static void
test_Set_and_Get(TestBatch *batch) {
unsigned i, max;
const uint32_t three = 3;
const uint32_t seventeen = 17;
BitVector *bit_vec = BitVec_new(8);
BitVec_Set(bit_vec, three);
TEST_TRUE(batch, BitVec_Get_Capacity(bit_vec) < seventeen,
"set below cap");
BitVec_Set(bit_vec, seventeen);
TEST_TRUE(batch, BitVec_Get_Capacity(bit_vec) > seventeen,
"set above cap causes BitVector to grow");
for (i = 0, max = BitVec_Get_Capacity(bit_vec); i < max; i++) {
if (i == three || i == seventeen) {
TEST_TRUE(batch, BitVec_Get(bit_vec, i), "set/get %d", i);
}
else {
TEST_FALSE(batch, BitVec_Get(bit_vec, i), "get %d", i);
}
}
TEST_FALSE(batch, BitVec_Get(bit_vec, i), "out of range get");
DECREF(bit_vec);
}
static void
test_Set_and_Get(TestBatchRunner *runner) {
const size_t three = 3;
const size_t seventeen = 17;
BitVector *bit_vec = BitVec_new(8);
BitVec_Set(bit_vec, three);
TEST_TRUE(runner, BitVec_Get_Capacity(bit_vec) < seventeen,
"set below cap");
BitVec_Set(bit_vec, seventeen);
TEST_TRUE(runner, BitVec_Get_Capacity(bit_vec) > seventeen,
"set above cap causes BitVector to grow");
size_t i, max;
for (i = 0, max = BitVec_Get_Capacity(bit_vec); i < max; i++) {
if (i == three || i == seventeen) {
TEST_TRUE(runner, BitVec_Get(bit_vec, i), "set/get %u", (unsigned)i);
}
else {
TEST_FALSE(runner, BitVec_Get(bit_vec, i), "get %u", (unsigned)i);
}
}
TEST_FALSE(runner, BitVec_Get(bit_vec, i), "out of range get");
DECREF(bit_vec);
}
static void
test_Mimic(TestBatch *batch) {
int foo;
for (foo = 0; foo <= 17; foo++) {
int bar;
for (bar = 0; bar <= 17; bar++) {
int i;
BitVector *foo_vec = BitVec_new(0);
BitVector *bar_vec = BitVec_new(0);
BitVec_Set(foo_vec, foo);
BitVec_Set(bar_vec, bar);
BitVec_Mimic(foo_vec, (Obj*)bar_vec);
for (i = 0; i <= 17; i++) {
if (BitVec_Get(foo_vec, i) && i != bar) { break; }
}
TEST_INT_EQ(batch, i, 18, "Mimic(%d, %d)", foo, bar);
DECREF(foo_vec);
DECREF(bar_vec);
}
}
}
static void
test_To_Array(TestBatch *batch) {
uint64_t *source_ints = TestUtils_random_u64s(NULL, 20, 0, 200);
BitVector *bit_vec = BitVec_new(0);
I32Array *array;
long num_unique = 0;
long i;
// Unique the random ints.
Sort_quicksort(source_ints, 20, sizeof(uint64_t),
S_compare_u64s, NULL);
for (i = 0; i < 19; i++) {
if (source_ints[i] != source_ints[i + 1]) {
source_ints[num_unique] = source_ints[i];
num_unique++;
}
}
// Set bits.
for (i = 0; i < num_unique; i++) {
BitVec_Set(bit_vec, (uint32_t)source_ints[i]);
}
// Create the array and compare it to the source.
array = BitVec_To_Array(bit_vec);
for (i = 0; i < num_unique; i++) {
if (I32Arr_Get(array, i) != (int32_t)source_ints[i]) { break; }
}
TEST_INT_EQ(batch, i, num_unique, "To_Array (%ld == %ld)", i,
num_unique);
DECREF(array);
DECREF(bit_vec);
FREEMEM(source_ints);
}
static void
test_To_Array(TestBatchRunner *runner) {
uint64_t *source_ints = TestUtils_random_u64s(NULL, 20, 0, 200);
BitVector *bit_vec = BitVec_new(0);
I32Array *array;
unsigned num_unique = 0;
// Unique the random ints.
