void mecab2njd(NJD * njd, char **feature, int size)
{
int i;
NJDNode *node;
for (i = 0; i < size; i++) {
node = (NJDNode *) calloc(1, sizeof(NJDNode));
NJDNode_initialize(node);
NJDNode_load(node, feature[i]);
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;
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;
}
}