bool
PhraseQuery_Equals_IMP(PhraseQuery *self, Obj *other) {
if ((PhraseQuery*)other == self) { return true; }
if (!Obj_is_a(other, PHRASEQUERY)) { return false; }
PhraseQueryIVARS *const ivars = PhraseQuery_IVARS(self);
PhraseQueryIVARS *const ovars = PhraseQuery_IVARS((PhraseQuery*)other);
if (ivars->boost != ovars->boost) { return false; }
if (ivars->field && !ovars->field) { return false; }
if (!ivars->field && ovars->field) { return false; }
if (ivars->field && !Str_Equals(ivars->field, (Obj*)ovars->field)) {
return false;
}
if (!Vec_Equals(ovars->terms, (Obj*)ivars->terms)) { return false; }
return true;
}
void
PhraseQuery_Serialize_IMP(PhraseQuery *self, OutStream *outstream) {
PhraseQueryIVARS *const ivars = PhraseQuery_IVARS(self);
OutStream_Write_F32(outstream, ivars->boost);
Freezer_serialize_string(ivars->field, outstream);
Freezer_serialize_varray(ivars->terms, outstream);
}
void
PhraseQuery_Destroy_IMP(PhraseQuery *self) {
PhraseQueryIVARS *const ivars = PhraseQuery_IVARS(self);
DECREF(ivars->terms);
DECREF(ivars->field);
SUPER_DESTROY(self, PHRASEQUERY);
}
PhraseCompiler*
PhraseCompiler_init(PhraseCompiler *self, PhraseQuery *parent,
Searcher *searcher, float boost) {
PhraseCompilerIVARS *const ivars = PhraseCompiler_IVARS(self);
PhraseQueryIVARS *const parent_ivars = PhraseQuery_IVARS(parent);
Schema *schema = Searcher_Get_Schema(searcher);
Similarity *sim = Schema_Fetch_Sim(schema, parent_ivars->field);
Vector *terms = parent_ivars->terms;
// Try harder to find a Similarity if necessary.
if (!sim) { sim = Schema_Get_Similarity(schema); }
// Init.
Compiler_init((Compiler*)self, (Query*)parent, searcher, sim, boost);
// Store IDF for the phrase.
ivars->idf = 0;
for (uint32_t i = 0, max = Vec_Get_Size(terms); i < max; i++) {
Obj *term = Vec_Fetch(terms, i);
int32_t doc_max = Searcher_Doc_Max(searcher);
int32_t doc_freq = Searcher_Doc_Freq(searcher, parent_ivars->field, term);
ivars->idf += Sim_IDF(sim, doc_freq, doc_max);
}
// Calculate raw weight.
ivars->raw_weight = ivars->idf * ivars->boost;
return self;
}
Compiler*
PhraseQuery_Make_Compiler_IMP(PhraseQuery *self, Searcher *searcher,
float boost, bool subordinate) {
PhraseQueryIVARS *const ivars = PhraseQuery_IVARS(self);
if (Vec_Get_Size(ivars->terms) == 1) {
// Optimize for one-term "phrases".
Obj *term = Vec_Fetch(ivars->terms, 0);
TermQuery *term_query = TermQuery_new(ivars->field, term);
TermCompiler *term_compiler;
TermQuery_Set_Boost(term_query, ivars->boost);
term_compiler
= (TermCompiler*)TermQuery_Make_Compiler(term_query, searcher,
boost, subordinate);
DECREF(term_query);
return (Compiler*)term_compiler;
}
else {
PhraseCompiler *compiler
= PhraseCompiler_new(self, searcher, boost);
if (!subordinate) {
PhraseCompiler_Normalize(compiler);
}
return (Compiler*)compiler;
}
}
Obj*
PhraseQuery_Dump_IMP(PhraseQuery *self) {
PhraseQueryIVARS *ivars = PhraseQuery_IVARS(self);
PhraseQuery_Dump_t super_dump
= SUPER_METHOD_PTR(PHRASEQUERY, LUCY_PhraseQuery_Dump);
Hash *dump = (Hash*)CERTIFY(super_dump(self), HASH);
Hash_Store_Utf8(dump, "field", 5, Freezer_dump((Obj*)ivars->field));
Hash_Store_Utf8(dump, "terms", 5, Freezer_dump((Obj*)ivars->terms));
return (Obj*)dump;
}
static PhraseQuery*
S_do_init(PhraseQuery *self, String *field, Vector *terms, float boost) {
Query_init((Query*)self, boost);
PhraseQueryIVARS *const ivars = PhraseQuery_IVARS(self);
for (uint32_t i = 0, max = Vec_Get_Size(terms); i < max; i++) {
CERTIFY(Vec_Fetch(terms, i), OBJ);
}
ivars->field = field;
ivars->terms = terms;
return self;
}
Matcher*
PhraseCompiler_Make_Matcher_IMP(PhraseCompiler *self, SegReader *reader,
bool need_score) {
UNUSED_VAR(need_score);
PhraseCompilerIVARS *const ivars = PhraseCompiler_IVARS(self);
PhraseQueryIVARS *const parent_ivars
= PhraseQuery_IVARS((PhraseQuery*)ivars->parent);
Vector *const terms = parent_ivars->terms;
uint32_t num_terms = Vec_Get_Size(terms);
// Bail if there are no terms.
if (!num_terms) { return NULL; }
// Bail unless field is valid and posting type supports positions.
