/**
* Computes the bag distance of two strings. The distance approximates
* and lower bounds the Levenshtein distance.
* @param x first string
* @param y second string
* @return Bag distance
*/
float dist_bag_compare(hstring_t x, hstring_t y)
{
float d = 0;
bag_t *xh, *yh, *xb, *yb;
xh = bag_create(x);
yh = bag_create(y);
int missing = y.len;
for (xb = xh; xb != NULL; xb = xb->hh.next) {
HASH_FIND(hh, yh, &(xb->sym), sizeof(sym_t), yb);
if (!yb) {
d += xb->cnt;
} else {
d += fabs(xb->cnt - yb->cnt);
missing -= yb->cnt;
}
}
d += missing;
bag_destroy(xh);
bag_destroy(yh);
return lnorm(n, d, x, y);
}
bag_t *generate_index(FILE *input, int min_word_len)
{
bag_t *index = bag_create(entry_cmp);
if (index) {
char word[LINE_LENGTH] = "";
entry_t new_word, *existing_entry;
bag_elem_t new_entry;
unsigned page = 0;
while (get_word(input, word, &page))
{
new_word.entry_word = word;
// check if the length of the word is long enough
if(strlen(word) >= min_word_len)
{
existing_entry = bag_contains(index, &new_word);
if(existing_entry != NULL) // if the word is already in index
{
entry_add(existing_entry, page); // add the location to the list of locations for that word
}
else // if the word isn't in the index
{
new_entry = entry_create(word, page); // create the entry
bag_insert(index, new_entry); // add the location
}
}
}
}
return index;
}
Bucket* bucket_create(
Mailbin* mailbin,
valueID unallocatedvalueids_start,
valueID unallocatedvalueids_end,
fileID bucketfileID
) {
Bucket* this =
alloct(Bucket);
this->unallocatedvalueids_start =
unallocatedvalueids_start;
this->unallocatedvalueids_end =
unallocatedvalueids_end;
this->unallocatedvalueids_flag =
tflag_create(); /// this->un..flag+
ref(this, this->unallocatedvalueids_flag); /// this->un..flag+
release(this->unallocatedvalueids_flag); /// this->un..flag-
this->rocks =
bag_create(
rock_serialize,
rock_deserialize,
bucketfileID
); /// this->rocks+
ref(this, this->rocks); /// this->rocks+
release(this->rocks); /// this->rocks-
this->mailman_mirrorrock =
mailbin_allocate_mailman(
mailbin,
MESSAGE_TYPE_MIRROR_ROCKS,
this,
message_mirrorrocks_process
);/// this->mailman_mirrorrock+
ref(this, this->mailman_mirrorrock); /// this->mailman_mirrorrock+
release(this->mailman_mirrorrock); /// this->mailman_mirrorrock-
return this;
}
bag_elem_t entry_create(const char *word, unsigned page)
{
// Allocate the memory for the new entry
entry_t *new_entry = malloc(sizeof(entry_t));
// Copy the word into a new string and put it in the entry.
new_entry -> entry_word = malloc((strlen(word) + 1) * sizeof(char));
strcpy(new_entry -> entry_word, word);
// Create the page index bag to hold the page numbers.
new_entry->page_index = bag_create(page_cmp);
page_entry *new_page = malloc(sizeof(page_entry));
*new_page = page;
// add the page to the page index.
bag_insert(new_entry->page_index, new_page);
return new_entry;
}
