git.oblomov.eu Git - wine/blob - dlls/dwrite/analyzer.c

   1 /*
   2  *    Text analyzer
   3  *
   4  * Copyright 2012 Nikolay Sivov for CodeWeavers
   5  *
   6  * This library is free software; you can redistribute it and/or
   7  * modify it under the terms of the GNU Lesser General Public
   8  * License as published by the Free Software Foundation; either
   9  * version 2.1 of the License, or (at your option) any later version.
  10  *
  11  * This library is distributed in the hope that it will be useful,
  12  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  13  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
  14  * Lesser General Public License for more details.
  15  *
  16  * You should have received a copy of the GNU Lesser General Public
  17  * License along with this library; if not, write to the Free Software
  18  * Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301, USA
  19  */
  20
  21 #define COBJMACROS
  22
  23 #include "dwrite.h"
  24 #include "dwrite_private.h"
  25
  26 #include "wine/debug.h"
  27
  28 WINE_DEFAULT_DEBUG_CHANNEL(dwrite);
  29
  30 enum scriptcode {
  31     Script_Arabic = 0,
  32     Script_Armenian = 1,
  33     Script_Bengali = 3,
  34     Script_Canadian = 8,
  35     Script_Cherokee = 11,
  36     Script_Controls = 12,
  37     Script_Coptic = 13,
  38     Script_Cyrillic = 16,
  39     Script_Devanagari = 18,
  40     Script_Ethiopic = 19,
  41     Script_Georgian = 20,
  42     Script_Greek = 23,
  43     Script_Gujarati = 24,
  44     Script_Gurmukhi = 25,
  45     Script_Hangul = 27,
  46     Script_Hebrew = 29,
  47     Script_Kannada = 32,
  48     Script_Khmer = 36,
  49     Script_Lao = 37,
  50     Script_Latin  = 38,
  51     Script_Malayalam = 44,
  52     Script_Mongolian = 45,
  53     Script_Myanmar = 46,
  54     Script_NKo = 48,
  55     Script_Ogham = 49,
  56     Script_Oriya = 53,
  57     Script_Runic = 58,
  58     Script_Sinhala = 61,
  59     Script_Syriac = 64,
  60     Script_Tamil = 68,
  61     Script_Telugu = 69,
  62     Script_Thaana = 70,
  63     Script_Thai = 71,
  64     Script_Tibetan = 72,
  65     Script_Symbol = 77,
  66     Script_Unknown = (UINT16)-1
  67 };
  68
  69 struct script_range {
  70     UINT16 script;
  71     DWORD first;
  72     DWORD last;
  73 };
  74
  75 static const struct script_range script_ranges[] = {
  76     /* C0 Controls: U+0000–U+001F */
  77     /* ASCII punctuation and symbols: U+0020–U+002F */
  78     /* ASCII digits: U+0030–U+0039 */
  79     /* ASCII punctuation and symbols: U+003A–U+0040 */
  80     { Script_Symbol, 0x00, 0x040 },
  81     /* Latin uppercase: U+0041–U+005A */
  82     { Script_Latin, 0x41, 0x5a },
  83     /* ASCII punctuation and symbols: U+005B–U+0060 */
  84     { Script_Symbol, 0x5b, 0x060 },
  85     /* Latin lowercase: U+0061–U+007A */
  86     { Script_Latin, 0x61, 0x7a },
  87     /* ASCII punctuation and symbols, control char DEL: U+007B–U+007F */
  88     { Script_Symbol, 0x7b, 0x7f },
  89     /* C1 Controls: U+0080–U+009F */
  90     { Script_Controls, 0x80, 0x9f },
  91     /* Latin-1 Supplement: U+00A0–U+00FF */
  92     /* Latin Extended-A: U+0100–U+017F */
  93     /* Latin Extended-B: U+0180–U+024F */
  94     /* IPA Extensions: U+0250–U+02AF */
  95     /* Spacing Modifier Letters: U+02B0–U+02FF */
  96     { Script_Latin, 0xa0, 0x2ff },
  97     /* Combining Diacritical Marks: U+0300–U+036F */
  98     { Script_Symbol, 0x300, 0x36f },
  99     /* Greek: U+0370–U+03E1 */
 100     { Script_Greek, 0x370, 0x3e1 },
 101     /* Coptic: U+03E2–U+03Ef */
 102     { Script_Coptic, 0x3e2, 0x3ef },
