// Insert Ref with hash key Key into global Hash_Table .
// Ref represents string S .
static
void
Hash_Insert(String_Ref_t Ref, uint64 Key, char * S) {
String_Ref_t H_Ref;
char * T;
int Shift;
unsigned char Key_Check;
int64 Ct, Probe, Sub;
int i;
Sub = HASH_FUNCTION (Key);
Shift = HASH_CHECK_FUNCTION (Key);
Hash_Check_Array [Sub] |= (((Check_Vector_t) 1) << Shift);
Key_Check = KEY_CHECK_FUNCTION (Key);
Probe = PROBE_FUNCTION (Key);
Ct = 0;
do {
for (i = 0; i < Hash_Table [Sub] . Entry_Ct; i ++)
if (Hash_Table [Sub] . Check [i] == Key_Check) {
H_Ref = Hash_Table [Sub] . Entry [i];
T = Data + String_Start [getStringRefStringNum(H_Ref)] + getStringRefOffset(H_Ref);
if (strncmp (S, T, Kmer_Len) == 0) {
if (getStringRefLast(H_Ref)) {
Extra_Ref_Ct ++;
}
Next_Ref [(String_Start [getStringRefStringNum(Ref)] + getStringRefOffset(Ref)) / (HASH_KMER_SKIP + 1)] = H_Ref;
Extra_Ref_Ct ++;
setStringRefLast(Ref, TRUELY_ZERO);
Hash_Table [Sub] . Entry [i] = Ref;
if (Hash_Table [Sub] . Hits [i] < HIGHEST_KMER_LIMIT)
Hash_Table [Sub] . Hits [i] ++;
return;
}
}
if (i != Hash_Table [Sub] . Entry_Ct) {
fprintf (stderr, "i = %d Sub = " F_S64 " Entry_Ct = %d\n",
i, Sub, Hash_Table [Sub] . Entry_Ct);
}
assert (i == Hash_Table [Sub] . Entry_Ct);
if (Hash_Table [Sub] . Entry_Ct < ENTRIES_PER_BUCKET) {
setStringRefLast(Ref, TRUELY_ONE);
Hash_Table [Sub] . Entry [i] = Ref;
Hash_Table [Sub] . Check [i] = Key_Check;
Hash_Table [Sub] . Entry_Ct ++;
Hash_Entries ++;
Hash_Table [Sub] . Hits [i] = 1;
return;
}
Sub = (Sub + Probe) % HASH_TABLE_SIZE;
} while (++ Ct < HASH_TABLE_SIZE);
fprintf (stderr, "ERROR: Hash table full\n");
assert (FALSE);
}
void bfilter_set(struct bloom_filter *bf, const void *buf, unsigned long buflen)
{
uint64_t hash;
int i;
for (i = 0; i < bf->nfunc; i++) {
hash = HASH_FUNCTION(buf, buflen, bf->seeds[i]);
bfilter_set_hash(bf, hash);
}
}
int bfilter_get(struct bloom_filter *bf, const void *buf, unsigned long buflen)
{
uint64_t hash;
int i;
for (i = 0; i < bf->nfunc; i++) {
hash = HASH_FUNCTION(buf, buflen, bf->seeds[i]);
if (bfilter_get_hash(bf, hash) == 0)
return 0;
}
return 1;
}
/*
* Adds a new hash entry to the hash table. Returns != 0 when there is no
* space left in the table.
*/
int hash_add (hash_t * hash, const hash_data_t * data)
{
int cell, free;
int rc = FALSE;
#if defined(MPI_HAS_INIT_THREAD_C) || defined(MPI_HAS_INIT_THREAD_F)
pthread_mutex_lock(&hash_lock);
#endif
cell = HASH_FUNCTION (data->key);
if (hash->table[cell].ovf_link == HASH_FREE)
{
hash->table[cell].ovf_link = HASH_NULL; /* No longer free */
hash->table[cell].data = *data;
}
else
{
/*
* Get a free overflow cell
*/
if ((free = hash->ovf_free) == HASH_NULL)
{
fprintf (stderr, PACKAGE_NAME": hash_add: No space left in hash table. Size is %d+%d\n", HASH_TABLE_SIZE, HASH_OVERFLOW_SIZE);
rc = TRUE;
}
else
{
hash->ovf_free = hash->overflow[free].next;
/*
* Insert free into overflow list of cell
*/
hash->overflow[free].next = hash->table[cell].ovf_link;
hash->table[cell].ovf_link = free;
/*
* Update user data
*/
hash->overflow[free].data = *data;
}
}
#if defined(MPI_HAS_INIT_THREAD_C) || defined(MPI_HAS_INIT_THREAD_F)
pthread_mutex_unlock(&hash_lock);
#endif
return rc;
}
PUBLIC HTAtom * HTAtom_for ARGS1(CONST char *, string)
{
int hash;
HTAtom * a;
/* First time around, clear hash table
*/
/*
* Memory leak fixed.
