*/
+#include "pinyin_parser2.h"
#include <ctype.h>
#include <assert.h>
+#include <stdio.h>
#include <string.h>
#include "stl_lite.h"
-#include "pinyin_parser2.h"
#include "pinyin_phrase2.h"
#include "pinyin_custom2.h"
#include "chewing_key.h"
using namespace pinyin;
-static bool check_pinyin_options(guint32 options, const pinyin_index_item_t * item) {
+static bool check_pinyin_options(pinyin_option_t options, const pinyin_index_item_t * item) {
guint32 flags = item->m_flags;
assert (flags & IS_PINYIN);
return true;
}
-static bool check_chewing_options(guint32 options, const chewing_index_item_t * item) {
+static bool check_chewing_options(pinyin_option_t options, const chewing_index_item_t * item) {
guint32 flags = item->m_flags;
assert (flags & IS_CHEWING);
}
-/* methods for Chewing Keys to access pinyin parser table. */
-const char * ChewingKeyRest::get_pinyin_string(){
- if (m_table_index == 0)
- return NULL;
+gint ChewingKey::get_table_index() {
+ assert(m_initial < CHEWING_NUMBER_OF_INITIALS);
+ assert(m_middle < CHEWING_NUMBER_OF_MIDDLES);
+ assert(m_final < CHEWING_NUMBER_OF_FINALS);
- /* check end boundary. */
- assert(m_table_index < G_N_ELEMENTS(content_table));
- return content_table[m_table_index].m_pinyin_str;
+ gint index = chewing_key_table[(m_initial * CHEWING_NUMBER_OF_MIDDLES + m_middle) * CHEWING_NUMBER_OF_FINALS + m_final];
+ return index == -1 ? 0 : index;
}
-const char * ChewingKeyRest::get_chewing_string(){
- if (m_table_index == 0)
- return NULL;
+gchar * ChewingKey::get_pinyin_string() {
+ assert(m_tone < CHEWING_NUMBER_OF_TONES);
+ gint index = get_table_index();
+ assert(index < G_N_ELEMENTS(content_table));
+ const content_table_item_t & item = content_table[index];
+
+ if (CHEWING_ZERO_TONE == m_tone) {
+ return g_strdup(item.m_pinyin_str);
+ } else {
+ return g_strdup_printf("%s%d", item.m_pinyin_str, m_tone);
+ }
+}
- /* check end boundary. */
- assert(m_table_index < G_N_ELEMENTS(content_table));
- return content_table[m_table_index].m_chewing_str;
+gchar * ChewingKey::get_chewing_string() {
+ assert(m_tone < CHEWING_NUMBER_OF_TONES);
+ gint index = get_table_index();
+ assert(index < G_N_ELEMENTS(content_table));
+ const content_table_item_t & item = content_table[index];
+
+ if (CHEWING_ZERO_TONE == m_tone) {
+ return g_strdup(item.m_chewing_str);
+ } else {
+ return g_strdup_printf("%s%s", item.m_chewing_str,
+ chewing_tone_table[m_tone]);
+ }
}
return 0 > strcmp(lhs.m_pinyin_input, rhs.m_pinyin_input);
}
-static inline bool search_pinyin_index(guint32 options, const char * pinyin,
- ChewingKey & key,
- ChewingKeyRest & key_rest){
+static inline bool search_pinyin_index(pinyin_option_t options,
+ const char * pinyin,
+ ChewingKey & key){
pinyin_index_item_t item;
memset(&item, 0, sizeof(item));
item.m_pinyin_input = pinyin;
if (!check_pinyin_options(options, index))
return false;
- key_rest.m_table_index = index->m_table_index;
- key = content_table[key_rest.m_table_index].m_chewing_key;
+ key = content_table[index->m_table_index].m_chewing_key;
+ assert(key.get_table_index() == index->m_table_index);
return true;
}
return 0 > strcmp(lhs.m_chewing_input, rhs.m_chewing_input);
}
-static inline bool search_chewing_index(guint32 options, const char * chewing,
- ChewingKey & key,
- ChewingKeyRest & key_rest){
+static inline bool search_chewing_index(pinyin_option_t options,
