simdjson 0.1.0
Sign up to get free protection for your applications and to get access to all the features.
- checksums.yaml +7 -0
- data/.clang-format +5 -0
- data/.gitignore +14 -0
- data/.gitmodules +3 -0
- data/.rubocop.yml +9 -0
- data/.travis.yml +7 -0
- data/Gemfile +4 -0
- data/LICENSE.txt +21 -0
- data/README.md +39 -0
- data/Rakefile +32 -0
- data/benchmark/apache_builds.json +4421 -0
- data/benchmark/demo.json +15 -0
- data/benchmark/github_events.json +1390 -0
- data/benchmark/run_benchmark.rb +30 -0
- data/ext/simdjson/extconf.rb +22 -0
- data/ext/simdjson/simdjson.cpp +76 -0
- data/ext/simdjson/simdjson.hpp +6 -0
- data/lib/simdjson/version.rb +3 -0
- data/lib/simdjson.rb +2 -0
- data/simdjson.gemspec +35 -0
- data/vendor/.gitkeep +0 -0
- data/vendor/simdjson/AUTHORS +3 -0
- data/vendor/simdjson/CMakeLists.txt +63 -0
- data/vendor/simdjson/CONTRIBUTORS +27 -0
- data/vendor/simdjson/Dockerfile +10 -0
- data/vendor/simdjson/LICENSE +201 -0
- data/vendor/simdjson/Makefile +203 -0
- data/vendor/simdjson/Notes.md +85 -0
- data/vendor/simdjson/README.md +581 -0
- data/vendor/simdjson/amalgamation.sh +158 -0
- data/vendor/simdjson/benchmark/CMakeLists.txt +8 -0
- data/vendor/simdjson/benchmark/benchmark.h +223 -0
- data/vendor/simdjson/benchmark/distinctuseridcompetition.cpp +347 -0
- data/vendor/simdjson/benchmark/linux/linux-perf-events.h +93 -0
- data/vendor/simdjson/benchmark/minifiercompetition.cpp +181 -0
- data/vendor/simdjson/benchmark/parse.cpp +393 -0
- data/vendor/simdjson/benchmark/parseandstatcompetition.cpp +305 -0
- data/vendor/simdjson/benchmark/parsingcompetition.cpp +298 -0
- data/vendor/simdjson/benchmark/statisticalmodel.cpp +208 -0
- data/vendor/simdjson/dependencies/jsoncppdist/json/json-forwards.h +344 -0
- data/vendor/simdjson/dependencies/jsoncppdist/json/json.h +2366 -0
- data/vendor/simdjson/dependencies/jsoncppdist/jsoncpp.cpp +5418 -0
- data/vendor/simdjson/doc/apache_builds.jsonparseandstat.png +0 -0
- data/vendor/simdjson/doc/gbps.png +0 -0
- data/vendor/simdjson/doc/github_events.jsonparseandstat.png +0 -0
- data/vendor/simdjson/doc/twitter.jsonparseandstat.png +0 -0
- data/vendor/simdjson/doc/update-center.jsonparseandstat.png +0 -0
- data/vendor/simdjson/images/halvarflake.png +0 -0
- data/vendor/simdjson/images/logo.png +0 -0
- data/vendor/simdjson/include/simdjson/common_defs.h +102 -0
- data/vendor/simdjson/include/simdjson/isadetection.h +152 -0
- data/vendor/simdjson/include/simdjson/jsoncharutils.h +301 -0
- data/vendor/simdjson/include/simdjson/jsonformatutils.h +202 -0
- data/vendor/simdjson/include/simdjson/jsonioutil.h +32 -0
- data/vendor/simdjson/include/simdjson/jsonminifier.h +30 -0
- data/vendor/simdjson/include/simdjson/jsonparser.h +250 -0
- data/vendor/simdjson/include/simdjson/numberparsing.h +587 -0
- data/vendor/simdjson/include/simdjson/padded_string.h +70 -0
- data/vendor/simdjson/include/simdjson/parsedjson.h +544 -0
- data/vendor/simdjson/include/simdjson/portability.h +172 -0
- data/vendor/simdjson/include/simdjson/simdjson.h +44 -0
- data/vendor/simdjson/include/simdjson/simdjson_version.h +13 -0
- data/vendor/simdjson/include/simdjson/simdprune_tables.h +35074 -0
- data/vendor/simdjson/include/simdjson/simdutf8check_arm64.h +180 -0