* 05-29-94 Lynx 2-3-1 Garrett Arch Blythe
*/
if (!initialised) {
int i;
for (i = 0; i < HASH_SIZE; i++)
hash_table[i] = (HTAtom *) 0;
initialised = YES;
atexit(free_atoms);
}
/* Generate hash function
*/
hash = HASH_FUNCTION(string);
/* Search for the string in the list
*/
for (a = hash_table[hash]; a; a = a->next) {
if (0 == strcasecomp(a->name, string)) {
/* CTRACE(tfp, "HTAtom: Old atom %p for `%s'\n", a, string); */
return a; /* Found: return it */
}
}
/* Generate a new entry
*/
a = (HTAtom *)malloc(sizeof(*a));
if (a == NULL)
outofmem(__FILE__, "HTAtom_for");
a->name = (char *)malloc(strlen(string)+1);
if (a->name == NULL)
outofmem(__FILE__, "HTAtom_for");
strcpy(a->name, string);
a->next = hash_table[hash]; /* Put onto the head of list */
hash_table[hash] = a;
#ifdef NOT_DEFINED
CTRACE(tfp, "HTAtom: New atom %p for `%s'\n", a, string);
#endif /* NOT_DEFINED */
return a;
}
// Set the empty bit to true for the hash table entry
// corresponding to string s whose hash key is key .
// Also set global String_Info . left/right_end_screened
// true if the entry occurs near the left/right end, resp.,
// of the string in the hash table. If not found, add an
// entry to the hash table and mark it empty.
static
void
Hash_Mark_Empty(uint64 key, char * s) {
String_Ref_t h_ref;
char * t;
unsigned char key_check;
int64 ct, probe;
int64 sub;
int i, shift;
sub = HASH_FUNCTION (key);
key_check = KEY_CHECK_FUNCTION (key);
probe = PROBE_FUNCTION (key);
ct = 0;
do {
for (i = 0; i < Hash_Table [sub] . Entry_Ct; i ++)
if (Hash_Table [sub] . Check [i] == key_check) {
h_ref = Hash_Table [sub] . Entry [i];
t = Data + String_Start [getStringRefStringNum(h_ref)] + getStringRefOffset(h_ref);
if (strncmp (s, t, Kmer_Len) == 0) {
if (! getStringRefEmpty(Hash_Table [sub] . Entry [i]))
Mark_Screened_Ends_Chain (Hash_Table [sub] . Entry [i]);
setStringRefEmpty(Hash_Table [sub] . Entry [i], TRUELY_ONE);
return;
}
}
assert (i == Hash_Table [sub] . Entry_Ct);
if (Hash_Table [sub] . Entry_Ct < ENTRIES_PER_BUCKET) {
// Not found
if (Use_Hopeless_Check) {
Hash_Table [sub] . Entry [i] = Add_Extra_Hash_String (s);
setStringRefEmpty(Hash_Table [sub] . Entry [i], TRUELY_ONE);
Hash_Table [sub] . Check [i] = key_check;
Hash_Table [sub] . Entry_Ct ++;
Hash_Table [sub] . Hits [i] = 0;
Hash_Entries ++;
shift = HASH_CHECK_FUNCTION (key);
Hash_Check_Array [sub] |= (((Check_Vector_t) 1) << shift);
}
return;
}
sub = (sub + probe) % HASH_TABLE_SIZE;
} while (++ ct < HASH_TABLE_SIZE);
fprintf (stderr, "ERROR: Hash table full\n");
assert (FALSE);
}
/*
* Searches for key in the hash table. Returns NULL when the key is not
* found in the table.
*/
hash_data_t *hash_search (const hash_t * hash, MPI_Request key)
{
int cell, ovf;
cell = HASH_FUNCTION (key);
if (hash->table[cell].ovf_link == HASH_FREE)
return NULL;
if (hash->table[cell].data.key == key)
return (hash_data_t *) & hash->table[cell].data;
/*
* Look for key in overflow list
*/
ovf = hash->table[cell].ovf_link;
while (ovf != HASH_NULL)
{
if (hash->overflow[ovf].data.key == key)
return (hash_data_t *) & hash->overflow[ovf].data;
ovf = hash->overflow[ovf].next;
}
return NULL;
}
/* The address has no anchor tag, for sure.
