harfbuzz/src/hb-icu.cc

400 lines
12 KiB
C++
Raw Normal View History

2009-11-03 22:35:10 +01:00
/*
2011-04-21 23:14:28 +02:00
* Copyright © 2009 Red Hat, Inc.
* Copyright © 2009 Keith Stribley
* Copyright © 2011 Google, Inc.
2009-11-03 22:35:10 +01:00
*
2010-04-22 06:11:43 +02:00
* This is part of HarfBuzz, a text shaping library.
2009-11-03 22:35:10 +01:00
*
* Permission is hereby granted, without written agreement and without
* license or royalty fees, to use, copy, modify, and distribute this
* software and its documentation for any purpose, provided that the
* above copyright notice and the following two paragraphs appear in
* all copies of this software.
*
* IN NO EVENT SHALL THE COPYRIGHT HOLDER BE LIABLE TO ANY PARTY FOR
* DIRECT, INDIRECT, SPECIAL, INCIDENTAL, OR CONSEQUENTIAL DAMAGES
* ARISING OUT OF THE USE OF THIS SOFTWARE AND ITS DOCUMENTATION, EVEN
* IF THE COPYRIGHT HOLDER HAS BEEN ADVISED OF THE POSSIBILITY OF SUCH
* DAMAGE.
*
* THE COPYRIGHT HOLDER SPECIFICALLY DISCLAIMS ANY WARRANTIES, INCLUDING,
* BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND
* FITNESS FOR A PARTICULAR PURPOSE. THE SOFTWARE PROVIDED HEREUNDER IS
* ON AN "AS IS" BASIS, AND THE COPYRIGHT HOLDER HAS NO OBLIGATION TO
* PROVIDE MAINTENANCE, SUPPORT, UPDATES, ENHANCEMENTS, OR MODIFICATIONS.
*
* Red Hat Author(s): Behdad Esfahbod
2011-04-21 23:14:28 +02:00
* Google Author(s): Behdad Esfahbod
2009-11-03 22:35:10 +01:00
*/
#include "hb-private.hh"
2009-11-03 22:35:10 +01:00
#include "hb-icu.h"
2011-04-20 08:00:47 +02:00
#include "hb-unicode-private.hh"
2009-11-03 22:35:10 +01:00
#include <unicode/uchar.h>
#include <unicode/unorm2.h>
#include <unicode/ustring.h>
#include <unicode/utf16.h>
2012-08-13 16:52:52 +02:00
#include <unicode/uversion.h>
2010-07-23 21:11:18 +02:00
hb_script_t
hb_icu_script_to_script (UScriptCode script)
2009-11-03 22:35:10 +01:00
{
if (unlikely (script == USCRIPT_INVALID_CODE))
return HB_SCRIPT_INVALID;
return hb_script_from_string (uscript_getShortName (script), -1);
2009-11-03 22:35:10 +01:00
}
UScriptCode
hb_icu_script_from_script (hb_script_t script)
2009-11-03 22:35:10 +01:00
{
if (unlikely (script == HB_SCRIPT_INVALID))
return USCRIPT_INVALID_CODE;
2010-11-17 20:35:34 +01:00
for (unsigned int i = 0; i < USCRIPT_CODE_LIMIT; i++)
if (unlikely (hb_icu_script_to_script ((UScriptCode) i) == script))
return (UScriptCode) i;
return USCRIPT_UNKNOWN;
}
2012-08-01 22:23:44 +02:00
static hb_unicode_combining_class_t
2011-07-21 04:16:13 +02:00
hb_icu_unicode_combining_class (hb_unicode_funcs_t *ufuncs HB_UNUSED,
hb_codepoint_t unicode,
void *user_data HB_UNUSED)
{
2012-08-01 22:23:44 +02:00
return (hb_unicode_combining_class_t) u_getCombiningClass (unicode);
}
static unsigned int