- data/vendor/simdjson/include/simdjson/simdutf8check_haswell.h +198 -0
- data/vendor/simdjson/include/simdjson/simdutf8check_westmere.h +169 -0
- data/vendor/simdjson/include/simdjson/stage1_find_marks.h +121 -0
- data/vendor/simdjson/include/simdjson/stage1_find_marks_arm64.h +210 -0
- data/vendor/simdjson/include/simdjson/stage1_find_marks_flatten.h +93 -0
- data/vendor/simdjson/include/simdjson/stage1_find_marks_flatten_haswell.h +95 -0
- data/vendor/simdjson/include/simdjson/stage1_find_marks_haswell.h +210 -0
- data/vendor/simdjson/include/simdjson/stage1_find_marks_macros.h +239 -0
- data/vendor/simdjson/include/simdjson/stage1_find_marks_westmere.h +194 -0
- data/vendor/simdjson/include/simdjson/stage2_build_tape.h +85 -0
- data/vendor/simdjson/include/simdjson/stringparsing.h +105 -0
- data/vendor/simdjson/include/simdjson/stringparsing_arm64.h +56 -0
- data/vendor/simdjson/include/simdjson/stringparsing_haswell.h +43 -0
- data/vendor/simdjson/include/simdjson/stringparsing_macros.h +88 -0
- data/vendor/simdjson/include/simdjson/stringparsing_westmere.h +41 -0
- data/vendor/simdjson/jsonexamples/small/jsoniter_scala/README.md +4 -0
- data/vendor/simdjson/scripts/dumpsimplestats.sh +11 -0
- data/vendor/simdjson/scripts/issue150.sh +14 -0
- data/vendor/simdjson/scripts/javascript/README.md +3 -0
- data/vendor/simdjson/scripts/javascript/generatelargejson.js +19 -0
- data/vendor/simdjson/scripts/minifier.sh +11 -0
- data/vendor/simdjson/scripts/parseandstat.sh +24 -0
- data/vendor/simdjson/scripts/parser.sh +11 -0
- data/vendor/simdjson/scripts/parsingcompdata.sh +26 -0
- data/vendor/simdjson/scripts/plotparse.sh +98 -0
- data/vendor/simdjson/scripts/selectparser.sh +11 -0
- data/vendor/simdjson/scripts/setupfortesting/disablehyperthreading.sh +15 -0
- data/vendor/simdjson/scripts/setupfortesting/powerpolicy.sh +32 -0
- data/vendor/simdjson/scripts/setupfortesting/setupfortesting.sh +6 -0
- data/vendor/simdjson/scripts/setupfortesting/turboboost.sh +51 -0
- data/vendor/simdjson/scripts/testjson2json.sh +99 -0
- data/vendor/simdjson/scripts/transitions/Makefile +10 -0
- data/vendor/simdjson/scripts/transitions/generatetransitions.cpp +20 -0
- data/vendor/simdjson/singleheader/README.md +1 -0
- data/vendor/simdjson/singleheader/amalgamation_demo.cpp +20 -0
- data/vendor/simdjson/singleheader/simdjson.cpp +1652 -0
- data/vendor/simdjson/singleheader/simdjson.h +39692 -0
- data/vendor/simdjson/src/CMakeLists.txt +67 -0
- data/vendor/simdjson/src/jsonioutil.cpp +35 -0
- data/vendor/simdjson/src/jsonminifier.cpp +285 -0
- data/vendor/simdjson/src/jsonparser.cpp +91 -0
- data/vendor/simdjson/src/parsedjson.cpp +323 -0
- data/vendor/simdjson/src/parsedjsoniterator.cpp +272 -0
- data/vendor/simdjson/src/simdjson.cpp +30 -0
- data/vendor/simdjson/src/stage1_find_marks.cpp +41 -0
- data/vendor/simdjson/src/stage2_build_tape.cpp +567 -0
- data/vendor/simdjson/style/clang-format-check.sh +25 -0
- data/vendor/simdjson/style/clang-format.sh +25 -0
- data/vendor/simdjson/style/run-clang-format.py +326 -0
- data/vendor/simdjson/tape.md +134 -0
- data/vendor/simdjson/tests/CMakeLists.txt +25 -0
- data/vendor/simdjson/tests/allparserscheckfile.cpp +192 -0
- data/vendor/simdjson/tests/basictests.cpp +75 -0
- data/vendor/simdjson/tests/jsoncheck.cpp +136 -0