+ const char * chewing,
+ ChewingKey & key){
chewing_index_item_t item;
memset(&item, 0, sizeof(item));
item.m_chewing_input = chewing;
if (!check_chewing_options(options, index))
return false;
- key_rest.m_table_index = index->m_table_index;
- key = content_table[key_rest.m_table_index].m_chewing_key;
+ key = content_table[index->m_table_index].m_chewing_key;
+ assert(key.get_table_index() == index->m_table_index);
return true;
}
}
-bool FullPinyinParser2::parse_one_key (guint32 options, ChewingKey & key,
- ChewingKeyRest & key_rest,
+bool FullPinyinParser2::parse_one_key (pinyin_option_t options,
+ ChewingKey & key,
const char * pinyin, int len) const {
/* "'" are not accepted in parse_one_key. */
- assert(NULL == strchr(pinyin, '\''));
gchar * input = g_strndup(pinyin, len);
+ assert(NULL == strchr(input, '\''));
guint16 tone = CHEWING_ZERO_TONE; guint16 tone_pos = 0;
guint16 parsed_len = len;
- key = ChewingKey(); key_rest = ChewingKeyRest();
+ key = ChewingKey();
if (options & USE_TONE) {
/* find the tone in the last character. */
/* Note: optimize here? */
input[parsed_len] = '\0';
- if (!search_pinyin_index(options, input, key, key_rest))
- --parsed_len;
+ if (!search_pinyin_index(options, input, key)) {
+ g_free(input);
+ return false;
+ }
if (options & USE_TONE) {
/* post processing tone. */
}
}
- key_rest.m_raw_begin = 0; key_rest.m_raw_end = parsed_len;
g_free(input);
return parsed_len == len;
}
-int FullPinyinParser2::parse (guint32 options, ChewingKeyVector & keys,
+int FullPinyinParser2::parse (pinyin_option_t options, ChewingKeyVector & keys,
ChewingKeyRestVector & key_rests,
const char *str, int len) const {
int i;
gchar * input = g_strndup(str, len);
parse_value_t * curstep = NULL, * nextstep = NULL;
- for (i = 0; i < len; ) {
+ for (i = 0; i < len; ++i) {
if (input[i] == '\'') {
curstep = &g_array_index(m_parse_steps, parse_value_t, i);
nextstep = &g_array_index(m_parse_steps, parse_value_t, i + 1);
break;
}
next_sep = k;
- i = next_sep;
}
/* dynamic programming here. */
- for (size_t m = i; m < next_sep; ++m) {
+ /* for (size_t m = i; m < next_sep; ++m) */
+ {
+ size_t m = i;
curstep = &g_array_index(m_parse_steps, parse_value_t, m);
size_t try_len = std_lite::min
(m + max_full_pinyin_length, next_sep);
ChewingKey key; ChewingKeyRest rest;
bool parsed = parse_one_key
- (options, key, rest, onepinyin, onepinyinlen);
+ (options, key, onepinyin, onepinyinlen);
rest.m_raw_begin = m; rest.m_raw_end = n;
if (!parsed)
continue;
+
+ //printf("onepinyin:%s len:%d\n", onepinyin, onepinyinlen);
+
value.m_key = key; value.m_key_rest = rest;
value.m_num_keys = curstep->m_num_keys + 1;
value.m_parsed_len = curstep->m_parsed_len + onepinyinlen;
value.m_last_step = m;
/* save next step */
+ /* no previous result */
if (-1 == nextstep->m_last_step)
*nextstep = value;
+ /* prefer the longest pinyin */
if (value.m_parsed_len > nextstep->m_parsed_len)
*nextstep = value;
+ /* prefer the shortest keys with the same pinyin length */
if (value.m_parsed_len == nextstep->m_parsed_len &&
value.m_num_keys < nextstep->m_num_keys)
*nextstep = value;
+
+ /* handle with the same pinyin length and the number of keys */
+ if (value.m_parsed_len == nextstep->m_parsed_len &&
+ value.m_num_keys == nextstep->m_num_keys) {
+
+#if 0
+ /* prefer the complete pinyin with shengmu
+ * over without shengmu,
+ * ex: "kaneiji" -> "ka'nei'ji".
