faiss 0.2.0 → 0.2.1
Sign up to get free protection for your applications and to get access to all the features.
- checksums.yaml +4 -4
- data/CHANGELOG.md +4 -0
- data/lib/faiss/version.rb +1 -1
- data/vendor/faiss/faiss/AutoTune.cpp +292 -291
- data/vendor/faiss/faiss/AutoTune.h +55 -56
- data/vendor/faiss/faiss/Clustering.cpp +334 -195
- data/vendor/faiss/faiss/Clustering.h +88 -35
- data/vendor/faiss/faiss/IVFlib.cpp +171 -195
- data/vendor/faiss/faiss/IVFlib.h +48 -51
- data/vendor/faiss/faiss/Index.cpp +85 -103
- data/vendor/faiss/faiss/Index.h +54 -48
- data/vendor/faiss/faiss/Index2Layer.cpp +139 -164
- data/vendor/faiss/faiss/Index2Layer.h +22 -22
- data/vendor/faiss/faiss/IndexBinary.cpp +45 -37
- data/vendor/faiss/faiss/IndexBinary.h +140 -132
- data/vendor/faiss/faiss/IndexBinaryFlat.cpp +73 -53
- data/vendor/faiss/faiss/IndexBinaryFlat.h +29 -24
- data/vendor/faiss/faiss/IndexBinaryFromFloat.cpp +46 -43
- data/vendor/faiss/faiss/IndexBinaryFromFloat.h +16 -15
- data/vendor/faiss/faiss/IndexBinaryHNSW.cpp +215 -232
- data/vendor/faiss/faiss/IndexBinaryHNSW.h +25 -24
- data/vendor/faiss/faiss/IndexBinaryHash.cpp +182 -177
- data/vendor/faiss/faiss/IndexBinaryHash.h +41 -34
- data/vendor/faiss/faiss/IndexBinaryIVF.cpp +489 -461
- data/vendor/faiss/faiss/IndexBinaryIVF.h +97 -68
- data/vendor/faiss/faiss/IndexFlat.cpp +116 -147
- data/vendor/faiss/faiss/IndexFlat.h +35 -46
- data/vendor/faiss/faiss/IndexHNSW.cpp +372 -348
- data/vendor/faiss/faiss/IndexHNSW.h +57 -41
- data/vendor/faiss/faiss/IndexIVF.cpp +474 -454
- data/vendor/faiss/faiss/IndexIVF.h +146 -113
- data/vendor/faiss/faiss/IndexIVFFlat.cpp +248 -250
- data/vendor/faiss/faiss/IndexIVFFlat.h +48 -51
- data/vendor/faiss/faiss/IndexIVFPQ.cpp +457 -516
- data/vendor/faiss/faiss/IndexIVFPQ.h +74 -66
- data/vendor/faiss/faiss/IndexIVFPQFastScan.cpp +406 -372
- data/vendor/faiss/faiss/IndexIVFPQFastScan.h +82 -57
- data/vendor/faiss/faiss/IndexIVFPQR.cpp +104 -102
- data/vendor/faiss/faiss/IndexIVFPQR.h +33 -28
- data/vendor/faiss/faiss/IndexIVFSpectralHash.cpp +125 -133
- data/vendor/faiss/faiss/IndexIVFSpectralHash.h +19 -21
- data/vendor/faiss/faiss/IndexLSH.cpp +75 -96
- data/vendor/faiss/faiss/IndexLSH.h +21 -26
- data/vendor/faiss/faiss/IndexLattice.cpp +42 -56
- data/vendor/faiss/faiss/IndexLattice.h +11 -16
- data/vendor/faiss/faiss/IndexNNDescent.cpp +231 -0
- data/vendor/faiss/faiss/IndexNNDescent.h +72 -0
- data/vendor/faiss/faiss/IndexNSG.cpp +303 -0
- data/vendor/faiss/faiss/IndexNSG.h +85 -0
- data/vendor/faiss/faiss/IndexPQ.cpp +405 -464
- data/vendor/faiss/faiss/IndexPQ.h +64 -67
- data/vendor/faiss/faiss/IndexPQFastScan.cpp +143 -170
- data/vendor/faiss/faiss/IndexPQFastScan.h +46 -32
- data/vendor/faiss/faiss/IndexPreTransform.cpp +120 -150
- data/vendor/faiss/faiss/IndexPreTransform.h +33 -36
- data/vendor/faiss/faiss/IndexRefine.cpp +115 -131
- data/vendor/faiss/faiss/IndexRefine.h +22 -23
- data/vendor/faiss/faiss/IndexReplicas.cpp +147 -153
- data/vendor/faiss/faiss/IndexReplicas.h +62 -56
- data/vendor/faiss/faiss/IndexResidual.cpp +291 -0
- data/vendor/faiss/faiss/IndexResidual.h +152 -0
- data/vendor/faiss/faiss/IndexScalarQuantizer.cpp +120 -155
- data/vendor/faiss/faiss/IndexScalarQuantizer.h +41 -45
- data/vendor/faiss/faiss/IndexShards.cpp +256 -240
- data/vendor/faiss/faiss/IndexShards.h +85 -73
- data/vendor/faiss/faiss/MatrixStats.cpp +112 -97
