3 * Library to deal with pinyin.
5 * Copyright (C) 2011 Peng Wu <alexepico@gmail.com>
7 * This program is free software; you can redistribute it and/or modify
8 * it under the terms of the GNU General Public License as published by
9 * the Free Software Foundation; either version 2 of the License, or
10 * (at your option) any later version.
12 * This program is distributed in the hope that it will be useful,
13 * but WITHOUT ANY WARRANTY; without even the implied warranty of
14 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
15 * GNU General Public License for more details.
17 * You should have received a copy of the GNU General Public License
18 * along with this program; if not, write to the Free Software
19 * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA.
23 #include "pinyin_parser2.h"
29 #include "pinyin_phrase2.h"
30 #include "pinyin_custom2.h"
31 #include "chewing_key.h"
32 #include "pinyin_parser_table.h"
33 #include "double_pinyin_table.h"
34 #include "chewing_table.h"
37 using namespace pinyin;
39 static bool check_pinyin_options(pinyin_option_t options, const pinyin_index_item_t * item) {
40 guint32 flags = item->m_flags;
41 assert (flags & IS_PINYIN);
43 /* handle incomplete pinyin. */
44 if (flags & PINYIN_INCOMPLETE) {
45 if (!(options & PINYIN_INCOMPLETE))
49 /* handle correct pinyin, currently only one flag per item. */
50 flags &= PINYIN_CORRECT_ALL;
51 options &= PINYIN_CORRECT_ALL;
54 if ((flags & options) != flags)
61 static bool check_chewing_options(pinyin_option_t options, const chewing_index_item_t * item) {
62 guint32 flags = item->m_flags;
63 assert (flags & IS_CHEWING);
65 /* handle incomplete chewing. */
66 if (flags & CHEWING_INCOMPLETE) {
67 if (!(options & CHEWING_INCOMPLETE))
75 /* methods for Chewing Keys to access pinyin parser table. */
76 const char * ChewingKeyRest::get_pinyin_string(){
77 if (m_table_index == 0)
80 /* check end boundary. */
81 assert(m_table_index < G_N_ELEMENTS(content_table));
82 return content_table[m_table_index].m_pinyin_str;
85 const char * ChewingKeyRest::get_chewing_string(){
86 if (m_table_index == 0)
89 /* check end boundary. */
90 assert(m_table_index < G_N_ELEMENTS(content_table));
91 return content_table[m_table_index].m_chewing_str;
97 /* internal information for pinyin parsers. */
100 ChewingKeyRest m_key_rest;
114 const guint16 max_full_pinyin_length = 7; /* include tone. */
116 const guint16 max_double_pinyin_length = 3; /* include tone. */
118 const guint16 max_chewing_length = 4; /* include tone. */
120 static bool compare_pinyin_less_than(const pinyin_index_item_t & lhs,
121 const pinyin_index_item_t & rhs){
122 return 0 > strcmp(lhs.m_pinyin_input, rhs.m_pinyin_input);
125 static inline bool search_pinyin_index(pinyin_option_t options, const char * pinyin,
127 ChewingKeyRest & key_rest){
128 pinyin_index_item_t item;
129 memset(&item, 0, sizeof(item));
130 item.m_pinyin_input = pinyin;
132 std_lite::pair<const pinyin_index_item_t *,
133 const pinyin_index_item_t *> range;
134 range = std_lite::equal_range
135 (pinyin_index, pinyin_index + G_N_ELEMENTS(pinyin_index),
136 item, compare_pinyin_less_than);
138 guint16 range_len = range.second - range.first;
139 assert(range_len <= 1);
140 if (range_len == 1) {
141 const pinyin_index_item_t * index = range.first;
143 if (!check_pinyin_options(options, index))
146 key_rest.m_table_index = index->m_table_index;
147 key = content_table[key_rest.m_table_index].m_chewing_key;
