Add Yyjson benchmarks

benchmark/CMakeLists.txt

@@ -47,10 +47,12 @@ if (TARGET benchmark::benchmark)
   add_subdirectory(largerandom)
   add_executable(bench_parse_call bench_parse_call.cpp)
   add_executable(bench_dom_api bench_dom_api.cpp)
+  if (SIMDJSON_EXCEPTIONS)
     add_executable(bench_ondemand bench_ondemand.cpp)
     if (TARGET yyjson)
       target_link_libraries(bench_ondemand PRIVATE yyjson)
-  endif ()
+    endif (TARGET yyjson)
+  endif (SIMDJSON_EXCEPTIONS)
 endif()
 
 include(checkperf.cmake)

benchmark/bench_ondemand.cpp

@@ -30,6 +30,10 @@ SIMDJSON_POP_DISABLE_WARNINGS
 // // yyjson
 #ifdef SIMDJSON_COMPETITION_YYJSON
 #include "partial_tweets/yyjson.h"
+#include "largerandom/yyjson.h"
+#include "kostya/yyjson.h"
+#include "distinctuserid/yyjson.h"
+#include "find_tweet/yyjson.h"
 #endif
 
 

benchmark/distinctuserid/yyjson.h

@@ -0,0 +1,49 @@
+#pragma once
+
+#include "distinctuserid.h"
+
+namespace distinct_user_id {
+
+class Yyjson {
+public:
+  simdjson_really_inline const std::vector<int64_t> &Result() { return ids; }
+  simdjson_really_inline size_t ItemCount() { return ids.size(); }
+
+private:
+  std::vector<int64_t> ids{};
+
+public:
+  simdjson_really_inline bool Run(const padded_string &json) {
+    ids.clear();
+
+    // Walk the document, parsing the tweets as we go
+    yyjson_doc *doc = yyjson_read(json.data(), json.size(), 0);
+    if (!doc) { return false; }
+    yyjson_val *root = yyjson_doc_get_root(doc);
+    yyjson_val *statuses = yyjson_obj_get(root, "statuses");
+
+    size_t tweet_idx, tweets_max;
+    yyjson_val *tweet;
+    yyjson_arr_foreach(statuses, tweet_idx, tweets_max, tweet) {
+      auto user = yyjson_obj_get(tweet, "user");
+      auto id = yyjson_obj_get(user, "id");
+      ids.push_back(yyjson_get_sint(id));
+
+      // Not all tweets have a "retweeted_status", but when they do
+      // we want to go and find the user within.
+      auto retweet = yyjson_obj_get(tweet, "retweeted_status");
+      if (retweet) {
+        user = yyjson_obj_get(retweet, "user");
+        id = yyjson_obj_get(user, "id");
+        ids.push_back(yyjson_get_sint(id));
+      }
+    }
+    remove_duplicates(ids);
+    return true;
+  }
+
+};
+
+BENCHMARK_TEMPLATE(DistinctUserID, Yyjson);
+
+} // namespace partial_tweets

benchmark/find_tweet/yyjson.h

@@ -0,0 +1,41 @@
+#pragma once
+
+#include "find_tweet.h"
+
+namespace find_tweet {
+
+class Yyjson {
+public:
+  simdjson_really_inline std::string_view Result() { return text; }
+  simdjson_really_inline size_t ItemCount() { return 1; }
+
+private:
+  std::string_view text{};
+
+public:
+  simdjson_really_inline bool Run(const padded_string &json) {
+    text = "";
+
+    // Walk the document, parsing the tweets as we go
+    yyjson_doc *doc = yyjson_read(json.data(), json.size(), 0);
+    if (!doc) { return false; }
+    yyjson_val *root = yyjson_doc_get_root(doc);
+    yyjson_val *statuses = yyjson_obj_get(root, "statuses");
+
+    size_t tweet_idx, tweets_max;
+    yyjson_val *tweet;
+    yyjson_arr_foreach(statuses, tweet_idx, tweets_max, tweet) {
+      auto id = yyjson_obj_get(tweet, "id");
+      if (yyjson_get_uint(id) == TWEET_ID) {
+        auto _text = yyjson_obj_get(tweet, "text");
+        text = yyjson_get_str(_text);
+        return true;
+      }
+    }
+    return false;
+  }
+};
+
+BENCHMARK_TEMPLATE(FindTweet, Yyjson);
+
+} // namespace find_tweet

benchmark/kostya/yyjson.h

@@ -0,0 +1,42 @@
+#pragma once
+
+#include "kostya.h"
+
+namespace kostya {
+
+class Yyjson {
+public:
+  simdjson_really_inline const std::vector<my_point> &Result() { return container; }
+  simdjson_really_inline size_t ItemCount() { return container.size(); }
+
+private:
+  ondemand::parser parser{};
+  std::vector<my_point> container{};
+
+  simdjson_really_inline double get_double(yyjson_val *obj, std::string_view key) {
+    yyjson_val *val = yyjson_obj_getn(obj, key.data(), key.length());
+    return yyjson_get_real(val);
+  }
+
+public:
+  simdjson_really_inline bool Run(const padded_string &json) {
+    container.clear();
+
+    // Walk the document, parsing the tweets as we go
+    yyjson_doc *doc = yyjson_read(json.data(), json.size(), 0);
+    if (!doc) { return false; }
+    yyjson_val *root = yyjson_doc_get_root(doc);
+    yyjson_val *coords = yyjson_obj_get(root, "coordinates");
+
+    size_t idx, max;
+    yyjson_val *coord;
+    yyjson_arr_foreach(coords, idx, max, coord) {
+      container.emplace_back(my_point{get_double(coord, "x"), get_double(coord, "y"), get_double(coord, "z")});
+    }
+    return true;
+  }
+};
+
+BENCHMARK_TEMPLATE(Kostya, Yyjson);
+
+} // namespace kostya

benchmark/largerandom/largerandom.h

@@ -1,7 +1,5 @@
 #pragma once
 
-#if SIMDJSON_EXCEPTIONS
-
 //
 // Interface
 //
@@ -72,5 +70,3 @@ template<typename T> static void LargeRandom(benchmark::State &state) {
 }
 
 } // namespace largerandom
-
-#endif // SIMDJSON_EXCEPTIONS

benchmark/largerandom/yyjson.h

@@ -0,0 +1,41 @@
+#pragma once
+
+#include "largerandom.h"
+
+namespace largerandom {
+
+class Yyjson {
+public:
+  simdjson_really_inline const std::vector<my_point> &Result() { return container; }
+  simdjson_really_inline size_t ItemCount() { return container.size(); }
+
+private:
+  ondemand::parser parser{};
+  std::vector<my_point> container{};
+
+  simdjson_really_inline double get_double(yyjson_val *obj, std::string_view key) {
+    yyjson_val *val = yyjson_obj_getn(obj, key.data(), key.length());
+    return yyjson_get_real(val);
+  }
+
+public:
+  simdjson_really_inline bool Run(const padded_string &json) {
+    container.clear();
+
+    // Walk the document, parsing the tweets as we go
+    yyjson_doc *doc = yyjson_read(json.data(), json.size(), 0);
+    if (!doc) { return false; }
+    yyjson_val *coords = yyjson_doc_get_root(doc);
+
+    size_t idx, max;
+    yyjson_val *coord;
+    yyjson_arr_foreach(coords, idx, max, coord) {
+      container.emplace_back(my_point{get_double(coord, "x"), get_double(coord, "y"), get_double(coord, "z")});
+    }
+    return true;
+  }
+};
+
+BENCHMARK_TEMPLATE(LargeRandom, Yyjson);
+
+} // namespace kostya