- data/vendor/faiss/faiss/MatrixStats.h +7 -10
- data/vendor/faiss/faiss/MetaIndexes.cpp +135 -157
- data/vendor/faiss/faiss/MetaIndexes.h +40 -34
- data/vendor/faiss/faiss/MetricType.h +7 -7
- data/vendor/faiss/faiss/VectorTransform.cpp +652 -474
- data/vendor/faiss/faiss/VectorTransform.h +61 -89
- data/vendor/faiss/faiss/clone_index.cpp +77 -73
- data/vendor/faiss/faiss/clone_index.h +4 -9
- data/vendor/faiss/faiss/gpu/GpuAutoTune.cpp +33 -38
- data/vendor/faiss/faiss/gpu/GpuAutoTune.h +11 -9
- data/vendor/faiss/faiss/gpu/GpuCloner.cpp +197 -170
- data/vendor/faiss/faiss/gpu/GpuCloner.h +53 -35
- data/vendor/faiss/faiss/gpu/GpuClonerOptions.cpp +12 -14
- data/vendor/faiss/faiss/gpu/GpuClonerOptions.h +27 -25
- data/vendor/faiss/faiss/gpu/GpuDistance.h +116 -112
- data/vendor/faiss/faiss/gpu/GpuFaissAssert.h +1 -2
- data/vendor/faiss/faiss/gpu/GpuIndex.h +134 -137
- data/vendor/faiss/faiss/gpu/GpuIndexBinaryFlat.h +76 -73
- data/vendor/faiss/faiss/gpu/GpuIndexFlat.h +173 -162
- data/vendor/faiss/faiss/gpu/GpuIndexIVF.h +67 -64
- data/vendor/faiss/faiss/gpu/GpuIndexIVFFlat.h +89 -86
- data/vendor/faiss/faiss/gpu/GpuIndexIVFPQ.h +150 -141
- data/vendor/faiss/faiss/gpu/GpuIndexIVFScalarQuantizer.h +101 -103
- data/vendor/faiss/faiss/gpu/GpuIndicesOptions.h +17 -16
- data/vendor/faiss/faiss/gpu/GpuResources.cpp +116 -128
- data/vendor/faiss/faiss/gpu/GpuResources.h +182 -186
- data/vendor/faiss/faiss/gpu/StandardGpuResources.cpp +433 -422
- data/vendor/faiss/faiss/gpu/StandardGpuResources.h +131 -130
- data/vendor/faiss/faiss/gpu/impl/InterleavedCodes.cpp +468 -456
- data/vendor/faiss/faiss/gpu/impl/InterleavedCodes.h +25 -19
- data/vendor/faiss/faiss/gpu/impl/RemapIndices.cpp +22 -20
- data/vendor/faiss/faiss/gpu/impl/RemapIndices.h +9 -8
- data/vendor/faiss/faiss/gpu/perf/IndexWrapper-inl.h +39 -44
- data/vendor/faiss/faiss/gpu/perf/IndexWrapper.h +16 -14
- data/vendor/faiss/faiss/gpu/perf/PerfClustering.cpp +77 -71
- data/vendor/faiss/faiss/gpu/perf/PerfIVFPQAdd.cpp +109 -88
- data/vendor/faiss/faiss/gpu/perf/WriteIndex.cpp +75 -64
- data/vendor/faiss/faiss/gpu/test/TestCodePacking.cpp +230 -215
- data/vendor/faiss/faiss/gpu/test/TestGpuIndexBinaryFlat.cpp +80 -86
- data/vendor/faiss/faiss/gpu/test/TestGpuIndexFlat.cpp +284 -277
- data/vendor/faiss/faiss/gpu/test/TestGpuIndexIVFFlat.cpp +416 -416
- data/vendor/faiss/faiss/gpu/test/TestGpuIndexIVFPQ.cpp +611 -517
- data/vendor/faiss/faiss/gpu/test/TestGpuIndexIVFScalarQuantizer.cpp +166 -164
- data/vendor/faiss/faiss/gpu/test/TestGpuMemoryException.cpp +61 -53
- data/vendor/faiss/faiss/gpu/test/TestUtils.cpp +274 -238
- data/vendor/faiss/faiss/gpu/test/TestUtils.h +73 -57
- data/vendor/faiss/faiss/gpu/test/demo_ivfpq_indexing_gpu.cpp +47 -50
- data/vendor/faiss/faiss/gpu/utils/DeviceUtils.h +79 -72
- data/vendor/faiss/faiss/gpu/utils/StackDeviceMemory.cpp +140 -146
- data/vendor/faiss/faiss/gpu/utils/StackDeviceMemory.h +69 -71
- data/vendor/faiss/faiss/gpu/utils/StaticUtils.h +21 -16
- data/vendor/faiss/faiss/gpu/utils/Timer.cpp +25 -29
- data/vendor/faiss/faiss/gpu/utils/Timer.h +30 -29
- data/vendor/faiss/faiss/impl/AdditiveQuantizer.cpp +270 -0
- data/vendor/faiss/faiss/impl/AdditiveQuantizer.h +115 -0
- data/vendor/faiss/faiss/impl/AuxIndexStructures.cpp +90 -120
- data/vendor/faiss/faiss/impl/AuxIndexStructures.h +81 -65