+ */
+ if ((value.m_key.m_initial != CHEWING_ZERO_INITIAL &&
+ !(value.m_key.m_middle == CHEWING_ZERO_MIDDLE &&
+ value.m_key.m_final == CHEWING_ZERO_FINAL)) &&
+ nextstep->m_key.m_initial == CHEWING_ZERO_INITIAL)
+ *nextstep = value;
+
+ /* prefer the complete pinyin 'er'
+ * over the in-complete pinyin 'r',
+ * ex: "xierqi" -> "xi'er'qi."
+ */
+ if ((value.m_key.m_initial == CHEWING_ZERO_INITIAL &&
+ value.m_key.m_middle == CHEWING_ZERO_MIDDLE &&
+ value.m_key.m_final == CHEWING_ER) &&
+ (nextstep->m_key.m_initial == CHEWING_R &&
+ nextstep->m_key.m_middle == CHEWING_ZERO_MIDDLE &&
+ nextstep->m_key.m_final == CHEWING_ZERO_FINAL))
+ *nextstep = value;
+#endif
+
+ /* prefer the 'a' at the end of clause,
+ * ex: "zheyanga$" -> "zhe'yang'a$".
+ */
+ if (value.m_parsed_len == len &&
+ (nextstep->m_key.m_initial != CHEWING_ZERO_INITIAL &&
+ nextstep->m_key.m_final == CHEWING_A) &&
+ (value.m_key.m_initial == CHEWING_ZERO_INITIAL &&
+ value.m_key.m_middle == CHEWING_ZERO_MIDDLE &&
+ value.m_key.m_final == CHEWING_A))
+ *nextstep = value;
+ }
}
}
}
/* post processing for re-split table. */
if (options & USE_RESPLIT_TABLE) {
- post_process(options, keys, key_rests);
+ post_process2(options, keys, key_rests, str, len);
}
g_free(input);
gint16 pos = curstep->m_num_keys - 1;
/* skip "'" */
- if (0 != curstep->m_key_rest.m_table_index) {
+ if (0 != curstep->m_key.get_table_index()) {
ChewingKey * key = &g_array_index(keys, ChewingKey, pos);
ChewingKeyRest * rest = &g_array_index
(key_rests, ChewingKeyRest, pos);
return parsed_len;
}
-
-bool FullPinyinParser2::post_process(guint32 options,
- ChewingKeyVector & keys,
- ChewingKeyRestVector & key_rests) const {
+bool FullPinyinParser2::post_process2(pinyin_option_t options,
+ ChewingKeyVector & keys,
+ ChewingKeyRestVector & key_rests,
+ const char * str,
+ int len) const {
int i;
assert(keys->len == key_rests->len);
- gint16 num_keys = keys->len;
+ gint num_keys = keys->len;
ChewingKey * cur_key = NULL, * next_key = NULL;
ChewingKeyRest * cur_rest = NULL, * next_rest = NULL;
- guint16 cur_tone = CHEWING_ZERO_TONE, next_tone = CHEWING_ZERO_TONE;
+ guint16 next_tone = CHEWING_ZERO_TONE;
for (i = 0; i < num_keys - 1; ++i) {
cur_rest = &g_array_index(key_rests, ChewingKeyRest, i);
cur_key = &g_array_index(keys, ChewingKey, i);
next_key = &g_array_index(keys, ChewingKey, i + 1);
+ /* some tone here */
+ if (CHEWING_ZERO_TONE != cur_key->m_tone)
+ continue;
+
+ /* back up tone */
if (options & USE_TONE) {
- cur_tone = cur_key->m_tone;
next_tone = next_key->m_tone;
- cur_key->m_tone = next_key->m_tone = CHEWING_ZERO_TONE;
+ if (CHEWING_ZERO_TONE != next_tone) {
+ next_key->m_tone = CHEWING_ZERO_TONE;
+ next_rest->m_raw_end --;
+ }
}
/* lookup re-split table */
- size_t k;
const resplit_table_item_t * item = NULL;
- for (k = 0; k < G_N_ELEMENTS(resplit_table); ++k) {
- item = resplit_table + k;
+
+ item = retrieve_resplit_item_by_original_pinyins
+ (options, cur_key, cur_rest, next_key, next_rest, str, len);
+
+ if (item) {
/* no ops */
if (item->m_orig_freq >= item->m_new_freq)
continue;
- /* use pinyin_exact_compare2 here. */
