json_schemer 1.0.3 → 2.1.1

data/lib/json_schemer/format.rb

@@ -1,126 +1,148 @@
 # frozen_string_literal: true
 module JSONSchemer
   module Format
-    include Email
-    include Hostname
-    include URITemplate
+    # https://datatracker.ietf.org/doc/html/draft-bhutton-json-schema-validation-01#section-7.3
+    DATE_TIME = proc do |instance, _format|
+      !instance.is_a?(String) || valid_date_time?(instance)
+    end
+    DATE = proc do |instance, _format|
+      !instance.is_a?(String) || valid_date_time?("#{instance}T04:05:06.123456789+07:00")
+    end
+    TIME = proc do |instance, _format|
+      !instance.is_a?(String) || valid_date_time?("2001-02-03T#{instance}")
+    end
+    DURATION = proc do |instance, _format|
+      !instance.is_a?(String) || valid_duration?(instance)
+    end
+    # https://datatracker.ietf.org/doc/html/draft-bhutton-json-schema-validation-01#section-7.3.2
+    EMAIL = proc do |instance, _format|
+      !instance.is_a?(String) || instance.ascii_only? && valid_email?(instance)
+    end
+    IDN_EMAIL = proc do |instance, _format|
+      !instance.is_a?(String) || valid_email?(instance)
+    end
+    # https://datatracker.ietf.org/doc/html/draft-bhutton-json-schema-validation-01#section-7.3.3
+    HOSTNAME = proc do |instance, _format|
+      !instance.is_a?(String) || instance.ascii_only? && valid_hostname?(instance)
+    end
+    IDN_HOSTNAME = proc do |instance, _format|
+      !instance.is_a?(String) || valid_hostname?(instance)
+    end
+    # https://datatracker.ietf.org/doc/html/draft-bhutton-json-schema-validation-01#section-7.3.4
+    IPV4 = proc do |instance, _format|
+      !instance.is_a?(String) || valid_ip?(instance, Socket::AF_INET)
+    end
+    IPV6 = proc do |instance, _format|
+      !instance.is_a?(String) || valid_ip?(instance, Socket::AF_INET6)
+    end
+    # https://datatracker.ietf.org/doc/html/draft-bhutton-json-schema-validation-01#section-7.3.5
+    URI = proc do |instance, _format|
+      !instance.is_a?(String) || valid_uri?(instance)
+    end
+    URI_REFERENCE = proc do |instance, _format|
+      !instance.is_a?(String) || valid_uri_reference?(instance)
+    end
+    IRI = proc do |instance, _format|
+      !instance.is_a?(String) || valid_uri?(iri_escape(instance))
+    end
+    IRI_REFERENCE = proc do |instance, _format|
+      !instance.is_a?(String) || valid_uri_reference?(iri_escape(instance))
+    end
+    UUID = proc do |instance, _format|
+      !instance.is_a?(String) || valid_uuid?(instance)
+    end
+    # https://datatracker.ietf.org/doc/html/draft-bhutton-json-schema-validation-01#section-7.3.6
+    URI_TEMPLATE = proc do |instance, _format|
+      !instance.is_a?(String) || valid_uri_template?(instance)
+    end
+    # https://datatracker.ietf.org/doc/html/draft-bhutton-json-schema-validation-01#section-7.3.7
+    JSON_POINTER = proc do |instance, _format|
+      !instance.is_a?(String) || valid_json_pointer?(instance)
+    end
+    RELATIVE_JSON_POINTER = proc do |instance, _format|
+      !instance.is_a?(String) || valid_relative_json_pointer?(instance)
+    end
+    # https://datatracker.ietf.org/doc/html/draft-bhutton-json-schema-validation-01#section-7.3.8
+    REGEX = proc do |instance, _format|
+      !instance.is_a?(String) || valid_regex?(instance)
+    end
 
-    JSON_POINTER_REGEX_STRING = '(\/([^~\/]|~[01])*)*'
-    JSON_POINTER_REGEX = /\A#{JSON_POINTER_REGEX_STRING}\z/.freeze
-    RELATIVE_JSON_POINTER_REGEX = /\A(0|[1-9]\d*)(#|#{JSON_POINTER_REGEX_STRING})?\z/.freeze
     DATE_TIME_OFFSET_REGEX = /(Z|[\+\-]([01][0-9]|2[0-3]):[0-5][0-9])\z/i.freeze
-    HOUR_24_REGEX = /T24/.freeze
-    LEAP_SECOND_REGEX = /T\d{2}:\d{2}:6/.freeze
+    DATE_TIME_SEPARATOR_CHARACTER_CLASS = '[Tt\s]'
+    HOUR_24_REGEX = /#{DATE_TIME_SEPARATOR_CHARACTER_CLASS}24:/.freeze
+    LEAP_SECOND_REGEX = /#{DATE_TIME_SEPARATOR_CHARACTER_CLASS}\d{2}:\d{2}:6/.freeze
     IP_REGEX = /\A[\h:.]+\z/.freeze
     INVALID_QUERY_REGEX = /\s/.freeze
+    IRI_ESCAPE_REGEX = /[^[:ascii:]]/
+    UUID_REGEX = /\A\h{8}-\h{4}-\h{4}-[89AB]\h{3}-\h{12}\z/i
+    NIL_UUID = '00000000-0000-0000-0000-000000000000'
+    ASCII_8BIT_TO_PERCENT_ENCODED = 256.times.each_with_object({}) do |byte, out|
+      out[-byte.chr] = -sprintf('%%%02X', byte)
+    end.freeze
 
