harfbuzz/src/hb-buffer.cc

674 lines
15 KiB
C++
Raw Normal View History

/*
2011-04-21 21:14:28 +00:00
* Copyright © 1998-2004 David Turner and Werner Lemberg
* Copyright © 2004,2007,2009,2010 Red Hat, Inc.
* Copyright © 2011 Google, Inc.
*
* This is part of HarfBuzz, a text shaping library.
*
* 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): Owen Taylor, Behdad Esfahbod
* Google Author(s): Behdad Esfahbod
*/
#include "hb-buffer-private.hh"
2009-05-25 07:39:11 +00:00
#include <string.h>
2010-07-23 19:11:18 +00:00
HB_BEGIN_DECLS
2009-08-02 02:53:04 +00:00
2009-08-02 02:19:06 +00:00
static hb_buffer_t _hb_buffer_nil = {
HB_OBJECT_HEADER_STATIC,
2009-11-03 20:15:07 +00:00
&_hb_unicode_funcs_default,
{
HB_DIRECTION_INVALID,
HB_SCRIPT_INVALID,
NULL,
},
2009-08-02 02:19:06 +00:00
};
/* Here is how the buffer works internally:
*
2010-05-15 02:13:38 +00:00
* There are two info pointers: info and out_info. They always have
* the same allocated size, but different lengths.
*
* As an optimization, both info and out_info may point to the
* same piece of memory, which is owned by info. This remains the
* case as long as out_len doesn't exceed i at any time.
2010-05-15 02:13:38 +00:00
* In that case, swap() is no-op and the glyph operations operate
* mostly in-place.
*
* As soon as out_info gets longer than info, out_info is moved over
2010-05-15 02:13:38 +00:00
* to an alternate buffer (which we reuse the pos buffer for!), and its
* current contents (out_len entries) are copied to the new place.
* This should all remain transparent to the user. swap() then
* switches info and out_info.
*/
2009-05-25 08:04:24 +00:00
static hb_bool_t
_hb_buffer_enlarge (hb_buffer_t *buffer, unsigned int size)
{
if (unlikely (buffer->in_error))
return FALSE;
unsigned int new_allocated = buffer->allocated;
hb_glyph_position_t *new_pos;
hb_glyph_info_t *new_info;
bool separate_out;
separate_out = buffer->out_info != buffer->info;
while (size > new_allocated)
new_allocated += (new_allocated >> 1) + 32;
2011-02-28 18:13:52 +00:00
ASSERT_STATIC (sizeof (buffer->info[0]) == sizeof (buffer->pos[0]));
bool overflows = new_allocated >= ((unsigned int) -1) / sizeof (buffer->info[0]);
if (unlikely (overflows)) {
new_pos = NULL;
new_info = NULL;
} else {
new_pos = (hb_glyph_position_t *) realloc (buffer->pos, new_allocated * sizeof (buffer->pos[0]));
new_info = (hb_glyph_info_t *) realloc (buffer->info, new_allocated * sizeof (buffer->info[0]));
}
if (unlikely (!new_pos || !new_info))
buffer->in_error = TRUE;
if (likely (new_pos))
buffer->pos = new_pos;
if (likely (new_info))
buffer->info = new_info;
buffer->out_info = separate_out ? (hb_glyph_info_t *) buffer->pos : buffer->info;
if (likely (!buffer->in_error))
buffer->allocated = new_allocated;
return likely (!buffer->in_error);
}
static inline hb_bool_t
_hb_buffer_ensure (hb_buffer_t *buffer, unsigned int size)
{
return likely (size <= buffer->allocated) ? TRUE : _hb_buffer_enlarge (buffer, size);
}
2010-07-23 19:11:18 +00:00
static inline hb_bool_t
_hb_buffer_ensure_separate (hb_buffer_t *buffer, unsigned int size)
2009-05-25 08:04:24 +00:00
{
if (unlikely (!_hb_buffer_ensure (buffer, size))) return FALSE;
2010-05-15 03:20:16 +00:00
if (buffer->out_info == buffer->info)
{
assert (buffer->have_output);
buffer->out_info = (hb_glyph_info_t *) buffer->pos;