154 static bool compare_chewing_less_than(const chewing_index_item_t & lhs,
155 const chewing_index_item_t & rhs){
156 return 0 > strcmp(lhs.m_chewing_input, rhs.m_chewing_input);
159 static inline bool search_chewing_index(pinyin_option_t options, const char * chewing,
161 ChewingKeyRest & key_rest){
162 chewing_index_item_t item;
163 memset(&item, 0, sizeof(item));
164 item.m_chewing_input = chewing;
166 std_lite::pair<const chewing_index_item_t *,
167 const chewing_index_item_t *> range;
168 range = std_lite::equal_range
169 (chewing_index, chewing_index + G_N_ELEMENTS(chewing_index),
170 item, compare_chewing_less_than);
172 guint16 range_len = range.second - range.first;
173 assert (range_len <= 1);
175 if (range_len == 1) {
176 const chewing_index_item_t * index = range.first;
178 if (!check_chewing_options(options, index))
181 key_rest.m_table_index = index->m_table_index;
182 key = content_table[key_rest.m_table_index].m_chewing_key;
189 /* Full Pinyin Parser */
190 FullPinyinParser2::FullPinyinParser2 (){
191 m_parse_steps = g_array_new(TRUE, FALSE, sizeof(parse_value_t));
195 bool FullPinyinParser2::parse_one_key (pinyin_option_t options, ChewingKey & key,
196 ChewingKeyRest & key_rest,
197 const char * pinyin, int len) const {
198 /* "'" are not accepted in parse_one_key. */
199 gchar * input = g_strndup(pinyin, len);
200 assert(NULL == strchr(input, '\''));
202 guint16 tone = CHEWING_ZERO_TONE; guint16 tone_pos = 0;
203 guint16 parsed_len = len;
204 key = ChewingKey(); key_rest = ChewingKeyRest();
206 if (options & USE_TONE) {
207 /* find the tone in the last character. */
208 char chr = input[parsed_len - 1];
209 if ( '0' < chr && chr <= '5' ) {
212 tone_pos = parsed_len;
216 /* parse pinyin core staff here. */
218 /* Note: optimize here? */
219 input[parsed_len] = '\0';
220 if (!search_pinyin_index(options, input, key, key_rest)) {
225 if (options & USE_TONE) {
226 /* post processing tone. */
227 if ( parsed_len == tone_pos ) {
228 if (tone != CHEWING_ZERO_TONE) {
235 key_rest.m_raw_begin = 0; key_rest.m_raw_end = parsed_len;
237 return parsed_len == len;
241 int FullPinyinParser2::parse (pinyin_option_t options, ChewingKeyVector & keys,
242 ChewingKeyRestVector & key_rests,
243 const char *str, int len) const {
246 g_array_set_size(keys, 0);
247 g_array_set_size(key_rests, 0);
249 /* init m_parse_steps, and prepare dynamic programming. */
250 int step_len = len + 1;
251 g_array_set_size(m_parse_steps, 0);
253 for (i = 0; i < step_len; ++i) {
254 g_array_append_val(m_parse_steps, value);
258 gchar * input = g_strndup(str, len);
259 parse_value_t * curstep = NULL, * nextstep = NULL;
261 for (i = 0; i < len; ++i) {
262 if (input[i] == '\'') {
263 curstep = &g_array_index(m_parse_steps, parse_value_t, i);
264 nextstep = &g_array_index(m_parse_steps, parse_value_t, i + 1);
266 /* propagate current step into next step. */
267 nextstep->m_key = ChewingKey();
268 nextstep->m_key_rest = ChewingKeyRest();
269 nextstep->m_num_keys = curstep->m_num_keys;
270 nextstep->m_parsed_len = curstep->m_parsed_len + 1;
271 nextstep->m_last_step = i;
276 /* forward to next "'" */
277 if ( 0 == next_sep ) {
279 for (k = i; k < len; ++k) {
280 if (input[k] == '\'')
286 /* dynamic programming here. */
287 for (size_t m = i; m < next_sep; ++m) {
288 curstep = &g_array_index(m_parse_steps, parse_value_t, m);
289 size_t try_len = std_lite::min
290 (m + max_full_pinyin_length, next_sep);
