1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
|
// Copyright Yahoo. Licensed under the terms of the Apache 2.0 license. See LICENSE in the project root.
#include "flagattribute.h"
#include "load_utils.hpp"
#include "multinumericattribute.h"
#include "multi_numeric_flag_search_context.h"
#include <vespa/searchcommon/attribute/config.h>
#include <vespa/searchlib/common/bitvector.h>
#include <vespa/searchlib/query/query_term_simple.h>
#include <vespa/log/log.h>
LOG_SETUP(".searchlib.attribute.flag_attribute");
namespace search {
using queryeval::SearchIterator;
namespace {
template <class FA, typename T>
class SaveBits
{
vespalib::ConstArrayRef<T> _map;
FA &_fa;
public:
SaveBits(vespalib::ConstArrayRef<T> map, FA &fa)
: _map(map),
_fa(fa)
{
}
void
save(uint32_t e, uint32_t docId, int32_t weight)
{
(void) weight;
assert(e < _map.size());
_fa.setNewBVValue(docId, _map[e]);
}
};
}
template <typename B>
FlagAttributeT<B>::FlagAttributeT(const vespalib::string & baseFileName, const AttributeVector::Config & cfg) :
B(baseFileName, cfg),
_bitVectorHolder(),
_bitVectorStore(256),
_bitVectors(256),
_bitVectorSize(cfg.getGrowStrategy().getInitialCapacity())
{
}
template <typename B>
std::unique_ptr<attribute::SearchContext>
FlagAttributeT<B>::getSearch(QueryTermSimple::UP qTerm, const attribute::SearchContextParams &) const
{
return std::make_unique<attribute::MultiNumericFlagSearchContext<typename B::BaseType, typename B::WType>>(std::move(qTerm), *this, this->_mvMapping.make_read_view(this->getCommittedDocIdLimit()), _bitVectors);
}
template <typename B>
void FlagAttributeT<B>::clearOldValues(DocId doc)
{
const typename B::WType * values(nullptr);
for (uint32_t i(0), m(this->get(doc, values)); i < m; i++) {
BitVector * bv = _bitVectors[getOffset(multivalue::get_value(values[i]))].load_relaxed();
if (bv != nullptr) {
bv->clearBitAndMaintainCount(doc);
}
}
}
template <typename B>
bool
FlagAttributeT<B>::onLoadEnumerated(ReaderBase &attrReader)
{
using TT = multivalue::ValueType_t<typename B::WType>;
uint32_t numDocs = attrReader.getNumIdx() - 1;
uint64_t numValues = attrReader.getNumValues();
uint64_t enumCount = attrReader.getEnumCount();
assert(numValues == enumCount);
(void) enumCount;
this->setNumDocs(numDocs);
this->setCommittedDocIdLimit(numDocs);
if (numValues > 0)
_bitVectorSize = numDocs;
auto udatBuffer = attribute::LoadUtils::loadUDAT(*this);
assert((udatBuffer->size() % sizeof(TT)) == 0);
vespalib::ConstArrayRef<TT> map(reinterpret_cast<const TT *>(udatBuffer->buffer()),
udatBuffer->size() / sizeof(TT));
SaveBits<FlagAttributeT<B>, TT> saver(map, *this);
uint32_t maxvc = attribute::loadFromEnumeratedMultiValue(this->_mvMapping, attrReader, map, vespalib::ConstArrayRef<uint32_t>(), saver);
this->checkSetMaxValueCount(maxvc);
return true;
}
template <typename B>
bool FlagAttributeT<B>::onLoad(vespalib::Executor * executor)
{
for (size_t i(0), m(_bitVectors.size()); i < m; i++) {
_bitVectorStore[i].reset();
_bitVectors[i].store_relaxed(nullptr);
}
_bitVectorSize = 0;
return B::onLoad(executor);
}
template <typename B>
void FlagAttributeT<B>::setNewValues(DocId doc, const std::vector<typename B::WType> & values)