- if (0 == pinyin_exact_compare2(item->m_orig_keys,
- cur_key, 2))
- break;
+ /* do re-split */
+ const char * onepinyin = str + cur_rest->m_raw_begin;
+ size_t len = strlen(item->m_new_keys[0]);
- }
+ assert(parse_one_key(options, *cur_key, onepinyin, len));
+ cur_rest->m_raw_end = cur_rest->m_raw_begin + len;
- /* find the match */
- if (k < G_N_ELEMENTS(resplit_table)) {
- /* do re-split */
- item = resplit_table + k;
- *cur_key = item->m_new_keys[0];
- *next_key = item->m_new_keys[1];
- /* assumes only moved one char in gen_all_resplit script. */
- cur_rest->m_raw_end --;
- next_rest->m_raw_begin --;
+ next_rest->m_raw_begin = cur_rest->m_raw_end;
+ onepinyin = str + next_rest->m_raw_begin;
+ len = strlen(item->m_new_keys[1]);
+
+ assert(parse_one_key(options, *next_key, onepinyin, len));
}
- /* save back tones */
+ /* restore tones */
if (options & USE_TONE) {
- cur_key->m_tone = cur_tone;
- next_key->m_tone = next_tone;
+ if (CHEWING_ZERO_TONE != next_tone) {
+ next_key->m_tone = next_tone;
+ next_rest->m_raw_end ++;
+ }
}
}
return true;
}
+const divided_table_item_t * FullPinyinParser2::retrieve_divided_item
+(pinyin_option_t options, ChewingKey * key, ChewingKeyRest * rest,
+ const char * str, int len) const {
+
+ /* lookup divided table */
+ size_t k;
+ const divided_table_item_t * item = NULL;
+ for (k = 0; k < G_N_ELEMENTS(divided_table); ++k) {
+ item = divided_table + k;
+
+ const char * onepinyin = str + rest->m_raw_begin;
+ size_t len = strlen(item->m_orig_key);
+
+ if (rest->length() != len)
+ continue;
+
+ if (0 == strncmp(onepinyin, item->m_orig_key, len))
+ break;
+ }
+
+ /* found the match */
+ if (k < G_N_ELEMENTS(divided_table)) {
+ /* do divided */
+ item = divided_table + k;
+ return item;
+ }
+
+ return NULL;
+}
+
+
+const resplit_table_item_t * FullPinyinParser2::retrieve_resplit_item_by_original_pinyins
+(pinyin_option_t options,
+ ChewingKey * cur_key, ChewingKeyRest * cur_rest,
+ ChewingKey * next_key, ChewingKeyRest * next_rest,
+ const char * str, int len) const{
+ /* lookup re-split table */
+ size_t k;
+ const resplit_table_item_t * item = NULL;
+
+ for (k = 0; k < G_N_ELEMENTS(resplit_table); ++k) {
+ item = resplit_table + k;
+
+ const char * onepinyin = str + cur_rest->m_raw_begin;
+ size_t len = strlen(item->m_orig_keys[0]);
+
+ if (cur_rest->length() != len)
+ continue;
+
+ if (0 != strncmp(onepinyin, item->m_orig_keys[0], len))
+ continue;
+
+ onepinyin = str + next_rest->m_raw_begin;
+ len = strlen(item->m_orig_keys[1]);
+
+ if (next_rest->length() != len)
+ continue;
+
+ if (0 == strncmp(onepinyin, item->m_orig_keys[1], len))
+ break;
+ }
+
+ /* found the match */
+ if (k < G_N_ELEMENTS(resplit_table)) {
+ item = resplit_table + k;
+ return item;
+ }
+
+ return NULL;
+}
+
+const resplit_table_item_t * FullPinyinParser2::retrieve_resplit_item_by_resplit_pinyins
+(pinyin_option_t options,
+ ChewingKey * cur_key, ChewingKeyRest * cur_rest,
+ ChewingKey * next_key, ChewingKeyRest * next_rest,
+ const char * str, int len) const {
+ /* lookup divide table */
+ size_t k;
+ const resplit_table_item_t * item = NULL;
+
+ for (k = 0; k < G_N_ELEMENTS(resplit_table); ++k) {