- data/vendor/faiss/faiss/impl/FaissAssert.h +73 -58
- data/vendor/faiss/faiss/impl/FaissException.cpp +56 -48
- data/vendor/faiss/faiss/impl/FaissException.h +41 -29
- data/vendor/faiss/faiss/impl/HNSW.cpp +595 -611
- data/vendor/faiss/faiss/impl/HNSW.h +179 -200
- data/vendor/faiss/faiss/impl/LocalSearchQuantizer.cpp +672 -0
- data/vendor/faiss/faiss/impl/LocalSearchQuantizer.h +172 -0
- data/vendor/faiss/faiss/impl/NNDescent.cpp +487 -0
- data/vendor/faiss/faiss/impl/NNDescent.h +154 -0
- data/vendor/faiss/faiss/impl/NSG.cpp +682 -0
- data/vendor/faiss/faiss/impl/NSG.h +199 -0
- data/vendor/faiss/faiss/impl/PolysemousTraining.cpp +484 -454
- data/vendor/faiss/faiss/impl/PolysemousTraining.h +52 -55
- data/vendor/faiss/faiss/impl/ProductQuantizer-inl.h +26 -47
- data/vendor/faiss/faiss/impl/ProductQuantizer.cpp +469 -459
- data/vendor/faiss/faiss/impl/ProductQuantizer.h +76 -87
- data/vendor/faiss/faiss/impl/ResidualQuantizer.cpp +448 -0
- data/vendor/faiss/faiss/impl/ResidualQuantizer.h +130 -0
- data/vendor/faiss/faiss/impl/ResultHandler.h +96 -132
- data/vendor/faiss/faiss/impl/ScalarQuantizer.cpp +648 -701
- data/vendor/faiss/faiss/impl/ScalarQuantizer.h +48 -46
- data/vendor/faiss/faiss/impl/ThreadedIndex-inl.h +129 -131
- data/vendor/faiss/faiss/impl/ThreadedIndex.h +61 -55
- data/vendor/faiss/faiss/impl/index_read.cpp +547 -479
- data/vendor/faiss/faiss/impl/index_write.cpp +497 -407
- data/vendor/faiss/faiss/impl/io.cpp +75 -94
- data/vendor/faiss/faiss/impl/io.h +31 -41
- data/vendor/faiss/faiss/impl/io_macros.h +40 -29
- data/vendor/faiss/faiss/impl/lattice_Zn.cpp +137 -186
- data/vendor/faiss/faiss/impl/lattice_Zn.h +40 -51
- data/vendor/faiss/faiss/impl/platform_macros.h +29 -8
- data/vendor/faiss/faiss/impl/pq4_fast_scan.cpp +77 -124
- data/vendor/faiss/faiss/impl/pq4_fast_scan.h +39 -48
- data/vendor/faiss/faiss/impl/pq4_fast_scan_search_1.cpp +41 -52
- data/vendor/faiss/faiss/impl/pq4_fast_scan_search_qbs.cpp +80 -117
- data/vendor/faiss/faiss/impl/simd_result_handlers.h +109 -137
- data/vendor/faiss/faiss/index_factory.cpp +269 -218
- data/vendor/faiss/faiss/index_factory.h +6 -7
- data/vendor/faiss/faiss/index_io.h +23 -26
- data/vendor/faiss/faiss/invlists/BlockInvertedLists.cpp +67 -75
- data/vendor/faiss/faiss/invlists/BlockInvertedLists.h +22 -24
- data/vendor/faiss/faiss/invlists/DirectMap.cpp +96 -112
- data/vendor/faiss/faiss/invlists/DirectMap.h +29 -33
- data/vendor/faiss/faiss/invlists/InvertedLists.cpp +307 -364
- data/vendor/faiss/faiss/invlists/InvertedLists.h +151 -151
- data/vendor/faiss/faiss/invlists/InvertedListsIOHook.cpp +29 -34
- data/vendor/faiss/faiss/invlists/InvertedListsIOHook.h +17 -18
- data/vendor/faiss/faiss/invlists/OnDiskInvertedLists.cpp +257 -293
- data/vendor/faiss/faiss/invlists/OnDiskInvertedLists.h +50 -45
- data/vendor/faiss/faiss/python/python_callbacks.cpp +23 -26
- data/vendor/faiss/faiss/python/python_callbacks.h +9 -16
- data/vendor/faiss/faiss/utils/AlignedTable.h +79 -44
- data/vendor/faiss/faiss/utils/Heap.cpp +40 -48
- data/vendor/faiss/faiss/utils/Heap.h +186 -209
- data/vendor/faiss/faiss/utils/WorkerThread.cpp +67 -76
- data/vendor/faiss/faiss/utils/WorkerThread.h +32 -33
- data/vendor/faiss/faiss/utils/distances.cpp +301 -310
- data/vendor/faiss/faiss/utils/distances.h +133 -118
- data/vendor/faiss/faiss/utils/distances_simd.cpp +456 -516
