perf: optimize vocabulary lookup with bitset

syedazeez337 · syedazeez337 · commit b9d603ab87e4 · 2025-11-11T12:13:12.000+05:30
Signed-off-by: Azeez Syed &lt;syedazeez337@gmail.com&gt;
diff --git a/benchmark/e2e/runner.cc b/benchmark/e2e/runner.cc
@@ -12,8 +12,9 @@
 #include <sourcemeta/blaze/compiler.h>
 #include <sourcemeta/blaze/evaluator.h>
 
+// clang-format off
 #define REGISTER_E2E_COMPILER(name, directory_name)                            \
-  static auto E2E_Compiler_##name(benchmark::State &state)->void {             \
+  static auto E2E_Compiler_##name(benchmark::State &state) -> void {           \
     const std::filesystem::path directory{CURRENT_DIRECTORY                    \
                                           "/e2e/" directory_name};             \
     const auto schema{sourcemeta::core::read_json(directory / "schema.json")}; \
@@ -30,7 +31,7 @@
   BENCHMARK(E2E_Compiler_##name)
 
 #define REGISTER_E2E_EVALUATOR(name, directory_name)                           \
-  static auto E2E_Evaluator_##name(benchmark::State &state)->void {            \
+  static auto E2E_Evaluator_##name(benchmark::State &state) -> void {          \
     const std::filesystem::path directory{CURRENT_DIRECTORY                    \
                                           "/e2e/" directory_name};             \
     const auto schema{sourcemeta::core::read_json(directory / "schema.json")}; \
@@ -56,6 +57,7 @@
     }                                                                          \
   }                                                                            \
   BENCHMARK(E2E_Evaluator_##name)
+// clang-format on
 
 REGISTER_E2E_COMPILER(adaptivecard, "adaptivecard");
 REGISTER_E2E_COMPILER(ansible_meta, "ansible-meta");
diff --git a/src/compiler/compile.cc b/src/compiler/compile.cc
@@ -10,6 +10,7 @@
 #include <utility>   // std::move, std::pair
 
 #include "compile_helpers.h"
+#include "vocabulary_lookup.h"
 
 namespace {
 
@@ -33,6 +34,12 @@ auto compile_subschema(const sourcemeta::blaze::Context &context,
   }
 
   Instructions steps;
+  // Reserve capacity to avoid reallocations - typical schemas have 3-10
+  // keywords
+  if (schema_context.schema.is_object()) {
+    steps.reserve(schema_context.schema.size());
+  }
+
   for (const auto &entry : sourcemeta::core::SchemaKeywordIterator{
            schema_context.schema, context.walker, context.resolver,
            default_dialect}) {
@@ -285,11 +292,14 @@ auto compile(const sourcemeta::core::JSON &schema,
   ///////////////////////////////////////////////////////////////////
 
   Instructions compiler_template;
+  // Reserve initial capacity for the template
+  compiler_template.reserve(32);
+
   if (uses_dynamic_scopes &&
-      (schema_context.vocabularies.contains(
-           "https://json-schema.org/draft/2019-09/vocab/core") ||
-       schema_context.vocabularies.contains(
-           "https://json-schema.org/draft/2020-12/vocab/core"))) {
+      (has_vocabulary(schema_context.vocabularies,
+                      "https://json-schema.org/draft/2019-09/vocab/core") ||
+       has_vocabulary(schema_context.vocabularies,
+                      "https://json-schema.org/draft/2020-12/vocab/core"))) {
     for (const auto &entry : context.frame.locations()) {
       // We are only trying to find dynamic anchors
       if (entry.second.type !=
diff --git a/src/compiler/compile_helpers.h b/src/compiler/compile_helpers.h
@@ -11,22 +11,30 @@
 #include <utility>   // std::declval, std::move
 #include <variant>   // std::visit
 
+#include "vocabulary_lookup.h"
+
 namespace sourcemeta::blaze {
 
+// Static empty string for DynamicContext with no keyword
+inline const std::string &empty_keyword() {
+  static const std::string value{""};
+  return value;
+}
+
 inline auto relative_dynamic_context() -> DynamicContext {
-  return {"", sourcemeta::core::empty_pointer, sourcemeta::core::empty_pointer,
-          false};
+  return {empty_keyword(), sourcemeta::core::empty_pointer,
+          sourcemeta::core::empty_pointer, false};
 }
 
 inline auto relative_dynamic_context(const DynamicContext &dynamic_context)
     -> DynamicContext {
-  return {"", sourcemeta::core::empty_pointer, sourcemeta::core::empty_pointer,
-          dynamic_context.property_as_target};
+  return {empty_keyword(), sourcemeta::core::empty_pointer,
+          sourcemeta::core::empty_pointer, dynamic_context.property_as_target};
 }
 
 inline auto property_relative_dynamic_context() -> DynamicContext {
-  return {"", sourcemeta::core::empty_pointer, sourcemeta::core::empty_pointer,
-          true};
+  return {empty_keyword(), sourcemeta::core::empty_pointer,
+          sourcemeta::core::empty_pointer, true};
 }
 