- data/vendor/simdjson/tests/numberparsingcheck.cpp +224 -0
- data/vendor/simdjson/tests/pointercheck.cpp +38 -0
- data/vendor/simdjson/tests/singleheadertest.cpp +22 -0
- data/vendor/simdjson/tests/stringparsingcheck.cpp +408 -0
- data/vendor/simdjson/tools/CMakeLists.txt +3 -0
- data/vendor/simdjson/tools/cmake/FindCTargets.cmake +15 -0
- data/vendor/simdjson/tools/cmake/FindOptions.cmake +52 -0
- data/vendor/simdjson/tools/json2json.cpp +112 -0
- data/vendor/simdjson/tools/jsonpointer.cpp +93 -0
- data/vendor/simdjson/tools/jsonstats.cpp +143 -0
- data/vendor/simdjson/tools/minify.cpp +21 -0
- data/vendor/simdjson/tools/release.py +125 -0
- data/vendor/simdjson/windows/dirent_portable.h +1043 -0
- metadata +273 -0
@@ -0,0 +1,298 @@
|
|
1
|
+
#include "simdjson/jsonparser.h"
|
2
|
+
#ifndef _MSC_VER
|
3
|
+
#include "linux-perf-events.h"
|
4
|
+
#include <unistd.h>
|
5
|
+
#ifdef __linux__
|
6
|
+
#include <libgen.h>
|
7
|
+
#endif //__linux__
|
8
|
+
#endif // _MSC_VER
|
9
|
+
|
10
|
+
#include <memory>
|
11
|
+
|
12
|
+
#include "benchmark.h"
|
13
|
+
|
14
|
+
// #define RAPIDJSON_SSE2 // bad for performance
|
15
|
+
// #define RAPIDJSON_SSE42 // bad for performance
|
16
|
+
#include "rapidjson/document.h"
|
17
|
+
#include "rapidjson/reader.h"
|
18
|
+
#include "rapidjson/stringbuffer.h"
|
19
|
+
#include "rapidjson/writer.h"
|
20
|
+
|
21
|
+
#include "sajson.h"
|
22
|
+
|
23
|
+
#include <nlohmann/json.hpp>
|
24
|
+
using json = nlohmann::json;
|
25
|
+
|
26
|
+
#ifdef ALLPARSER
|
27
|
+
|
28
|
+
#include "fastjson.cpp"
|
29
|
+
#include "fastjson_dom.cpp"
|
30
|
+
#include "gason.cpp"
|
31
|
+
|
32
|
+
#include "json11.cpp"
|
33
|
+
extern "C" {
|
34
|
+
#include "cJSON.c"
|
35
|
+
#include "cJSON.h"
|
36
|
+
#include "jsmn.c"
|
37
|
+
#include "jsmn.h"
|
38
|
+
#include "ujdecode.h"
|
39
|
+
#include "ultrajsondec.c"
|
40
|
+
}
|
41
|
+
|
42
|
+
#include "jsoncpp.cpp"
|
43
|
+
#include "json/json.h"
|
44
|
+
|
45
|
+
#endif
|
46
|
+
|
47
|
+
using namespace rapidjson;
|
48
|
+
|
49
|
+
#ifdef ALLPARSER
|
50
|
+
// fastjson has a tricky interface
|
51
|
+
void on_json_error(void *, const fastjson::ErrorContext &ec) {
|
52
|
+
// std::cerr<<"ERROR: "<<ec.mesg<<std::endl;
|
53
|
+
}
|
54
|
+
bool fastjson_parse(const char *input) {
|
55
|
+
fastjson::Token token;
|
56
|
+
fastjson::dom::Chunk chunk;
|
57
|
+
return fastjson::dom::parse_string(input, &token, &chunk, 0, &on_json_error,
|
58
|
+
NULL);
|
59
|
+
}
|
60
|
+
// end of fastjson stuff
|
61
|
+
#endif
|
62
|
+
|
63
|
+
int main(int argc, char *argv[]) {
|
64
|
+
bool verbose = false;
|
65
|
+
bool just_data = false;
|
66
|
+
int c;
|
67
|
+
while ((c = getopt(argc, argv, "vt")) != -1)
|
68
|
+
switch (c) {
|
69
|
+
case 't':
|
70
|
+
just_data = true;
|
71
|
+
break;
|
72
|
+
case 'v':
|
73
|
+
verbose = true;
|
74
|
+
break;
|
75
|
+
default:
|
76
|
+
abort();
|
77
|
+
}
|
78
|
+
if (optind >= argc) {
|
79
|
+
std::cerr << "Usage: " << argv[0] << " <jsonfile>" << std::endl;
|
80
|
+
std::cerr << "Or " << argv[0] << " -v <jsonfile>" << std::endl;
|
81
|
+
std::cerr << "The '-t' flag outputs a table. " << std::endl;
|
82
|
+
exit(1);
|
83
|
+
}
|
84
|
+
const char *filename = argv[optind];
|
85
|
+
if (optind + 1 < argc) {
|
86
|
+
std::cerr << "warning: ignoring everything after " << argv[optind + 1]