2011-07-21 04:16:13 +02:00
hb_icu_unicode_eastasian_width (hb_unicode_funcs_t *ufuncs HB_UNUSED,
hb_codepoint_t unicode,
void *user_data HB_UNUSED)
{
switch (u_getIntPropertyValue(unicode, UCHAR_EAST_ASIAN_WIDTH))
{
case U_EA_WIDE:
case U_EA_FULLWIDTH:
return 2;
case U_EA_NEUTRAL:
case U_EA_AMBIGUOUS:
case U_EA_HALFWIDTH:
case U_EA_NARROW:
return 1;
}
return 1;
}
static hb_unicode_general_category_t
2011-07-21 04:16:13 +02:00
hb_icu_unicode_general_category (hb_unicode_funcs_t *ufuncs HB_UNUSED,
hb_codepoint_t unicode,
void *user_data HB_UNUSED)
{
switch (u_getIntPropertyValue(unicode, UCHAR_GENERAL_CATEGORY))
{
case U_UNASSIGNED: return HB_UNICODE_GENERAL_CATEGORY_UNASSIGNED;
case U_UPPERCASE_LETTER: return HB_UNICODE_GENERAL_CATEGORY_UPPERCASE_LETTER;
case U_LOWERCASE_LETTER: return HB_UNICODE_GENERAL_CATEGORY_LOWERCASE_LETTER;
case U_TITLECASE_LETTER: return HB_UNICODE_GENERAL_CATEGORY_TITLECASE_LETTER;
case U_MODIFIER_LETTER: return HB_UNICODE_GENERAL_CATEGORY_MODIFIER_LETTER;
case U_OTHER_LETTER: return HB_UNICODE_GENERAL_CATEGORY_OTHER_LETTER;
case U_NON_SPACING_MARK: return HB_UNICODE_GENERAL_CATEGORY_NON_SPACING_MARK;
case U_ENCLOSING_MARK: return HB_UNICODE_GENERAL_CATEGORY_ENCLOSING_MARK;
case U_COMBINING_SPACING_MARK: return HB_UNICODE_GENERAL_CATEGORY_SPACING_MARK;
case U_DECIMAL_DIGIT_NUMBER: return HB_UNICODE_GENERAL_CATEGORY_DECIMAL_NUMBER;
case U_LETTER_NUMBER: return HB_UNICODE_GENERAL_CATEGORY_LETTER_NUMBER;
case U_OTHER_NUMBER: return HB_UNICODE_GENERAL_CATEGORY_OTHER_NUMBER;
case U_SPACE_SEPARATOR: return HB_UNICODE_GENERAL_CATEGORY_SPACE_SEPARATOR;
case U_LINE_SEPARATOR: return HB_UNICODE_GENERAL_CATEGORY_LINE_SEPARATOR;
case U_PARAGRAPH_SEPARATOR: return HB_UNICODE_GENERAL_CATEGORY_PARAGRAPH_SEPARATOR;
case U_CONTROL_CHAR: return HB_UNICODE_GENERAL_CATEGORY_CONTROL;
case U_FORMAT_CHAR: return HB_UNICODE_GENERAL_CATEGORY_FORMAT;
case U_PRIVATE_USE_CHAR: return HB_UNICODE_GENERAL_CATEGORY_PRIVATE_USE;
case U_SURROGATE: return HB_UNICODE_GENERAL_CATEGORY_SURROGATE;
case U_DASH_PUNCTUATION: return HB_UNICODE_GENERAL_CATEGORY_DASH_PUNCTUATION;
case U_START_PUNCTUATION: return HB_UNICODE_GENERAL_CATEGORY_OPEN_PUNCTUATION;
case U_END_PUNCTUATION: return HB_UNICODE_GENERAL_CATEGORY_CLOSE_PUNCTUATION;
case U_CONNECTOR_PUNCTUATION: return HB_UNICODE_GENERAL_CATEGORY_CONNECT_PUNCTUATION;
case U_OTHER_PUNCTUATION: return HB_UNICODE_GENERAL_CATEGORY_OTHER_PUNCTUATION;
case U_MATH_SYMBOL: return HB_UNICODE_GENERAL_CATEGORY_MATH_SYMBOL;
case U_CURRENCY_SYMBOL: return HB_UNICODE_GENERAL_CATEGORY_CURRENCY_SYMBOL;
case U_MODIFIER_SYMBOL: return HB_UNICODE_GENERAL_CATEGORY_MODIFIER_SYMBOL;
case U_OTHER_SYMBOL: return HB_UNICODE_GENERAL_CATEGORY_OTHER_SYMBOL;