memcpy (buffer->out_info, buffer->info, buffer->out_len * sizeof (buffer->out_info[0]));
}
2010-05-15 03:20:16 +00:00
return TRUE;
}
/* Public API */
hb_buffer_t *
2009-08-02 02:19:06 +00:00
hb_buffer_create (unsigned int pre_alloc_size)
{
hb_buffer_t *buffer;
2009-08-04 02:43:02 +00:00
if (!HB_OBJECT_DO_CREATE (hb_buffer_t, buffer))
2009-08-02 02:19:06 +00:00
return &_hb_buffer_nil;
2009-08-02 02:19:06 +00:00
if (pre_alloc_size)
_hb_buffer_ensure (buffer, pre_alloc_size);
hb_buffer_reset (buffer);
2009-11-03 20:15:07 +00:00
2009-08-02 02:19:06 +00:00
return buffer;
}
2009-08-02 02:19:06 +00:00
hb_buffer_t *
hb_buffer_reference (hb_buffer_t *buffer)
{
HB_OBJECT_DO_REFERENCE (buffer);
}
2009-07-28 19:43:34 +00:00
void
2009-08-02 02:19:06 +00:00
hb_buffer_destroy (hb_buffer_t *buffer)
{
2009-08-02 02:19:06 +00:00
HB_OBJECT_DO_DESTROY (buffer);
2009-11-03 20:15:07 +00:00
hb_unicode_funcs_destroy (buffer->unicode);
free (buffer->info);
free (buffer->pos);
2009-08-02 02:19:06 +00:00
free (buffer);
}
2009-11-03 20:15:07 +00:00
void
hb_buffer_set_unicode_funcs (hb_buffer_t *buffer,
hb_unicode_funcs_t *unicode)
{
2009-12-20 15:25:18 +00:00
if (!unicode)
unicode = &_hb_unicode_funcs_default;
2009-12-20 15:25:18 +00:00
2009-11-03 20:15:07 +00:00
hb_unicode_funcs_reference (unicode);
hb_unicode_funcs_destroy (buffer->unicode);
buffer->unicode = unicode;
}
hb_unicode_funcs_t *
hb_buffer_get_unicode_funcs (hb_buffer_t *buffer)
{
return buffer->unicode;
}
void
hb_buffer_set_direction (hb_buffer_t *buffer,
hb_direction_t direction)
{
buffer->props.direction = direction;
2009-11-03 20:15:07 +00:00
}
hb_direction_t
hb_buffer_get_direction (hb_buffer_t *buffer)
{
return buffer->props.direction;
2009-11-03 20:15:07 +00:00
}
2009-11-05 01:29:54 +00:00
void
hb_buffer_set_script (hb_buffer_t *buffer,
hb_script_t script)
{
buffer->props.script = script;
2009-11-05 01:29:54 +00:00
}
hb_script_t
hb_buffer_get_script (hb_buffer_t *buffer)
{
return buffer->props.script;
2009-11-05 01:29:54 +00:00
}
void
hb_buffer_set_language (hb_buffer_t *buffer,
hb_language_t language)
{
buffer->props.language = language;
2009-11-05 01:29:54 +00:00
}
hb_language_t
hb_buffer_get_language (hb_buffer_t *buffer)
{
return buffer->props.language;
2009-11-05 01:29:54 +00:00
}
2009-11-03 20:15:07 +00:00
void
hb_buffer_reset (hb_buffer_t *buffer)
{
hb_unicode_funcs_destroy (buffer->unicode);
buffer->unicode = _hb_buffer_nil.unicode;
buffer->props = _hb_buffer_nil.props;
buffer->have_output = FALSE;
buffer->have_positions = FALSE;
2010-05-15 03:20:16 +00:00
buffer->in_error = FALSE;
buffer->i = 0;
buffer->len = 0;
buffer->out_len = 0;
buffer->serial = 0;
buffer->out_info = buffer->info;
}
2010-05-15 03:20:16 +00:00
hb_bool_t
hb_buffer_pre_allocate (hb_buffer_t *buffer, unsigned int size)
2009-07-28 19:43:34 +00:00
{
return _hb_buffer_ensure (buffer, size);
2009-07-28 19:43:34 +00:00
}
hb_bool_t
hb_buffer_allocation_successful (hb_buffer_t *buffer)
{
return !buffer->in_error;
}
void
hb_buffer_add (hb_buffer_t *buffer,
hb_codepoint_t codepoint,
hb_mask_t mask,
unsigned int cluster)
{
hb_glyph_info_t *glyph;
if (unlikely (!_hb_buffer_ensure (buffer, buffer->len + 1))) return;
glyph = &buffer->info[buffer->len];
memset (glyph, 0, sizeof (*glyph));
2009-08-02 02:53:04 +00:00