 inline auto schema_resource_id(const std::vector<std::string> &resources,
@@ -222,7 +230,7 @@ inline auto find_adjacent(const Context &context,
 
     if (std::any_of(vocabularies.cbegin(), vocabularies.cend(),
                     [&subschema_vocabularies](const auto &vocabulary) {
-                      return subschema_vocabularies.contains(vocabulary);
+                      return has_vocabulary(subschema_vocabularies, vocabulary);
                     }) &&
         subschema.type() == type) {
       result.emplace_back(subschema);
diff --git a/src/compiler/default_compiler.cc b/src/compiler/default_compiler.cc
@@ -5,6 +5,7 @@
 #include "default_compiler_draft4.h"
 #include "default_compiler_draft6.h"
 #include "default_compiler_draft7.h"
+#include "vocabulary_lookup.h"
 
 #include <cassert>       // assert
 #include <string>        // std::string
@@ -39,7 +40,8 @@ auto sourcemeta::blaze::default_schema_compiler(
       "http://json-schema.org/draft-06/hyper-schema#",
       "http://json-schema.org/draft-04/schema#",
       "http://json-schema.org/draft-04/hyper-schema#"};
-  for (const auto &vocabulary : schema_context.vocabularies) {
+  for (const auto &vocabulary :
+       schema_context.vocabularies.all_vocabularies()) {
     if (!SUPPORTED_VOCABULARIES.contains(vocabulary.first) &&
         vocabulary.second) {
       throw sourcemeta::core::SchemaVocabularyError(
@@ -50,22 +52,22 @@ auto sourcemeta::blaze::default_schema_compiler(
   using namespace sourcemeta::blaze;
 
 #define COMPILE(vocabulary, _keyword, handler)                                 \
-  if (schema_context.vocabularies.contains(vocabulary) &&                      \
+  if (has_vocabulary(schema_context.vocabularies, vocabulary) &&               \
       dynamic_context.keyword == (_keyword)) {                                 \
     return internal::handler(context, schema_context, dynamic_context,         \
                              current);                                         \
   }
 
 #define COMPILE_ANY(vocabulary_1, vocabulary_2, _keyword, handler)             \
-  if ((schema_context.vocabularies.contains(vocabulary_1) ||                   \
-       schema_context.vocabularies.contains(vocabulary_2)) &&                  \
+  if ((has_vocabulary(schema_context.vocabularies, vocabulary_1) ||            \
+       has_vocabulary(schema_context.vocabularies, vocabulary_2)) &&           \
       dynamic_context.keyword == (_keyword)) {                                 \
     return internal::handler(context, schema_context, dynamic_context,         \
                              current);                                         \
   }
 
 #define STOP_IF_SIBLING_KEYWORD(vocabulary, _keyword)                          \
-  if (schema_context.vocabularies.contains(vocabulary) &&                      \
+  if (has_vocabulary(schema_context.vocabularies, vocabulary) &&               \
       schema_context.schema.is_object() &&                                     \
       schema_context.schema.defines(_keyword)) {                               \
     return {};                                                                 \
diff --git a/src/compiler/include/sourcemeta/blaze/compiler.h b/src/compiler/include/sourcemeta/blaze/compiler.h
@@ -51,7 +51,7 @@ struct SchemaContext {
 /// disposal to implement a keyword
 struct DynamicContext {
   /// The schema keyword
-  const std::string keyword;
+  const std::string &keyword;
   /// The schema base keyword path
   const sourcemeta::core::Pointer &base_schema_location;
   /// The base instance location that the keyword must be evaluated to
diff --git a/src/compiler/unevaluated.cc b/src/compiler/unevaluated.cc
@@ -1,7 +1,10 @@
 #include <sourcemeta/blaze/compiler.h>
 
+#include "vocabulary_lookup.h"
+
 namespace {
 using namespace sourcemeta::core;
+using sourcemeta::blaze::has_vocabulary;
 
 auto find_adjacent_dependencies(
     const JSON::String &current, const JSON &schema, const SchemaFrame &frame,
@@ -22,7 +25,8 @@ auto find_adjacent_dependencies(
       continue;
     } else if (keywords.contains(property.first)) {
       // In 2019-09, `additionalItems` takes no effect without `items`
-      if (subschema_vocabularies.contains(
+      if (has_vocabulary(
+              subschema_vocabularies,
               "https://json-schema.org/draft/2019-09/vocab/applicator") &&
           property.first == "additionalItems" && !subschema.defines("items")) {
         continue;
@@ -154,9 +158,11 @@ auto unevaluated(const JSON &schema, const SchemaFrame &frame,
       const auto keyword_uri{frame.uri(entry.second, {pair.first})};
       SchemaUnevaluatedEntry unevaluated;
 
