InverterEntry*
InvEntry_init(InverterEntry *self, Schema *schema, String *field,
int32_t field_num) {
InverterEntryIVARS *const ivars = InvEntry_IVARS(self);
ivars->field_num = field_num;
ivars->field = field ? Str_Clone(field) : NULL;
ivars->inversion = NULL;
if (schema) {
ivars->analyzer
= (Analyzer*)INCREF(Schema_Fetch_Analyzer(schema, field));
ivars->sim = (Similarity*)INCREF(Schema_Fetch_Sim(schema, field));
ivars->type = (FieldType*)INCREF(Schema_Fetch_Type(schema, field));
if (!ivars->type) { THROW(ERR, "Unknown field: '%o'", field); }
uint8_t prim_id = FType_Primitive_ID(ivars->type);
switch (prim_id & FType_PRIMITIVE_ID_MASK) {
case FType_TEXT:
ivars->value = NULL;
break;
case FType_BLOB:
ivars->value = (Obj*)ViewBB_new(NULL, 0);
break;
case FType_INT32:
ivars->value = (Obj*)Int32_new(0);
break;
case FType_INT64:
ivars->value = (Obj*)Int64_new(0);
break;
case FType_FLOAT32:
ivars->value = (Obj*)Float32_new(0);
break;
case FType_FLOAT64:
ivars->value = (Obj*)Float64_new(0);
break;
default:
THROW(ERR, "Unrecognized primitive id: %i8", prim_id);
}
ivars->indexed = FType_Indexed(ivars->type);
if (ivars->indexed && FType_Is_A(ivars->type, NUMERICTYPE)) {
THROW(ERR, "Field '%o' spec'd as indexed, but numerical types cannot "
"be indexed yet", field);
}
if (FType_Is_A(ivars->type, FULLTEXTTYPE)) {
ivars->highlightable
= FullTextType_Highlightable((FullTextType*)ivars->type);
}
}
return self;
}
NumericSortCache*
NumSortCache_init(NumericSortCache *self, const CharBuf *field,
FieldType *type, int32_t cardinality, int32_t doc_max,
int32_t null_ord, int32_t ord_width, InStream *ord_in,
InStream *dat_in) {
// Validate.
if (!type || !FType_Sortable(type) || !FType_Is_A(type, NUMERICTYPE)) {
DECREF(self);
THROW(ERR, "'%o' isn't a sortable NumericType field", field);
}
// Mmap ords and super-init.
int64_t ord_len = InStream_Length(ord_in);
void *ords = InStream_Buf(ord_in, (size_t)ord_len);
SortCache_init((SortCache*)self, field, type, ords, cardinality, doc_max,
null_ord, ord_width);
// Assign.
self->ord_in = (InStream*)INCREF(ord_in);
self->dat_in = (InStream*)INCREF(dat_in);
// Validate ord file length.
double BITS_PER_BYTE = 8.0;
double docs_per_byte = BITS_PER_BYTE / self->ord_width;
double max_ords = ord_len * docs_per_byte;
if (max_ords < self->doc_max + 1) {
DECREF(self);
THROW(ERR, "Conflict between ord count max %f64 and doc_max %i32 for "
"field %o", max_ords, self->doc_max, field);
}
ABSTRACT_CLASS_CHECK(self, NUMERICSORTCACHE);
return self;
}
void
Indexer_delete_by_term(Indexer *self, CharBuf *field, Obj *term)
{
Schema *schema = self->schema;
FieldType *type = Schema_Fetch_Type(schema, field);
// Raise exception if the field isn't indexed.
if (!type || !FType_Indexed(type))
THROW(ERR, "%o is not an indexed field", field);
// Analyze term if appropriate, then zap.
if (FType_Is_A(type, FULLTEXTTYPE)) {
CERTIFY(term, CHARBUF);
{
Analyzer *analyzer = Schema_Fetch_Analyzer(schema, field);
VArray *terms = Analyzer_Split(analyzer, (CharBuf*)term);
Obj *analyzed_term = VA_Fetch(terms, 0);
if (analyzed_term) {
DelWriter_Delete_By_Term(self->del_writer, field,
analyzed_term);
}
DECREF(terms);
}
}
else {
DelWriter_Delete_By_Term(self->del_writer, field, term);
}
}
void
HLWriter_Add_Inverted_Doc_IMP(HighlightWriter *self, Inverter *inverter,
int32_t doc_id) {
HighlightWriterIVARS *const ivars = HLWriter_IVARS(self);
OutStream *dat_out = S_lazy_init(self);
OutStream *ix_out = ivars->ix_out;
int64_t filepos = OutStream_Tell(dat_out);
uint32_t num_highlightable = 0;
int32_t expected = (int32_t)(OutStream_Tell(ix_out) / 8);
// Verify doc id.
if (doc_id != expected) {
THROW(ERR, "Expected doc id %i32 but got %i32", expected, doc_id);
}
// Write index data.
OutStream_Write_I64(ix_out, filepos);
// Count, then write number of highlightable fields.
Inverter_Iterate(inverter);
while (Inverter_Next(inverter)) {
FieldType *type = Inverter_Get_Type(inverter);
if (FType_Is_A(type, FULLTEXTTYPE)
&& FullTextType_Highlightable((FullTextType*)type)
) {
num_highlightable++;
}
}
OutStream_Write_C32(dat_out, num_highlightable);
Inverter_Iterate(inverter);
while (Inverter_Next(inverter)) {
FieldType *type = Inverter_Get_Type(inverter);
if (FType_Is_A(type, FULLTEXTTYPE)
&& FullTextType_Highlightable((FullTextType*)type)
) {
String *field = Inverter_Get_Field_Name(inverter);
Inversion *inversion = Inverter_Get_Inversion(inverter);
ByteBuf *tv_buf = HLWriter_TV_Buf(self, inversion);
Freezer_serialize_string(field, dat_out);
Freezer_serialize_bytebuf(tv_buf, dat_out);
DECREF(tv_buf);
}
}
}