|
87
|
+
<< std::endl;
|
88
|
+
}
|
89
|
+
simdjson::padded_string p;
|
90
|
+
try {
|
91
|
+
simdjson::get_corpus(filename).swap(p);
|
92
|
+
} catch (const std::exception &e) { // caught by reference to base
|
93
|
+
std::cout << "Could not load the file " << filename << std::endl;
|
94
|
+
return EXIT_FAILURE;
|
95
|
+
}
|
96
|
+
|
97
|
+
if (verbose) {
|
98
|
+
std::cout << "Input has ";
|
99
|
+
if (p.size() > 1024 * 1024)
|
100
|
+
std::cout << p.size() / (1024 * 1024) << " MB ";
|
101
|
+
else if (p.size() > 1024)
|
102
|
+
std::cout << p.size() / 1024 << " KB ";
|
103
|
+
else
|
104
|
+
std::cout << p.size() << " B ";
|
105
|
+
std::cout << std::endl;
|
106
|
+
}
|
107
|
+
simdjson::ParsedJson pj;
|
108
|
+
bool allocok = pj.allocate_capacity(p.size(), 1024);
|
109
|
+
|
110
|
+
if (!allocok) {
|
111
|
+
std::cerr << "can't allocate memory" << std::endl;
|
112
|
+
return EXIT_FAILURE;
|
113
|
+
}
|
114
|
+
int repeat = (p.size() < 1 * 1000 * 1000 ? 1000 : 10);
|
115
|
+
int volume = p.size();
|
116
|
+
if (just_data) {
|
117
|
+
printf("%-42s %20s %20s %20s %20s \n", "name", "cycles_per_byte",
|
118
|
+
"cycles_per_byte_err", "gb_per_s", "gb_per_s_err");
|
119
|
+
}
|
120
|
+
if (!just_data)
|
121
|
+
BEST_TIME("simdjson (dynamic mem) ", build_parsed_json(p).is_valid(), true,
|
122
|
+
, repeat, volume, !just_data);
|
123
|
+
// (static alloc)
|
124
|
+
BEST_TIME("simdjson ", json_parse(p, pj), simdjson::SUCCESS, , repeat, volume,
|
125
|
+
!just_data);
|
126
|
+
|
127
|
+
rapidjson::Document d;
|
128
|
+
|
129
|
+
char *buffer = (char *)malloc(p.size() + 1);
|
130
|
+
memcpy(buffer, p.data(), p.size());
|
131
|
+
buffer[p.size()] = '\0';
|
132
|
+
#ifndef ALLPARSER
|
133
|
+
if (!just_data)
|
134
|
+
#endif
|
135
|
+
BEST_TIME("RapidJSON ",
|
136
|
+
d.Parse<kParseValidateEncodingFlag>((const char *)buffer)
|
137
|
+
.HasParseError(),
|
138
|
+
false, memcpy(buffer, p.data(), p.size()), repeat, volume,
|
139
|
+
!just_data);
|
140
|
+
BEST_TIME("RapidJSON (insitu)",
|
141
|
+
d.ParseInsitu<kParseValidateEncodingFlag>(buffer).HasParseError(),
|
142
|
+
false,
|
143
|
+
memcpy(buffer, p.data(), p.size()) && (buffer[p.size()] = '\0'),
|
144
|
+
repeat, volume, !just_data);
|
145
|
+
#ifndef ALLPARSER
|
146
|
+
if (!just_data)
|
147
|
+
#endif
|
148
|
+
BEST_TIME("sajson (dynamic mem)",
|
149
|
+
sajson::parse(sajson::dynamic_allocation(),
|
150
|
+
sajson::mutable_string_view(p.size(), buffer))
|
151
|
+
.is_valid(),
|
152
|
+
true, memcpy(buffer, p.data(), p.size()), repeat, volume,
|
153
|
+
!just_data);
|
154
|
+
|
155
|
+
size_t ast_buffer_size = p.size();
|
156
|
+
size_t *ast_buffer = (size_t *)malloc(ast_buffer_size * sizeof(size_t));
|
157
|
+
// (static alloc, insitu)
|
158
|
+
BEST_TIME(
|
159
|
+
"sajson",
|
160
|
+
sajson::parse(sajson::bounded_allocation(ast_buffer, ast_buffer_size),
|
161
|
+
sajson::mutable_string_view(p.size(), buffer))
|
162
|
+
.is_valid(),
|
163
|
+
true, memcpy(buffer, p.data(), p.size()), repeat, volume, !just_data);
|
164
|
+
size_t expected = json::parse(p.data(), p.data() + p.size()).size();
|
165
|
+
BEST_TIME("nlohmann-json", json::parse(buffer, buffer + p.size()).size(),
|
166
|
+
expected, memcpy(buffer, p.data(), p.size()), repeat, volume,
|
167
|
+
!just_data);
|
168
|
+
|
169
