static void convert_numerative_pron(const char *list[], NJDNode * node1, NJDNode * node2)
{
int i, j;
int type = 0;
const char *str = NJDNode_get_string(node1);
char buff[MAXBUFLEN];
if (strcmp(str, "*") == 0)
return;
for (i = 0; list[i] != NULL; i += 2) {
if (strcmp(list[i], str) == 0) {
type = atoi(list[i + 1]);
break;
}
}
if (type == 1) {
for (i = 0; njd_set_digit_rule_voiced_sound_symbol_list[i] != NULL; i += 2) {
str = NJDNode_get_pron(node2);
j = strtopcmp(str, njd_set_digit_rule_voiced_sound_symbol_list[i]);
if (j >= 0) {
strcpy(buff, njd_set_digit_rule_voiced_sound_symbol_list[i + 1]);
strcat(buff, &str[j]);
NJDNode_set_pron(node2, buff);
break;
}
}
} else if (type == 2) {
for (i = 0; njd_set_digit_rule_semivoiced_sound_symbol_list[i] != NULL; i += 2) {
str = NJDNode_get_pron(node2);
j = strtopcmp(str, njd_set_digit_rule_semivoiced_sound_symbol_list[i]);
if (j >= 0) {
strcpy(buff, njd_set_digit_rule_semivoiced_sound_symbol_list[i + 1]);
strcat(buff, &str[j]);
NJDNode_set_pron(node2, buff);
break;
}
}
}
}
static void convert_digit_pron(const char *list[], NJDNode * node)
{
int i;
const char *str = NJDNode_get_string(node);
if (strcmp(str, "*") == 0)
return;
for (i = 0; list[i] != NULL; i += 4) {
if (strcmp(list[i], str) == 0) {
NJDNode_set_pron(node, (char *) list[i + 1]);
NJDNode_set_acc(node, atoi(list[i + 2]));
NJDNode_set_mora_size(node, atoi(list[i + 3]));
return;
}
}
}
void NJDNode_copy(NJDNode * node1, NJDNode * node2)
{
NJDNode_set_string(node1, node2->string);
NJDNode_set_pos(node1, node2->pos);
NJDNode_set_pos_group1(node1, node2->pos_group1);
NJDNode_set_pos_group2(node1, node2->pos_group2);
NJDNode_set_pos_group3(node1, node2->pos_group3);
NJDNode_set_ctype(node1, node2->ctype);
NJDNode_set_cform(node1, node2->cform);
NJDNode_set_orig(node1, node2->orig);
NJDNode_set_read(node1, node2->read);
NJDNode_set_pron(node1, node2->pron);
NJDNode_set_acc(node1, node2->acc);
NJDNode_set_mora_size(node1, node2->mora_size);
NJDNode_set_chain_rule(node1, node2->chain_rule);
NJDNode_set_chain_flag(node1, node2->chain_flag);
}
void njd_set_pronunciation(NJD * njd)
{
NJDNode *node;
const char *str;
int i, j = 0;
int pos;
int len;
for (node = njd->head; node != NULL; node = node->next) {
if (NJDNode_get_mora_size(node) == 0) {
NJDNode_set_read(node, NULL);
NJDNode_set_pron(node, NULL);
if (strcmp(NJDNode_get_pos(node), NJD_SET_PRONUNCIATION_KIGOU) == 0 || strcmp(NJDNode_get_pos_group1(node), NJD_SET_PRONUNCIATION_KAZU) == 0) { /* for symbol */
for (i = 0; njd_set_pronunciation_symbol_list[i] != NULL; i += 2)
if (strcmp(NJDNode_get_string(node), njd_set_pronunciation_symbol_list[i]) == 0) {
NJDNode_set_read(node, (char *) njd_set_pronunciation_symbol_list[i + 1]);
NJDNode_set_pron(node, (char *) njd_set_pronunciation_symbol_list[i + 1]);
break;
}
} else if (strcmp(NJDNode_get_pron(node), "*") == 0) { /* for others */
str = NJDNode_get_string(node);
len = strlen(str);
for (pos = 0; pos < len;) {
for (i = 0, j = 0; njd_set_pronunciation_list[i] != NULL; i += 3) {
j = strtopcmp(&str[pos], njd_set_pronunciation_list[i]);
if (j > 0)
break;
}
if (j > 0) {
pos += j;
NJDNode_add_read(node, (char *) njd_set_pronunciation_list[i + 1]);
NJDNode_add_pron(node, (char *) njd_set_pronunciation_list[i + 1]);