 103     /* Greek: U+03F0–U+03FF */
 104     { Script_Greek, 0x3f0, 0x3ff },
 105     /* Cyrillic: U+0400–U+04FF */
 106     /* Cyrillic Supplement: U+0500–U+052F */
 107     /* Cyrillic Supplement range is incomplete cause it's based on Unicode 5.2
 108        that doesn't define some Abkhaz and Azerbaijani letters, we support Unicode 6.0 range here */
 109     { Script_Cyrillic, 0x400, 0x52f },
 110     /* Armenian: U+0530–U+058F */
 111     { Script_Armenian, 0x530, 0x58f },
 112     /* Hebrew: U+0590–U+05FF */
 113     { Script_Hebrew, 0x590, 0x5ff },
 114     /* Arabic: U+0600–U+06FF */
 115     { Script_Arabic, 0x600, 0x6ff },
 116     /* Syriac: U+0600–U+06FF */
 117     { Script_Syriac, 0x700, 0x74f },
 118     /* Arabic Supplement: U+0750–U+077F */
 119     { Script_Arabic, 0x750, 0x77f },
 120     /* Thaana: U+0780–U+07BF */
 121     { Script_Thaana, 0x780, 0x7bf },
 122     /* N'Ko: U+07C0–U+07FF */
 123     { Script_NKo, 0x7c0, 0x7ff },
 124     /* Devanagari: U+0900–U+097F */
 125     { Script_Devanagari, 0x900, 0x97f },
 126     /* Bengali: U+0980–U+09FF */
 127     { Script_Bengali, 0x980, 0x9ff },
 128     /* Gurmukhi: U+0A00–U+0A7F */
 129     { Script_Gurmukhi, 0xa00, 0xa7f },
 130     /* Gujarati: U+0A80–U+0AFF */
 131     { Script_Gujarati, 0xa80, 0xaff },
 132     /* Oriya: U+0B00–U+0B7F */
 133     { Script_Oriya, 0xb00, 0xb7f },
 134     /* Tamil: U+0B80–U+0BFF */
 135     { Script_Tamil, 0xb80, 0xbff },
 136     /* Telugu: U+0C00–U+0C7F */
 137     { Script_Telugu, 0xc00, 0xc7f },
 138     /* Kannada: U+0C80–U+0CFF */
 139     { Script_Kannada, 0xc80, 0xcff },
 140     /* Malayalam: U+0D00–U+0D7F */
 141     { Script_Malayalam, 0xd00, 0xd7f },
 142     /* Sinhala: U+0D80–U+0DFF */
 143     { Script_Sinhala, 0xd80, 0xdff },
 144     /* Thai: U+0E00–U+0E7F */
 145     { Script_Thai, 0xe00, 0xe7f },
 146     /* Lao: U+0E80–U+0EFF */
 147     { Script_Lao, 0xe80, 0xeff },
 148     /* Tibetan: U+0F00–U+0FFF */
 149     { Script_Tibetan, 0xf00, 0xfff },
 150     /* Myanmar: U+1000–U+109F */
 151     { Script_Myanmar, 0x1000, 0x109f },
 152     /* Georgian: U+10A0–U+10FF */
 153     { Script_Georgian, 0x10a0, 0x10ff },
 154     /* Hangul Jamo: U+1100–U+11FF */
 155     { Script_Hangul, 0x1100, 0x11ff },
 156     /* Ethiopic: U+1200–U+137F */
 157     /* Ethiopic Extensions: U+1380–U+139F */
 158     { Script_Ethiopic, 0x1200, 0x139f },
 159     /* Cherokee: U+13A0–U+13FF */
 160     { Script_Cherokee, 0x13a0, 0x13ff },
 161     /* Canadian Aboriginal Syllabics: U+1400–U+167F */
 162     { Script_Canadian, 0x1400, 0x167f },
 163     /* Ogham: U+1680–U+169F */
 164     { Script_Ogham, 0x1680, 0x169f },
 165     /* Runic: U+16A0–U+16F0 */
 166     { Script_Runic, 0x16a0, 0x16f0 },
 167     /* Khmer: U+1780–U+17FF */
 168     { Script_Khmer, 0x1780, 0x17ff },
 169     /* Mongolian: U+1800–U+18AF */
 170     { Script_Mongolian, 0x1800, 0x18af},
 171     /* unsupported range */
 172     { Script_Unknown }
 173 };
 174
 175 static UINT16 get_char_script( WCHAR c )
 176 {
 177     DWORD ch = c;
 178     int i;
 179
 180     for (i = 0; i < sizeof(script_ranges)/sizeof(struct script_range); i++)
 181     {
 182         const struct script_range *range = &script_ranges[i];
 183         if (range->script == Script_Unknown || (range->first <= ch && range->last >= ch))
 184             return range->script;
 185     }
 186
 187     return Script_Unknown;
 188 }
 189
 190 static HRESULT analyze_script(const WCHAR *text, UINT32 len, IDWriteTextAnalysisSink *sink)
 191 {
 192     DWRITE_SCRIPT_ANALYSIS sa;
 193     UINT32 pos, i, length;
 194