|
+
#ifdef ALLPARSER
|
170
|
+
std::string json11err;
|
171
|
+
BEST_TIME("dropbox (json11) ",
|
172
|
+
((json11::Json::parse(buffer, json11err).is_null()) ||
|
173
|
+
(!json11err.empty())),
|
174
|
+
false, memcpy(buffer, p.data(), p.size()), repeat, volume,
|
175
|
+
!just_data);
|
176
|
+
|
177
|
+
BEST_TIME("fastjson ", fastjson_parse(buffer), true,
|
178
|
+
memcpy(buffer, p.data(), p.size()), repeat, volume, !just_data);
|
179
|
+
JsonValue value;
|
180
|
+
JsonAllocator allocator;
|
181
|
+
char *endptr;
|
182
|
+
BEST_TIME("gason ", jsonParse(buffer, &endptr, &value, allocator),
|
183
|
+
JSON_OK, memcpy(buffer, p.data(), p.size()), repeat, volume,
|
184
|
+
!just_data);
|
185
|
+
void *state;
|
186
|
+
BEST_TIME("ultrajson ",
|
187
|
+
(UJDecode(buffer, p.size(), NULL, &state) == NULL), false,
|
188
|
+
memcpy(buffer, p.data(), p.size()), repeat, volume, !just_data);
|
189
|
+
|
190
|
+
{
|
191
|
+
std::unique_ptr<jsmntok_t[]> tokens =
|
192
|
+
std::make_unique<jsmntok_t[]>(p.size());
|
193
|
+
jsmn_parser parser;
|
194
|
+
jsmn_init(&parser);
|
195
|
+
memcpy(buffer, p.data(), p.size());
|
196
|
+
buffer[p.size()] = '\0';
|
197
|
+
BEST_TIME(
|
198
|
+
"jsmn ",
|
199
|
+
(jsmn_parse(&parser, buffer, p.size(), tokens.get(), p.size()) > 0),
|
200
|
+
true, jsmn_init(&parser), repeat, volume, !just_data);
|
201
|
+
}
|
202
|
+
memcpy(buffer, p.data(), p.size());
|
203
|
+
buffer[p.size()] = '\0';
|
204
|
+
cJSON *tree = cJSON_Parse(buffer);
|
205
|
+
BEST_TIME("cJSON ", ((tree = cJSON_Parse(buffer)) != NULL), true,
|
206
|
+
cJSON_Delete(tree), repeat, volume, !just_data);
|
207
|
+
cJSON_Delete(tree);
|
208
|
+
|
209
|
+
Json::CharReaderBuilder b;
|
210
|
+
Json::CharReader *json_cpp_reader = b.newCharReader();
|
211
|
+
Json::Value root;
|
212
|
+
Json::String errs;
|
213
|
+
BEST_TIME("jsoncpp ",
|
214
|
+
json_cpp_reader->parse(buffer, buffer + volume, &root, &errs), true,
|
215
|
+
, repeat, volume, !just_data);
|
216
|
+
delete json_cpp_reader;
|
217
|
+
#endif
|
218
|
+
if (!just_data)
|
219
|
+
BEST_TIME("memcpy ",
|
220
|
+
(memcpy(buffer, p.data(), p.size()) == buffer), true, , repeat,
|
221
|
+
volume, !just_data);
|
222
|
+
#ifdef __linux__
|
223
|
+
if (!just_data) {
|
224
|
+
printf("\n \n <doing additional analysis with performance counters (Linux "
|
225
|
+
"only)>\n");
|
226
|
+
std::vector<int> evts;
|
227
|
+
evts.push_back(PERF_COUNT_HW_CPU_CYCLES);
|
228
|
+
evts.push_back(PERF_COUNT_HW_INSTRUCTIONS);
|
229
|
+
evts.push_back(PERF_COUNT_HW_BRANCH_MISSES);
|
230
|
+
evts.push_back(PERF_COUNT_HW_CACHE_REFERENCES);
|
231
|
+
evts.push_back(PERF_COUNT_HW_CACHE_MISSES);
|
232
|
+
LinuxEvents<PERF_TYPE_HARDWARE> unified(evts);
|
233
|
+
std::vector<unsigned long long> results;
|
234
|
+
std::vector<unsigned long long> stats;
|
235
|
+
results.resize(evts.size());
|
236
|
+
stats.resize(evts.size());
|
237
|
+
std::fill(stats.begin(), stats.end(), 0); // unnecessary
|
238
|
+
for (int i = 0; i < repeat; i++) {
|
239
|
+
unified.start();
|
240
|
+
if (json_parse(p, pj) != simdjson::SUCCESS)
|
241
|
+
printf("bug\n");
|
242
|
+
unified.end(results);
|
243
|
+
std::transform(stats.begin(), stats.end(), results.begin(), stats.begin(),
|
244
|
+
std::plus<unsigned long long>());
|
245
|
+
}
|
246
|
+