NJDNode_add_mora_size(node, atoi(njd_set_pronunciation_list[i + 2]));
} else {
pos++;
}
}
}
}
}
NJD_remove_silent_node(njd);
for (node = njd->head; node != NULL; node = node->next) {
if (node->next != NULL
&& strcmp(NJDNode_get_pron(node->next), NJD_SET_PRONUNCIATION_U) == 0
&& strcmp(NJDNode_get_pos(node->next), NJD_SET_PRONUNCIATION_JODOUSHI) == 0
&& (strcmp(NJDNode_get_pos(node), NJD_SET_PRONUNCIATION_DOUSHI) == 0
|| strcmp(NJDNode_get_pos(node), NJD_SET_PRONUNCIATION_JODOUSHI) == 0)
&& NJDNode_get_mora_size(node) > 0) {
NJDNode_set_pron(node->next, NJD_SET_PRONUNCIATION_CHOUON);
}
if (node->next != NULL
&& strcmp(NJDNode_get_pos(node), NJD_SET_PRONUNCIATION_JODOUSHI) == 0
&& strcmp(NJDNode_get_string(node->next), NJD_SET_PRONUNCIATION_QUESTION) == 0) {
if (strcmp(NJDNode_get_string(node), NJD_SET_PRONUNCIATION_DESU_STR) == 0)
NJDNode_set_pron(node, NJD_SET_PRONUNCIATION_DESU_PRON);
else if (strcmp(NJDNode_get_string(node), NJD_SET_PRONUNCIATION_MASU_STR) == 0)
NJDNode_set_pron(node, NJD_SET_PRONUNCIATION_MASU_PRON);
}
}
}
void njd_set_digit(NJD * njd)
{
int i, j;
NJDNode *s = NULL;
NJDNode *e = NULL;
NJDNode *node;
int find = 0;
/* convert digit sequence */
for (node = njd->head; node != NULL; node = node->next) {
if (find == 0 && strcmp(NJDNode_get_pos_group1(node), NJD_SET_DIGIT_KAZU) == 0)
find = 1;
if (get_digit(node, 1) >= 0) {
if (s == NULL)
s = node;
if (node == njd->tail)
e = node;
} else {
if (s != NULL)
e = node->prev;
}
if (s != NULL && e != NULL) {
convert_digit_sequence(s, e);
s = e = NULL;
}
}
if (find == 0)
return;
NJD_remove_silent_node(njd);
if (njd->head == NULL)
return;
for (node = njd->head->next; node != NULL && node->next != NULL; node = node->next) {
if (strcmp(NJDNode_get_string(node), "*") != 0
&& strcmp(NJDNode_get_string(node->prev), "*") != 0
&& (strcmp(NJDNode_get_string(node), NJD_SET_DIGIT_TEN1) == 0
|| strcmp(NJDNode_get_string(node), NJD_SET_DIGIT_TEN2) == 0)
&& strcmp(NJDNode_get_pos_group1(node->prev), NJD_SET_DIGIT_KAZU) == 0
&& strcmp(NJDNode_get_pos_group1(node->next), NJD_SET_DIGIT_KAZU) == 0) {
NJDNode_load(node, NJD_SET_DIGIT_TEN_FEATURE);
NJDNode_set_chain_flag(node, 1);
if (strcmp(NJDNode_get_string(node->prev), NJD_SET_DIGIT_ZERO1) == 0
|| strcmp(NJDNode_get_string(node->prev), NJD_SET_DIGIT_ZERO2) == 0) {
NJDNode_set_pron(node->prev, NJD_SET_DIGIT_ZERO_BEFORE_DP);
NJDNode_set_mora_size(node->prev, 2);
} else if (strcmp(NJDNode_get_string(node->prev), NJD_SET_DIGIT_TWO) == 0) {
NJDNode_set_pron(node->prev, NJD_SET_DIGIT_TWO_BEFORE_DP);
NJDNode_set_mora_size(node->prev, 2);
} else if (strcmp(NJDNode_get_string(node->prev), NJD_SET_DIGIT_FIVE) == 0) {
NJDNode_set_pron(node->prev, NJD_SET_DIGIT_FIVE_BEFORE_DP);
NJDNode_set_mora_size(node->prev, 2);
}
}
}
for (node = njd->head->next; node != NULL; node = node->next) {
if (strcmp(NJDNode_get_pos_group1(node->prev), NJD_SET_DIGIT_KAZU) == 0) {
if (strcmp(NJDNode_get_pos_group2(node), NJD_SET_DIGIT_JOSUUSHI) == 0
|| strcmp(NJDNode_get_pos_group1(node), NJD_SET_DIGIT_FUKUSHIKANOU) == 0) {
/* convert digit pron */
if (search_numerative_class(njd_set_digit_rule_numerative_class1b, node) == 1)
convert_digit_pron(njd_set_digit_rule_conv_table1b, node->prev);
