sa.script;
+        if (sa.script == Script_Symbol) sa.script = script;
+        if (script    == Script_Symbol) script = sa.script;
         /* this is a length of a sequence to be reported next */
         if (sa.script == script) length++;
 
diff --git a/dlls/dwrite/tests/analyzer.c b/dlls/dwrite/tests/analyzer.c
index d68cbcd..c610357 100644
--- a/dlls/dwrite/tests/analyzer.c
+++ b/dlls/dwrite/tests/analyzer.c
@@ -399,8 +399,11 @@ struct sa_test {
 enum scriptcode {
     Script_Arabic = 0,
     Script_C1Controls = 12,
+    Script_Coptic = 13,
+    Script_Cyrillic = 16,
+    Script_Greek = 23,
     Script_Latin  = 38,
-    Script_Latin_Symb = 77
+    Script_Symbol = 77
 };
 
 static struct sa_test sa_tests[] = {
@@ -415,16 +418,16 @@ static struct sa_test sa_tests[] = {
     },
     {
       {' ',' ',' ',' ','!','$','[','^','{','~',0}, 1,
-          { { 0, 10, { Script_Latin_Symb, DWRITE_SCRIPT_SHAPES_DEFAULT } }}
+          { { 0, 10, { Script_Symbol, DWRITE_SCRIPT_SHAPES_DEFAULT } }}
     },
     {
       {' ',' ',' ','1','2',' ',0}, 1,
-          { { 0, 6, { Script_Latin_Symb, DWRITE_SCRIPT_SHAPES_DEFAULT } }}
+          { { 0, 6, { Script_Symbol, DWRITE_SCRIPT_SHAPES_DEFAULT } }}
     },
     {
       /* digits only */
       {'1','2',0}, 1,
-          { { 0, 2, { Script_Latin_Symb, DWRITE_SCRIPT_SHAPES_DEFAULT } }}
+          { { 0, 2, { Script_Symbol, DWRITE_SCRIPT_SHAPES_DEFAULT } }}
     },
     {
       /* Arabic */
@@ -484,6 +487,24 @@ static struct sa_test sa_tests[] = {
       {0x2b0,0x2ba,0x2d7,0x2dd,0x2ef,0x2ff,0}, 1,
           { { 0, 6, { Script_Latin, DWRITE_SCRIPT_SHAPES_DEFAULT } }}
     },
+    {
+      /* Combining Diacritical Marks */
+      {0x300,0x320,0x340,0x345,0x350,0x36f,0}, 1,
+          { { 0, 6, { Script_Symbol, DWRITE_SCRIPT_SHAPES_DEFAULT } }}
+    },
+    {
+      /* Greek and Coptic */
+      {0x370,0x388,0x3d8,0x3e1,0x3e2,0x3fa,0x3ff,0}, 3,
+          { { 0, 4, { Script_Greek, DWRITE_SCRIPT_SHAPES_DEFAULT } },
+            { 4, 1, { Script_Coptic, DWRITE_SCRIPT_SHAPES_DEFAULT } },
+            { 5, 2, { Script_Greek, DWRITE_SCRIPT_SHAPES_DEFAULT } }
+          }
+    },
+    {
+      /* Cyrillic and Cyrillic Supplement */
+      {0x400,0x40f,0x410,0x44f,0x450,0x45f,0x460,0x481,0x48a,0x4f0,0x4fa,0x4ff,0x500,0x510,0x520,0}, 1,
+          { { 0, 15, { Script_Cyrillic, DWRITE_SCRIPT_SHAPES_DEFAULT } }}
+    },
     /* keep this as end marker */
     { {0} }
 };
-- 
1.7.10.4