+ item = resplit_table + k;
+
+ const char * onepinyin = str + cur_rest->m_raw_begin;
+ size_t len = strlen(item->m_new_keys[0]);
+
+ if (cur_rest->length() != len)
+ continue;
+
+ if (0 != strncmp(onepinyin, item->m_new_keys[0], len))
+ continue;
+
+ onepinyin = str + next_rest->m_raw_begin;
+ len = strlen(item->m_new_keys[1]);
+
+ if (next_rest->length() != len)
+ continue;
+
+ if (0 == strncmp(onepinyin, item->m_new_keys[1], len))
+ break;
+ }
+
+ /* found the match */
+ if (k < G_N_ELEMENTS(resplit_table)) {
+ item = resplit_table + k;
+ return item;
+ }
+
+ return NULL;
+}
+
#define IS_KEY(x) (('a' <= x && x <= 'z') || x == ';')
-bool DoublePinyinParser2::parse_one_key(guint32 options, ChewingKey & key,
- ChewingKeyRest & key_rest,
+bool DoublePinyinParser2::parse_one_key(pinyin_option_t options,
+ ChewingKey & key,
const char *str, int len) const {
+ options &= ~(PINYIN_CORRECT_ALL|PINYIN_AMB_ALL);
if (1 == len) {
if (!(options & PINYIN_INCOMPLETE))
if (NULL == sheng || strcmp(sheng, "'") == 0)
return false;
- if (search_pinyin_index(options, sheng, key, key_rest)) {
- key_rest.m_raw_begin = 0; key_rest.m_raw_end = len;
+ if (search_pinyin_index(options, sheng, key)) {
return true;
} else {
return false;
}
ChewingTone tone = CHEWING_ZERO_TONE;
- options &= ~(PINYIN_CORRECT_ALL|PINYIN_AMB_ALL);
+ options &= ~(PINYIN_INCOMPLETE|CHEWING_INCOMPLETE);
+ options |= PINYIN_CORRECT_UE_VE | PINYIN_CORRECT_V_U;
/* parse tone */
if (3 == len) {
charid = ch == ';' ? 26 : ch - 'a';
/* first yunmu */
const char * yun = m_yunmu_table[charid].m_yunmus[0];
+ if (NULL == yun)
+ return false;
+
gchar * pinyin = g_strdup_printf("%s%s", sheng, yun);
- if (search_pinyin_index(options, pinyin, key, key_rest)) {
- key_rest.m_raw_begin = 0; key_rest.m_raw_end = len;
+ if (search_pinyin_index(options, pinyin, key)) {
key.m_tone = tone;
g_free(pinyin);
return true;
/* second yunmu */
yun = m_yunmu_table[charid].m_yunmus[1];
+ if (NULL == yun)
+ return false;
+
pinyin = g_strdup_printf("%s%s", sheng, yun);
- if (search_pinyin_index(options, pinyin, key, key_rest)) {
- key_rest.m_raw_begin = 0; key_rest.m_raw_end = len;
+ if (search_pinyin_index(options, pinyin, key)) {
key.m_tone = tone;
g_free(pinyin);
return true;
/* only 'a'-'z' and ';' are accepted here. */
-int DoublePinyinParser2::parse(guint32 options, ChewingKeyVector & keys,
+int DoublePinyinParser2::parse(pinyin_option_t options, ChewingKeyVector & keys,
ChewingKeyRestVector & key_rests,
const char *str, int len) const {
g_array_set_size(keys, 0);
int maximum_len = 0; int i;
/* probe the longest possible double pinyin string. */
for (i = 0; i < len; ++i) {
- if (!IS_KEY(str[i]))
+ const char ch = str[i];
+ if (!(IS_KEY(ch) || ('0' < ch && ch <= '5')))
break;
}
maximum_len = i;
ChewingKey key; ChewingKeyRest key_rest;
for (; i > 0; --i) {
- bool success = parse_one_key(options, key, key_rest, cur_str, i);
+ bool success = parse_one_key(options, key, cur_str, i);
if (success)
break;
}
/* the chewing string must be freed with g_free. */
static bool search_chewing_symbols(const chewing_symbol_item_t * symbol_table,
