[WIP] JSON-LD performance improvements

Signed-off-by: Juan Cruz Viotti <jv@jviotti.com>

Author: jviotti

benchmark/micro/2020_12.cc

@@ -1,6 +1,8 @@
 #include <benchmark/benchmark.h>
 
 #include <cassert>    // assert
+#include <cstddef>    // std::size_t
+#include <cstdint>    // std::int64_t
 #include <filesystem> // std::filesystem::path
 
 #include <sourcemeta/blaze/foundation.h>
@@ -214,6 +216,84 @@ static void Micro_2020_12_Simple_Output_Annotations(benchmark::State &state) {
   }
 }
 
+static void
+Micro_2020_12_Simple_Output_Annotation_Dropping(benchmark::State &state) {
+  const sourcemeta::core::JSON schema{sourcemeta::core::parse_json(R"JSON({
+    "$schema": "https://json-schema.org/draft/2020-12/schema",
+    "type": "array",
+    "items": {
+      "oneOf": [
+        {
+          "type": "string",
+          "title": "String branch",
+          "description": "This branch matches string values"
+        },
+        {
+          "type": "integer",
+          "title": "Integer branch",
+          "description": "This branch matches integer values"
+        },
+        {
+          "type": "boolean",
+          "title": "Boolean branch",
+          "description": "This branch matches boolean values"
+        },
+        {
+          "type": "object",
+          "title": "Object branch",
+          "description": "This branch matches object values"
+        },
+        {
+          "type": "array",
+          "title": "Array branch",
+          "description": "This branch matches array values"
+        }
+      ]
+    }
+  })JSON")};
+
+  // Every element matches exactly one branch, so the array validates, but the
+  // four losing branches each collect a title and a description annotation that
+  // are then dropped. Exhaustive mode forces every branch to run, so this
+  // stresses the annotation collection and retraction path heavily
+  auto instance{sourcemeta::core::JSON::make_array()};
+  for (std::size_t index = 0; index < 512; index += 1) {
+    switch (index % 5) {
+      case 0:
+        instance.push_back(
+            sourcemeta::core::JSON{sourcemeta::core::JSON::String{"text"}});
+        break;
+      case 1:
+        instance.push_back(
+            sourcemeta::core::JSON{static_cast<std::int64_t>(index)});
+        break;
+      case 2:
+        instance.push_back(sourcemeta::core::JSON{index % 2 == 0});
+        break;
+      case 3:
+        instance.push_back(sourcemeta::core::JSON::make_object());
+        break;
+      default:
+        instance.push_back(sourcemeta::core::JSON::make_array());
+        break;
+    }
+  }
+
+  const auto schema_template{
+      sourcemeta::blaze::compile(schema, sourcemeta::blaze::schema_walker,
+                                 sourcemeta::blaze::schema_resolver,
+                                 sourcemeta::blaze::default_schema_compiler,
+                                 sourcemeta::blaze::Mode::Exhaustive)};
+  sourcemeta::blaze::Evaluator evaluator;
+  for (auto _ : state) {
+    sourcemeta::blaze::SimpleOutput output{instance};
+    auto result{
+        evaluator.validate(schema_template, instance, std::ref(output))};
+    assert(result);
+    benchmark::DoNotOptimize(result);
+  }
+}
+
 static void
 Micro_2020_12_Compile_NonCircular_Shared_Refs(benchmark::State &state) {
   const sourcemeta::core::JSON schema{sourcemeta::core::parse_json(R"JSON({
@@ -397,6 +477,7 @@ BENCHMARK(Micro_2020_12_Dynamic_Ref);
 BENCHMARK(Micro_2020_12_Dynamic_Ref_Single);
 BENCHMARK(Micro_2020_12_Simple_Output_Mask);
 BENCHMARK(Micro_2020_12_Simple_Output_Annotations);
+BENCHMARK(Micro_2020_12_Simple_Output_Annotation_Dropping);
 BENCHMARK(Micro_2020_12_Compile_NonCircular_Shared_Refs);
 BENCHMARK(Micro_2020_12_Exhaustive_Deep_Numeric);
 BENCHMARK(Micro_2020_12_Exhaustive_Deep_Numeric_SimpleOutput);

src/output/include/sourcemeta/blaze/output_simple.h

@@ -16,7 +16,6 @@
 #include <map>     // std::map
 #include <ostream> // std::ostream
 #include <string>  // std::string
-#include <tuple>   // std::tie
 #include <utility> // std::pair
 #include <vector>  // std::vector
 
@@ -83,6 +82,14 @@ class SOURCEMETA_BLAZE_OUTPUT_EXPORT SimpleOutput {
     std::reference_wrapper<const std::string> schema_location;
   };
 
+  /// A single collected annotation, in evaluation order
+  struct AnnotationEntry {
+    sourcemeta::core::WeakPointer instance_location;
+    sourcemeta::core::WeakPointer evaluate_path;
+    std::reference_wrapper<const std::string> schema_location;
+    sourcemeta::core::JSON value;
+  };
+
   auto operator()(const EvaluationType type, const bool result,
                   const Instruction &step,
                   const InstructionExtra &step_metadata,
@@ -97,12 +104,6 @@ class SOURCEMETA_BLAZE_OUTPUT_EXPORT SimpleOutput {
   [[nodiscard]] auto cbegin() const -> const_iterator;
   [[nodiscard]] auto cend() const -> const_iterator;
 
-  /// Access annotations that were collected during evaluation, indexed by
-  /// instance location and evaluation path
-  [[nodiscard]] auto annotations() const -> const auto & {
-    return this->annotations_;
-  }
-
   /// Move out the collected error entries, leaving this output empty. Useful to
   /// take ownership of the trace without copying when the output is no longer
   /// needed
@@ -114,22 +115,12 @@ class SOURCEMETA_BLAZE_OUTPUT_EXPORT SimpleOutput {
     return result;
   }
 
-  // NOLINTNEXTLINE(bugprone-exception-escape)
-  struct Location {
-    auto operator<(const Location &other) const noexcept -> bool {
-      // Perform a lexicographical comparison
-      return std::tie(this->instance_location, this->evaluate_path,
-                      this->schema_location.get()) <
-             std::tie(other.instance_location, other.evaluate_path,
-                      other.schema_location.get());
-    }
-
-    // NOLINTBEGIN(cppcoreguidelines-avoid-const-or-ref-data-members)
-    const sourcemeta::core::WeakPointer instance_location;
-    const sourcemeta::core::WeakPointer evaluate_path;
-    const std::reference_wrapper<const std::string> schema_location;
-    // NOLINTEND(cppcoreguidelines-avoid-const-or-ref-data-members)
-  };
+  /// Access the annotations collected during evaluation, as a flat log in
+  /// evaluation order. Consumers that need them grouped by location index this
+  /// log themselves
+  [[nodiscard]] auto annotations() const -> const auto & {
+    return this->annotations_;
+  }
 
 private:
 // Exporting symbols that depends on the standard C++ library is considered
@@ -148,7 +139,7 @@ class SOURCEMETA_BLAZE_OUTPUT_EXPORT SimpleOutput {
       std::pair<sourcemeta::core::WeakPointer, sourcemeta::core::WeakPointer>,
       std::vector<Entry>>
       masked_traces;
-  std::map<Location, std::vector<sourcemeta::core::JSON>> annotations_;
+  std::vector<AnnotationEntry> annotations_;
 #if defined(_MSC_VER)
 #pragma warning(default : 4251)
 #endif