291 for (size_t n = m + 1; n < try_len + 1; ++n) {
292 nextstep = &g_array_index(m_parse_steps, parse_value_t, n);
295 const char * onepinyin = input + m;
296 gint16 onepinyinlen = n - m;
297 value = parse_value_t();
299 ChewingKey key; ChewingKeyRest rest;
300 bool parsed = parse_one_key
301 (options, key, rest, onepinyin, onepinyinlen);
302 rest.m_raw_begin = m; rest.m_raw_end = n;
305 value.m_key = key; value.m_key_rest = rest;
306 value.m_num_keys = curstep->m_num_keys + 1;
307 value.m_parsed_len = curstep->m_parsed_len + onepinyinlen;
308 value.m_last_step = m;
311 if (-1 == nextstep->m_last_step)
313 if (value.m_parsed_len > nextstep->m_parsed_len)
315 if (value.m_parsed_len == nextstep->m_parsed_len &&
316 value.m_num_keys < nextstep->m_num_keys)
322 /* final step for back tracing. */
323 gint16 parsed_len = final_step(step_len, keys, key_rests);
325 /* post processing for re-split table. */
326 if (options & USE_RESPLIT_TABLE) {
327 post_process(options, keys, key_rests);
334 int FullPinyinParser2::final_step(size_t step_len, ChewingKeyVector & keys,
335 ChewingKeyRestVector & key_rests) const{
337 gint16 parsed_len = 0;
338 parse_value_t * curstep = NULL;
340 /* find longest match, which starts from the beginning of input. */
341 for (i = step_len - 1; i >= 0; --i) {
342 curstep = &g_array_index(m_parse_steps, parse_value_t, i);
343 if (i == curstep->m_parsed_len)
346 /* prepare saving. */
347 parsed_len = curstep->m_parsed_len;
348 gint16 num_keys = curstep->m_num_keys;
349 g_array_set_size(keys, num_keys);
350 g_array_set_size(key_rests, num_keys);
352 /* save the match. */
353 while (curstep->m_last_step != -1) {
354 gint16 pos = curstep->m_num_keys - 1;
357 if (0 != curstep->m_key_rest.m_table_index) {
358 ChewingKey * key = &g_array_index(keys, ChewingKey, pos);
359 ChewingKeyRest * rest = &g_array_index
360 (key_rests, ChewingKeyRest, pos);
361 *key = curstep->m_key; *rest = curstep->m_key_rest;
365 curstep = &g_array_index(m_parse_steps, parse_value_t,
366 curstep->m_last_step);
372 bool FullPinyinParser2::post_process(pinyin_option_t options,
373 ChewingKeyVector & keys,
374 ChewingKeyRestVector & key_rests) const {
376 assert(keys->len == key_rests->len);
377 gint16 num_keys = keys->len;
379 ChewingKey * cur_key = NULL, * next_key = NULL;
380 ChewingKeyRest * cur_rest = NULL, * next_rest = NULL;
381 guint16 cur_tone = CHEWING_ZERO_TONE, next_tone = CHEWING_ZERO_TONE;
383 for (i = 0; i < num_keys - 1; ++i) {
384 cur_rest = &g_array_index(key_rests, ChewingKeyRest, i);
385 next_rest = &g_array_index(key_rests, ChewingKeyRest, i + 1);
388 if (cur_rest->m_raw_end != next_rest->m_raw_begin)
391 cur_key = &g_array_index(keys, ChewingKey, i);
392 next_key = &g_array_index(keys, ChewingKey, i + 1);
394 if (options & USE_TONE) {
395 cur_tone = cur_key->m_tone;
396 next_tone = next_key->m_tone;
397 cur_key->m_tone = next_key->m_tone = CHEWING_ZERO_TONE;
400 /* lookup re-split table */
402 const resplit_table_item_t * item = NULL;
403 for (k = 0; k < G_N_ELEMENTS(resplit_table); ++k) {
404 item = resplit_table + k;
406 if (item->m_orig_freq >= item->m_new_freq)
409 /* use pinyin_exact_compare2 here. */
410 if (0 == pinyin_exact_compare2(item->m_orig_keys,
417 if (k < G_N_ELEMENTS(resplit_table)) {
419 item = resplit_table + k;
420 *cur_key = item->m_new_keys[0];
421 *next_key = item->m_new_keys[1];
422 /* assumes only moved one char in gen_all_resplit script. */
