static Block
blocks_sort( /* Sort 'b1' on 'live_bytes'. */
Block b1)
{
int len = block_count(b1);
if (len < 2)
return b1;
{
Block b2 = blocks_split(b1, len / 2);
return blocks_merge(blocks_sort(b1), blocks_sort(b2)
);
}
}
/* Convert blocks into windows whilst ensuring blocks which
* transverse buffers are accounted for */
int process_block(char *block) {
long sz, bsz;
char *regex_b, *b = NULL;
bool alloced = false;
window_t *w = NULL;
extern int store_count;
/* previous block didn't terminate so we need to merge it
* with the current block */
if(blocks_require_merge()) {
if((sz = blocks_merge(block, &b)) == RERROR)
return RERROR;
alloced = true;
} else {
b = block;
sz = strlen(block);
}
regex_b = regex_creplace(b);
bsz = strlen(regex_b);
if(win_init(&w, regex_b, bsz)==RERROR) {
setlocation(LOCATION, __func__);
return RERROR;
}
free(regex_b);
// need this for ngram computation over blocks
w->mcount = store_count;
lgprintf(LG_INFO, "merged window size: %d\n", w->wc);
win_store_add(w);
/* this block does not terminate so we need to store it so
* that blocks from the next buffer can be merged with it. */
if(!block_match(b[sz-1]))
blocks_store(w);
else
blocks_clear();
if(alloced) {
free(b);
alloced = false;
}
return 0;
}