else if (search_numerative_class(njd_set_digit_rule_numerative_class1c1, node) == 1)
convert_digit_pron(njd_set_digit_rule_conv_table1c1, node->prev);
else if (search_numerative_class(njd_set_digit_rule_numerative_class1c2, node) == 1)
convert_digit_pron(njd_set_digit_rule_conv_table1c2, node->prev);
else if (search_numerative_class(njd_set_digit_rule_numerative_class1d, node) == 1)
convert_digit_pron(njd_set_digit_rule_conv_table1d, node->prev);
else if (search_numerative_class(njd_set_digit_rule_numerative_class1e, node) == 1)
convert_digit_pron(njd_set_digit_rule_conv_table1e, node->prev);
else if (search_numerative_class(njd_set_digit_rule_numerative_class1f, node) == 1)
convert_digit_pron(njd_set_digit_rule_conv_table1f, node->prev);
else if (search_numerative_class(njd_set_digit_rule_numerative_class1g, node) == 1)
convert_digit_pron(njd_set_digit_rule_conv_table1g, node->prev);
else if (search_numerative_class(njd_set_digit_rule_numerative_class1h, node) == 1)
convert_digit_pron(njd_set_digit_rule_conv_table1h, node->prev);
else if (search_numerative_class(njd_set_digit_rule_numerative_class1i, node) == 1)
convert_digit_pron(njd_set_digit_rule_conv_table1i, node->prev);
else if (search_numerative_class(njd_set_digit_rule_numerative_class1j, node) == 1)
convert_digit_pron(njd_set_digit_rule_conv_table1j, node->prev);
else if (search_numerative_class(njd_set_digit_rule_numerative_class1k, node) == 1)
convert_digit_pron(njd_set_digit_rule_conv_table1k, node->prev);
/* convert numerative pron */
if (search_numerative_class(njd_set_digit_rule_numerative_class2b, node) == 1)
convert_numerative_pron(njd_set_digit_rule_conv_table2b, node->prev, node);
else if (search_numerative_class(njd_set_digit_rule_numerative_class2c, node) == 1)
convert_numerative_pron(njd_set_digit_rule_conv_table2c, node->prev, node);
else if (search_numerative_class(njd_set_digit_rule_numerative_class2d, node) == 1)
convert_numerative_pron(njd_set_digit_rule_conv_table2d, node->prev, node);
else if (search_numerative_class(njd_set_digit_rule_numerative_class2e, node) == 1)
convert_numerative_pron(njd_set_digit_rule_conv_table2e, node->prev, node);
else if (search_numerative_class(njd_set_digit_rule_numerative_class2f, node) == 1)
convert_numerative_pron(njd_set_digit_rule_conv_table2f, node->prev, node);
/* modify accent phrase */
NJDNode_set_chain_flag(node->prev, 0);
NJDNode_set_chain_flag(node, 1);
}
}
}
for (node = njd->head->next; node != NULL; node = node->next) {
if (strcmp(NJDNode_get_pos_group1(node->prev), NJD_SET_DIGIT_KAZU) == 0) {
if (strcmp(NJDNode_get_pos_group1(node), NJD_SET_DIGIT_KAZU) == 0
&& NJDNode_get_string(node->prev) != NULL && NJDNode_get_string(node) != NULL) {
/* modify accent phrase */
find = 0;
for (i = 0; njd_set_digit_rule_numeral_list4[i] != NULL; i++) {
if (strcmp(NJDNode_get_string(node->prev), njd_set_digit_rule_numeral_list4[i]) == 0) {
for (j = 0; njd_set_digit_rule_numeral_list5[j] != NULL; j++) {
if (strcmp(NJDNode_get_string(node), njd_set_digit_rule_numeral_list5[j]) == 0) {
NJDNode_set_chain_flag(node->prev, 0);
NJDNode_set_chain_flag(node, 1);
find = 1;
break;
}
}
break;
}
}
if (find == 0) {
for (i = 0; njd_set_digit_rule_numeral_list5[i] != NULL; i++) {
