schemurai 2.2.0 → 2.2.2

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: 01b0c89354b90232337525b7fe75c063053fecd2965036bb7ff600d2b4e823cb
-  data.tar.gz: f45549ac8cf41eae0c2b28220ca93fe87042cec2aa7c7926382a00f8f65c105d
+  metadata.gz: 163dbc9593611cce4f28183766bd5e17a08387a54883c17a502a198eb010ad36
+  data.tar.gz: aa3d5f90db227194179023b31043cf36b4597e4af5f1e2209b2846e9c36f9044
 SHA512:
-  metadata.gz: 618d494971d7a1745075dae9b9e6155562f9c0f0fb058e923cd457f26bb062be2858db9f2ffaf711d035ee83490179e005869106f3ae29ef0c8011512bed2af9
-  data.tar.gz: 6ecd49a7320b6b936067808a2a0bd07d6e497418427d0e1612d624dd703a71c6eae99f9ae27ab0a94c2946c3a6495fdc2c11bf665b7d59a9fad628f285af0f5e
+  metadata.gz: 5f5ba215dde91e1bf364ff6ebbbebb7ea40b284e3908cfcb578d48aad2174c34e5cf6d9fd121947bb54f8b2cf0fad626e629ed94d6b4a52ab41ba84ad7dfbc55
+  data.tar.gz: 4efa97250aad095e414eb230310ae580bbd029b834aa11322cab8da0b19df7a1fb6703240ee0439fd4abe788fcfa47c918ac92806bef174d38d0110bb35cc934

data/lib/schemurai/evaluator.rb

@@ -4,10 +4,13 @@ require "json"
 require "base64"
 require_relative "evaluation"
 require_relative "error_message"
+require_relative "unique_items"
 
 module Schemurai
   module Internal
     class Evaluator
+      include UniqueItems
+
       MISSING_SEGMENT = Object.new.freeze
 
       def backend = :ruby
@@ -226,11 +229,7 @@ module Schemurai
         length = value.length
         return false if schema.key?("maxItems") && length > schema["maxItems"]
         return false if schema.key?("minItems") && length < schema["minItems"]
-        if schema["uniqueItems"]
-          value.each_with_index do |item, index|
-            return false if value[0...index].any? { |previous| json_equal?(previous, item) }
-          end
-        end
+        return false if schema["uniqueItems"] && !unique_items?(value)
 
         prefix_items = schema["prefixItems"]
         if prefix_items.is_a?(Array)
@@ -555,12 +554,7 @@ module Schemurai
         limit(schema, "maxItems", value.length) { |actual, expected| actual <= expected }
         limit(schema, "minItems", value.length) { |actual, expected| actual >= expected }
 
-        if schema["uniqueItems"]
-          duplicate = value.each_with_index.any? do |item, index|
-            value[0...index].any? { |previous| json_equal?(previous, item) }
-          end
-          add_error("uniqueItems") { ErrorMessage.unique_items } if duplicate
-        end
+        add_error("uniqueItems") { ErrorMessage.unique_items } if schema["uniqueItems"] && !unique_items?(value)
 
         prefix_items = schema["prefixItems"]
         if prefix_items.is_a?(Array)

data/lib/schemurai/unique_items.rb

@@ -0,0 +1,66 @@
+# frozen_string_literal: true
+
+module Schemurai
+  module Internal
+    module UniqueItems
+      private def unique_items?(values)
+        # Avoid building an index for small arrays and early duplicates.
+        if values.length <= 16
+          values.each_with_index do |item, index|
+            index.times { |previous| return false if json_equal?(values[previous], item) }
+          end
+          return true
+        end
+
+        buckets = {}
+        # Store a single index until a collision needs a bucket array.
+        index = 0
+        while index < values.length
+          item = values[index]
+          fingerprint = json_fingerprint(item)
+          if (bucket = buckets[fingerprint])
+            if bucket.is_a?(Integer)
+              return false if json_equal?(values[bucket], item)
+
+              buckets[fingerprint] = [bucket, index]
+            else
+              return false if bucket.any? { |previous| json_equal?(values[previous], item) }
+
+              bucket << index
+            end
+          else
+            buckets[fingerprint] = index
+          end
+          index += 1
+        end
+        true
+      end
+
+      private def json_fingerprint(value)
+        case value
+        when Numeric
+          # JSON numbers compare across Integer/Float. Converting integral
+          # floats to integers also preserves precision for large integers.
+          return value.hash if value.is_a?(Integer)
+
+          number = value.to_f
+          (number.finite? && number == number.to_i) ? number.to_i.hash : number.hash
+        when Array
+          # Mask before multiplication to keep intermediate values immediate
+          # integers; hash each step to retain order without temporary arrays.
+          fingerprint = value.class.hash
+          value.each { |item| fingerprint = (((fingerprint & 0x1fffffff) * 31) ^ json_fingerprint(item)).hash }
+          fingerprint
+        when Hash
+          # Object key order is immaterial. Hash collisions are resolved using
+          # json_equal?, so this fingerprint never decides equality itself.
+          entries = 0
+          value.each_pair { |key, item| entries ^= (((key.hash & 0x1fffffff) * 31) ^ json_fingerprint(item)).hash }
+          (value.class.hash ^ value.length ^ entries).hash
+        else
+          value.hash
+        end
+      end
+    end
+  end
+end