glyph->codepoint = codepoint;
glyph->mask = mask;
glyph->cluster = cluster;
buffer->len++;
}
/* HarfBuzz-Internal API */
void
_hb_buffer_clear_output (hb_buffer_t *buffer)
{
buffer->have_output = TRUE;
buffer->have_positions = FALSE;
buffer->out_len = 0;
buffer->out_info = buffer->info;
}
void
_hb_buffer_clear_positions (hb_buffer_t *buffer)
{
buffer->have_output = FALSE;
buffer->have_positions = TRUE;
memset (buffer->pos, 0, sizeof (buffer->pos[0]) * buffer->len);
}
2009-08-11 03:40:28 +00:00
void
2009-05-25 08:04:24 +00:00
_hb_buffer_swap (hb_buffer_t *buffer)
{
2009-05-25 08:04:24 +00:00
unsigned int tmp;
assert (buffer->have_output);
2010-05-15 03:20:16 +00:00
if (unlikely (buffer->in_error)) return;
if (buffer->out_info != buffer->info)
{
hb_glyph_info_t *tmp_string;
tmp_string = buffer->info;
buffer->info = buffer->out_info;
buffer->out_info = tmp_string;
buffer->pos = (hb_glyph_position_t *) buffer->out_info;
}
tmp = buffer->len;
buffer->len = buffer->out_len;
buffer->out_len = tmp;
buffer->i = 0;
}
void
2010-10-27 20:57:01 +00:00
_hb_buffer_replace_glyphs_be16 (hb_buffer_t *buffer,
unsigned int num_in,
unsigned int num_out,
const uint16_t *glyph_data_be)
{
if (buffer->out_info != buffer->info ||
buffer->out_len + num_out > buffer->i + num_in)
{
if (unlikely (!_hb_buffer_ensure_separate (buffer, buffer->out_len + num_out)))
2010-05-15 03:20:16 +00:00
return;
}
hb_glyph_info_t orig_info = buffer->info[buffer->i];
for (unsigned int i = 0; i < num_out; i++)
{
hb_glyph_info_t *info = &buffer->out_info[buffer->out_len + i];
*info = orig_info;
info->codepoint = hb_be_uint16 (glyph_data_be[i]);
}
buffer->i += num_in;
buffer->out_len += num_out;
}
2009-05-25 07:39:11 +00:00
2009-08-11 03:40:28 +00:00
void
2010-10-27 20:57:01 +00:00
_hb_buffer_replace_glyph (hb_buffer_t *buffer,
hb_codepoint_t glyph_index)
{
hb_glyph_info_t *info;
2009-05-25 08:04:24 +00:00
if (buffer->out_info != buffer->info)
{
if (unlikely (!_hb_buffer_ensure (buffer, buffer->out_len + 1))) return;
buffer->out_info[buffer->out_len] = buffer->info[buffer->i];
}
else if (buffer->out_len != buffer->i)
buffer->out_info[buffer->out_len] = buffer->info[buffer->i];
info = &buffer->out_info[buffer->out_len];
2009-08-02 02:53:04 +00:00
info->codepoint = glyph_index;
2009-05-25 08:04:24 +00:00
buffer->i++;
buffer->out_len++;
}
2009-08-11 03:40:28 +00:00
void
2009-05-25 08:04:24 +00:00
_hb_buffer_next_glyph (hb_buffer_t *buffer)
2009-05-17 10:03:42 +00:00
{
if (buffer->have_output)
{
if (buffer->out_info != buffer->info)
{
if (unlikely (!_hb_buffer_ensure (buffer, buffer->out_len + 1))) return;
buffer->out_info[buffer->out_len] = buffer->info[buffer->i];
}
else if (buffer->out_len != buffer->i)
buffer->out_info[buffer->out_len] = buffer->info[buffer->i];
buffer->out_len++;
}
buffer->i++;
}
2010-05-21 16:31:45 +00:00
void
_hb_buffer_reset_masks (hb_buffer_t *buffer,
hb_mask_t mask)
2010-05-21 16:31:45 +00:00
{
unsigned int count = buffer->len;
for (unsigned int i = 0; i < count; i++)
2010-10-13 21:17:00 +00:00
buffer->info[i].mask = mask;
2010-05-21 16:31:45 +00:00
}
void
_hb_buffer_add_masks (hb_buffer_t *buffer,
hb_mask_t mask)
{
unsigned int count = buffer->len;
for (unsigned int i = 0; i < count; i++)
buffer->info[i].mask |= mask;
}
void
_hb_buffer_set_masks (hb_buffer_t *buffer,
hb_mask_t value,