- data/vendor/faiss/faiss/utils/extra_distances-inl.h +117 -0
- data/vendor/faiss/faiss/utils/extra_distances.cpp +113 -232
- data/vendor/faiss/faiss/utils/extra_distances.h +30 -29
- data/vendor/faiss/faiss/utils/hamming-inl.h +260 -209
- data/vendor/faiss/faiss/utils/hamming.cpp +375 -469
- data/vendor/faiss/faiss/utils/hamming.h +62 -85
- data/vendor/faiss/faiss/utils/ordered_key_value.h +16 -18
- data/vendor/faiss/faiss/utils/partitioning.cpp +393 -318
- data/vendor/faiss/faiss/utils/partitioning.h +26 -21
- data/vendor/faiss/faiss/utils/quantize_lut.cpp +78 -66
- data/vendor/faiss/faiss/utils/quantize_lut.h +22 -20
- data/vendor/faiss/faiss/utils/random.cpp +39 -63
- data/vendor/faiss/faiss/utils/random.h +13 -16
- data/vendor/faiss/faiss/utils/simdlib.h +4 -2
- data/vendor/faiss/faiss/utils/simdlib_avx2.h +88 -85
- data/vendor/faiss/faiss/utils/simdlib_emulated.h +226 -165
- data/vendor/faiss/faiss/utils/simdlib_neon.h +832 -0
- data/vendor/faiss/faiss/utils/utils.cpp +304 -287
- data/vendor/faiss/faiss/utils/utils.h +53 -48
- metadata +20 -2
@@ -8,150 +8,141 @@
|
|
8
8
|
// -*- c++ -*-
|
9
9
|
|
10
10
|
#include <algorithm>
|
11
|
-
#include <cstring>
|
12
11
|
#include <cassert>
|
12
|
+
#include <cstring>
|
13
13
|
|
14
|
-
#include <faiss/impl/io.h>
|
15
14
|
#include <faiss/impl/FaissAssert.h>
|
16
|
-
|
15
|
+
#include <faiss/impl/io.h>
|
17
16
|
|
18
17
|
namespace faiss {
|
19
18
|
|
20
|
-
|
21
19
|
/***********************************************************************
|
22
20
|
* IO functions
|
23
21
|
***********************************************************************/
|
24
22
|
|
25
|
-
|
26
|
-
|
27
|
-
{
|
28
|
-
FAISS_THROW_MSG ("IOReader does not support memory mapping");
|
23
|
+
int IOReader::fileno() {
|
24
|
+
FAISS_THROW_MSG("IOReader does not support memory mapping");
|
29
25
|
}
|
30
26
|
|
31
|
-
int IOWriter::fileno
|
32
|
-
|
33
|
-
FAISS_THROW_MSG ("IOWriter does not support memory mapping");
|
27
|
+
int IOWriter::fileno() {
|
28
|
+
FAISS_THROW_MSG("IOWriter does not support memory mapping");
|
34
29
|
}
|
35
30
|
|
36
31
|
/***********************************************************************
|
37
32
|
* IO Vector
|
38
33
|
***********************************************************************/
|
39
34
|
|
40
|
-
|
41
|
-
size_t VectorIOWriter::operator()(
|
42
|
-
const void *ptr, size_t size, size_t nitems)
|
43
|
-
{
|
35
|
+
size_t VectorIOWriter::operator()(const void* ptr, size_t size, size_t nitems) {
|
44
36
|
size_t bytes = size * nitems;
|
45
37
|
if (bytes > 0) {
|
46
38
|
size_t o = data.size();
|
47
39
|
data.resize(o + bytes);
|
48
|
-
memcpy
|
40
|
+
memcpy(&data[o], ptr, size * nitems);
|
49
41
|
}
|
50
42
|
return nitems;
|
51
43
|
}
|
52
44
|
|
53
|
-
size_t VectorIOReader::operator()(
|
54
|
-
|
55
|
-
|
56
|
-
if (rp >= data.size()) return 0;
|
45
|
+
size_t VectorIOReader::operator()(void* ptr, size_t size, size_t nitems) {
|
46
|
+
if (rp >= data.size())
|
47
|
+
return 0;
|
57
48
|
size_t nremain = (data.size() - rp) / size;
|
58
|
-
if (nremain < nitems)
|
49
|
+
if (nremain < nitems)
|
50
|
+
nitems = nremain;
|
59
51
|
if (size * nitems > 0) {
|
60
|
-
memcpy
|
52
|
+
memcpy(ptr, &data[rp], size * nitems);
|
61
53
|
rp += size * nitems;
|
62
54
|
}
|
63
55
|
return nitems;
|
64
56
|
}
|
65
57
|
|
66
|
-
|
67
|
-
|
68
|
-
|
69
58
|