case U_INITIAL_PUNCTUATION: return HB_UNICODE_GENERAL_CATEGORY_INITIAL_PUNCTUATION;
case U_FINAL_PUNCTUATION: return HB_UNICODE_GENERAL_CATEGORY_FINAL_PUNCTUATION;
}
return HB_UNICODE_GENERAL_CATEGORY_UNASSIGNED;
}
static hb_codepoint_t
2011-07-21 04:16:13 +02:00
hb_icu_unicode_mirroring (hb_unicode_funcs_t *ufuncs HB_UNUSED,
hb_codepoint_t unicode,
void *user_data HB_UNUSED)
{
return u_charMirror(unicode);
}
static hb_script_t
2011-07-21 04:16:13 +02:00
hb_icu_unicode_script (hb_unicode_funcs_t *ufuncs HB_UNUSED,
hb_codepoint_t unicode,
void *user_data HB_UNUSED)
{
UErrorCode status = U_ZERO_ERROR;
UScriptCode scriptCode = uscript_getScript(unicode, &status);
2012-01-19 04:32:52 +01:00
if (unlikely (U_FAILURE (status)))
return HB_SCRIPT_UNKNOWN;
return hb_icu_script_to_script (scriptCode);
2009-11-03 22:35:10 +01:00
}
#if U_ICU_VERSION_MAJOR_NUM >= 49
static const UNormalizer2 *normalizer;
#endif
2011-07-21 04:16:13 +02:00
static hb_bool_t
hb_icu_unicode_compose (hb_unicode_funcs_t *ufuncs HB_UNUSED,
hb_codepoint_t a,
hb_codepoint_t b,
hb_codepoint_t *ab,
void *user_data HB_UNUSED)
{
#if U_ICU_VERSION_MAJOR_NUM >= 49
{
UChar32 ret = unorm2_composePair (normalizer, a, b);
if (ret < 0) return false;
*ab = ret;
return true;
}
#endif
/* We don't ifdef-out the fallback code such that compiler always
* sees it and makes sure it's compilable. */
2011-07-21 05:19:49 +02:00
UChar utf16[4], normalized[5];
2012-08-07 22:57:48 +02:00
unsigned int len;
2011-07-21 05:19:49 +02:00
hb_bool_t ret, err;
UErrorCode icu_err;
len = 0;
2012-06-06 02:35:40 +02:00
err = false;
2011-07-21 05:19:49 +02:00
U16_APPEND (utf16, len, ARRAY_LENGTH (utf16), a, err);
2012-06-06 02:35:40 +02:00
if (err) return false;
2011-07-21 05:19:49 +02:00
U16_APPEND (utf16, len, ARRAY_LENGTH (utf16), b, err);
2012-06-06 02:35:40 +02:00
if (err) return false;
2011-07-21 05:19:49 +02:00
icu_err = U_ZERO_ERROR;
len = unorm2_normalize (unorm2_getNFCInstance (&icu_err), utf16, len, normalized, ARRAY_LENGTH (normalized), &icu_err);
2012-01-19 04:32:52 +01:00
if (U_FAILURE (icu_err))
2012-06-06 02:35:40 +02:00
return false;
if (u_countChar32 (normalized, len) == 1) {
2011-07-21 05:19:49 +02:00
U16_GET_UNSAFE (normalized, 0, *ab);
2012-06-06 02:35:40 +02:00
ret = true;
2011-07-21 05:19:49 +02:00
} else {
2012-06-06 02:35:40 +02:00
ret = false;
2011-07-21 05:19:49 +02:00
}
return ret;
2011-07-21 04:16:13 +02:00
}
static hb_bool_t
hb_icu_unicode_decompose (hb_unicode_funcs_t *ufuncs HB_UNUSED,
hb_codepoint_t ab,
hb_codepoint_t *a,
hb_codepoint_t *b,
void *user_data HB_UNUSED)
{
#if U_ICU_VERSION_MAJOR_NUM >= 49
{
UChar decomposed[4];
int len;
UErrorCode icu_err = U_ZERO_ERROR;
len = unorm2_getRawDecomposition (normalizer, ab, decomposed,