Similarity *sim = PhraseCompiler_Get_Similarity(self);
Posting *posting = Sim_Make_Posting(sim);
if (posting == NULL || !Obj_is_a((Obj*)posting, SCOREPOSTING)) {
DECREF(posting);
return NULL;
}
DECREF(posting);
// Bail if there's no PostingListReader for this segment.
PostingListReader *const plist_reader
= (PostingListReader*)SegReader_Fetch(
reader, Class_Get_Name(POSTINGLISTREADER));
if (!plist_reader) { return NULL; }
// Look up each term.
Vector *plists = Vec_new(num_terms);
for (uint32_t i = 0; i < num_terms; i++) {
Obj *term = Vec_Fetch(terms, i);
PostingList *plist
= PListReader_Posting_List(plist_reader, parent_ivars->field, term);
// Bail if any one of the terms isn't in the index.
if (!plist || !PList_Get_Doc_Freq(plist)) {
DECREF(plist);
DECREF(plists);
return NULL;
}
Vec_Push(plists, (Obj*)plist);
}
Matcher *retval
= (Matcher*)PhraseMatcher_new(sim, plists, (Compiler*)self);
DECREF(plists);
return retval;
}
Obj*
PhraseQuery_Load_IMP(PhraseQuery *self, Obj *dump) {
Hash *source = (Hash*)CERTIFY(dump, HASH);
PhraseQuery_Load_t super_load
= SUPER_METHOD_PTR(PHRASEQUERY, LUCY_PhraseQuery_Load);
PhraseQuery *loaded = (PhraseQuery*)super_load(self, dump);
PhraseQueryIVARS *loaded_ivars = PhraseQuery_IVARS(loaded);
Obj *field = CERTIFY(Hash_Fetch_Utf8(source, "field", 5), OBJ);
loaded_ivars->field
= (String*)CERTIFY(Freezer_load(field), STRING);
Obj *terms = CERTIFY(Hash_Fetch_Utf8(source, "terms", 5), OBJ);
loaded_ivars->terms
= (Vector*)CERTIFY(Freezer_load(terms), VECTOR);
return (Obj*)loaded;
}
String*
PhraseQuery_To_String_IMP(PhraseQuery *self) {
PhraseQueryIVARS *const ivars = PhraseQuery_IVARS(self);
uint32_t num_terms = Vec_Get_Size(ivars->terms);
CharBuf *buf = CB_new_from_str(ivars->field);
CB_Cat_Trusted_Utf8(buf, ":\"", 2);
for (uint32_t i = 0; i < num_terms; i++) {
Obj *term = Vec_Fetch(ivars->terms, i);
String *term_string = Obj_To_String(term);
CB_Cat(buf, term_string);
DECREF(term_string);
if (i < num_terms - 1) {
CB_Cat_Trusted_Utf8(buf, " ", 1);
}
}
CB_Cat_Trusted_Utf8(buf, "\"", 1);
String *retval = CB_Yield_String(buf);
DECREF(buf);
return retval;
}
Vector*
PhraseCompiler_Highlight_Spans_IMP(PhraseCompiler *self, Searcher *searcher,
DocVector *doc_vec, String *field) {
PhraseCompilerIVARS *const ivars = PhraseCompiler_IVARS(self);
PhraseQueryIVARS *const parent_ivars
= PhraseQuery_IVARS((PhraseQuery*)ivars->parent);
Vector *const terms = parent_ivars->terms;
Vector *const spans = Vec_new(0);
const uint32_t num_terms = Vec_Get_Size(terms);
UNUSED_VAR(searcher);
// Bail if no terms or field doesn't match.
if (!num_terms) { return spans; }
if (!Str_Equals(field, (Obj*)parent_ivars->field)) { return spans; }
Vector *term_vectors = Vec_new(num_terms);
BitVector *posit_vec = BitVec_new(0);
BitVector *other_posit_vec = BitVec_new(0);
for (uint32_t i = 0; i < num_terms; i++) {
Obj *term = Vec_Fetch(terms, i);
TermVector *term_vector
= DocVec_Term_Vector(doc_vec, field, (String*)term);
// Bail if any term is missing.
if (!term_vector) {
break;
}
Vec_Push(term_vectors, (Obj*)term_vector);
if (i == 0) {
// Set initial positions from first term.
I32Array *positions = TV_Get_Positions(term_vector);
for (uint32_t j = I32Arr_Get_Size(positions); j > 0; j--) {
BitVec_Set(posit_vec, I32Arr_Get(positions, j - 1));
}
}
else {
// Filter positions using logical "and".
I32Array *positions = TV_Get_Positions(term_vector);
BitVec_Clear_All(other_posit_vec);
for (uint32_t 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.
uint32_t num_tvs = Vec_Get_Size(term_vectors);
if (num_tvs == num_terms) {
TermVector *first_tv = (TermVector*)Vec_Fetch(term_vectors, 0);
TermVector *last_tv
= (TermVector*)Vec_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;
float weight = PhraseCompiler_Get_Weight(self);
uint32_t i = 0;
// Add only those starts/ends that belong to a valid position.
for (uint32_t 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;
for (uint32_t 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 (uint32_t 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;
}
}
Vec_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;
}
Vector*
PhraseQuery_Get_Terms_IMP(PhraseQuery *self) {
return PhraseQuery_IVARS(self)->terms;
}
String*
PhraseQuery_Get_Field_IMP(PhraseQuery *self) {
return PhraseQuery_IVARS(self)->field;
}
