TermVector*
TV_Deserialize_IMP(TermVector *self, InStream *instream) {
String *field = Freezer_read_string(instream);
String *text = Freezer_read_string(instream);
size_t num_pos = InStream_Read_C64(instream);
// Read positional data.
int32_t *posits = (int32_t*)MALLOCATE(num_pos * sizeof(int32_t));
int32_t *starts = (int32_t*)MALLOCATE(num_pos * sizeof(int32_t));
int32_t *ends = (int32_t*)MALLOCATE(num_pos * sizeof(int32_t));
for (size_t i = 0; i < num_pos; i++) {
posits[i] = InStream_Read_C32(instream);
starts[i] = InStream_Read_C32(instream);
ends[i] = InStream_Read_C32(instream);
}
I32Array *positions = I32Arr_new_steal(posits, num_pos);
I32Array *start_offsets = I32Arr_new_steal(starts, num_pos);
I32Array *end_offsets = I32Arr_new_steal(ends, num_pos);
TV_init(self, field, text, positions, start_offsets, end_offsets);
DECREF(positions);
DECREF(start_offsets);
DECREF(end_offsets);
DECREF(text);
DECREF(field);
return self;
}
static void
S_read_entry(LexIndex *self) {
LexIndexIVARS *const ivars = LexIndex_IVARS(self);
InStream *ix_in = ivars->ix_in;
TermInfo *const tinfo = ivars->tinfo;
int64_t offset = (int64_t)NumUtil_decode_bigend_u64(ivars->offsets + ivars->tick);
InStream_Seek(ix_in, offset);
TermStepper_Read_Key_Frame(ivars->term_stepper, ix_in);
int32_t doc_freq = InStream_Read_C32(ix_in);
TInfo_Set_Doc_Freq(tinfo, doc_freq);
TInfo_Set_Post_FilePos(tinfo, InStream_Read_C64(ix_in));
int64_t skip_filepos = doc_freq >= ivars->skip_interval
? InStream_Read_C64(ix_in)
: 0;
TInfo_Set_Skip_FilePos(tinfo, skip_filepos);
TInfo_Set_Lex_FilePos(tinfo, InStream_Read_C64(ix_in));
}
void
MatchTInfoStepper_read_delta(MatchTermInfoStepper *self, InStream *instream) {
TermInfo *const tinfo = (TermInfo*)self->value;
// Read doc freq.
tinfo->doc_freq = InStream_Read_C32(instream);
// Adjust postings file pointer.
tinfo->post_filepos += InStream_Read_C64(instream);
// Maybe read skip pointer.
if (tinfo->doc_freq >= self->skip_interval) {
tinfo->skip_filepos = InStream_Read_C64(instream);
}
else {
tinfo->skip_filepos = 0;
}
}
void
MatchTInfoStepper_Read_Delta_IMP(MatchTermInfoStepper *self, InStream *instream) {
MatchTermInfoStepperIVARS *const ivars = MatchTInfoStepper_IVARS(self);
TermInfoIVARS *const tinfo_ivars = TInfo_IVARS((TermInfo*)ivars->value);
// Read doc freq.
tinfo_ivars->doc_freq = InStream_Read_C32(instream);
// Adjust postings file pointer.
tinfo_ivars->post_filepos += InStream_Read_C64(instream);
// Maybe read skip pointer.
if (tinfo_ivars->doc_freq >= ivars->skip_interval) {
tinfo_ivars->skip_filepos = InStream_Read_C64(instream);
}
else {
tinfo_ivars->skip_filepos = 0;
}
}
HitDoc*
DefDocReader_Fetch_Doc_IMP(DefaultDocReader *self, int32_t doc_id) {
DefaultDocReaderIVARS *const ivars = DefDocReader_IVARS(self);
Schema *const schema = ivars->schema;
InStream *const dat_in = ivars->dat_in;
InStream *const ix_in = ivars->ix_in;
Hash *const fields = Hash_new(1);
int64_t start;
uint32_t num_fields;
uint32_t field_name_cap = 31;
char *field_name = (char*)MALLOCATE(field_name_cap + 1);
// Get data file pointer from index, read number of fields.
InStream_Seek(ix_in, (int64_t)doc_id * 8);
start = InStream_Read_U64(ix_in);
InStream_Seek(dat_in, start);
num_fields = InStream_Read_C32(dat_in);
// Decode stored data and build up the doc field by field.
while (num_fields--) {
uint32_t field_name_len;
Obj *value;
FieldType *type;
// Read field name.
field_name_len = InStream_Read_C32(dat_in);
if (field_name_len > field_name_cap) {
field_name_cap = field_name_len;
field_name = (char*)REALLOCATE(field_name,
field_name_cap + 1);
}
InStream_Read_Bytes(dat_in, field_name, field_name_len);
// Find the Field's FieldType.
