static cst_val* word_to_phones(const cst_item *word)
{
cst_val*phones=NULL;
const char *name=item_feat_string(word, "name");
ustring32_t letters=ustring32_alloc(0);
if(letters==NULL) return NULL;
ustring32_assign8(letters,(const uint8_t*)name);
if(ustring32_empty(letters))
{
ustring32_free(letters);
return NULL;
}
unsigned int flags=classify_characters(ustring32_str(letters),ustring32_length(letters));
int variant=item_feat_int(item_parent(item_as(word,"Token")),"variant");
if((flags&cs_lc)&&cst_streq(ffeature_string(word,"gpos"),"content"))
{
if(variant==variant_pseudo_english)
phones=ustring32_lts_apply(letters,&en_consonants_lts);
else phones=ustring32_lts_apply(letters,&ru_consonants_lts);
item_set_int(word,"no_vr",1);
}
else if((variant==variant_pseudo_english)&&(flags&cs_en))
{
cst_val *en_phones=lex_lookup(en_lex,name,(cst_streq(name,"a")?"n":NULL));
if(en_phones)
{
phones=ru_lts_apply(en_phones,&ru_en_lts);
delete_val(en_phones);
}
item_set_int(word,"no_pl",1);
}
else
{
const ru_dict_entry *e=bsearch(name,ru_dict,ru_dict_size,sizeof(ru_dict_entry),compare_entries);
if(e!=NULL)
{
if(e->stress > 0)
ustring32_set(letters,e->stress-1,1105);
else
item_set_int(word,"stressed_syl_num",e->stress);
}
phones=ustring32_lts_apply(letters,&ru_lts);
}
ustring32_free(letters);
return phones;
}
/* Creates a buffer with an embedding level for every character in the
* given text. Also determines the base level and returns it in
* *baseDir if *baseDir does not initially contain a valid direction.
*/
static fz_bidi_level *
create_levels(fz_context *ctx,
const uint32_t *text,
size_t len,
fz_bidi_direction *baseDir,
int resolveWhiteSpace,
int flags)
{
fz_bidi_level *levels;
fz_bidi_chartype *types = NULL;
fz_bidi_level baseLevel;
levels = fz_malloc(ctx, len * sizeof(*levels));
fz_var(types);
fz_try(ctx)
{
types = fz_malloc(ctx, len * sizeof(fz_bidi_chartype));
classify_characters(text, types, len, flags);
if (*baseDir != FZ_BIDI_LTR && *baseDir != FZ_BIDI_RTL)
{
/* Derive the base level from the text and
* update *baseDir in case the caller wants to know.
*/
baseLevel = base_level_from_text(types, len);
*baseDir = ODD(baseLevel)==1 ? FZ_BIDI_RTL : FZ_BIDI_LTR;
}
else
{
baseLevel = (fz_bidi_level)*baseDir;
}
{
/* Replace tab with base direction, i.e. make tab appear as
* 'strong left' if the base direction is left-to-right and
* 'strong right' if base direction is right-to-left. This
* allows Layout to implicitly treat tabs as 'segment separators'.
*/
size_t i;
for (i = 0u; i < len; i++)
{
if (text[i]=='\t')
{
types[i] = (*baseDir == FZ_BIDI_RTL) ? BDI_R : BDI_L;
}
}
}
/* Look for quotation marks. Classify them as RLE or LRE
* or leave them alone, depending on what follows them.
*/
classify_quoted_blocks(text, types, len);
/* Work out the levels and character types... */
(void)fz_bidi_resolve_explicit(baseLevel, BDI_N, types, levels, len, 0);
fz_bidi_resolve_weak(ctx, baseLevel, types, levels, len);
fz_bidi_resolve_neutrals(baseLevel,types, levels, len);
fz_bidi_resolve_implicit(types, levels, len);
classify_characters(text, types, len, BIDI_CLASSIFY_WHITE_SPACE);
if (resolveWhiteSpace)
{
/* resolve whitespace */
fz_bidi_resolve_whitespace(baseLevel, types, levels, len);
}
/* The levels buffer now has odd and even numbers indicating
* rtl or ltr characters, respectively.
*/
#ifdef DEBUG_BIDI_VERBOSE
fprintf(stderr, "Levels: ");
{
size_t i;
for (i = 0; i < len; i++)
{
fprintf(stderr, "%d", levels[i]>9?0:levels[i]);
}
fprintf(stderr, "\n");
}
#endif
}
fz_always(ctx)
{
fz_free(ctx, types);
}
fz_catch(ctx)
{
fz_free(ctx, levels);
fz_rethrow(ctx);
}
return levels;
}