if (strcmp(NJDNode_get_string(node->prev), njd_set_digit_rule_numeral_list5[i]) ==
0) {
for (j = 0; njd_set_digit_rule_numeral_list4[j] != NULL; j++) {
if (strcmp(NJDNode_get_string(node), njd_set_digit_rule_numeral_list4[j]) ==
0) {
NJDNode_set_chain_flag(node, 0);
break;
}
}
break;
}
}
}
}
if (search_numerative_class(njd_set_digit_rule_numeral_list8, node) == 1)
convert_digit_pron(njd_set_digit_rule_numeral_list9, node->prev);
if (search_numerative_class(njd_set_digit_rule_numeral_list10, node) == 1)
convert_digit_pron(njd_set_digit_rule_numeral_list11, node->prev);
if (search_numerative_class(njd_set_digit_rule_numeral_list6, node) == 1)
convert_numerative_pron(njd_set_digit_rule_numeral_list7, node->prev, node);
}
}
for (node = njd->head; node != NULL; node = node->next) {
if (node->next != NULL &&
strcmp(NJDNode_get_string(node->next), "*") != 0 &&
strcmp(NJDNode_get_pos_group1(node), NJD_SET_DIGIT_KAZU) == 0 &&
(node->prev == NULL
|| strcmp(NJDNode_get_pos_group1(node->prev), NJD_SET_DIGIT_KAZU) != 0)
&& (strcmp(NJDNode_get_pos_group2(node->next), NJD_SET_DIGIT_JOSUUSHI) == 0
|| strcmp(NJDNode_get_pos_group1(node->next), NJD_SET_DIGIT_FUKUSHIKANOU) == 0)) {
/* convert class3 */
for (i = 0; njd_set_digit_rule_numerative_class3[i] != NULL; i += 2) {
if (strcmp(NJDNode_get_string(node->next), njd_set_digit_rule_numerative_class3[i]) == 0
&& strcmp(NJDNode_get_read(node->next),
njd_set_digit_rule_numerative_class3[i + 1]) == 0) {
for (j = 0; njd_set_digit_rule_conv_table3[j] != NULL; j += 4) {
if (strcmp(NJDNode_get_string(node), njd_set_digit_rule_conv_table3[j]) == 0) {
NJDNode_set_read(node, (char *) njd_set_digit_rule_conv_table3[j + 1]);
NJDNode_set_pron(node, (char *) njd_set_digit_rule_conv_table3[j + 1]);
NJDNode_set_acc(node, atoi(njd_set_digit_rule_conv_table3[j + 2]));
NJDNode_set_mora_size(node, atoi(njd_set_digit_rule_conv_table3[j + 3]));
break;
}
}
break;
}
}
/* person */
if (strcmp(NJDNode_get_string(node->next), NJD_SET_DIGIT_NIN) == 0) {
for (i = 0; njd_set_digit_rule_conv_table4[i] != NULL; i += 2) {
if (strcmp(NJDNode_get_string(node), njd_set_digit_rule_conv_table4[i]) == 0) {
NJDNode_load(node, (char *) njd_set_digit_rule_conv_table4[i + 1]);
NJDNode_set_pron(node->next, NULL);
break;
}
}
}
/* the day of month */
if (strcmp(NJDNode_get_string(node->next), NJD_SET_DIGIT_NICHI) == 0
&& strcmp(NJDNode_get_string(node), "*") != 0) {
if (node->prev != NULL
&& strstr(NJDNode_get_string(node->prev), NJD_SET_DIGIT_GATSU) != NULL
&& strcmp(NJDNode_get_string(node), NJD_SET_DIGIT_ONE) == 0) {
NJDNode_load(node, NJD_SET_DIGIT_TSUITACHI);
NJDNode_set_pron(node->next, NULL);
} else {
for (i = 0; njd_set_digit_rule_conv_table5[i] != NULL; i += 2) {
if (strcmp(NJDNode_get_string(node), njd_set_digit_rule_conv_table5[i]) == 0) {
NJDNode_load(node, (char *) njd_set_digit_rule_conv_table5[i + 1]);
NJDNode_set_pron(node->next, NULL);
break;
}
}
}
} else if (strcmp(NJDNode_get_string(node->next), NJD_SET_DIGIT_NICHIKAN) == 0) {
for (i = 0; njd_set_digit_rule_conv_table6[i] != NULL; i += 2) {
if (strcmp(NJDNode_get_string(node), njd_set_digit_rule_conv_table6[i]) == 0) {
NJDNode_load(node, (char *) njd_set_digit_rule_conv_table6[i + 1]);
NJDNode_set_pron(node->next, NULL);