- const char key, char ** chewing) {
+ const char key, const char ** chewing) {
+ *chewing = NULL;
/* just iterate the table, as we only have < 50 items. */
while (symbol_table->m_input != '\0') {
if (symbol_table->m_input == key) {
- *chewing = g_strdup(symbol_table->m_chewing);
+ *chewing = symbol_table->m_chewing;
return true;
}
symbol_table ++;
static bool search_chewing_tones(const chewing_tone_item_t * tone_table,
const char key, char * tone) {
+ *tone = CHEWING_ZERO_TONE;
/* just iterate the table, as we only have < 10 items. */
while (tone_table->m_input != '\0') {
if (tone_table->m_input == key) {
}
-bool ChewingParser2::parse_one_key(guint32 options, ChewingKey & key, ChewingKeyRest & key_rest, const char *str, int len) const {
+bool ChewingParser2::parse_one_key(pinyin_option_t options,
+ ChewingKey & key,
+ const char *str, int len) const {
+ options &= ~(PINYIN_CORRECT_ALL|PINYIN_AMB_ALL);
char tone = CHEWING_ZERO_TONE;
int symbols_len = len;
}
int i;
- gchar * chewing = NULL, * onechar = NULL;
+ gchar * chewing = NULL; const char * onechar = NULL;
/* probe the possible chewing map in the rest of str. */
for (i = 0; i < symbols_len; ++i) {
if (!search_chewing_symbols(m_symbol_table, str[i], &onechar)) {
- g_free(onechar);
g_free(chewing);
return false;
}
chewing = g_strconcat(chewing, onechar, NULL);
g_free(tmp);
}
- g_free(onechar);
}
/* search the chewing in the chewing index table. */
- if (search_chewing_index(options, chewing, key, key_rest)) {
- key_rest.m_raw_begin = 0; key_rest.m_raw_end = len;
+ if (chewing && search_chewing_index(options, chewing, key)) {
/* save back tone if available. */
key.m_tone = tone;
g_free(chewing);
/* only characters in chewing keyboard scheme are accepted here. */
-int ChewingParser2::parse(guint32 options, ChewingKeyVector & keys,
+int ChewingParser2::parse(pinyin_option_t options, ChewingKeyVector & keys,
ChewingKeyRestVector & key_rests,
const char *str, int len) const {
g_array_set_size(keys, 0);
int maximum_len = 0; int i;
/* probe the longest possible chewing string. */
for (i = 0; i < len; ++i) {
- if (!in_chewing_scheme(str[i]))
+ if (!in_chewing_scheme(options, str[i], NULL))
break;
}
maximum_len = i;
ChewingKey key; ChewingKeyRest key_rest;
for (; i > 0; --i) {
- bool success = parse_one_key(options, key, key_rest, cur_str, i);
+ bool success = parse_one_key(options, key, cur_str, i);
if (success)
break;
}
}
-bool ChewingParser2::in_chewing_scheme(const char key) const {
- gchar * chewing = NULL;
+bool ChewingParser2::in_chewing_scheme(pinyin_option_t options,
+ const char key, const char ** symbol)
+ const {
+ const gchar * chewing = NULL;
char tone = CHEWING_ZERO_TONE;
- bool retval = search_chewing_symbols(m_symbol_table, key, &chewing) ||
- search_chewing_tones(m_tone_table, key, &tone);
- g_free(chewing);
+ if (search_chewing_symbols(m_symbol_table, key, &chewing)) {
+ if (symbol)
+ *symbol = chewing;
+ return true;
+ }
- return retval;
+ if (!(options & USE_TONE))
+ return false;
+
+ if (search_chewing_tones(m_tone_table, key, &tone)) {
+ if (symbol)
+ *symbol = chewing_tone_table[tone];
+ return true;
+ }
+
+ return false;
}