printf("simdjson : cycles %10.0f instructions %10.0f branchmisses %10.0f "
|
247
|
+
"cacheref %10.0f cachemisses %10.0f bytespercachemiss %10.0f "
|
248
|
+
"inspercycle %10.1f insperbyte %10.1f\n",
|
249
|
+
stats[0] * 1.0 / repeat, stats[1] * 1.0 / repeat,
|
250
|
+
stats[2] * 1.0 / repeat, stats[3] * 1.0 / repeat,
|
251
|
+
stats[4] * 1.0 / repeat, volume * repeat * 1.0 / stats[2],
|
252
|
+
stats[1] * 1.0 / stats[0], stats[1] * 1.0 / (volume * repeat));
|
253
|
+
|
254
|
+
std::fill(stats.begin(), stats.end(), 0);
|
255
|
+
for (int i = 0; i < repeat; i++) {
|
256
|
+
memcpy(buffer, p.data(), p.size());
|
257
|
+
buffer[p.size()] = '\0';
|
258
|
+
unified.start();
|
259
|
+
if (d.ParseInsitu<kParseValidateEncodingFlag>(buffer).HasParseError() !=
|
260
|
+
false)
|
261
|
+
printf("bug\n");
|
262
|
+
unified.end(results);
|
263
|
+
std::transform(stats.begin(), stats.end(), results.begin(), stats.begin(),
|
264
|
+
std::plus<unsigned long long>());
|
265
|
+
}
|
266
|
+
printf("RapidJSON: cycles %10.0f instructions %10.0f branchmisses %10.0f "
|
267
|
+
"cacheref %10.0f cachemisses %10.0f bytespercachemiss %10.0f "
|
268
|
+
"inspercycle %10.1f insperbyte %10.1f\n",
|
269
|
+
stats[0] * 1.0 / repeat, stats[1] * 1.0 / repeat,
|
270
|
+
stats[2] * 1.0 / repeat, stats[3] * 1.0 / repeat,
|
271
|
+
stats[4] * 1.0 / repeat, volume * repeat * 1.0 / stats[2],
|
272
|
+
stats[1] * 1.0 / stats[0], stats[1] * 1.0 / (volume * repeat));
|
273
|
+
|
274
|
+
std::fill(stats.begin(), stats.end(), 0); // unnecessary
|
275
|
+
for (int i = 0; i < repeat; i++) {
|
276
|
+
memcpy(buffer, p.data(), p.size());
|
277
|
+
unified.start();
|
278
|
+
if (sajson::parse(sajson::bounded_allocation(ast_buffer, ast_buffer_size),
|
279
|
+
sajson::mutable_string_view(p.size(), buffer))
|
280
|
+
.is_valid() != true)
|
281
|
+
printf("bug\n");
|
282
|
+
unified.end(results);
|
283
|
+
std::transform(stats.begin(), stats.end(), results.begin(), stats.begin(),
|
284
|
+
std::plus<unsigned long long>());
|
285
|
+
}
|
286
|
+
printf("sajson : cycles %10.0f instructions %10.0f branchmisses %10.0f "
|
287
|
+
"cacheref %10.0f cachemisses %10.0f bytespercachemiss %10.0f "
|
288
|
+
"inspercycle %10.1f insperbyte %10.1f\n",
|
289
|
+
stats[0] * 1.0 / repeat, stats[1] * 1.0 / repeat,
|
290
|
+
stats[2] * 1.0 / repeat, stats[3] * 1.0 / repeat,
|
291
|
+
stats[4] * 1.0 / repeat, volume * repeat * 1.0 / stats[2],
|
292
|
+
stats[1] * 1.0 / stats[0], stats[1] * 1.0 / (volume * repeat));
|
293
|
+
}
|
294
|
+
#endif // __linux__
|
295
|
+
|
296
|
+
free(ast_buffer);
|
297
|
+
free(buffer);
|
298
|
+
}
|
@@ -0,0 +1,208 @@
|
|
1
|
+
#include <iostream>
|
2
|
+
#ifndef _MSC_VER
|
3
|
+
#include <unistd.h>
|
4
|
+
#endif
|
5
|
+
#include "simdjson/jsonioutil.h"
|
6
|
+
#include "simdjson/jsonparser.h"
|
7
|
+
#ifdef __linux__
|
8
|
+
#include "linux-perf-events.h"
|
9
|
+
#endif
|
10
|
+
|
11
|
+
size_t count_nonasciibytes(const uint8_t *input, size_t length) {
|
12
|
+
size_t count = 0;
|
13
|
+
for (size_t i = 0; i < length; i++) {
|
14
|
+
count += input[i] >> 7;
|
15
|
+
}
|
16
|
+
return count;
|
17
|
+
}
|
18
|
+
|
19
|
+
size_t count_backslash(const uint8_t *input, size_t length) {
|
20
|
+
size_t count = 0;
|
21
|
+
for (size_t i = 0; i < length; i++) {
|
22
|
+
count += (input[i] == '\\') ? 1 : 0;