break;
}
}
}
}
}
for (node = njd->head; node != NULL; node = node->next) {
if ((node->prev == NULL
|| strcmp(NJDNode_get_pos_group1(node->prev), NJD_SET_DIGIT_KAZU) != 0)
&& node->next != NULL && node->next->next != NULL) {
if (strcmp(NJDNode_get_string(node), NJD_SET_DIGIT_TEN) == 0
&& strcmp(NJDNode_get_string(node->next), NJD_SET_DIGIT_FOUR) == 0) {
if (strcmp(NJDNode_get_string(node->next->next), NJD_SET_DIGIT_NICHI) == 0) {
NJDNode_load(node, NJD_SET_DIGIT_JUYOKKA);
NJDNode_set_pron(node->next, NULL);
NJDNode_set_pron(node->next->next, NULL);
} else if (strcmp(NJDNode_get_string(node->next->next), NJD_SET_DIGIT_NICHIKAN) == 0) {
NJDNode_load(node, NJD_SET_DIGIT_JUYOKKAKAN);
NJDNode_set_pron(node->next, NULL);
NJDNode_set_pron(node->next->next, NULL);
}
} else if (strcmp(NJDNode_get_string(node), NJD_SET_DIGIT_TWO) == 0
&& strcmp(NJDNode_get_string(node->next), NJD_SET_DIGIT_TEN) == 0) {
if (strcmp(NJDNode_get_string(node->next->next), NJD_SET_DIGIT_NICHI) == 0) {
NJDNode_load(node, NJD_SET_DITIT_HATSUKA);
NJDNode_set_pron(node->next, NULL);
NJDNode_set_pron(node->next->next, NULL);
} else if (strcmp(NJDNode_get_string(node->next->next), NJD_SET_DIGIT_NICHIKAN) == 0) {
NJDNode_load(node, NJD_SET_DIGIT_HATSUKAKAN);
NJDNode_set_pron(node->next, NULL);
NJDNode_set_pron(node->next->next, NULL);
} else if (strcmp(NJDNode_get_string(node->next->next), NJD_SET_DIGIT_FOUR) == 0
&& node->next->next->next != NULL) {
if (strcmp(NJDNode_get_string(node->next->next->next), NJD_SET_DIGIT_NICHI) == 0) {
NJDNode_load(node, NJD_SET_DIGIT_NIJU);
NJDNode_load(node->next, NJD_SET_DITIT_YOKKA);
NJDNode_set_pron(node->next->next, NULL);
NJDNode_set_pron(node->next->next->next, NULL);
} else if (strcmp(NJDNode_get_string(node->next->next->next), NJD_SET_DIGIT_NICHIKAN)
== 0) {
NJDNode_load(node, NJD_SET_DIGIT_NIJU);
NJDNode_load(node->next, NJD_SET_DIGIT_YOKKAKAN);
NJDNode_set_pron(node->next->next, NULL);
NJDNode_set_pron(node->next->next->next, NULL);
}
}
}
}
}
NJD_remove_silent_node(njd);
if (njd->head == NULL)
return;
}
static void convert_digit_sequence(NJDNode * start, NJDNode * end)
{
NJDNode *node;
NJDNode *newnode;
int digit;
int place = 0;
int index;
int size = 0;
int have = 0;
for (node = start; node != end->next; node = node->next)
size++;
if (size <= 1)
return;
if (get_digit_sequence_score(start, end) < 0) {
for (node = start, size = 0; node != end->next; node = node->next) {
if (strcmp(NJDNode_get_string(node), NJD_SET_DIGIT_ZERO1) == 0
|| strcmp(NJDNode_get_string(node), NJD_SET_DIGIT_ZERO2) == 0) {
NJDNode_set_pron(node, NJD_SET_DIGIT_ZERO_AFTER_DP);
NJDNode_set_mora_size(node, 2);
} else if (strcmp(NJDNode_get_string(node), NJD_SET_DIGIT_TWO) == 0) {
NJDNode_set_pron(node, NJD_SET_DIGIT_TWO_AFTER_DP);
NJDNode_set_mora_size(node, 2);
} else if (strcmp(NJDNode_get_string(node), NJD_SET_DIGIT_FIVE) == 0) {
NJDNode_set_pron(node, NJD_SET_DIGIT_FIVE_AFTER_DP);
NJDNode_set_mora_size(node, 2);
}
NJDNode_set_chain_rule(node, NULL);
if (size % 2 == 0) {
NJDNode_set_chain_flag(node, 0);
} else {
NJDNode_set_chain_flag(node, 1);
NJDNode_set_acc(node->prev, 3);
}
size++;
}
return;
}
index = size % 4;
if (index == 0)
index = 4;
if (size > index)
place = (size - index) / 4;