data/lib/schemurai/version.rb

@@ -2,5 +2,5 @@
 
 module Schemurai
   # The installed Schemurai version.
-  VERSION = "2.2.0"
+  VERSION = "2.2.2"
 end

data/lib/schemurai/vm/evaluator.rb

@@ -4,6 +4,7 @@ require "base64"
 require "json"
 require_relative "../evaluation"
 require_relative "../error_message"
+require_relative "../unique_items"
 require_relative "compiler"
 
 module Schemurai
@@ -49,6 +50,8 @@ module Schemurai
     private_constant :EvaluationBuffer
 
     class Evaluator
+      include Internal::UniqueItems
+
       MISSING_SEGMENT = Object.new.freeze
       DECIMAL_CACHE_LIMIT = 16
 
@@ -552,17 +555,7 @@ module Schemurai
         length = value.length
         return false if rules.max_items && length > rules.max_items
         return false if rules.min_items && length < rules.min_items
-        if rules.unique
-          index = 1
-          while index < length
-            previous_index = 0
-            while previous_index < index
-              return false if json_equal?(value[previous_index], value[index])
-              previous_index += 1
-            end
-            index += 1
-          end
-        end
+        return false if rules.unique && !unique_items?(value)
 
         if (prefix_items = rules.prefix_items)
           prefix_items.each_with_index do |child, index|
@@ -616,25 +609,10 @@ module Schemurai
 
           add_error("minItems") { Internal::ErrorMessage.size("minItems", rules.min_items, value.length) }
         end
-        if rules.unique
-          duplicate = false
-          index = 1
-          while index < value.length && !duplicate
-            previous_index = 0
-            while previous_index < index
-              if json_equal?(value[previous_index], value[index])
-                duplicate = true
-                break
-              end
-              previous_index += 1
-            end
-            index += 1
-          end
-          if duplicate
-            return Evaluation.invalid unless @errors
+        if rules.unique && !unique_items?(value)
+          return Evaluation.invalid unless @errors
 
-            add_error("uniqueItems") { Internal::ErrorMessage.unique_items }
-          end
+          add_error("uniqueItems") { Internal::ErrorMessage.unique_items }
         end
 
         if (prefix_items = rules.prefix_items)

data/sig/schemurai.rbs

@@ -1,13 +1,6 @@
 module Schemurai
   # A JSON-shaped Ruby value accepted as a schema value or validation instance.
-  type json_value =
-    nil
-    | bool
-    | Integer
-    | Float
-    | String
-    | Array[json_value]
-    | Hash[String, json_value]
+  type json_value = nil | bool | Integer | Float | String | Array[json_value] | Hash[String, json_value]
 
   # A JSON Schema root.
   type schema = bool | Hash[String, json_value]
@@ -50,12 +43,7 @@ module Schemurai
     def self.[]: (String keyword, String instance_path, String schema_path, String message) -> ValidationError
                | (keyword: String, instance_path: String, schema_path: String, message: String) -> ValidationError
 
-    def to_h: () -> {
-      keyword: String,
-      instance_path: String,
-      schema_path: String,
-      message: String
-    }
+    def to_h: () -> { keyword: String, instance_path: String, schema_path: String, message: String }
   end
 
   class Result
@@ -68,7 +56,9 @@ module Schemurai
 
   interface _Evaluator
     def backend: () -> backend
+
     def validate: (json_value instance) -> Result
+
     def valid?: (json_value instance) -> bool
   end
 
@@ -82,76 +72,29 @@ module Schemurai
   class SchemaRegistry
     attr_reader backend: backend
 