|
23
|
+
}
|
24
|
+
return count;
|
25
|
+
}
|
26
|
+
|
27
|
+
struct stat_s {
|
28
|
+
size_t integer_count;
|
29
|
+
size_t float_count;
|
30
|
+
size_t string_count;
|
31
|
+
size_t backslash_count;
|
32
|
+
size_t non_ascii_byte_count;
|
33
|
+
size_t object_count;
|
34
|
+
size_t array_count;
|
35
|
+
size_t null_count;
|
36
|
+
size_t true_count;
|
37
|
+
size_t false_count;
|
38
|
+
size_t byte_count;
|
39
|
+
size_t structural_indexes_count;
|
40
|
+
bool valid;
|
41
|
+
};
|
42
|
+
|
43
|
+
using stat_t = struct stat_s;
|
44
|
+
|
45
|
+
stat_t simdjson_compute_stats(const simdjson::padded_string &p) {
|
46
|
+
stat_t answer;
|
47
|
+
simdjson::ParsedJson pj = simdjson::build_parsed_json(p);
|
48
|
+
answer.valid = pj.is_valid();
|
49
|
+
if (!answer.valid) {
|
50
|
+
return answer;
|
51
|
+
}
|
52
|
+
answer.backslash_count =
|
53
|
+
count_backslash(reinterpret_cast<const uint8_t *>(p.data()), p.size());
|
54
|
+
answer.non_ascii_byte_count = count_nonasciibytes(
|
55
|
+
reinterpret_cast<const uint8_t *>(p.data()), p.size());
|
56
|
+
answer.byte_count = p.size();
|
57
|
+
answer.integer_count = 0;
|
58
|
+
answer.float_count = 0;
|
59
|
+
answer.object_count = 0;
|
60
|
+
answer.array_count = 0;
|
61
|
+
answer.null_count = 0;
|
62
|
+
answer.true_count = 0;
|
63
|
+
answer.false_count = 0;
|
64
|
+
answer.string_count = 0;
|
65
|
+
answer.structural_indexes_count = pj.n_structural_indexes;
|
66
|
+
size_t tape_idx = 0;
|
67
|
+
uint64_t tape_val = pj.tape[tape_idx++];
|
68
|
+
uint8_t type = (tape_val >> 56);
|
69
|
+
size_t how_many = 0;
|
70
|
+
assert(type == 'r');
|
71
|
+
how_many = tape_val & JSON_VALUE_MASK;
|
72
|
+
for (; tape_idx < how_many; tape_idx++) {
|
73
|
+
tape_val = pj.tape[tape_idx];
|
74
|
+
// uint64_t payload = tape_val & JSON_VALUE_MASK;
|
75
|
+
type = (tape_val >> 56);
|
76
|
+
switch (type) {
|
77
|
+
case 'l': // we have a long int
|
78
|
+
answer.integer_count++;
|
79
|
+
tape_idx++; // skipping the integer
|
80
|
+
break;
|
81
|
+
case 'd': // we have a double
|
82
|
+
answer.float_count++;
|
83
|
+
tape_idx++; // skipping the double
|
84
|
+
break;
|
85
|
+
case 'n': // we have a null
|
86
|
+
answer.null_count++;
|
87
|
+
break;
|
88
|
+
case 't': // we have a true
|
89
|
+
answer.true_count++;
|
90
|
+
break;
|
91
|
+
case 'f': // we have a false
|
92
|
+
answer.false_count++;
|
93
|
+
break;
|
94
|
+
case '{': // we have an object
|
95
|
+
answer.object_count++;
|
96
|
+
break;
|
97
|
+
case '}': // we end an object
|
98
|
+
break;
|
99
|
+
case '[': // we start an array
|
100
|
+
answer.array_count++;
|
101
|
+
break;
|
102
|
+
case ']': // we end an array
|
103
|
+
break;
|
104
|
+
case '"': // we have a string
|
105
|
+
answer.string_count++;
|
106
|
+
break;
|
107
|
+
default:
|
108
|
+
break; // ignore
|
109
|
+
}
|
110
|
+
}
|
111
|
+
return answer;
|
112
|
+
}
|
113
|
+
|
114
|
+
int main(int argc, char *argv[]) {
|
115
|
+
#ifndef _MSC_VER
|
116
|
+
int c;
|
117
|
+
while ((c = getopt(argc, argv, "")) != -1) {
|
118
|
+
switch (c) {
|
119
|
+
|
120
|
+
default:
|
121
|
+
abort();
|
122
|
+
}
|
123
|
+
}
|
124
|
+
#else
|
125
|
+
int optind = 1;
|
126
|
+
#endif
|
127
|
+
if (optind >= argc) {
|
128
|
+
std::cerr << "Reads json, prints stats. " << std::endl;
|
129
|