/***********************************************************************
|
70
59
|
* IO File
|
71
60
|
***********************************************************************/
|
72
61
|
|
62
|
+
FileIOReader::FileIOReader(FILE* rf) : f(rf) {}
|
73
63
|
|
74
|
-
|
75
|
-
FileIOReader::FileIOReader(FILE *rf): f(rf) {}
|
76
|
-
|
77
|
-
FileIOReader::FileIOReader(const char * fname)
|
78
|
-
{
|
64
|
+
FileIOReader::FileIOReader(const char* fname) {
|
79
65
|
name = fname;
|
80
66
|
f = fopen(fname, "rb");
|
81
|
-
FAISS_THROW_IF_NOT_FMT
|
82
|
-
|
67
|
+
FAISS_THROW_IF_NOT_FMT(
|
68
|
+
f, "could not open %s for reading: %s", fname, strerror(errno));
|
83
69
|
need_close = true;
|
84
70
|
}
|
85
71
|
|
86
|
-
FileIOReader::~FileIOReader()
|
72
|
+
FileIOReader::~FileIOReader() {
|
87
73
|
if (need_close) {
|
88
74
|
int ret = fclose(f);
|
89
|
-
if (ret != 0) {// we cannot raise and exception in the destructor
|
90
|
-
fprintf(stderr,
|
91
|
-
|
75
|
+
if (ret != 0) { // we cannot raise and exception in the destructor
|
76
|
+
fprintf(stderr,
|
77
|
+
"file %s close error: %s",
|
78
|
+
name.c_str(),
|
79
|
+
strerror(errno));
|
92
80
|
}
|
93
81
|
}
|
94
82
|
}
|
95
83
|
|
96
|
-
size_t FileIOReader::operator()(void
|
84
|
+
size_t FileIOReader::operator()(void* ptr, size_t size, size_t nitems) {
|
97
85
|
return fread(ptr, size, nitems, f);
|
98
86
|
}
|
99
87
|
|
100
|
-
int FileIOReader::fileno()
|
101
|
-
return ::fileno
|
88
|
+
int FileIOReader::fileno() {
|
89
|
+
return ::fileno(f);
|
102
90
|
}
|
103
91
|
|
92
|
+
FileIOWriter::FileIOWriter(FILE* wf) : f(wf) {}
|
104
93
|
|
105
|
-
FileIOWriter::FileIOWriter(
|
106
|
-
|
107
|
-
FileIOWriter::FileIOWriter(const char * fname)
|
108
|
-
{
|
94
|
+
FileIOWriter::FileIOWriter(const char* fname) {
|
109
95
|
name = fname;
|
110
96
|
f = fopen(fname, "wb");
|
111
|
-
FAISS_THROW_IF_NOT_FMT
|
112
|
-
|
97
|
+
FAISS_THROW_IF_NOT_FMT(
|
98
|
+
f, "could not open %s for writing: %s", fname, strerror(errno));
|
113
99
|
need_close = true;
|
114
100
|
}
|
115
101
|
|
116
|
-
FileIOWriter::~FileIOWriter()
|
102
|
+
FileIOWriter::~FileIOWriter() {
|
117
103
|
if (need_close) {
|
118
104
|
int ret = fclose(f);
|
119
105
|
if (ret != 0) {
|
120
106
|
// we cannot raise and exception in the destructor
|
121
|
-
fprintf(stderr,
|
122
|
-
|
107
|
+
fprintf(stderr,
|
108
|
+
"file %s close error: %s",
|
109
|
+
name.c_str(),
|
110
|
+
strerror(errno));
|
123
111
|
}
|
124
112
|
}
|
125
113
|
}
|
126
114
|
|
127
|
-
size_t FileIOWriter::operator()(const void
|
115
|
+
size_t FileIOWriter::operator()(const void* ptr, size_t size, size_t nitems) {
|
128
116
|
return fwrite(ptr, size, nitems, f);
|
129
117
|
}
|
130
118
|
|
131
|
-
int FileIOWriter::fileno()
|
132
|
-
return ::fileno
|
119
|
+
int FileIOWriter::fileno() {
|
120
|
+
return ::fileno(f);
|
133
121
|
}
|
134
122
|
|
135
123
|
/***********************************************************************
|
136
124
|
* IO buffer
|
137
125
|
***********************************************************************/
|
138
126
|
|
139
|
-
BufferedIOReader::BufferedIOReader(IOReader
|
140
|
-
|
141
|
-
|
142
|
-
|
143
|
-
|
127
|
+
BufferedIOReader::BufferedIOReader(IOReader* reader, size_t bsz)
|
128
|
+
: reader(reader),
|
129
|
+
bsz(bsz),
|
130
|
+
ofs(0),
|
131
|
+
ofs2(0),
|
132
|
+
b0(0),
|
133
|
+
b1(0),
|
134
|
+
buffer(bsz) {}
|
144
135
|
|
145
|
-