qsort(source_ints, 20, sizeof(uint64_t), S_compare_u64s);
for (unsigned i = 0; i < 19; i++) {
if (source_ints[i] != source_ints[i + 1]) {
source_ints[num_unique] = source_ints[i];
num_unique++;
}
}
// Set bits.
for (unsigned i = 0; i < num_unique; i++) {
BitVec_Set(bit_vec, (size_t)source_ints[i]);
}
// Create the array and compare it to the source.
array = BitVec_To_Array(bit_vec);
unsigned i;
for (i = 0; i < num_unique; i++) {
if (I32Arr_Get(array, (size_t)i) != (int32_t)source_ints[i]) { break; }
}
TEST_UINT_EQ(runner, i, num_unique, "To_Array (%u == %u)", i,
num_unique);
DECREF(array);
DECREF(bit_vec);
FREEMEM(source_ints);
}
void
DefDelWriter_Delete_By_Term_IMP(DefaultDeletionsWriter *self,
String *field, Obj *term) {
DefaultDeletionsWriterIVARS *const ivars = DefDelWriter_IVARS(self);
for (uint32_t i = 0, max = VA_Get_Size(ivars->seg_readers); i < max; i++) {
SegReader *seg_reader = (SegReader*)VA_Fetch(ivars->seg_readers, i);
PostingListReader *plist_reader
= (PostingListReader*)SegReader_Fetch(
seg_reader, Class_Get_Name(POSTINGLISTREADER));
BitVector *bit_vec = (BitVector*)VA_Fetch(ivars->bit_vecs, i);
PostingList *plist = plist_reader
? PListReader_Posting_List(plist_reader, field, term)
: NULL;
int32_t doc_id;
int32_t num_zapped = 0;
// Iterate through postings, marking each doc as deleted.
if (plist) {
while (0 != (doc_id = PList_Next(plist))) {
num_zapped += !BitVec_Get(bit_vec, doc_id);
BitVec_Set(bit_vec, doc_id);
}
if (num_zapped) { ivars->updated[i] = true; }
DECREF(plist);
}
}
}
void
DefDelWriter_Delete_By_Query_IMP(DefaultDeletionsWriter *self, Query *query) {
DefaultDeletionsWriterIVARS *const ivars = DefDelWriter_IVARS(self);
Compiler *compiler = Query_Make_Compiler(query, (Searcher*)ivars->searcher,
Query_Get_Boost(query), false);
for (uint32_t i = 0, max = VA_Get_Size(ivars->seg_readers); i < max; i++) {
SegReader *seg_reader = (SegReader*)VA_Fetch(ivars->seg_readers, i);
BitVector *bit_vec = (BitVector*)VA_Fetch(ivars->bit_vecs, i);
Matcher *matcher = Compiler_Make_Matcher(compiler, seg_reader, false);
if (matcher) {
int32_t doc_id;
int32_t num_zapped = 0;
// Iterate through matches, marking each doc as deleted.
while (0 != (doc_id = Matcher_Next(matcher))) {
num_zapped += !BitVec_Get(bit_vec, doc_id);
BitVec_Set(bit_vec, doc_id);
}
if (num_zapped) { ivars->updated[i] = true; }
DECREF(matcher);
}
}
DECREF(compiler);
}
static SeriesMatcher*
S_make_series_matcher(I32Array *doc_ids, I32Array *offsets, int32_t doc_max) {
int32_t num_doc_ids = I32Arr_Get_Size(doc_ids);
int32_t num_matchers = I32Arr_Get_Size(offsets);
VArray *matchers = VA_new(num_matchers);
int32_t tick = 0;
int32_t i;
// Divvy up doc_ids by segment into BitVectors.
for (i = 0; i < num_matchers; i++) {
int32_t offset = I32Arr_Get(offsets, i);
int32_t max = i == num_matchers - 1
? doc_max + 1
: I32Arr_Get(offsets, i + 1);
BitVector *bit_vec = BitVec_new(max - offset);
while (tick < num_doc_ids) {
int32_t doc_id = I32Arr_Get(doc_ids, tick);
if (doc_id > max) { break; }
else { tick++; }
BitVec_Set(bit_vec, doc_id - offset);
}
VA_Push(matchers, (Obj*)BitVecMatcher_new(bit_vec));
DECREF(bit_vec);
}
SeriesMatcher *series_matcher = SeriesMatcher_new(matchers, offsets);
DECREF(matchers);
return series_matcher;
}
// Valgrind only - detect off-by-one error.