index--;
if (place > 17)
return;
for (node = start; node != end->next; node = node->next) {
digit = get_digit(node, 0);
if (index == 0) {
if (digit == 0) {
NJDNode_set_pron(node, NULL);
NJDNode_set_acc(node, 0);
NJDNode_set_mora_size(node, 0);
} else {
have = 1;
}
if (have == 1) {
if (place > 0) {
newnode = (NJDNode *) calloc(1, sizeof(NJDNode));
NJDNode_initialize(newnode);
NJDNode_load(newnode, (char *) njd_set_digit_rule_numeral_list3[place]);
node = NJDNode_insert(node, node->next, newnode);
}
have = 0;
}
place--;
} else {
if (digit <= 1) {
NJDNode_set_pron(node, NULL);
NJDNode_set_acc(node, 0);
NJDNode_set_mora_size(node, 0);
}
if (digit > 0) {
newnode = (NJDNode *) calloc(1, sizeof(NJDNode));
NJDNode_initialize(newnode);
NJDNode_load(newnode, (char *) njd_set_digit_rule_numeral_list2[index]);
node = NJDNode_insert(node, node->next, newnode);
have = 1;
}
}
index--;
if (index < 0)
index = 4 - 1;
}
}
void NJD_load_from_fp(NJD * njd, FILE * fp)
{
NJDNode *node = NULL;
char string[MAXBUFLEN];
char pos[MAXBUFLEN];
char pos_group1[MAXBUFLEN];
char pos_group2[MAXBUFLEN];
char pos_group3[MAXBUFLEN];
char ctype[MAXBUFLEN];
char cform[MAXBUFLEN];
char orig[MAXBUFLEN];
char read[MAXBUFLEN];
char pron[MAXBUFLEN];
char acc[MAXBUFLEN];
char mora_size[MAXBUFLEN];
char chain_rule[MAXBUFLEN];
char chain_flag[MAXBUFLEN];
if (fp == NULL) {
fprintf(stderr, "WARNING: NJD_load_from_fp() in njd.c: File pointer should not be null.");
return;
}
while (1) {
get_token_from_fp(fp, string, ',');
if (get_token_from_fp(fp, pos, ',') <= 0)
break;
if (get_token_from_fp(fp, pos_group1, ',') <= 0)
break;
if (get_token_from_fp(fp, pos_group2, ',') <= 0)
break;
if (get_token_from_fp(fp, pos_group3, ',') <= 0)
break;
if (get_token_from_fp(fp, ctype, ',') <= 0)
break;
if (get_token_from_fp(fp, cform, ',') <= 0)
break;
get_token_from_fp(fp, orig, ',');
get_token_from_fp(fp, read, ',');
get_token_from_fp(fp, pron, ',');
if (get_token_from_fp(fp, acc, '/') <= 0)
break;
if (get_token_from_fp(fp, mora_size, ',') <= 0)
break;
get_token_from_fp(fp, chain_rule, ',');
if (get_token_from_fp(fp, chain_flag, ',') <= 0)
break;
node = (NJDNode *) calloc(1, sizeof(NJDNode));
NJDNode_initialize(node);
NJDNode_set_string(node, string);
NJDNode_set_pos(node, pos);
NJDNode_set_pos_group1(node, pos_group1);
NJDNode_set_pos_group2(node, pos_group2);
NJDNode_set_pos_group3(node, pos_group3);
NJDNode_set_ctype(node, ctype);
NJDNode_set_cform(node, cform);
NJDNode_set_orig(node, orig);
NJDNode_set_read(node, read);
NJDNode_set_pron(node, pron);
NJDNode_set_acc(node, atoi(acc));
NJDNode_set_mora_size(node, atoi(mora_size));
NJDNode_set_chain_rule(node, chain_rule);
NJDNode_set_chain_flag(node, atoi(chain_flag));
NJD_push_node(njd, node);
}
}
void NJD_load(NJD * njd, char *str)
{
int i = 0;
NJDNode *node = NULL;
char string[MAXBUFLEN];
char pos[MAXBUFLEN];
char pos_group1[MAXBUFLEN];
char pos_group2[MAXBUFLEN];
char pos_group3[MAXBUFLEN];
char ctype[MAXBUFLEN];
char cform[MAXBUFLEN];
char orig[MAXBUFLEN];
char read[MAXBUFLEN];
char pron[MAXBUFLEN];
char acc[MAXBUFLEN];
char mora_size[MAXBUFLEN];
char chain_rule[MAXBUFLEN];
char chain_flag[MAXBUFLEN];
if (strlen(str) < 1) {
fprintf(stderr, "WARNING: NJD_load() in njd.c: Input string should not be empty.");