size_t BufferedIOReader::operator()(void
|
146
|
-
{
|
136
|
+
size_t BufferedIOReader::operator()(void* ptr, size_t unitsize, size_t nitems) {
|
147
137
|
size_t size = unitsize * nitems;
|
148
|
-
if (size == 0)
|
149
|
-
|
138
|
+
if (size == 0)
|
139
|
+
return 0;
|
140
|
+
char* dst = (char*)ptr;
|
150
141
|
size_t nb;
|
151
142
|
|
152
143
|
{ // first copy available bytes
|
153
144
|
nb = std::min(b1 - b0, size);
|
154
|
-
memcpy
|
145
|
+
memcpy(dst, buffer.data() + b0, nb);
|
155
146
|
b0 += nb;
|
156
147
|
dst += nb;
|
157
148
|
size -= nb;
|
@@ -159,7 +150,7 @@ size_t BufferedIOReader::operator()(void *ptr, size_t unitsize, size_t nitems)
|
|
159
150
|
|
160
151
|
// while we would like to have more data
|
161
152
|
while (size > 0) {
|
162
|
-
assert
|
153
|
+
assert(b0 == b1); // buffer empty on input
|
163
154
|
// try to read from main reader
|
164
155
|
b0 = 0;
|
165
156
|
b1 = (*reader)(buffer.data(), 1, bsz);
|
@@ -172,7 +163,7 @@ size_t BufferedIOReader::operator()(void *ptr, size_t unitsize, size_t nitems)
|
|
172
163
|
|
173
164
|
// copy remaining bytes
|
174
165
|
size_t nb2 = std::min(b1, size);
|
175
|
-
memcpy
|
166
|
+
memcpy(dst, buffer.data(), nb2);
|
176
167
|
b0 = nb2;
|
177
168
|
nb += nb2;
|
178
169
|
dst += nb2;
|
@@ -182,22 +173,22 @@ size_t BufferedIOReader::operator()(void *ptr, size_t unitsize, size_t nitems)
|
|
182
173
|
return nb / unitsize;
|
183
174
|
}
|
184
175
|
|
176
|
+
BufferedIOWriter::BufferedIOWriter(IOWriter* writer, size_t bsz)
|
177
|
+
: writer(writer), bsz(bsz), ofs2(0), b0(0), buffer(bsz) {}
|
185
178
|
|
186
|
-
BufferedIOWriter::
|
187
|
-
|
188
|
-
|
189
|
-
|
190
|
-
|
191
|
-
size_t BufferedIOWriter::operator()(const void *ptr, size_t unitsize, size_t nitems)
|
192
|
-
{
|
179
|
+
size_t BufferedIOWriter::operator()(
|
180
|
+
const void* ptr,
|
181
|
+
size_t unitsize,
|
182
|
+
size_t nitems) {
|
193
183
|
size_t size = unitsize * nitems;
|
194
|
-
if (size == 0)
|
195
|
-
|
184
|
+
if (size == 0)
|
185
|
+
return 0;
|
186
|
+
const char* src = (const char*)ptr;
|
196
187
|
size_t nb;
|
197
188
|
|
198
189
|
{ // copy as many bytes as possible to buffer
|
199
190
|
nb = std::min(bsz - b0, size);
|
200
|
-
memcpy
|
191
|
+
memcpy(buffer.data() + b0, src, nb);
|
201
192
|
b0 += nb;
|
202
193
|
src += nb;
|
203
194
|
size -= nb;
|
@@ -207,15 +198,15 @@ size_t BufferedIOWriter::operator()(const void *ptr, size_t unitsize, size_t nit
|
|
207
198
|
// now we need to flush to add more bytes
|
208
199
|
size_t ofs = 0;
|
209
200
|
do {
|
210
|
-
assert
|
201
|
+
assert(ofs < 10000000);
|
211
202
|
size_t written = (*writer)(buffer.data() + ofs, 1, bsz - ofs);
|
212
203
|
FAISS_THROW_IF_NOT(written > 0);
|
213
204
|
ofs += written;
|
214
|
-
} while(ofs != bsz);
|
205
|
+
} while (ofs != bsz);
|
215
206
|
|
216
207
|
// copy src to buffer
|
217
208
|
size_t nb1 = std::min(bsz, size);
|
218
|
-
memcpy
|
209
|
+
memcpy(buffer.data(), src, nb1);
|
219
210
|
b0 = nb1;
|
220
211
|
nb += nb1;
|
221
212
|
src += nb1;
|
@@ -225,31 +216,25 @@ size_t BufferedIOWriter::operator()(const void *ptr, size_t unitsize, size_t nit
|
|
225
216
|
return nb / unitsize;
|
226
217
|
}
|
227
218
|
|
228
|
-
BufferedIOWriter::~BufferedIOWriter()
|
229
|
-
{
|
219
|
+
BufferedIOWriter::~BufferedIOWriter() {
|
230
220
|
size_t ofs = 0;
|
231
|
-
while(ofs != b0) {
|
221
|
+
while (ofs != b0) {
|