static void
test_off_by_one_error() {
for (unsigned cap = 5; cap <= 24; cap++) {
BitVector *bit_vec = BitVec_new(cap);
BitVec_Set(bit_vec, cap - 2);
DECREF(bit_vec);
}
}
static void
test_Clone(TestBatchRunner *runner) {
BitVector *self = BitVec_new(30);
BitVector *twin;
BitVec_Set(self, 2);
BitVec_Set(self, 3);
BitVec_Set(self, 10);
BitVec_Set(self, 20);
twin = BitVec_Clone(self);
size_t i;
for (i = 0; i < 50; i++) {
if (BitVec_Get(self, i) != BitVec_Get(twin, i)) { break; }
}
TEST_UINT_EQ(runner, i, 50, "Clone");
TEST_UINT_EQ(runner, BitVec_Count(twin), 4, "clone Count");
DECREF(self);
DECREF(twin);
}
static void
test_Clone(TestBatch *batch) {
int i;
BitVector *self = BitVec_new(30);
BitVector *twin;
BitVec_Set(self, 2);
BitVec_Set(self, 3);
BitVec_Set(self, 10);
BitVec_Set(self, 20);
twin = BitVec_Clone(self);
for (i = 0; i < 50; i++) {
if (BitVec_Get(self, i) != BitVec_Get(twin, i)) { break; }
}
TEST_INT_EQ(batch, i, 50, "Clone");
TEST_INT_EQ(batch, BitVec_Count(twin), 4, "clone Count");
DECREF(self);
DECREF(twin);
}
static BitVector*
S_create_set(int set_num) {
unsigned nums_1[] = { 1, 2, 3, 10, 20, 30, 0 };
unsigned nums_2[] = { 2, 3, 4, 5, 6, 7, 8, 9, 10,
25, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35, 0
};
unsigned *nums = set_num == 1 ? nums_1 : nums_2;
BitVector *bit_vec = BitVec_new(31);
for (unsigned i = 0; nums[i] != 0; i++) {
BitVec_Set(bit_vec, nums[i]);
}
return bit_vec;
}
static void
test_Mimic(TestBatchRunner *runner) {
for (unsigned foo = 0; foo <= 17; foo++) {
for (unsigned bar = 0; bar <= 17; bar++) {
BitVector *foo_vec = BitVec_new(0);
BitVector *bar_vec = BitVec_new(0);
BitVec_Set(foo_vec, foo);
BitVec_Set(bar_vec, bar);
BitVec_Mimic(foo_vec, (Obj*)bar_vec);
unsigned i;
for (i = 0; i <= 17; i++) {
if (BitVec_Get(foo_vec, i) && i != bar) { break; }
}
TEST_UINT_EQ(runner, i, 18, "Mimic(%u, %u)", foo, bar);
DECREF(foo_vec);
DECREF(bar_vec);
}
}
}
void
DefDelWriter_Delete_By_Doc_ID_IMP(DefaultDeletionsWriter *self, int32_t doc_id) {
DefaultDeletionsWriterIVARS *const ivars = DefDelWriter_IVARS(self);
uint32_t sub_tick = PolyReader_sub_tick(ivars->seg_starts, doc_id);
BitVector *bit_vec = (BitVector*)VA_Fetch(ivars->bit_vecs, sub_tick);
uint32_t offset = I32Arr_Get(ivars->seg_starts, sub_tick);
int32_t seg_doc_id = doc_id - offset;
if (!BitVec_Get(bit_vec, seg_doc_id)) {
ivars->updated[sub_tick] = true;
BitVec_Set(bit_vec, seg_doc_id);
}
}
static void
test_Next_Hit(TestBatchRunner *runner) {
for (int i = 24; i <= 33; i++) {
BitVector *bit_vec = BitVec_new(64);
BitVec_Set(bit_vec, (size_t)i);
TEST_INT_EQ(runner, BitVec_Next_Hit(bit_vec, 0), i,
"Next_Hit for 0 is %d", i);
TEST_INT_EQ(runner, BitVec_Next_Hit(bit_vec, 0), i,
"Next_Hit for 1 is %d", i);
for (int probe = 15; probe <= i; probe++) {
TEST_INT_EQ(runner, BitVec_Next_Hit(bit_vec, (size_t)probe), i,
"Next_Hit for %d is %d", probe, i);
}
for (int probe = i + 1; probe <= i + 9; probe++) {
TEST_INT_EQ(runner, BitVec_Next_Hit(bit_vec, (size_t)probe), -1,
"no Next_Hit for %d when max is %d", probe, i);
}
DECREF(bit_vec);
}
}
DefaultDeletionsWriter*
DefDelWriter_init(DefaultDeletionsWriter *self, Schema *schema,
Snapshot *snapshot, Segment *segment,
PolyReader *polyreader) {
DataWriter_init((DataWriter*)self, schema, snapshot, segment, polyreader);
DefaultDeletionsWriterIVARS *const ivars = DefDelWriter_IVARS(self);
ivars->seg_readers = PolyReader_Seg_Readers(polyreader);
uint32_t num_seg_readers = VA_Get_Size(ivars->seg_readers);
ivars->seg_starts = PolyReader_Offsets(polyreader);
ivars->bit_vecs = VA_new(num_seg_readers);
ivars->updated = (bool*)CALLOCATE(num_seg_readers, sizeof(bool));
ivars->searcher = IxSearcher_new((Obj*)polyreader);
ivars->name_to_tick = Hash_new(num_seg_readers);
// Materialize a BitVector of deletions for each segment.