{
B::setNewValues(doc, values);
if (_bitVectorSize == 0) { // attribute being loaded
_bitVectorSize = this->getNumDocs();
}
for (uint32_t i(0), m(values.size()); i < m; i++) {
typename B::WType value = values[i];
uint32_t offset = getOffset(value);
BitVector * bv = _bitVectors[offset].load_relaxed();
if (bv == nullptr) {
assert(_bitVectorSize >= this->getNumDocs());
_bitVectorStore[offset] = std::make_shared<GrowableBitVector>(_bitVectorSize, _bitVectorSize, _bitVectorHolder);
_bitVectors[offset].store_release(&_bitVectorStore[offset]->writer());
bv = _bitVectors[offset].load_relaxed();
ensureGuardBit(*bv);
}
bv->setBitAndMaintainCount(doc);
}
}
template <typename B>
void
FlagAttributeT<B>::setNewBVValue(DocId doc, multivalue::ValueType_t<typename B::WType> value)
{
uint32_t offset = getOffset(value);
BitVector * bv = _bitVectors[offset].load_relaxed();
if (bv == nullptr) {
assert(_bitVectorSize >= this->getNumDocs());
_bitVectorStore[offset] = std::make_shared<GrowableBitVector>(_bitVectorSize, _bitVectorSize, _bitVectorHolder);
_bitVectors[offset].store_release(&_bitVectorStore[offset]->writer());
bv = _bitVectors[offset].load_relaxed();
ensureGuardBit(*bv);
}
bv->setBitAndMaintainCount(doc);
}
template <typename B>
bool
FlagAttributeT<B>::onAddDoc(DocId doc)
{
bool retval = false;
if (doc >= _bitVectorSize) {
resizeBitVectors(this->getNumDocs());
retval = true;
} else {
ensureGuardBit();
}
std::atomic_thread_fence(std::memory_order_release);
clearGuardBit(doc);
return retval;
}
template <typename B>
void
FlagAttributeT<B>::onAddDocs(DocId docidLimit)
{
if (docidLimit > _bitVectorSize) {
resizeBitVectors(docidLimit);
}
}
template <typename B>
void
FlagAttributeT<B>::ensureGuardBit(BitVector & bv)
{
if (this->getNumDocs() < bv.size()) {
bv.setBit(this->getNumDocs()); // add guard bit to avoid scanning to the end during search
}
}
template <typename B>
void
FlagAttributeT<B>::ensureGuardBit()
{
for (const auto &wrapper: _bitVectors) {
BitVector *bv = wrapper.load_relaxed();
if (bv != nullptr) {
ensureGuardBit(*bv);
}
}
}
template <typename B>
void
FlagAttributeT<B>::clearGuardBit(DocId doc)
{
for (const auto &wrapper: _bitVectors) {
BitVector *bv = wrapper.load_relaxed();
if (bv != nullptr) {
bv->clearBit(doc); // clear guard bit and start using this doc id
}
}
}
template <typename B>
void
FlagAttributeT<B>::resizeBitVectors(uint32_t neededSize)
{
const GrowStrategy & gs = this->getConfig().getGrowStrategy();
uint32_t newSize = neededSize + (neededSize * gs.getGrowFactor()) + gs.getGrowDelta();
for (size_t i(0), m(_bitVectors.size()); i < m; i++) {
BitVector *bv = _bitVectors[i].load_relaxed();
if (bv != nullptr) {
if (_bitVectorStore[i]->extend(newSize)) {
_bitVectors[i].store_release(&_bitVectorStore[i]->writer());
bv = _bitVectors[i].load_relaxed();
}
ensureGuardBit(*bv);
}
}
_bitVectorSize = newSize;
_bitVectorHolder.assign_generation(this->getCurrentGeneration());
}
template <typename B>
void
FlagAttributeT<B>::reclaim_memory(vespalib::GenerationHandler::generation_t oldest_used_gen)
{
B::reclaim_memory(oldest_used_gen);
_bitVectorHolder.reclaim(oldest_used_gen);
}
template class FlagAttributeT<FlagBaseImpl>;
}
|
