Vector*
TermCompiler_Highlight_Spans_IMP(TermCompiler *self, Searcher *searcher,
DocVector *doc_vec, String *field) {
TermCompilerIVARS *const ivars = TermCompiler_IVARS(self);
TermQueryIVARS *const parent_ivars
= TermQuery_IVARS((TermQuery*)ivars->parent);
Vector *spans = Vec_new(0);
TermVector *term_vector;
I32Array *starts, *ends;
UNUSED_VAR(searcher);
if (!Str_Equals(parent_ivars->field, (Obj*)field)) { return spans; }
// Add all starts and ends.
term_vector
= DocVec_Term_Vector(doc_vec, field, (String*)parent_ivars->term);
if (!term_vector) { return spans; }
starts = TV_Get_Start_Offsets(term_vector);
ends = TV_Get_End_Offsets(term_vector);
for (size_t i = 0, max = I32Arr_Get_Size(starts); i < max; i++) {
int32_t start = I32Arr_Get(starts, i);
int32_t length = I32Arr_Get(ends, i) - start;
Vec_Push(spans,
(Obj*)Span_new(start, length, TermCompiler_Get_Weight(self)));
}
DECREF(term_vector);
return spans;
}
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;
}