for (uint32_t i = 0; i < num_seg_readers; i++) {
SegReader *seg_reader = (SegReader*)VA_Fetch(ivars->seg_readers, i);
BitVector *bit_vec = BitVec_new(SegReader_Doc_Max(seg_reader));
DeletionsReader *del_reader
= (DeletionsReader*)SegReader_Fetch(
seg_reader, Class_Get_Name(DELETIONSREADER));
Matcher *seg_dels = del_reader
? DelReader_Iterator(del_reader)
: NULL;
if (seg_dels) {
int32_t del;
while (0 != (del = Matcher_Next(seg_dels))) {
BitVec_Set(bit_vec, del);
}
DECREF(seg_dels);
}
VA_Store(ivars->bit_vecs, i, (Obj*)bit_vec);
Hash_Store(ivars->name_to_tick,
(Obj*)SegReader_Get_Seg_Name(seg_reader),
(Obj*)Int32_new(i));
}
return self;
}
static void
test_Count(TestBatch *batch) {
int i;
int shuffled[64];
BitVector *bit_vec = BitVec_new(64);
for (i = 0; i < 64; i++) { shuffled[i] = i; }
for (i = 0; i < 64; i++) {
int shuffle_pos = rand() % 64;
int temp = shuffled[shuffle_pos];
shuffled[shuffle_pos] = shuffled[i];
shuffled[i] = temp;
}
for (i = 0; i < 64; i++) {
BitVec_Set(bit_vec, shuffled[i]);
if (BitVec_Count(bit_vec) != (uint32_t)(i + 1)) { break; }
}
TEST_INT_EQ(batch, i, 64, "Count() returns the right number of bits");
DECREF(bit_vec);
}
static void
test_Count(TestBatchRunner *runner) {
unsigned shuffled[64];
BitVector *bit_vec = BitVec_new(64);
for (unsigned i = 0; i < 64; i++) { shuffled[i] = i; }
for (unsigned i = 0; i < 64; i++) {
unsigned shuffle_pos = (unsigned)rand() % 64;
unsigned temp = shuffled[shuffle_pos];
shuffled[shuffle_pos] = shuffled[i];
shuffled[i] = temp;
}
unsigned i;
for (i = 0; i < 64; i++) {
BitVec_Set(bit_vec, shuffled[i]);
if (BitVec_Count(bit_vec) != (uint32_t)(i + 1)) { break; }
}
TEST_UINT_EQ(runner, i, 64, "Count() returns the right number of bits");
DECREF(bit_vec);
}
static void
test_Next_Hit(TestBatch *batch) {
int i;
for (i = 24; i <= 33; i++) {
int probe;
BitVector *bit_vec = BitVec_new(64);
BitVec_Set(bit_vec, i);
TEST_INT_EQ(batch, BitVec_Next_Hit(bit_vec, 0), i,
"Next_Hit for 0 is %d", i);
TEST_INT_EQ(batch, BitVec_Next_Hit(bit_vec, 0), i,
"Next_Hit for 1 is %d", i);
for (probe = 15; probe <= i; probe++) {
TEST_INT_EQ(batch, BitVec_Next_Hit(bit_vec, probe), i,
"Next_Hit for %d is %d", probe, i);
}
for (probe = i + 1; probe <= i + 9; probe++) {
TEST_INT_EQ(batch, BitVec_Next_Hit(bit_vec, probe), -1,
"no Next_Hit for %d when max is %d", probe, i);
}
DECREF(bit_vec);
}
}
VArray*
PhraseCompiler_highlight_spans(PhraseCompiler *self, Searcher *searcher,
DocVector *doc_vec, const CharBuf *field)
{
PhraseQuery *const parent = (PhraseQuery*)self->parent;
VArray *const terms = parent->terms;
VArray *const spans = VA_new(0);
VArray *term_vectors;
BitVector *posit_vec;
BitVector *other_posit_vec;
uint32_t i;
const uint32_t num_terms = VA_Get_Size(terms);
uint32_t num_tvs;
UNUSED_VAR(searcher);
// Bail if no terms or field doesn't match.