 195     if (!len) return S_OK;
 196
 197     sa.script = get_char_script(*text);
 198     sa.shapes = DWRITE_SCRIPT_SHAPES_DEFAULT;
 199
 200     pos = 0;
 201     length = 1;
 202
 203     for (i = 1; i < len; i++)
 204     {
 205         UINT16 script = get_char_script(text[i]);
 206
 207         /* Script_Latin_Symb script type is ignored when preceded or followed by another script */
 208         if (sa.script == Script_Symbol) sa.script = script;
 209         if (script    == Script_Symbol) script = sa.script;
 210         /* this is a length of a sequence to be reported next */
 211         if (sa.script == script) length++;
 212
 213         if (sa.script != script)
 214         {
 215             HRESULT hr = IDWriteTextAnalysisSink_SetScriptAnalysis(sink, pos, length, &sa);
 216             if (FAILED(hr)) return hr;
 217             pos = i;
 218             length = 1;
 219             sa.script = script;
 220         }
 221     }
 222
 223     /* 1 length case or normal completion call */
 224     return IDWriteTextAnalysisSink_SetScriptAnalysis(sink, pos, length, &sa);
 225 }
 226
 227 static HRESULT WINAPI dwritetextanalyzer_QueryInterface(IDWriteTextAnalyzer *iface, REFIID riid, void **obj)
 228 {
 229     TRACE("(%s %p)\n", debugstr_guid(riid), obj);
 230
 231     if (IsEqualIID(riid, &IID_IUnknown) || IsEqualIID(riid, &IID_IDWriteTextAnalyzer))
 232     {
 233         *obj = iface;
 234         return S_OK;
 235     }
 236
 237     *obj = NULL;
 238     return E_NOINTERFACE;
 239
 240 }
 241
 242 static ULONG WINAPI dwritetextanalyzer_AddRef(IDWriteTextAnalyzer *iface)
 243 {
 244     return 2;
 245 }
 246
 247 static ULONG WINAPI dwritetextanalyzer_Release(IDWriteTextAnalyzer *iface)
 248 {
 249     return 1;
 250 }
 251
 252 static HRESULT WINAPI dwritetextanalyzer_AnalyzeScript(IDWriteTextAnalyzer *iface,
 253     IDWriteTextAnalysisSource* source, UINT32 position, UINT32 length, IDWriteTextAnalysisSink* sink)
 254 {
 255     const WCHAR *text;
 256     HRESULT hr;
 257     UINT32 len;
 258
 259     TRACE("(%p %u %u %p)\n", source, position, length, sink);
 260
 261     hr = IDWriteTextAnalysisSource_GetTextAtPosition(source, position, &text, &len);
 262     if (FAILED(hr)) return hr;
 263
 264     return analyze_script(text, len, sink);
 265 }
 266
 267 static HRESULT WINAPI dwritetextanalyzer_AnalyzeBidi(IDWriteTextAnalyzer *iface,
 268     IDWriteTextAnalysisSource* source, UINT32 position, UINT32 length, IDWriteTextAnalysisSink* sink)
 269 {
 270     FIXME("(%p %u %u %p): stub\n", source, position, length, sink);
 271     return E_NOTIMPL;
 272 }
 273
 274 static HRESULT WINAPI dwritetextanalyzer_AnalyzeNumberSubstitution(IDWriteTextAnalyzer *iface,
 275     IDWriteTextAnalysisSource* source, UINT32 position, UINT32 length, IDWriteTextAnalysisSink* sink)
 276 {
 277     FIXME("(%p %u %u %p): stub\n", source, position, length, sink);
 278     return E_NOTIMPL;
 279 }
 280
 281 static HRESULT WINAPI dwritetextanalyzer_AnalyzeLineBreakpoints(IDWriteTextAnalyzer *iface,
 282     IDWriteTextAnalysisSource* source, UINT32 position, UINT32 length, IDWriteTextAnalysisSink* sink)
 283 {
 284     FIXME("(%p %u %u %p): stub\n", source, position, length, sink);
 285     return E_NOTIMPL;
 286 }
 287
 288 static HRESULT WINAPI dwritetextanalyzer_GetGlyphs(IDWriteTextAnalyzer *iface,
 289     WCHAR const* text, UINT32 length, IDWriteFontFace* font_face, BOOL is_sideways,
 290     BOOL is_rtl, DWRITE_SCRIPT_ANALYSIS const* analysis, WCHAR const* locale,
 291     IDWriteNumberSubstitution* substitution, DWRITE_TYPOGRAPHIC_FEATURES const** features,