-      if ((subschema_vocabularies.contains(
+      if ((has_vocabulary(
+               subschema_vocabularies,
                "https://json-schema.org/draft/2020-12/vocab/unevaluated") &&
-           subschema_vocabularies.contains(
+           has_vocabulary(
+               subschema_vocabularies,
                "https://json-schema.org/draft/2020-12/vocab/applicator")) &&
           // NOLINTNEXTLINE(bugprone-branch-clone)
           pair.first == "unevaluatedProperties") {
@@ -167,29 +173,31 @@ auto unevaluated(const JSON &schema, const SchemaFrame &frame,
             entry.second, entry.second, true, unevaluated);
         result.emplace(keyword_uri, std::move(unevaluated));
       } else if (
-          (subschema_vocabularies.contains(
+          (has_vocabulary(
+               subschema_vocabularies,
                "https://json-schema.org/draft/2020-12/vocab/unevaluated") &&
-           subschema_vocabularies.contains(
+           has_vocabulary(
+               subschema_vocabularies,
                "https://json-schema.org/draft/2020-12/vocab/applicator")) &&
           pair.first == "unevaluatedItems") {
         find_adjacent_dependencies(
             pair.first, schema, frame, walker, resolver,
             {"prefixItems", "items", "contains", "unevaluatedItems"},
             entry.second, entry.second, true, unevaluated);
         result.emplace(keyword_uri, std::move(unevaluated));
-      } else if (subschema_vocabularies.contains(
-                     "https://json-schema.org/draft/2019-09/vocab/"
-                     "applicator") &&
+      } else if (has_vocabulary(subschema_vocabularies,
+                                "https://json-schema.org/draft/2019-09/vocab/"
+                                "applicator") &&
                  pair.first == "unevaluatedProperties") {
         find_adjacent_dependencies(
             pair.first, schema, frame, walker, resolver,
             {"properties", "patternProperties", "additionalProperties",
              "unevaluatedProperties"},
             entry.second, entry.second, true, unevaluated);
         result.emplace(keyword_uri, std::move(unevaluated));
-      } else if (subschema_vocabularies.contains(
-                     "https://json-schema.org/draft/2019-09/vocab/"
-                     "applicator") &&
+      } else if (has_vocabulary(subschema_vocabularies,
+                                "https://json-schema.org/draft/2019-09/vocab/"
+                                "applicator") &&
                  pair.first == "unevaluatedItems") {
         find_adjacent_dependencies(
             pair.first, schema, frame, walker, resolver,
diff --git a/src/compiler/vocabulary_lookup.h b/src/compiler/vocabulary_lookup.h
@@ -0,0 +1,20 @@
+#ifndef SOURCEMETA_BLAZE_COMPILER_VOCABULARY_LOOKUP_H_
+#define SOURCEMETA_BLAZE_COMPILER_VOCABULARY_LOOKUP_H_
+
+#include <sourcemeta/core/jsonschema.h>
+
+#include <string_view> // std::string_view
+
+namespace sourcemeta::blaze {
+
+// Optimized vocabulary lookup using bitwise operations
+// This directly uses the Vocabularies::contains() method which performs
+// O(1) bitwise AND operations for known vocabularies
+inline auto has_vocabulary(const sourcemeta::core::Vocabularies &vocabularies,
+                           std::string_view uri) -> bool {
+  return vocabularies.contains(uri);
+}
+
+} // namespace sourcemeta::blaze
+
+#endif
diff --git a/src/evaluator/evaluator.cc b/src/evaluator/evaluator.cc
@@ -65,16 +65,18 @@ auto Evaluator::validate(const Template &schema,
   assert(this->instance_location.empty());
   assert(this->resources.empty());
   this->labels.clear();
+  // Reserve capacity to avoid rehashing during evaluation
+  this->labels.reserve(16);
 
-  if (schema.track && schema.dynamic) {
+  if (schema.track && schema.dynamic) [[unlikely]] {
     this->evaluated_.clear();
     return complete::evaluate(instance, *this, schema, nullptr);
-  } else if (schema.track) {
+  } else if (schema.track) [[unlikely]] {
     this->evaluated_.clear();
     return track::evaluate(instance, *this, schema);
-  } else if (schema.dynamic) {
+  } else if (schema.dynamic) [[unlikely]] {
     return dynamic::evaluate(instance, *this, schema);
-  } else {
+  } else [[likely]] {
     return fast::evaluate(instance, *this, schema);
   }
 }
@@ -87,6 +89,7 @@ auto Evaluator::validate(const Template &schema,
   assert(this->instance_location.empty());
   assert(this->resources.empty());
   this->labels.clear();
+  this->labels.reserve(16);
   this->evaluated_.clear();
 
   return complete::evaluate(instance, *this, schema, callback);
diff --git a/vendor/core/src/core/jsonschema/include/sourcemeta/core/jsonschema_types.h b/vendor/core/src/core/jsonschema/include/sourcemeta/core/jsonschema_types.h
diff --git a/vendor/core/src/core/jsonschema/jsonschema.cc b/vendor/core/src/core/jsonschema/jsonschema.cc
diff --git a/vendor/core/src/extension/alterschema/alterschema.cc b/vendor/core/src/extension/alterschema/alterschema.cc