return;
}
while (1) {
get_token_from_string(str, &i, string, ',');
if (get_token_from_string(str, &i, pos, ',') <= 0)
break;
if (get_token_from_string(str, &i, pos_group1, ',') <= 0)
break;
if (get_token_from_string(str, &i, pos_group2, ',') <= 0)
break;
if (get_token_from_string(str, &i, pos_group3, ',') <= 0)
break;
if (get_token_from_string(str, &i, ctype, ',') <= 0)
break;
if (get_token_from_string(str, &i, cform, ',') <= 0)
break;
get_token_from_string(str, &i, orig, ',');
get_token_from_string(str, &i, read, ',');
get_token_from_string(str, &i, pron, ',');
if (get_token_from_string(str, &i, acc, '/') <= 0)
break;
if (get_token_from_string(str, &i, mora_size, ',') <= 0)
break;
get_token_from_string(str, &i, chain_rule, ',');
if (get_token_from_string(str, &i, chain_flag, ',') <= 0)
break;
node = (NJDNode *) calloc(1, sizeof(NJDNode));
NJDNode_initialize(node);
NJDNode_set_string(node, string);
NJDNode_set_pos(node, pos);
NJDNode_set_pos_group1(node, pos_group1);
NJDNode_set_pos_group2(node, pos_group2);
NJDNode_set_pos_group3(node, pos_group3);
NJDNode_set_ctype(node, ctype);
NJDNode_set_cform(node, cform);
NJDNode_set_orig(node, orig);
NJDNode_set_read(node, read);
NJDNode_set_pron(node, pron);
NJDNode_set_acc(node, atoi(acc));
NJDNode_set_mora_size(node, atoi(mora_size));
NJDNode_set_chain_rule(node, chain_rule);
NJDNode_set_chain_flag(node, atoi(chain_flag));
NJD_push_node(njd, node);
}
}
static void convert_digit_sequence(NJDNode * start, NJDNode * end)
{
NJDNode *node;
NJDNode *newnode;
int digit;
int place = 0;
int index;
int size = 0;
int have = 0;
if (get_digit_sequence_score(start, end) < 0)
return;
for (node = start; node != end->next; node = node->next)
size++;
if (size <= 1)
return;
index = size % 4;
if (index == 0)
index = 4;
if (size > index)
place = (size - index) / 4;
index--;
if (place > 17)
return;
for (node = start; node != end->next; node = node->next) {
digit = get_digit(node, 0);
if (index == 0) {
if (digit == 0) {
NJDNode_set_pron(node, NULL);
NJDNode_set_acc(node, 0);
NJDNode_set_mora_size(node, 0);
} else {
have = 1;
}
if (have == 1) {
if (place > 0) {
newnode = (NJDNode *) calloc(1, sizeof(NJDNode));
NJDNode_initialize(newnode);
NJDNode_load(newnode, (char *) njd_set_digit_rule_numeral_list3[place]);
node = NJDNode_insert(node, node->next, newnode);
}
have = 0;
}
place--;
} else {
if (digit <= 1) {
NJDNode_set_pron(node, NULL);
NJDNode_set_acc(node, 0);
NJDNode_set_mora_size(node, 0);
}
if (digit > 0) {
newnode = (NJDNode *) calloc(1, sizeof(NJDNode));
NJDNode_initialize(newnode);
NJDNode_load(newnode, (char *) njd_set_digit_rule_numeral_list2[index]);
node = NJDNode_insert(node, node->next, newnode);
have = 1;
}
}
index--;
if (index < 0)
index = 4 - 1;
}
}
void NJDNode_load(NJDNode * node, const char *str)
{
int i, j;
int index = 0;
char buff[MAXBUFLEN];
char buff_string[MAXBUFLEN];
char buff_orig[MAXBUFLEN];
char buff_read[MAXBUFLEN];
char buff_pron[MAXBUFLEN];
char buff_acc[MAXBUFLEN];
int count;
int index_string;
int index_orig;
int index_read;
int index_pron;
int index_acc;
NJDNode *prev = NULL;
/* load */
get_token_from_string(str, &index, buff_string, ',');
get_token_from_string(str, &index, buff, ',');