232
222
|
// printf("Destructor write %zd \n", b0 - ofs);
|
233
223
|
size_t written = (*writer)(buffer.data() + ofs, 1, b0 - ofs);
|
234
224
|
FAISS_THROW_IF_NOT(written > 0);
|
235
225
|
ofs += written;
|
236
226
|
}
|
237
|
-
|
238
227
|
}
|
239
228
|
|
240
|
-
|
241
|
-
|
242
|
-
|
243
|
-
|
244
|
-
uint32_t fourcc (const char sx[4]) {
|
245
|
-
FAISS_THROW_IF_NOT (4 == strlen(sx));
|
246
|
-
const unsigned char *x = (unsigned char*)sx;
|
229
|
+
uint32_t fourcc(const char sx[4]) {
|
230
|
+
FAISS_THROW_IF_NOT(4 == strlen(sx));
|
231
|
+
const unsigned char* x = (unsigned char*)sx;
|
247
232
|
return x[0] | x[1] << 8 | x[2] << 16 | x[3] << 24;
|
248
233
|
}
|
249
234
|
|
250
|
-
uint32_t fourcc
|
251
|
-
FAISS_THROW_IF_NOT
|
252
|
-
const unsigned char
|
235
|
+
uint32_t fourcc(const std::string& sx) {
|
236
|
+
FAISS_THROW_IF_NOT(sx.length() == 4);
|
237
|
+
const unsigned char* x = (unsigned char*)sx.c_str();
|
253
238
|
return x[0] | x[1] << 8 | x[2] << 16 | x[3] << 24;
|
254
239
|
}
|
255
240
|
|
@@ -264,7 +249,6 @@ std::string fourcc_inv(uint32_t x) {
|
|
264
249
|
return std::string(str);
|
265
250
|
}
|
266
251
|
|
267
|
-
|
268
252
|
std::string fourcc_inv_printable(uint32_t x) {
|
269
253
|
char cstr[5];
|
270
254
|
fourcc_inv(x, cstr);
|
@@ -282,7 +266,4 @@ std::string fourcc_inv_printable(uint32_t x) {
|
|
282
266
|
return str;
|
283
267
|
}
|
284
268
|
|
285
|
-
|
286
|
-
|
287
|
-
|
288
269
|
} // namespace faiss
|
@@ -16,25 +16,23 @@
|
|
16
16
|
|
17
17
|
#pragma once
|
18
18
|
|
19
|
-
#include <string>
|
20
19
|
#include <cstdio>
|
20
|
+
#include <string>
|
21
21
|
#include <vector>
|
22
22
|
|
23
23
|
#include <faiss/Index.h>
|
24
24
|
|
25
25
|
namespace faiss {
|
26
26
|
|
27
|
-
|
28
27
|
struct IOReader {
|
29
28
|
// name that can be used in error messages
|
30
29
|
std::string name;
|
31
30
|
|
32
31
|
// fread. Returns number of items read or 0 in case of EOF.
|
33
|
-
virtual size_t operator()(
|
34
|
-
void *ptr, size_t size, size_t nitems) = 0;
|
32
|
+
virtual size_t operator()(void* ptr, size_t size, size_t nitems) = 0;
|
35
33
|
|
36
34
|
// return a file number that can be memory-mapped
|
37
|
-
virtual int fileno
|
35
|
+
virtual int fileno();
|
38
36
|
|
39
37
|
virtual ~IOReader() {}
|
40
38
|
};
|
@@ -44,53 +42,51 @@ struct IOWriter {
|
|
44
42
|
std::string name;
|
45
43
|
|
46
44
|
// fwrite. Return number of items written
|
47
|
-
virtual size_t operator()(
|
48
|
-
const void *ptr, size_t size, size_t nitems) = 0;
|
45
|
+
virtual size_t operator()(const void* ptr, size_t size, size_t nitems) = 0;
|
49
46
|
|
50
47
|
// return a file number that can be memory-mapped
|
51
|
-
virtual int fileno
|
48
|
+
virtual int fileno();
|
52
49
|
|
53
50
|
virtual ~IOWriter() noexcept(false) {}
|
54
51
|
};
|
55
52
|
|
56
|
-
|
57
|
-
struct VectorIOReader:IOReader {
|
53
|
+
struct VectorIOReader : IOReader {
|
58
54
|
std::vector<uint8_t> data;
|
59
55
|
size_t rp = 0;
|
60
|
-
size_t operator()(void
|
56
|
+
size_t operator()(void* ptr, size_t size, size_t nitems) override;
|
61
57
|
};
|
62
58
|
|
63
|
-
struct VectorIOWriter:IOWriter {
|
59
|
+
struct VectorIOWriter : IOWriter {
|
64
60
|
std::vector<uint8_t> data;
|
65
|
-
size_t operator()(const void
|
61
|
+
size_t operator()(const void* ptr, size_t size, size_t nitems) override;
|
66
62
|
};
|
67
63
|
|
68
|
-
struct FileIOReader: IOReader {