-    def valid_spec_format?(data, format)
-      case format
-      when 'date-time'
-        valid_date_time?(data)
-      when 'date'
-        valid_date_time?("#{data}T04:05:06.123456789+07:00")
-      when 'time'
-        valid_date_time?("2001-02-03T#{data}")
-      when 'email'
-        data.ascii_only? && valid_email?(data)
-      when 'idn-email'
-        valid_email?(data)
-      when 'hostname'
-        data.ascii_only? && valid_hostname?(data)
-      when 'idn-hostname'
-        valid_hostname?(data)
-      when 'ipv4'
-        valid_ip?(data, Socket::AF_INET)
-      when 'ipv6'
-        valid_ip?(data, Socket::AF_INET6)
-      when 'uri'
-        valid_uri?(data)
-      when 'uri-reference'
-        valid_uri_reference?(data)
-      when 'iri'
-        valid_uri?(iri_escape(data))
-      when 'iri-reference'
-        valid_uri_reference?(iri_escape(data))
-      when 'uri-template'
-        valid_uri_template?(data)
-      when 'json-pointer'
-        valid_json_pointer?(data)
-      when 'relative-json-pointer'
-        valid_relative_json_pointer?(data)
-      when 'regex'
-        valid_regex?(data)
-      else
-        raise UnknownFormat, format
-      end
-    end
-
-    def valid_json?(data)
-      JSON.parse(data)
-      true
-    rescue JSON::ParserError
-      false
-    end
+    class << self
+      include Duration
+      include Email
+      include Hostname
+      include JSONPointer
+      include URITemplate
 
-    def valid_date_time?(data)
-      return false if HOUR_24_REGEX.match?(data)
-      datetime = DateTime.rfc3339(data)
-      return false if LEAP_SECOND_REGEX.match?(data) && datetime.new_offset.strftime('%H:%M') != '23:59'
-      DATE_TIME_OFFSET_REGEX.match?(data)
-    rescue ArgumentError
-      false
-    end
+      def percent_encode(data, regexp)
+        data = data.dup
+        data.force_encoding(Encoding::ASCII_8BIT)
+        data.gsub!(regexp, ASCII_8BIT_TO_PERCENT_ENCODED)
+        data.force_encoding(Encoding::US_ASCII)
+      end
 
-    def valid_ip?(data, family)
-      IPAddr.new(data, family)
-      IP_REGEX.match?(data)
-    rescue IPAddr::Error
-      false
-    end
+      def valid_date_time?(data)
+        return false if HOUR_24_REGEX.match?(data)
+        datetime = DateTime.rfc3339(data)
+        return false if LEAP_SECOND_REGEX.match?(data) && datetime.new_offset.strftime('%H:%M') != '23:59'
+        DATE_TIME_OFFSET_REGEX.match?(data)
+      rescue ArgumentError
+        false
+      end
 
-    def parse_uri_scheme(data)
-      scheme, _userinfo, _host, _port, _registry, _path, opaque, query, _fragment = URI::RFC3986_PARSER.split(data)
-      # URI::RFC3986_PARSER.parse allows spaces in these and I don't think it should
-      raise URI::InvalidURIError if INVALID_QUERY_REGEX.match?(query) || INVALID_QUERY_REGEX.match?(opaque)
-      scheme
-    end
+      def valid_ip?(data, family)
+        IPAddr.new(data, family)
+        IP_REGEX.match?(data)
+      rescue IPAddr::Error
+        false
+      end
 
-    def valid_uri?(data)
-      !!parse_uri_scheme(data)
-    rescue URI::InvalidURIError
-      false
-    end
+      def parse_uri_scheme(data)
+        scheme, _userinfo, _host, _port, _registry, _path, opaque, query, _fragment = ::URI::RFC3986_PARSER.split(data)
+        # ::URI::RFC3986_PARSER.parse allows spaces in these and I don't think it should
+        raise ::URI::InvalidURIError if INVALID_QUERY_REGEX.match?(query) || INVALID_QUERY_REGEX.match?(opaque)
+        scheme
+      end
 
-    def valid_uri_reference?(data)
-      parse_uri_scheme(data)
-      true
-    rescue URI::InvalidURIError
-      false
-    end
+      def valid_uri?(data)
+        !!parse_uri_scheme(data)
+      rescue ::URI::InvalidURIError
+        false
+      end
 