NJDNode_set_pos(node, buff);
get_token_from_string(str, &index, buff, ',');
NJDNode_set_pos_group1(node, buff);
get_token_from_string(str, &index, buff, ',');
NJDNode_set_pos_group2(node, buff);
get_token_from_string(str, &index, buff, ',');
NJDNode_set_pos_group3(node, buff);
get_token_from_string(str, &index, buff, ',');
NJDNode_set_ctype(node, buff);
get_token_from_string(str, &index, buff, ',');
NJDNode_set_cform(node, buff);
get_token_from_string(str, &index, buff_orig, ',');
get_token_from_string(str, &index, buff_read, ',');
get_token_from_string(str, &index, buff_pron, ',');
get_token_from_string(str, &index, buff_acc, ',');
get_token_from_string(str, &index, buff, ',');
NJDNode_set_chain_rule(node, buff);
get_token_from_string(str, &index, buff, ',');
if (strcmp(buff, "1") == 0)
NJDNode_set_chain_flag(node, 1);
else if (strcmp(buff, "0") == 0)
NJDNode_set_chain_flag(node, 0);
/* for symbol */
if (strstr(buff_acc, "*") != NULL || strstr(buff_acc, "/") == NULL) {
NJDNode_set_string(node, buff_string);
NJDNode_set_orig(node, buff_orig);
NJDNode_set_read(node, buff_read);
NJDNode_set_pron(node, buff_pron);
NJDNode_set_acc(node, 0);
NJDNode_set_mora_size(node, 0);
return;
}
/* count chained word */
for (i = 0, count = 0; buff_acc[i] != '\0'; i++)
if (buff_acc[i] == '/')
count++;
/* for single word */
if (count == 1) {
NJDNode_set_string(node, buff_string);
NJDNode_set_orig(node, buff_orig);
NJDNode_set_read(node, buff_read);
NJDNode_set_pron(node, buff_pron);
index_acc = 0;
get_token_from_string(buff_acc, &index_acc, buff, '/');
if (buff[0] == '\0') {
j = 0;
fprintf(stderr, "WARNING: NJDNode_load() in njd_node.c: Accent is empty.\n");
} else {
j = atoi(buff);
}
NJDNode_set_acc(node, j);
get_token_from_string(buff_acc, &index_acc, buff, ':');
if (buff[0] == '\0') {
j = 0;
fprintf(stderr, "WARNING: NJDNode_load() in njd_node.c: Mora size is empty.\n");
} else {
j = atoi(buff);
}
NJDNode_set_mora_size(node, j);
return;
}
/* parse chained word */
index_string = 0;
index_orig = 0;
index_read = 0;
index_pron = 0;
index_acc = 0;
for (i = 0; i < count; i++) {
if (i > 0) {
node = (NJDNode *) calloc(1, sizeof(NJDNode));
NJDNode_initialize(node);
NJDNode_copy(node, prev);
NJDNode_set_chain_flag(node, 0);
node->prev = prev;
prev->next = node;
}
/* orig */
get_token_from_string(buff_orig, &index_orig, buff, ':');
NJDNode_set_orig(node, buff);
/* string */
if (i + 1 < count) {
NJDNode_set_string(node, buff);
index_string += strlen(buff);
} else {
NJDNode_set_string(node, &buff_string[index_string]);
}
/* read */
get_token_from_string(buff_read, &index_read, buff, ':');
NJDNode_set_read(node, buff);
/* pron */
get_token_from_string(buff_pron, &index_pron, buff, ':');
NJDNode_set_pron(node, buff);
/* acc */
get_token_from_string(buff_acc, &index_acc, buff, '/');
if (buff[0] == '\0') {
j = 0;
fprintf(stderr, "WARNING: NJDNode_load() in njd_node.c: Accent is empty.\n");
} else {
j = atoi(buff);
}
NJDNode_set_acc(node, j);
/* mora size */
get_token_from_string(buff_acc, &index_acc, buff, ':');
if (buff[0] == '\0') {
j = 0;
fprintf(stderr, "WARNING: NJDNode_load() in njd_node.c: Mora size is empty.\n");
} else {
j = atoi(buff);
}
NJDNode_set_mora_size(node, j);
prev = node;
}
}