+
std::cerr << "Usage: " << argv[0] << " <jsonfile>" << std::endl;
|
130
|
+
|
131
|
+
exit(1);
|
132
|
+
}
|
133
|
+
const char *filename = argv[optind];
|
134
|
+
if (optind + 1 < argc) {
|
135
|
+
std::cerr << "warning: ignoring everything after " << argv[optind + 1]
|
136
|
+
<< std::endl;
|
137
|
+
}
|
138
|
+
simdjson::padded_string p;
|
139
|
+
try {
|
140
|
+
simdjson::get_corpus(filename).swap(p);
|
141
|
+
} catch (const std::exception &) { // caught by reference to base
|
142
|
+
std::cerr << "Could not load the file " << filename << std::endl;
|
143
|
+
return EXIT_FAILURE;
|
144
|
+
}
|
145
|
+
stat_t s = simdjson_compute_stats(p);
|
146
|
+
if (!s.valid) {
|
147
|
+
std::cerr << "not a valid JSON" << std::endl;
|
148
|
+
return EXIT_FAILURE;
|
149
|
+
}
|
150
|
+
|
151
|
+
printf("# integer_count float_count string_count backslash_count "
|
152
|
+
"non_ascii_byte_count object_count array_count null_count true_count "
|
153
|
+
"false_count byte_count structural_indexes_count ");
|
154
|
+
#ifdef __linux__
|
155
|
+
printf(" stage1_cycle_count stage1_instruction_count stage2_cycle_count "
|
156
|
+
" stage2_instruction_count stage3_cycle_count "
|
157
|
+
"stage3_instruction_count ");
|
158
|
+
#else
|
159
|
+
printf("(you are not under linux, so perf counters are disaabled)");
|
160
|
+
#endif
|
161
|
+
printf("\n");
|
162
|
+
printf("%zu %zu %zu %zu %zu %zu %zu %zu %zu %zu %zu %zu ", s.integer_count,
|
163
|
+
s.float_count, s.string_count, s.backslash_count,
|
164
|
+
s.non_ascii_byte_count, s.object_count, s.array_count, s.null_count,
|
165
|
+
s.true_count, s.false_count, s.byte_count, s.structural_indexes_count);
|
166
|
+
#ifdef __linux__
|
167
|
+
simdjson::ParsedJson pj;
|
168
|
+
bool allocok = pj.allocate_capacity(p.size());
|
169
|
+
if (!allocok) {
|
170
|
+
std::cerr << "failed to allocate memory" << std::endl;
|
171
|
+
return EXIT_FAILURE;
|
172
|
+
}
|
173
|
+
const uint32_t iterations = p.size() < 1 * 1000 * 1000 ? 1000 : 50;
|
174
|
+
std::vector<int> evts;
|
175
|
+
evts.push_back(PERF_COUNT_HW_CPU_CYCLES);
|
176
|
+
evts.push_back(PERF_COUNT_HW_INSTRUCTIONS);
|
177
|
+
LinuxEvents<PERF_TYPE_HARDWARE> unified(evts);
|
178
|
+
unsigned long cy1 = 0, cy2 = 0;
|
179
|
+
unsigned long cl1 = 0, cl2 = 0;
|
180
|
+
std::vector<unsigned long long> results;
|
181
|
+
results.resize(evts.size());
|
182
|
+
for (uint32_t i = 0; i < iterations; i++) {
|
183
|
+
unified.start();
|
184
|
+
// The default template is simdjson::Architecture::NATIVE.
|
185
|
+
bool isok = (simdjson::find_structural_bits<>(p.data(), p.size(), pj) ==
|
186
|
+
simdjson::SUCCESS);
|
187
|
+
unified.end(results);
|
188
|
+
|
189
|
+
cy1 += results[0];
|
190
|
+
cl1 += results[1];
|
191
|
+
|
192
|
+
unified.start();
|
193
|
+
isok =
|
194
|
+
isok && (simdjson::SUCCESS == unified_machine(p.data(), p.size(), pj));
|
195
|
+
unified.end(results);
|
196
|
+
|
197
|
+
cy2 += results[0];
|
198
|
+
cl2 += results[1];
|
199
|
+
if (!isok) {
|
200
|
+
std::cerr << "failure?" << std::endl;
|
201
|
+
}
|
202
|
+
}
|
203
|
+
printf("%f %f %f %f ", cy1 * 1.0 / iterations, cl1 * 1.0 / iterations,
|
204
|
+
cy2 * 1.0 / iterations, cl2 * 1.0 / iterations);
|
205
|
+
#endif // __linux__
|
206
|
+
printf("\n");
|
207
|
+
return EXIT_SUCCESS;
|
208
|
+
}
|