static int read_conllx(FILE* fp)
{
int num_sent = 0;
unsigned int line_num = 0;
char buff[BUF_SIZE];
const char* sep = "\t";
char *t, *line, *tofree;
size_t len;
const char** seq = xmalloc(sizeof(char *) * CONLLX_TOKEN_NUM_FIELDS);
struct blocks* blocks = new_blocks(256);
struct sentence* sent = sentence_new();
while (fgets(buff, sizeof(buff), fp)) {
line_num++;
if (buff[0] == '\n') {
latex_print_dep_tree(sent);
sentence_destroy(sent);
sent = sentence_new();
num_sent++;
continue;
}
len = strnlen(buff, BUF_SIZE);
tofree = line = strndup(buff, len-1);
blocks_add_elem(blocks, tofree);
int i = 0;
while ((t = strsep(&line, sep)) != NULL) {
seq[i] = t;
i++;
}
sentence_add_token(sent, token_new(seq, CONLLX_TOKEN_NUM_FIELDS));
}
destroy_blocks(blocks);
sentence_destroy(sent);
free(seq);
fprintf(stderr, "INFO: Number of sentences = %d\n", num_sent);
return 0;
}
struct big_bdescr* new_big_blocks(struct s_arena* arena, unsigned int count){
void* new_block = new_blocks(arena,count);
debug("new big block at %08x\n",(unsigned int)new_block);
init_big_block(new_block,count);
return (struct big_bdescr*)new_block;
}
struct single_bdescr* new_single_block(struct s_arena* arena){
struct single_bdescr* new_block = new_blocks(arena,1);
init_single_block(new_block);
debug("new single block at %08x\n",(unsigned int)new_block);
return new_block;
}