ARRAY_LENGTH (decomposed), &icu_err);
if (U_FAILURE (icu_err) || len < 0) return false;
len = u_countChar32 (decomposed, len);
if (len == 1) {
U16_GET_UNSAFE (decomposed, 0, *a);
*b = 0;
return *a != ab;
} else if (len == 2) {
len =0;
U16_NEXT_UNSAFE (decomposed, len, *a);
U16_NEXT_UNSAFE (decomposed, len, *b);
}
return true;
}
#endif
/* We don't ifdef-out the fallback code such that compiler always
* sees it and makes sure it's compilable. */
UChar utf16[2], normalized[2 * HB_UNICODE_MAX_DECOMPOSITION_LEN + 1];
2012-08-07 22:57:48 +02:00
unsigned int len;
2011-07-21 05:19:49 +02:00
hb_bool_t ret, err;
UErrorCode icu_err;
/* This function is a monster! Maybe it wasn't a good idea adding a
* pairwise decompose API... */
/* Watchout for the dragons. Err, watchout for macros changing len. */
2011-07-21 05:19:49 +02:00
len = 0;
2012-06-06 02:35:40 +02:00
err = false;
2011-07-21 05:19:49 +02:00
U16_APPEND (utf16, len, ARRAY_LENGTH (utf16), ab, err);
2012-06-06 02:35:40 +02:00
if (err) return false;
2011-07-21 05:19:49 +02:00
icu_err = U_ZERO_ERROR;
len = unorm2_normalize (unorm2_getNFDInstance (&icu_err), utf16, len, normalized, ARRAY_LENGTH (normalized), &icu_err);
2012-01-19 04:32:52 +01:00
if (U_FAILURE (icu_err))
2012-06-06 02:35:40 +02:00
return false;
2011-07-21 05:19:49 +02:00
len = u_countChar32 (normalized, len);
2011-07-21 05:19:49 +02:00
if (len == 1) {
U16_GET_UNSAFE (normalized, 0, *a);
*b = 0;
ret = *a != ab;
} else if (len == 2) {
len =0;
U16_NEXT_UNSAFE (normalized, len, *a);
U16_NEXT_UNSAFE (normalized, len, *b);
2011-07-21 05:19:49 +02:00
/* Here's the ugly part: if ab decomposes to a single character and
* that character decomposes again, we have to detect that and undo
* the second part :-(. */
UChar recomposed[20];
icu_err = U_ZERO_ERROR;
unorm2_normalize (unorm2_getNFCInstance (&icu_err), normalized, len, recomposed, ARRAY_LENGTH (recomposed), &icu_err);
2012-01-19 04:32:52 +01:00
if (U_FAILURE (icu_err))
2012-06-06 02:35:40 +02:00
return false;
hb_codepoint_t c;
U16_GET_UNSAFE (recomposed, 0, c);
if (c != *a && c != ab) {
*a = c;
2011-07-21 05:19:49 +02:00
*b = 0;
}
2012-06-06 02:35:40 +02:00
ret = true;
2011-07-21 05:19:49 +02:00
} else {
/* If decomposed to more than two characters, take the last one,
* and recompose the rest to get the first component. */
U16_PREV_UNSAFE (normalized, len, *b); /* Changes len in-place. */
UChar recomposed[18 * 2];
2011-07-21 05:19:49 +02:00
icu_err = U_ZERO_ERROR;
len = unorm2_normalize (unorm2_getNFCInstance (&icu_err), normalized, len, recomposed, ARRAY_LENGTH (recomposed), &icu_err);
2012-01-19 04:32:52 +01:00
if (U_FAILURE (icu_err))
2012-06-06 02:35:40 +02:00
return false;
2011-07-21 05:19:49 +02:00