-    def initialize: (
-      ?schemas: schemas,
-      ?validate_schema: bool,
-      ?backend: backend_selection
-    ) -> void
+    def initialize: (?schemas: schemas, ?validate_schema: bool, ?backend: backend_selection) -> void
 
     def validate_schema?: () -> bool
 
-    def compile: (
-      schema schema,
-      ?base_uri: String?,
-      ?content: bool,
-      ?format: bool
-    ) -> Validator
+    def compile: (schema schema, ?base_uri: String?, ?content: bool, ?format: bool) -> Validator
 
     def validate_schema: (schema schema) -> Result
     def valid_schema?: (schema schema) -> bool
     def make_shareable: () -> self
     def shareable?: () -> bool
 
-    def validator_for: (
-      String uri,
-      ?content: bool,
-      ?format: bool
-    ) -> Validator
+    def validator_for: (String uri, ?content: bool, ?format: bool) -> Validator
   end
 
   def self.backend: () -> backend
 
-  def self.compile: (
-    schema schema,
-    ?schemas: schemas,
-    ?base_uri: String?,
-    ?content: bool,
-    ?format: bool,
-    ?validate_schema: bool,
-    ?backend: backend_selection
-  ) -> Validator
-
-  def self.validate_schema: (
-    schema schema,
-    ?schemas: schemas,
-    ?backend: backend_selection
-  ) -> Result
-
-  def self.valid_schema?: (
-    schema schema,
-    ?schemas: schemas,
-    ?backend: backend_selection
-  ) -> bool
-
-  def self.validate: (
-    schema schema,
-    json_value instance,
-    ?schemas: schemas,
-    ?base_uri: String?,
-    ?content: bool,
-    ?format: bool,
-    ?validate_schema: bool,
-    ?backend: backend_selection
-  ) -> Result
-
-  def self.valid?: (
-    schema schema,
-    json_value instance,
-    ?schemas: schemas,
-    ?base_uri: String?,
-    ?content: bool,
-    ?format: bool,
-    ?validate_schema: bool,
-    ?backend: backend_selection
-  ) -> bool
+  def self.compile: (schema schema, ?schemas: schemas, ?base_uri: String?, ?content: bool, ?format: bool, ?validate_schema: bool, ?backend: backend_selection) -> Validator
+
+  def self.validate_schema: (schema schema, ?schemas: schemas, ?backend: backend_selection) -> Result
+
+  def self.valid_schema?: (schema schema, ?schemas: schemas, ?backend: backend_selection) -> bool
+
+  def self.validate: (schema schema, json_value instance, ?schemas: schemas, ?base_uri: String?, ?content: bool, ?format: bool, ?validate_schema: bool, ?backend: backend_selection) -> Result
+
+  def self.valid?: (schema schema, json_value instance, ?schemas: schemas, ?base_uri: String?, ?content: bool, ?format: bool, ?validate_schema: bool, ?backend: backend_selection) -> bool
 end

metadata

@@ -1,7 +1,7 @@
 --- !ruby/object:Gem::Specification
 name: schemurai
 version: !ruby/object:Gem::Version
-  version: 2.2.0
+  version: 2.2.2
 platform: ruby
 authors:
 - Kuya KOHARA (oakcask)
@@ -53,6 +53,7 @@ files:
 - lib/schemurai/schema_domain.rb
 - lib/schemurai/schema_graph.rb
 - lib/schemurai/schema_node.rb
+- lib/schemurai/unique_items.rb
 - lib/schemurai/version.rb
 - lib/schemurai/vm/compiler.rb
 - lib/schemurai/vm/evaluator.rb
@@ -63,7 +64,7 @@ licenses:
 metadata:
   homepage_uri: https://github.com/oakcask/schemurai
   source_code_uri: https://github.com/oakcask/schemurai/tree/main
-  documentation_uri: https://rubydoc.info/gems/schemurai/2.2.0
+  documentation_uri: https://rubydoc.info/gems/schemurai/2.2.2
   rubygems_mfa_required: 'true'
 rdoc_options:
 - "--main"
@@ -87,7 +88,7 @@ required_rubygems_version: !ruby/object:Gem::Requirement
     - !ruby/object:Gem::Version
       version: '0'
 requirements: []
-rubygems_version: 4.0.16
+rubygems_version: 4.0.20
 specification_version: 4
 summary: A fast, Ractor-safe, pure Ruby JSON Schema Draft 7+ validator
 test_files: []