I32Array*
DelWriter_generate_doc_map(DeletionsWriter *self, Matcher *deletions,
int32_t doc_max, int32_t offset) {
int32_t *doc_map = (int32_t*)CALLOCATE(doc_max + 1, sizeof(int32_t));
int32_t next_deletion = deletions ? Matcher_Next(deletions) : INT32_MAX;
UNUSED_VAR(self);
// 0 for a deleted doc, a new number otherwise
for (int32_t i = 1, new_doc_id = 1; i <= doc_max; i++) {
if (i == next_deletion) {
next_deletion = Matcher_Next(deletions);
}
else {
doc_map[i] = offset + new_doc_id++;
}
}
return I32Arr_new_steal(doc_map, doc_max + 1);
}
I32Array*
DelWriter_Generate_Doc_Map_IMP(DeletionsWriter *self, Matcher *deletions,
int32_t doc_max, int32_t offset) {
UNUSED_VAR(self);
if (doc_max < 0) {
THROW(ERR, "Negative doc_max is invalid: %i32", doc_max);
}
int32_t *doc_map = (int32_t*)CALLOCATE((size_t)doc_max + 1, sizeof(int32_t));
int32_t next_deletion = deletions ? Matcher_Next(deletions) : INT32_MAX;
// 0 for a deleted doc, a new number otherwise
for (int32_t i = 1, new_doc_id = 1; i <= doc_max; i++) {
if (i == next_deletion) {
next_deletion = Matcher_Next(deletions);
}
else {
doc_map[i] = offset + new_doc_id++;
}
}
return I32Arr_new_steal(doc_map, (size_t)doc_max + 1);
}
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 CFISH_INLINE int32_t
SI_top_next(ORMatcher *self, ORMatcherIVARS *ivars) {
HeapedMatcherDoc *const top_hmd = ivars->top_hmd;
top_hmd->doc = Matcher_Next(top_hmd->matcher);
return S_adjust_root(self, ivars);
}
int32_t
ReqOptMatcher_Next_IMP(RequiredOptionalMatcher *self) {
RequiredOptionalMatcherIVARS *const ivars = ReqOptMatcher_IVARS(self);
return Matcher_Next(ivars->req_matcher);
}
static INLINE int32_t
SI_top_next(ORMatcher *self) {
HeapedMatcherDoc *const top_hmd = self->top_hmd;
top_hmd->doc = Matcher_Next(top_hmd->matcher);
return S_adjust_root(self);
}
static bool
S_merge_updated_deletions(BackgroundMerger *self) {
BackgroundMergerIVARS *const ivars = BGMerger_IVARS(self);
Hash *updated_deletions = NULL;
PolyReader *new_polyreader
= PolyReader_open((Obj*)ivars->folder, NULL, NULL);
Vector *new_seg_readers
= PolyReader_Get_Seg_Readers(new_polyreader);
Vector *old_seg_readers
= PolyReader_Get_Seg_Readers(ivars->polyreader);
Hash *new_segs = Hash_new(Vec_Get_Size(new_seg_readers));
for (uint32_t i = 0, max = Vec_Get_Size(new_seg_readers); i < max; i++) {
SegReader *seg_reader = (SegReader*)Vec_Fetch(new_seg_readers, i);
String *seg_name = SegReader_Get_Seg_Name(seg_reader);
Hash_Store(new_segs, seg_name, INCREF(seg_reader));
}
for (uint32_t i = 0, max = Vec_Get_Size(old_seg_readers); i < max; i++) {
SegReader *seg_reader = (SegReader*)Vec_Fetch(old_seg_readers, i);
String *seg_name = SegReader_Get_Seg_Name(seg_reader);
// If this segment was merged away...
if (Hash_Fetch(ivars->doc_maps, seg_name)) {
SegReader *new_seg_reader
= (SegReader*)CERTIFY(
Hash_Fetch(new_segs, seg_name),
SEGREADER);
int32_t old_del_count = SegReader_Del_Count(seg_reader);
int32_t new_del_count = SegReader_Del_Count(new_seg_reader);
// ... were any new deletions applied against it?
if (old_del_count != new_del_count) {
DeletionsReader *del_reader
= (DeletionsReader*)SegReader_Obtain(
new_seg_reader,
Class_Get_Name(DELETIONSREADER));
if (!updated_deletions) {
updated_deletions = Hash_new(max);
}
Hash_Store(updated_deletions, seg_name,
(Obj*)DelReader_Iterator(del_reader));
}
}
}
DECREF(new_polyreader);
DECREF(new_segs);
if (!updated_deletions) {
return false;
}
else {
PolyReader *merge_polyreader
= PolyReader_open((Obj*)ivars->folder, ivars->snapshot, NULL);
Vector *merge_seg_readers
= PolyReader_Get_Seg_Readers(merge_polyreader);
Snapshot *latest_snapshot
= Snapshot_Read_File(Snapshot_new(), ivars->folder, NULL);
int64_t new_seg_num
= IxManager_Highest_Seg_Num(ivars->manager, latest_snapshot) + 1;
Segment *new_segment = Seg_new(new_seg_num);
SegWriter *seg_writer = SegWriter_new(ivars->schema, ivars->snapshot,
new_segment, merge_polyreader);
DeletionsWriter *del_writer = SegWriter_Get_Del_Writer(seg_writer);
int64_t merge_seg_num = Seg_Get_Number(ivars->segment);
uint32_t seg_tick = INT32_MAX;
int32_t offset = INT32_MAX;
SegWriter_Prep_Seg_Dir(seg_writer);
for (uint32_t i = 0, max = Vec_Get_Size(merge_seg_readers); i < max; i++) {
SegReader *seg_reader
= (SegReader*)Vec_Fetch(merge_seg_readers, i);
if (SegReader_Get_Seg_Num(seg_reader) == merge_seg_num) {
I32Array *offsets = PolyReader_Offsets(merge_polyreader);
seg_tick = i;
offset = I32Arr_Get(offsets, seg_tick);
DECREF(offsets);
}
}
if (offset == INT32_MAX) { THROW(ERR, "Failed sanity check"); }
HashIterator *iter = HashIter_new(updated_deletions);
while (HashIter_Next(iter)) {
String *seg_name = HashIter_Get_Key(iter);
Matcher *deletions = (Matcher*)HashIter_Get_Value(iter);
I32Array *doc_map
= (I32Array*)CERTIFY(
Hash_Fetch(ivars->doc_maps, seg_name),
I32ARRAY);
int32_t del;
while (0 != (del = Matcher_Next(deletions))) {
// Find the slot where the deleted doc resides in the
// rewritten segment. If the doc was already deleted when we
// were merging, do nothing.
int32_t remapped = I32Arr_Get(doc_map, del);
if (remapped) {
// It's a new deletion, so carry it forward and zap it in
// the rewritten segment.
DelWriter_Delete_By_Doc_ID(del_writer, remapped + offset);
}
}
}
DECREF(iter);
// Finish the segment and clean up.
DelWriter_Finish(del_writer);
SegWriter_Finish(seg_writer);
DECREF(seg_writer);
DECREF(new_segment);
DECREF(latest_snapshot);
DECREF(merge_polyreader);
DECREF(updated_deletions);
}
return true;
}