|
69
|
-
FILE
|
64
|
+
struct FileIOReader : IOReader {
|
65
|
+
FILE* f = nullptr;
|
70
66
|
bool need_close = false;
|
71
67
|
|
72
|
-
FileIOReader(FILE
|
68
|
+
FileIOReader(FILE* rf);
|
73
69
|
|
74
|
-
FileIOReader(const char
|
70
|
+
FileIOReader(const char* fname);
|
75
71
|
|
76
72
|
~FileIOReader() override;
|
77
73
|
|
78
|
-
size_t operator()(void
|
74
|
+
size_t operator()(void* ptr, size_t size, size_t nitems) override;
|
79
75
|
|
80
76
|
int fileno() override;
|
81
77
|
};
|
82
78
|
|
83
|
-
struct FileIOWriter: IOWriter {
|
84
|
-
FILE
|
79
|
+
struct FileIOWriter : IOWriter {
|
80
|
+
FILE* f = nullptr;
|
85
81
|
bool need_close = false;
|
86
82
|
|
87
|
-
FileIOWriter(FILE
|
83
|
+
FileIOWriter(FILE* wf);
|
88
84
|
|
89
|
-
FileIOWriter(const char
|
85
|
+
FileIOWriter(const char* fname);
|
90
86
|
|
91
87
|
~FileIOWriter() override;
|
92
88
|
|
93
|
-
size_t operator()(const void
|
89
|
+
size_t operator()(const void* ptr, size_t size, size_t nitems) override;
|
94
90
|
|
95
91
|
int fileno() override;
|
96
92
|
};
|
@@ -103,12 +99,9 @@ struct FileIOWriter: IOWriter {
|
|
103
99
|
* the read/write functions.
|
104
100
|
*******************************************************/
|
105
101
|
|
106
|
-
|
107
|
-
|
108
102
|
/** wraps an ioreader to make buffered reads to avoid too small reads */
|
109
|
-
struct BufferedIOReader: IOReader {
|
110
|
-
|
111
|
-
IOReader *reader;
|
103
|
+
struct BufferedIOReader : IOReader {
|
104
|
+
IOReader* reader;
|
112
105
|
size_t bsz;
|
113
106
|
size_t ofs; ///< offset in input stream
|
114
107
|
size_t ofs2; ///< number of bytes returned to caller
|
@@ -119,37 +112,34 @@ struct BufferedIOReader: IOReader {
|
|
119
112
|
* @param bsz buffer size (bytes). Reads will be done by batched of
|
120
113
|
* this size
|
121
114
|
*/
|
122
|
-
explicit BufferedIOReader(IOReader
|
115
|
+
explicit BufferedIOReader(IOReader* reader, size_t bsz = 1024 * 1024);
|
123
116
|
|
124
|
-
size_t operator()(void
|
117
|
+
size_t operator()(void* ptr, size_t size, size_t nitems) override;
|
125
118
|
};
|
126
119
|
|
127
|
-
|
128
|
-
|
129
|
-
|
130
|
-
IOWriter *writer;
|
120
|
+
struct BufferedIOWriter : IOWriter {
|
121
|
+
IOWriter* writer;
|
131
122
|
size_t bsz;
|
132
123
|
size_t ofs;
|
133
|
-
size_t ofs2;
|
134
|
-
size_t b0;
|
124
|
+
size_t ofs2; ///< number of bytes received from caller
|
125
|
+
size_t b0; ///< amount of data in buffer
|
135
126
|
std::vector<char> buffer;
|
136
127
|
|
137
|
-
explicit BufferedIOWriter(IOWriter
|
128
|
+
explicit BufferedIOWriter(IOWriter* writer, size_t bsz = 1024 * 1024);
|
138
129
|
|
139
|
-
size_t operator()(const void
|
130
|
+
size_t operator()(const void* ptr, size_t size, size_t nitems) override;
|
140
131
|
|
141
132
|
// flushes
|
142
133
|
~BufferedIOWriter() override;
|
143
134
|
};
|
144
135
|
|
145
136
|
/// cast a 4-character string to a uint32_t that can be written and read easily
|
146
|
-
uint32_t fourcc
|
147
|
-
uint32_t fourcc
|
137
|
+
uint32_t fourcc(const char sx[4]);
|
138
|
+
uint32_t fourcc(const std::string& sx);
|
148
139
|
|
149
140
|
// decoding of fourcc (int32 -> string)
|
150
141
|
void fourcc_inv(uint32_t x, char str[5]);
|
151
142
|
std::string fourcc_inv(uint32_t x);
|
152
143
|
std::string fourcc_inv_printable(uint32_t x);
|
153
144
|
|
154
|
-
|
155
145
|
} // namespace faiss
|