/* We expect that recomposed has exactly one character now. */
if (unlikely (u_countChar32 (recomposed, len) != 1))
return false;
2011-07-21 05:19:49 +02:00
U16_GET_UNSAFE (recomposed, 0, *a);
2012-06-06 02:35:40 +02:00
ret = true;
2011-07-21 05:19:49 +02:00
}
return ret;
2011-07-21 04:16:13 +02:00
}
static unsigned int
hb_icu_unicode_decompose_compatibility (hb_unicode_funcs_t *ufuncs HB_UNUSED,
hb_codepoint_t u,
hb_codepoint_t *decomposed,
void *user_data HB_UNUSED)
{
UChar utf16[2], normalized[2 * HB_UNICODE_MAX_DECOMPOSITION_LEN + 1];
2012-08-07 22:57:48 +02:00
unsigned int len;
int32_t utf32_len;
hb_bool_t err;
UErrorCode icu_err;
/* Copy @u into a UTF-16 array to be passed to ICU. */
len = 0;
err = false;
U16_APPEND (utf16, len, ARRAY_LENGTH (utf16), u, err);
if (err)
return 0;
/* Normalise the codepoint using NFKD mode. */
icu_err = U_ZERO_ERROR;
len = unorm2_normalize (unorm2_getNFKDInstance (&icu_err), utf16, len, normalized, ARRAY_LENGTH (normalized), &icu_err);
2017-10-23 00:23:38 +02:00
if (U_FAILURE (icu_err))
return 0;
/* Convert the decomposed form from UTF-16 to UTF-32. */
icu_err = U_ZERO_ERROR;
u_strToUTF32 ((UChar32*) decomposed, HB_UNICODE_MAX_DECOMPOSITION_LEN, &utf32_len, normalized, len, &icu_err);
2017-10-23 00:23:38 +02:00
if (U_FAILURE (icu_err))
return 0;
return utf32_len;
}
static hb_unicode_funcs_t *static_icu_funcs = nullptr;
#ifdef HB_USE_ATEXIT
static
void free_static_icu_funcs (void)
{
retry:
hb_unicode_funcs_t *icu_funcs = (hb_unicode_funcs_t *) hb_atomic_ptr_get (&static_icu_funcs);
if (!hb_atomic_ptr_cmpexch (&static_icu_funcs, icu_funcs, nullptr))
goto retry;
hb_unicode_funcs_destroy (icu_funcs);
}
#endif
hb_unicode_funcs_t *
hb_icu_get_unicode_funcs (void)
{
retry:
hb_unicode_funcs_t *funcs = (hb_unicode_funcs_t *) hb_atomic_ptr_get (&static_icu_funcs);
if (unlikely (!funcs))
{
#if U_ICU_VERSION_MAJOR_NUM >= 49
if (!hb_atomic_ptr_get (&normalizer)) {
UErrorCode icu_err = U_ZERO_ERROR;
/* We ignore failure in getNFCInstace(). */
(void) hb_atomic_ptr_cmpexch (&normalizer, nullptr, unorm2_getNFCInstance (&icu_err));
}
#endif
funcs = hb_unicode_funcs_create (nullptr);
#define HB_UNICODE_FUNC_IMPLEMENT(name) \
hb_unicode_funcs_set_##name##_func (funcs, hb_icu_unicode_##name, nullptr, nullptr);
HB_UNICODE_FUNCS_IMPLEMENT_CALLBACKS
2011-07-21 04:16:13 +02:00
#undef HB_UNICODE_FUNC_IMPLEMENT
hb_unicode_funcs_make_immutable (funcs);
if (!hb_atomic_ptr_cmpexch (&static_icu_funcs, nullptr, funcs)) {
hb_unicode_funcs_destroy (funcs);
goto retry;
}
2009-11-03 22:35:10 +01:00
#ifdef HB_USE_ATEXIT
atexit (free_static_icu_funcs); /* First person registers atexit() callback. */
#endif
};
2017-10-28 00:03:51 +02:00
return hb_unicode_funcs_reference (funcs);
2009-11-03 22:35:10 +01:00
}