hb_mask_t mask,
unsigned int cluster_start,
unsigned int cluster_end)
2010-05-21 16:31:45 +00:00
{
hb_mask_t not_mask = ~mask;
value &= mask;
2010-10-13 19:38:52 +00:00
if (!mask)
return;
2010-05-21 16:31:45 +00:00
if (cluster_start == 0 && cluster_end == (unsigned int)-1) {
unsigned int count = buffer->len;
for (unsigned int i = 0; i < count; i++)
buffer->info[i].mask = (buffer->info[i].mask & not_mask) | value;
2010-05-21 16:31:45 +00:00
return;
}
2010-10-07 05:21:19 +00:00
/* XXX can't bsearch since .cluster may not be sorted. */
2010-05-21 16:31:45 +00:00
/* Binary search to find the start position and go from there. */
unsigned int min = 0, max = buffer->len;
while (min < max)
{
unsigned int mid = min + ((max - min) / 2);
if (buffer->info[mid].cluster < cluster_start)
min = mid + 1;
else
max = mid;
}
unsigned int count = buffer->len;
for (unsigned int i = min; i < count && buffer->info[i].cluster < cluster_end; i++)
buffer->info[i].mask = (buffer->info[i].mask & not_mask) | value;
2010-05-21 16:31:45 +00:00
}
2009-08-02 02:19:06 +00:00
/* Public API again */
hb_bool_t
hb_buffer_set_length (hb_buffer_t *buffer,
unsigned int length)
{
if (!_hb_buffer_ensure (buffer, length))
return FALSE;
/* Wipe the new space */
if (length > buffer->len) {
memset (buffer->info + buffer->len, 0, sizeof (buffer->info[0]) * (length - buffer->len));
if (buffer->have_positions)
memset (buffer->pos + buffer->len, 0, sizeof (buffer->pos[0]) * (length - buffer->len));
}
buffer->len = length;
return TRUE;
}
2009-08-02 02:19:06 +00:00
unsigned int
hb_buffer_get_length (hb_buffer_t *buffer)
2009-08-02 02:19:06 +00:00
{
return buffer->len;
2009-08-02 02:19:06 +00:00
}
/* Return value valid as long as buffer not modified */
hb_glyph_info_t *
hb_buffer_get_glyph_infos (hb_buffer_t *buffer,
unsigned int *length)
2009-08-02 02:19:06 +00:00
{
if (length)
*length = buffer->len;
return (hb_glyph_info_t *) buffer->info;
2009-08-02 02:19:06 +00:00
}
/* Return value valid as long as buffer not modified */
hb_glyph_position_t *
hb_buffer_get_glyph_positions (hb_buffer_t *buffer,
unsigned int *length)
2009-08-02 02:19:06 +00:00
{
if (!buffer->have_positions)
_hb_buffer_clear_positions (buffer);
2009-08-02 02:19:06 +00:00
if (length)
*length = buffer->len;
return (hb_glyph_position_t *) buffer->pos;
2009-08-02 02:19:06 +00:00
}
2009-08-11 00:59:25 +00:00
2009-11-07 00:48:16 +00:00
static void
reverse_range (hb_buffer_t *buffer,
unsigned int start,
unsigned int end)
2009-08-11 00:59:25 +00:00
{
unsigned int i, j;
2009-11-07 00:48:16 +00:00
for (i = start, j = end - 1; i < j; i++, j--) {
hb_glyph_info_t t;
2009-08-11 00:59:25 +00:00
t = buffer->info[i];
buffer->info[i] = buffer->info[j];
buffer->info[j] = t;
2009-08-11 00:59:25 +00:00
}
if (buffer->pos) {
2009-11-07 00:48:16 +00:00
for (i = 0, j = end - 1; i < j; i++, j--) {
hb_glyph_position_t t;
2009-08-11 00:59:25 +00:00
t = buffer->pos[i];
buffer->pos[i] = buffer->pos[j];
buffer->pos[j] = t;
2009-08-11 00:59:25 +00:00
}
}
}
2009-08-11 02:47:57 +00:00
2009-11-07 00:48:16 +00:00
void
hb_buffer_reverse (hb_buffer_t *buffer)
{
if (unlikely (!buffer->len))
2009-11-07 00:48:16 +00:00
return;
reverse_range (buffer, 0, buffer->len);
2009-11-07 00:48:16 +00:00
}
void
hb_buffer_reverse_clusters (hb_buffer_t *buffer)
{