StackString *field_name_str
= SSTR_WRAP_UTF8(field_name, field_name_len);
type = Schema_Fetch_Type(schema, (String*)field_name_str);
// Read the field value.
switch (FType_Primitive_ID(type) & FType_PRIMITIVE_ID_MASK) {
case FType_TEXT: {
uint32_t value_len = InStream_Read_C32(dat_in);
char *buf = (char*)MALLOCATE(value_len + 1);
InStream_Read_Bytes(dat_in, buf, value_len);
buf[value_len] = '\0';
value = (Obj*)Str_new_steal_utf8(buf, value_len);
break;
}
case FType_BLOB: {
uint32_t value_len = InStream_Read_C32(dat_in);
char *buf = (char*)MALLOCATE(value_len);
InStream_Read_Bytes(dat_in, buf, value_len);
value = (Obj*)BB_new_steal_bytes(
buf, value_len, value_len);
break;
}
case FType_FLOAT32:
value = (Obj*)Float32_new(
InStream_Read_F32(dat_in));
break;
case FType_FLOAT64:
value = (Obj*)Float64_new(
InStream_Read_F64(dat_in));
break;
case FType_INT32:
value = (Obj*)Int32_new(
(int32_t)InStream_Read_C32(dat_in));
break;
case FType_INT64:
value = (Obj*)Int64_new(
(int64_t)InStream_Read_C64(dat_in));
break;
default:
value = NULL;
THROW(ERR, "Unrecognized type: %o", type);
}
// Store the value.
Hash_Store_Utf8(fields, field_name, field_name_len, value);
}
FREEMEM(field_name);
HitDoc *retval = HitDoc_new(fields, doc_id, 0.0);
DECREF(fields);
return retval;
}
Obj*
Freezer_deserialize(Obj *obj, InStream *instream) {
if (Obj_is_a(obj, STRING)) {
obj = (Obj*)Freezer_deserialize_string((String*)obj, instream);
}
else if (Obj_is_a(obj, BLOB)) {
obj = (Obj*)Freezer_deserialize_blob((Blob*)obj, instream);
}
else if (Obj_is_a(obj, VECTOR)) {
obj = (Obj*)Freezer_deserialize_varray((Vector*)obj, instream);
}
else if (Obj_is_a(obj, HASH)) {
obj = (Obj*)Freezer_deserialize_hash((Hash*)obj, instream);
}
else if (Obj_is_a(obj, INTEGER)) {
int64_t value = (int64_t)InStream_Read_C64(instream);
obj = (Obj*)Int_init((Integer*)obj, value);
}
else if (Obj_is_a(obj, FLOAT)) {
double value = InStream_Read_F64(instream);
obj = (Obj*)Float_init((Float*)obj, value);
}
else if (Obj_is_a(obj, BOOLEAN)) {
bool value = !!InStream_Read_U8(instream);
Obj *result = value ? INCREF(CFISH_TRUE) : INCREF(CFISH_FALSE);
// FIXME: This DECREF is essentially a no-op causing a
// memory leak.
DECREF(obj);
obj = result;
}
else if (Obj_is_a(obj, QUERY)) {
obj = (Obj*)Query_Deserialize((Query*)obj, instream);
}
else if (Obj_is_a(obj, DOC)) {
obj = (Obj*)Doc_Deserialize((Doc*)obj, instream);
}
else if (Obj_is_a(obj, DOCVECTOR)) {
obj = (Obj*)DocVec_Deserialize((DocVector*)obj, instream);
}
else if (Obj_is_a(obj, TERMVECTOR)) {
obj = (Obj*)TV_Deserialize((TermVector*)obj, instream);
}
else if (Obj_is_a(obj, SIMILARITY)) {
obj = (Obj*)Sim_Deserialize((Similarity*)obj, instream);
}
else if (Obj_is_a(obj, MATCHDOC)) {
obj = (Obj*)MatchDoc_Deserialize((MatchDoc*)obj, instream);
}
else if (Obj_is_a(obj, TOPDOCS)) {
obj = (Obj*)TopDocs_Deserialize((TopDocs*)obj, instream);
}
else if (Obj_is_a(obj, SORTSPEC)) {
obj = (Obj*)SortSpec_Deserialize((SortSpec*)obj, instream);
}
else if (Obj_is_a(obj, SORTRULE)) {
obj = (Obj*)SortRule_Deserialize((SortRule*)obj, instream);
}
else {
THROW(ERR, "Don't know how to deserialize a %o",
Obj_get_class_name(obj));
}
return obj;
}