-    def iri_escape(data)
-      data.gsub(/[^[:ascii:]]/) do |match|
-        us = match
-        tmp = +''
-        us.each_byte do |uc|
-          tmp << sprintf('%%%02X', uc)
-        end
-        tmp
-      end.force_encoding(Encoding::US_ASCII)
-    end
+      def valid_uri_reference?(data)
+        parse_uri_scheme(data)
+        true
+      rescue ::URI::InvalidURIError
+        false
+      end
 
-    def valid_json_pointer?(data)
-      JSON_POINTER_REGEX.match?(data)
-    end
+      def iri_escape(data)
+        Format.percent_encode(data, IRI_ESCAPE_REGEX)
+      end
 
-    def valid_relative_json_pointer?(data)
-      RELATIVE_JSON_POINTER_REGEX.match?(data)
-    end
+      def valid_regex?(data)
+        !!EcmaRegexp.ruby_equivalent(data)
+      rescue InvalidEcmaRegexp
+        false
+      end
 
-    def valid_regex?(data)
-      !!EcmaRegexp.ruby_equivalent(data)
-    rescue InvalidEcmaRegexp
-      false
+      def valid_uuid?(data)
+        UUID_REGEX.match?(data) || NIL_UUID == data
+      end
     end
   end
 end

data/lib/json_schemer/keyword.rb

@@ -0,0 +1,53 @@
+# frozen_string_literal: true
+module JSONSchemer
+  class Keyword
+    include Output
+
+    attr_reader :value, :parent, :root, :parsed
+
+    def initialize(value, parent, keyword, schema = parent)
+      @value = value
+      @parent = parent
+      @root = parent.root
+      @keyword = keyword
+      @schema = schema
+      @parsed = parse
+    end
+
+    def validate(_instance, _instance_location, _keyword_location, _context)
+      nil
+    end
+
+    def absolute_keyword_location
+      @absolute_keyword_location ||= "#{parent.absolute_keyword_location}/#{fragment_encode(escaped_keyword)}"
+    end
+
+    def schema_pointer
+      @schema_pointer ||= "#{parent.schema_pointer}/#{escaped_keyword}"
+    end
+
+    def error_key
+      keyword
+    end
+
+    def fetch(key)
+      parsed.fetch(parsed.is_a?(Array) ? key.to_i : key)
+    end
+
+    def parsed_schema
+      parsed.is_a?(Schema) ? parsed : nil
+    end
+
+  private
+
+    def parse
+      value
+    end
+
+    def subschema(value, keyword = nil, **options)
+      options[:base_uri] ||= schema.base_uri
+      options[:meta_schema] ||= schema.meta_schema
+      Schema.new(value, self, root, keyword, **options)
+    end
+  end
+end

data/lib/json_schemer/location.rb

@@ -0,0 +1,25 @@
+# frozen_string_literal: true
+module JSONSchemer
+  module Location
+    JSON_POINTER_TOKEN_ESCAPE_CHARS = { '~' => '~0', '/' => '~1' }
+    JSON_POINTER_TOKEN_ESCAPE_REGEX = Regexp.union(JSON_POINTER_TOKEN_ESCAPE_CHARS.keys)
+
+    class << self
+      def root
+        {}
+      end
+
+      def join(location, name)
+        location[name] ||= { :name => name, :parent => location }
+      end
+
+      def resolve(location)
+        location[:resolve] ||= location[:parent] ? "#{resolve(location[:parent])}/#{escape_json_pointer_token(location[:name])}" : ''
+      end
+
+      def escape_json_pointer_token(token)
+        token.gsub(JSON_POINTER_TOKEN_ESCAPE_REGEX, JSON_POINTER_TOKEN_ESCAPE_CHARS)
+      end
+    end
+  end
+end

data/lib/json_schemer/openapi.rb

@@ -0,0 +1,40 @@
+# frozen_string_literal: true
+module JSONSchemer
+  class OpenAPI
+    def initialize(document, **options)
+      @document = document
+
+      version = document['openapi']
+      case version
+      when /\A3\.1\.\d+\z/
+        @document_schema = JSONSchemer.openapi31_document
+        json_schema_dialect = document.fetch('jsonSchemaDialect') { OpenAPI31::BASE_URI.to_s }
+      when /\A3\.0\.\d+\z/
+        @document_schema = JSONSchemer.openapi30_document
+        json_schema_dialect = OpenAPI30::BASE_URI.to_s
+      else
+        raise UnsupportedOpenAPIVersion, version
+      end
+
+      meta_schema = META_SCHEMAS_BY_BASE_URI_STR[json_schema_dialect] || raise(UnsupportedMetaSchema, json_schema_dialect)
+
+      @schema = JSONSchemer.schema(@document, :meta_schema => meta_schema, **options)
+    end
+
+    def valid?
+      @document_schema.valid?(@document)
+    end
+
+    def validate(**options)
+      @document_schema.validate(@document, **options)
+    end
+
+    def ref(value)
+      @schema.ref(value)
+    end
+
+    def schema(name)
+      ref("#/components/schemas/#{name}")
+    end
+  end
+end