unsigned int i, start, count, last_cluster;
if (unlikely (!buffer->len))
2009-11-07 00:48:16 +00:00
return;
hb_buffer_reverse (buffer);
count = buffer->len;
2009-11-07 00:48:16 +00:00
start = 0;
last_cluster = buffer->info[0].cluster;
2009-11-07 00:48:16 +00:00
for (i = 1; i < count; i++) {
if (last_cluster != buffer->info[i].cluster) {
2009-11-07 00:48:16 +00:00
reverse_range (buffer, start, i);
start = i;
last_cluster = buffer->info[i].cluster;
2009-11-07 00:48:16 +00:00
}
}
reverse_range (buffer, start, i);
}
2009-08-11 02:47:57 +00:00
#define ADD_UTF(T) \
HB_STMT_START { \
const T *next = (const T *) text + item_offset; \
const T *end = next + item_length; \
while (next < end) { \
hb_codepoint_t u; \
const T *old_next = next; \
next = UTF_NEXT (next, end, u); \
hb_buffer_add (buffer, u, 1, old_next - (const T *) text); \
2009-08-11 02:47:57 +00:00
} \
} HB_STMT_END
#define UTF8_COMPUTE(Char, Mask, Len) \
if (Char < 128) { Len = 1; Mask = 0x7f; } \
else if ((Char & 0xe0) == 0xc0) { Len = 2; Mask = 0x1f; } \
else if ((Char & 0xf0) == 0xe0) { Len = 3; Mask = 0x0f; } \
else if ((Char & 0xf8) == 0xf0) { Len = 4; Mask = 0x07; } \
else Len = 0;
static inline const uint8_t *
hb_utf8_next (const uint8_t *text,
const uint8_t *end,
hb_codepoint_t *unicode)
{
uint8_t c = *text;
unsigned int mask, len;
2010-05-27 18:04:15 +00:00
/* TODO check for overlong sequences? also: optimize? */
2009-08-11 02:47:57 +00:00
UTF8_COMPUTE (c, mask, len);
if (unlikely (!len || (unsigned int) (end - text) < len)) {
2009-08-11 02:47:57 +00:00
*unicode = -1;
return text + 1;
} else {
hb_codepoint_t result;
unsigned int i;
result = c & mask;
for (i = 1; i < len; i++)
{
if (unlikely ((text[i] & 0xc0) != 0x80))
2009-08-11 02:47:57 +00:00
{
*unicode = -1;
return text + 1;
}
result <<= 6;
result |= (text[i] & 0x3f);
}
*unicode = result;
return text + len;
}
}
void
hb_buffer_add_utf8 (hb_buffer_t *buffer,
const char *text,
2010-04-29 17:56:44 +00:00
unsigned int text_length HB_UNUSED,
2009-08-11 02:47:57 +00:00
unsigned int item_offset,
unsigned int item_length)
{
#define UTF_NEXT(S, E, U) hb_utf8_next (S, E, &(U))
ADD_UTF (uint8_t);
#undef UTF_NEXT
}
static inline const uint16_t *
hb_utf16_next (const uint16_t *text,
const uint16_t *end,
hb_codepoint_t *unicode)
{
uint16_t c = *text++;
if (unlikely (c >= 0xd800 && c < 0xdc00)) {
2009-08-11 02:47:57 +00:00
/* high surrogate */
uint16_t l;
2011-04-13 19:02:40 +00:00
if (text < end && ((l = *text), likely (l >= 0xdc00 && l < 0xe000))) {
2009-08-11 02:47:57 +00:00
/* low surrogate */
*unicode = ((hb_codepoint_t) ((c) - 0xd800) * 0x400 + (l) - 0xdc00 + 0x10000);
text++;
} else
*unicode = -1;
} else
*unicode = c;
return text;
}
void
hb_buffer_add_utf16 (hb_buffer_t *buffer,
const uint16_t *text,
2010-04-29 17:56:44 +00:00
unsigned int text_length HB_UNUSED,
2009-08-11 02:47:57 +00:00
unsigned int item_offset,
unsigned int item_length)
{
#define UTF_NEXT(S, E, U) hb_utf16_next (S, E, &(U))
ADD_UTF (uint16_t);
#undef UTF_NEXT
}
void
hb_buffer_add_utf32 (hb_buffer_t *buffer,
const uint32_t *text,
2010-04-29 17:56:44 +00:00
unsigned int text_length HB_UNUSED,
2009-08-11 02:47:57 +00:00
unsigned int item_offset,
unsigned int item_length)
{
#define UTF_NEXT(S, E, U) ((U) = *(S), (S)+1)
ADD_UTF (uint32_t);
#undef UTF_NEXT
}
2010-07-23 19:11:18 +00:00
HB_END_DECLS