if (!num_terms) { return spans; }
if (!CB_Equals(field, (Obj*)parent->field)) { return spans; }
term_vectors = VA_new(num_terms);
posit_vec = BitVec_new(0);
other_posit_vec = BitVec_new(0);
for (i = 0; i < num_terms; i++) {
Obj *term = VA_Fetch(terms, i);
TermVector *term_vector
= DocVec_Term_Vector(doc_vec, field, (CharBuf*)term);
// Bail if any term is missing.
if (!term_vector)
break;
VA_Push(term_vectors, (Obj*)term_vector);
if (i == 0) {
// Set initial positions from first term.
uint32_t j;
I32Array *positions = TV_Get_Positions(term_vector);
for (j = I32Arr_Get_Size(positions); j > 0; j--) {
BitVec_Set(posit_vec, I32Arr_Get(positions, j - 1));
}
}
else {
// Filter positions using logical "and".
uint32_t j;
I32Array *positions = TV_Get_Positions(term_vector);
BitVec_Clear_All(other_posit_vec);
for (j = I32Arr_Get_Size(positions); j > 0; j--) {
int32_t pos = I32Arr_Get(positions, j - 1) - i;
if (pos >= 0) {
BitVec_Set(other_posit_vec, pos);
}
}
BitVec_And(posit_vec, other_posit_vec);
}
}
// Proceed only if all terms are present.
num_tvs = VA_Get_Size(term_vectors);
if (num_tvs == num_terms) {
TermVector *first_tv = (TermVector*)VA_Fetch(term_vectors, 0);
TermVector *last_tv
= (TermVector*)VA_Fetch(term_vectors, num_tvs - 1);
I32Array *tv_start_positions = TV_Get_Positions(first_tv);
I32Array *tv_end_positions = TV_Get_Positions(last_tv);
I32Array *tv_start_offsets = TV_Get_Start_Offsets(first_tv);
I32Array *tv_end_offsets = TV_Get_End_Offsets(last_tv);
uint32_t terms_max = num_terms - 1;
I32Array *valid_posits = BitVec_To_Array(posit_vec);
uint32_t num_valid_posits = I32Arr_Get_Size(valid_posits);
uint32_t j = 0;
uint32_t posit_tick;
float weight = PhraseCompiler_Get_Weight(self);
i = 0;
// Add only those starts/ends that belong to a valid position.
for (posit_tick = 0; posit_tick < num_valid_posits; posit_tick++) {
int32_t valid_start_posit = I32Arr_Get(valid_posits, posit_tick);
int32_t valid_end_posit = valid_start_posit + terms_max;
int32_t start_offset = 0, end_offset = 0;
uint32_t max;
for (max = I32Arr_Get_Size(tv_start_positions); i < max; i++) {
if (I32Arr_Get(tv_start_positions, i) == valid_start_posit) {
start_offset = I32Arr_Get(tv_start_offsets, i);
break;
}
}
for (max = I32Arr_Get_Size(tv_end_positions); j < max; j++) {
if (I32Arr_Get(tv_end_positions, j) == valid_end_posit) {
end_offset = I32Arr_Get(tv_end_offsets, j);
break;
}
}
VA_Push(spans, (Obj*)Span_new(start_offset,
end_offset - start_offset, weight) );
i++, j++;
}
DECREF(valid_posits);
}
DECREF(other_posit_vec);
DECREF(posit_vec);
DECREF(term_vectors);
return spans;
}