423 cur_rest->m_raw_end --;
424 next_rest->m_raw_begin --;
427 /* save back tones */
428 if (options & USE_TONE) {
429 cur_key->m_tone = cur_tone;
430 next_key->m_tone = next_tone;
437 #define IS_KEY(x) (('a' <= x && x <= 'z') || x == ';')
439 bool DoublePinyinParser2::parse_one_key(pinyin_option_t options, ChewingKey & key,
440 ChewingKeyRest & key_rest,
441 const char *str, int len) const {
444 if (!(options & PINYIN_INCOMPLETE))
451 int charid = ch == ';' ? 26 : ch - 'a';
452 const char * sheng = m_shengmu_table[charid].m_shengmu;
453 if (NULL == sheng || strcmp(sheng, "'") == 0)
456 if (search_pinyin_index(options, sheng, key, key_rest)) {
457 key_rest.m_raw_begin = 0; key_rest.m_raw_end = len;
464 ChewingTone tone = CHEWING_ZERO_TONE;
465 options &= ~(PINYIN_CORRECT_ALL|PINYIN_AMB_ALL);
469 if (!(options & USE_TONE))
472 if (!('0' < ch && ch <= '5'))
474 tone = (ChewingTone) (ch - '0');
477 if (2 == len || 3 == len) {
478 /* parse shengmu here. */
483 int charid = ch == ';' ? 26 : ch - 'a';
484 const char * sheng = m_shengmu_table[charid].m_shengmu;
487 if (strcmp(sheng, "'") == 0)
490 /* parse yunmu here. */
495 charid = ch == ';' ? 26 : ch - 'a';
497 const char * yun = m_yunmu_table[charid].m_yunmus[0];
498 gchar * pinyin = g_strdup_printf("%s%s", sheng, yun);
499 if (search_pinyin_index(options, pinyin, key, key_rest)) {
500 key_rest.m_raw_begin = 0; key_rest.m_raw_end = len;
508 yun = m_yunmu_table[charid].m_yunmus[1];
509 pinyin = g_strdup_printf("%s%s", sheng, yun);
510 if (search_pinyin_index(options, pinyin, key, key_rest)) {
511 key_rest.m_raw_begin = 0; key_rest.m_raw_end = len;
524 /* only 'a'-'z' and ';' are accepted here. */
525 int DoublePinyinParser2::parse(pinyin_option_t options, ChewingKeyVector & keys,
526 ChewingKeyRestVector & key_rests,
527 const char *str, int len) const {
528 g_array_set_size(keys, 0);
529 g_array_set_size(key_rests, 0);
531 int maximum_len = 0; int i;
532 /* probe the longest possible double pinyin string. */
533 for (i = 0; i < len; ++i) {
534 const char ch = str[i];
535 if (!(IS_KEY(ch) || ('0' < ch && ch <= '5')))
540 /* maximum forward match for double pinyin. */
542 while (parsed_len < maximum_len) {
543 const char * cur_str = str + parsed_len;
544 i = std_lite::min(maximum_len - parsed_len,
545 (int)max_double_pinyin_length);
547 ChewingKey key; ChewingKeyRest key_rest;
549 bool success = parse_one_key(options, key, key_rest, cur_str, i);
554 if (0 == i) /* no more possible double pinyins. */
557 key_rest.m_raw_begin = parsed_len; key_rest.m_raw_end = parsed_len + i;
560 /* save the pinyin */
561 g_array_append_val(keys, key);
562 g_array_append_val(key_rests, key_rest);
570 bool DoublePinyinParser2::set_scheme(DoublePinyinScheme scheme) {
573 case DOUBLE_PINYIN_ZRM:
574 m_shengmu_table = double_pinyin_zrm_sheng;
575 m_yunmu_table = double_pinyin_zrm_yun;
577 case DOUBLE_PINYIN_MS:
578 m_shengmu_table = double_pinyin_mspy_sheng;
579 m_yunmu_table = double_pinyin_mspy_yun;
581 case DOUBLE_PINYIN_ZIGUANG:
582 m_shengmu_table = double_pinyin_zgpy_sheng;
583 m_yunmu_table = double_pinyin_zgpy_yun;
585 case DOUBLE_PINYIN_ABC:
586 m_shengmu_table = double_pinyin_abc_sheng;
587 m_yunmu_table = double_pinyin_abc_yun;
589 case DOUBLE_PINYIN_PYJJ:
590 m_shengmu_table = double_pinyin_pyjj_sheng;
591 m_yunmu_table = double_pinyin_pyjj_yun;
593 case DOUBLE_PINYIN_XHE:
594 m_shengmu_table = double_pinyin_xhe_sheng;
595 m_yunmu_table = double_pinyin_xhe_yun;