 292     UINT32 const* feature_range_len, UINT32 feature_ranges, UINT32 max_glyph_count,
 293     UINT16* clustermap, DWRITE_SHAPING_TEXT_PROPERTIES* text_props, UINT16* glyph_indices,
 294     DWRITE_SHAPING_GLYPH_PROPERTIES* glyph_props, UINT32* actual_glyph_count)
 295 {
 296     FIXME("(%s:%u %p %d %d %p %s %p %p %p %u %u %p %p %p %p %p): stub\n", debugstr_wn(text, length),
 297         length, font_face, is_sideways, is_rtl, analysis, debugstr_w(locale), substitution, features, feature_range_len,
 298         feature_ranges, max_glyph_count, clustermap, text_props, glyph_indices, glyph_props, actual_glyph_count);
 299     return E_NOTIMPL;
 300 }
 301
 302 static HRESULT WINAPI dwritetextanalyzer_GetGlyphPlacements(IDWriteTextAnalyzer *iface,
 303     WCHAR const* text, UINT16 const* clustermap, DWRITE_SHAPING_TEXT_PROPERTIES* props,
 304     UINT32 text_len, UINT16 const* glyph_indices, DWRITE_SHAPING_GLYPH_PROPERTIES const* glyph_props,
 305     UINT32 glyph_count, IDWriteFontFace * font_face, FLOAT fontEmSize, BOOL is_sideways, BOOL is_rtl,
 306     DWRITE_SCRIPT_ANALYSIS const* analysis, WCHAR const* locale, DWRITE_TYPOGRAPHIC_FEATURES const** features,
 307     UINT32 const* feature_range_len, UINT32 feature_ranges, FLOAT* glyph_advances, DWRITE_GLYPH_OFFSET* glyph_offsets)
 308 {
 309     FIXME("(%s %p %p %u %p %p %u %p %f %d %d %p %s %p %p %u %p %p): stub\n", debugstr_w(text),
 310         clustermap, props, text_len, glyph_indices, glyph_props, glyph_count, font_face, fontEmSize, is_sideways,
 311         is_rtl, analysis, debugstr_w(locale), features, feature_range_len, feature_ranges, glyph_advances, glyph_offsets);
 312     return E_NOTIMPL;
 313 }
 314
 315 static HRESULT WINAPI dwritetextanalyzer_GetGdiCompatibleGlyphPlacements(IDWriteTextAnalyzer *iface,
 316     WCHAR const* text, UINT16 const* clustermap, DWRITE_SHAPING_TEXT_PROPERTIES* props,
 317     UINT32 text_len, UINT16 const* glyph_indices, DWRITE_SHAPING_GLYPH_PROPERTIES const* glyph_props,
 318     UINT32 glyph_count, IDWriteFontFace * font_face, FLOAT fontEmSize, FLOAT pixels_per_dip,
 319     DWRITE_MATRIX const* transform, BOOL use_gdi_natural, BOOL is_sideways, BOOL is_rtl,
 320     DWRITE_SCRIPT_ANALYSIS const* analysis, WCHAR const* locale, DWRITE_TYPOGRAPHIC_FEATURES const** features,
 321     UINT32 const* feature_range_lengths, UINT32 feature_ranges, FLOAT* glyph_advances, DWRITE_GLYPH_OFFSET* glyph_offsets)
 322 {
 323     FIXME("(%s %p %p %u %p %p %u %p %f %f %p %d %d %d %p %s %p %p %u %p %p): stub\n", debugstr_w(text),
 324         clustermap, props, text_len, glyph_indices, glyph_props, glyph_count, font_face, fontEmSize, pixels_per_dip,
 325         transform, use_gdi_natural, is_sideways, is_rtl, analysis, debugstr_w(locale), features, feature_range_lengths,
 326         feature_ranges, glyph_advances, glyph_offsets);
 327     return E_NOTIMPL;
 328 }
 329
 330 static const struct IDWriteTextAnalyzerVtbl textanalyzervtbl = {
 331     dwritetextanalyzer_QueryInterface,
 332     dwritetextanalyzer_AddRef,
 333     dwritetextanalyzer_Release,
 334     dwritetextanalyzer_AnalyzeScript,
 335     dwritetextanalyzer_AnalyzeBidi,
 336     dwritetextanalyzer_AnalyzeNumberSubstitution,
 337     dwritetextanalyzer_AnalyzeLineBreakpoints,
 338     dwritetextanalyzer_GetGlyphs,
 339     dwritetextanalyzer_GetGlyphPlacements,
 340     dwritetextanalyzer_GetGdiCompatibleGlyphPlacements
 341 };
 342
 343 static IDWriteTextAnalyzer textanalyzer = { &textanalyzervtbl };
 344
 345 HRESULT get_textanalyzer(IDWriteTextAnalyzer **ret)
 346 {
 347     *ret = &textanalyzer;
 348     return S_OK;
 349 }