*/
PRIVATE HTParentAnchor0 * HTAnchor_findAddress_in_adult_table ARGS1(
CONST DocAddress *, newdoc)
{
/*
** Check whether we have this node.
*/
int hash;
HTList * adults;
HTList *grownups;
HTParentAnchor0 * foundAnchor;
BOOL need_extra_info = (newdoc->post_data || newdoc->post_content_type ||
newdoc->bookmark || newdoc->isHEAD || newdoc->safe);
/*
* We need not free adult_table[] atexit -
* it should be perfectly empty after free'ing all HText's.
* (There is an error if it is not empty at exit). -LP
*/
/*
** Select list from hash table,
*/
hash = HASH_FUNCTION(newdoc->address);
adults = &(adult_table[hash]);
/*
** Search list for anchor.
*/
grownups = adults;
while (NULL != (foundAnchor =
(HTParentAnchor0 *)HTList_nextObject(grownups))) {
if (HTSEquivalent(foundAnchor->address, newdoc->address) &&
((!foundAnchor->info && !need_extra_info) ||
(foundAnchor->info &&
HTBEquivalent(foundAnchor->info->post_data, newdoc->post_data) &&
foundAnchor->info->isHEAD == newdoc->isHEAD)))
{
CTRACE((tfp, "Anchor %p with address `%s' already exists.\n",
(void *)foundAnchor, newdoc->address));
return foundAnchor;
}
}
/*
** Node not found: create new anchor.
*/
foundAnchor = HTParentAnchor0_new(newdoc->address, hash);
CTRACE((tfp, "New anchor %p has hash %d and address `%s'\n",
(void *)foundAnchor, hash, newdoc->address));
if (need_extra_info) {
/* rare case, create a big structure */
HTParentAnchor *p = HTParentAnchor_new(foundAnchor);
if (newdoc->post_data)
BStrCopy(p->post_data, newdoc->post_data);
if (newdoc->post_content_type)
StrAllocCopy(p->post_content_type,
newdoc->post_content_type);
if (newdoc->bookmark)
StrAllocCopy(p->bookmark, newdoc->bookmark);
p->isHEAD = newdoc->isHEAD;
p->safe = newdoc->safe;
}
HTList_linkObject(adults, foundAnchor, &foundAnchor->_add_adult);
return foundAnchor;
}
/*
* Removes entry key in the hash table. Returns != 0 when key is not found in
* the table.
*/
int hash_remove (hash_t * hash, MPI_Request key)
{
int cell, ovf, prev;
int rc = FALSE;
#if defined(MPI_HAS_INIT_THREAD_C) || defined(MPI_HAS_INIT_THREAD_F)
pthread_mutex_lock(&hash_lock);
#endif
cell = HASH_FUNCTION (key);
if (hash->table[cell].ovf_link == HASH_FREE)
{
#if SIZEOF_LONG == 8
fprintf (stderr, PACKAGE_NAME": hash_remove: Key %08lx not in hash table\n", (long) key);
#elif SIZEOF_LONG == 4
fprintf (stderr, PACKAGE_NAME": hash_remove: Key %04x not in hash table\n", (long) key);
#endif
rc = TRUE;
}
else if (hash->table[cell].data.key == key)
{
/*
* Remove the main entry
*/
if ((ovf = hash->table[cell].ovf_link) != HASH_NULL)
{
/*
* Bring 1st overflow to main entry
*/
hash->table[cell].data = hash->overflow[ovf].data;
hash->table[cell].ovf_link = hash->overflow[ovf].next;
/*
* Put freed ovf in free list
*/
hash->overflow[ovf].next = hash->ovf_free;
hash->ovf_free = ovf;
}
else
{
hash->table[cell].ovf_link = HASH_FREE; /* Mark as free */
}
}
else
{
/*
* Search key in overflow list
*/
ovf = hash->table[cell].ovf_link;
prev = HASH_NULL;
while (ovf != HASH_NULL && hash->overflow[ovf].data.key != key)
{
prev = ovf;
ovf = hash->overflow[ovf].next;
}
if (ovf == HASH_NULL) /* Not found */
{
#if SIZEOF_LONG == 8
fprintf (stderr, PACKAGE_NAME": hash_remove: Key %08lx not in hash table\n", (long) key);
#elif SIZEOF_LONG == 4
fprintf (stderr, PACKAGE_NAME": hash_remove: Key %04x not in hash table\n", (long) key);
#endif
rc = TRUE;
}
else
{
if (prev == HASH_NULL)
{
hash->table[cell].ovf_link = hash->overflow[ovf].next;
}
else
{
hash->overflow[prev].next = hash->overflow[ovf].next;
}
hash->overflow[ovf].next = hash->ovf_free;
hash->ovf_free = ovf;
}
}
#if defined(MPI_HAS_INIT_THREAD_C) || defined(MPI_HAS_INIT_THREAD_F)
pthread_mutex_unlock(&hash_lock);
#endif
return rc;
}