597 case DOUBLE_PINYIN_CUSTOMIZED:
601 return false; /* no such scheme. */
604 /* the chewing string must be freed with g_free. */
605 static bool search_chewing_symbols(const chewing_symbol_item_t * symbol_table,
606 const char key, char ** chewing) {
608 /* just iterate the table, as we only have < 50 items. */
609 while (symbol_table->m_input != '\0') {
610 if (symbol_table->m_input == key) {
611 *chewing = g_strdup(symbol_table->m_chewing);
619 static bool search_chewing_tones(const chewing_tone_item_t * tone_table,
620 const char key, char * tone) {
621 *tone = CHEWING_ZERO_TONE;
622 /* just iterate the table, as we only have < 10 items. */
623 while (tone_table->m_input != '\0') {
624 if (tone_table->m_input == key) {
625 *tone = tone_table->m_tone;
634 bool ChewingParser2::parse_one_key(pinyin_option_t options, ChewingKey & key, ChewingKeyRest & key_rest, const char *str, int len) const {
635 char tone = CHEWING_ZERO_TONE;
637 int symbols_len = len;
638 /* probe whether the last key is tone key in str. */
639 if (options & USE_TONE) {
640 char ch = str[len - 1];
641 /* remove tone from input */
642 if (search_chewing_tones(m_tone_table, ch, &tone))
647 gchar * chewing = NULL, * onechar = NULL;
649 /* probe the possible chewing map in the rest of str. */
650 for (i = 0; i < symbols_len; ++i) {
651 if (!search_chewing_symbols(m_symbol_table, str[i], &onechar)) {
658 chewing = g_strdup(onechar);
660 gchar * tmp = chewing;
661 chewing = g_strconcat(chewing, onechar, NULL);
667 /* search the chewing in the chewing index table. */
668 if (search_chewing_index(options, chewing, key, key_rest)) {
669 key_rest.m_raw_begin = 0; key_rest.m_raw_end = len;
670 /* save back tone if available. */
681 /* only characters in chewing keyboard scheme are accepted here. */
682 int ChewingParser2::parse(pinyin_option_t options, ChewingKeyVector & keys,
683 ChewingKeyRestVector & key_rests,
684 const char *str, int len) const {
685 g_array_set_size(keys, 0);
686 g_array_set_size(key_rests, 0);
688 int maximum_len = 0; int i;
689 /* probe the longest possible chewing string. */
690 for (i = 0; i < len; ++i) {
691 if (!in_chewing_scheme(str[i]))
696 /* maximum forward match for chewing. */
698 while (parsed_len < maximum_len) {
699 const char * cur_str = str + parsed_len;
700 i = std_lite::min(maximum_len - parsed_len,
701 (int)max_chewing_length);
703 ChewingKey key; ChewingKeyRest key_rest;
705 bool success = parse_one_key(options, key, key_rest, cur_str, i);
710 if (0 == i) /* no more possible chewings. */
713 key_rest.m_raw_begin = parsed_len; key_rest.m_raw_end = parsed_len + i;
716 /* save the pinyin. */
717 g_array_append_val(keys, key);
718 g_array_append_val(key_rests, key_rest);
725 bool ChewingParser2::set_scheme(ChewingScheme scheme) {
727 case CHEWING_STANDARD:
728 m_symbol_table = chewing_standard_symbols;
729 m_tone_table = chewing_standard_tones;
732 m_symbol_table = chewing_ibm_symbols;
733 m_tone_table = chewing_ibm_tones;
735 case CHEWING_GINYIEH:
736 m_symbol_table = chewing_ginyieh_symbols;
737 m_tone_table = chewing_ginyieh_tones;
740 m_symbol_table = chewing_eten_symbols;
741 m_tone_table = chewing_eten_tones;
749 bool ChewingParser2::in_chewing_scheme(const char key) const {
750 gchar * chewing = NULL;
751 char tone = CHEWING_ZERO_TONE;
753 bool retval = search_chewing_symbols(m_symbol_table, key, &chewing) ||
754 search_chewing_tones(m_tone_table, key, &tone);