weaseldb/benchmarks/bench_commit_request.cpp

#include "commit_request.hpp"
#include "json_commit_request_parser.hpp"
#include "test_data.hpp"

#include <nanobench.h>
#include <string>

using namespace weaseldb::test_data;

// JSON test data is now provided by test_data.hpp

int main() {
  // One-shot parsing benchmarks
  auto bench = ankerl::nanobench::Bench()
                   .title("CommitRequest One-Shot Parsing")
                   .unit("parse")
                   .warmup(100);

  // Simple JSON parsing
  bench.run("Simple JSON (3 fields)", [&] {
    CommitRequest request;
    JsonCommitRequestParser parser;
    std::string mutable_json = SIMPLE_JSON;
    bool result =
        parser.parse(request, mutable_json.data(), mutable_json.size());
    ankerl::nanobench::doNotOptimizeAway(result);
    ankerl::nanobench::doNotOptimizeAway(request.leader_id());
  });

  // Medium complexity JSON parsing
  bench.run("Medium JSON (2 preconditions, 2 operations)", [&] {
    CommitRequest request;
    JsonCommitRequestParser parser;
    std::string mutable_json = MEDIUM_JSON;
    bool result =
        parser.parse(request, mutable_json.data(), mutable_json.size());
    ankerl::nanobench::doNotOptimizeAway(result);
    ankerl::nanobench::doNotOptimizeAway(request.leader_id());
  });

  // Complex JSON parsing
  bench.run("Complex JSON (3 preconditions, 5 operations)", [&] {
    CommitRequest request;
    JsonCommitRequestParser parser;
    std::string mutable_json = COMPLEX_JSON;
    bool result =
        parser.parse(request, mutable_json.data(), mutable_json.size());
    ankerl::nanobench::doNotOptimizeAway(result);
    ankerl::nanobench::doNotOptimizeAway(request.leader_id());
  });

  // Large batch operations
  for (int num_ops : {10, 50, 100, 500}) {
    std::string large_json = generate_large_json(num_ops);
    bench.run("Large JSON (" + std::to_string(num_ops) + " operations)", [&] {
      CommitRequest request;
      JsonCommitRequestParser parser;
      std::string mutable_json = large_json;
      bool result =
          parser.parse(request, mutable_json.data(), mutable_json.size());
      ankerl::nanobench::doNotOptimizeAway(result);
      ankerl::nanobench::doNotOptimizeAway(request.leader_id());
    });
  }

  // Streaming parsing benchmarks
  auto streaming_bench = ankerl::nanobench::Bench()
                             .title("CommitRequest Streaming Parsing")
                             .unit("parse")
                             .warmup(50);

  // Streaming with different chunk sizes
  for (int chunk_size : {1, 8, 32, 128, 512}) {
    streaming_bench.run(
        "Streaming Medium JSON (chunk size " + std::to_string(chunk_size) + ")",
        [&] {
          CommitRequest request;
          JsonCommitRequestParser parser;
          std::string mutable_json = MEDIUM_JSON;

          parser.begin_streaming_parse(request);

          size_t offset = 0;
          CommitRequestParser::ParseStatus status =
              CommitRequestParser::ParseStatus::Incomplete;

          while (offset < mutable_json.size() &&
                 status == CommitRequestParser::ParseStatus::Incomplete) {
            size_t len = std::min(static_cast<size_t>(chunk_size),
                                  mutable_json.size() - offset);
            status =
                parser.parse_chunk(request, mutable_json.data() + offset, len);
            offset += len;
          }

          if (status == CommitRequestParser::ParseStatus::Incomplete) {
            status = parser.finish_streaming_parse(request);
          }

          ankerl::nanobench::doNotOptimizeAway(status);
          ankerl::nanobench::doNotOptimizeAway(request.leader_id());
        });
  }

  // Reset and reuse benchmarks
  auto reuse_bench = ankerl::nanobench::Bench()
                         .title("CommitRequest Reset and Reuse")
                         .unit("operation")
                         .warmup(50);

  reuse_bench.run("Parse -> Reset -> Parse cycle", [&] {
    static CommitRequest request; // Static to persist across invocations
    static JsonCommitRequestParser parser;

    std::string mutable_json1 = SIMPLE_JSON;
    bool result1 =
        parser.parse(request, mutable_json1.data(), mutable_json1.size());

    request.reset();

    std::string mutable_json2 = MEDIUM_JSON;
    bool result2 =
        parser.parse(request, mutable_json2.data(), mutable_json2.size());

    ankerl::nanobench::doNotOptimizeAway(result1);
    ankerl::nanobench::doNotOptimizeAway(result2);
    ankerl::nanobench::doNotOptimizeAway(request.leader_id());
  });

  // Base64 decoding performance
  auto base64_bench = ankerl::nanobench::Bench()
                          .title("Base64 Decoding Performance")
                          .unit("decode")
                          .warmup(50);

  // JSON with lots of base64 encoded data
  std::string base64_heavy_json = R"({
        "leader_id": "base64-test-leader",
        "read_version": 12345,
        "operations": [)";

  for (int i = 0; i < 20; ++i) {
    if (i > 0)
      base64_heavy_json += ",";
    base64_heavy_json += R"(
            {
                "type": "write",
                "key": "VGhpc0lzQUxvbmdCYXNlNjRFbmNvZGVkS2V5V2l0aExvdHNPZkRhdGFGb3JUZXN0aW5nUHVycG9zZXM=",
                "value": "VGhpc0lzQW5FdmVuTG9uZ2VyQmFzZTY0RW5jb2RlZFZhbHVlV2l0aEV2ZW5Nb3JlRGF0YUZvclRlc3RpbmdUaGVCYXNlNjREZWNvZGluZ1BlcmZvcm1hbmNlT2ZUaGVQYXJzZXI="
            })";
  }
  base64_heavy_json += R"(
        ]
    })";

  base64_bench.run(
      "Heavy Base64 JSON (20 operations with long encoded data)", [&] {
        CommitRequest request;
        JsonCommitRequestParser parser;
        std::string mutable_json = base64_heavy_json;
        bool result =
            parser.parse(request, mutable_json.data(), mutable_json.size());
        ankerl::nanobench::doNotOptimizeAway(result);
        ankerl::nanobench::doNotOptimizeAway(request.leader_id());
      });

  return 0;
}