openehr 2.3.0 → 2.3.2

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: 10d6fc971443c1c8014d60bebd609c71757870ba0bb0d710cc35f014981b7ca1
-  data.tar.gz: 93b4ce3dd288a4943c0b4112d6ec9cb55ee1438d0411f66a7c47d5d26ea8b4b9
+  metadata.gz: a3d85dcafa7ccea7c1ae4a7cd794297b9b7c4b6a6126bae494fb5a6d9538fffc
+  data.tar.gz: 34bfec35b2de56cc2c97a02d049f6233b011d3e93908db2b5065f5fc1ede068e
 SHA512:
-  metadata.gz: cdb1ec0e04bda30a3d15fed92ef940f0a4aacde6f4e6570f8f5c5f6b0c6e14320166f88834c0a657a1b37d6973edc63650ce641bc1f43d92043289a90b95f5ed
-  data.tar.gz: 4b42fece35848e073eb3111d5f7abd8b85d02986799433df8330ff65bbbf5c1b8d979cf2d16bee175d4faf4d908c7cf78479a8dd9adfa3c2c6035c84a51d59a5
+  metadata.gz: 9793f50126a31f5d84a21304cd55ecb1fcc3a65569ba9eb1c3c73de1e15ea991a7cc245cec6c37dde71ed8fa861df809d919e02e13baad328ee75d2f0b44a5c5
+  data.tar.gz: a79ea4ee8a7981000548c16405561f255f6c08b5a0fa413f7cc9301fa165894f1401ec0bcd9c7669f0f0018e3178cf479643212ae3feaa31c20225ef249cc35e

data/lib/openehr/am/openehr_profile/data_types/text.rb

@@ -42,6 +42,27 @@ module OpenEHR
               end
             end
           end
+
+          # C_CODE_REFERENCE is defined by Template.xsd (Template OM) as a
+          # C_CODE_PHRASE extension carrying a referenceSetUri. nil is accepted
+          # for tolerant parsing of real artifacts. valid_value? is inherited:
+          # without a terminology service, this class cannot expand the URI's
+          # reference set and must retain CCodePhrase's permissive behavior.
+          class CCodeReference < CCodePhrase
+            attr_reader :reference_set_uri
+
+            def initialize(args = { })
+              super
+              self.reference_set_uri = args[:reference_set_uri]
+            end
+
+            def reference_set_uri=(reference_set_uri)
+              if !reference_set_uri.nil? && reference_set_uri.empty?
+                raise ArgumentError, 'invalid reference_set_uri'
+              end
+              @reference_set_uri = reference_set_uri
+            end
+          end
         end # of Text
       end # of Data_Types
     end # of OpenEHR Profile

data/lib/openehr/parser/opt_parser.rb

@@ -40,7 +40,9 @@ module OpenEHR
       OCCURRENCE_PATH = '/occurrences'
 
       def parse
-        @opt = Nokogiri::XML::Document.parse(File.open(@filename))
+        @opt = File.open(@filename) do |file|
+          Nokogiri::XML::Document.parse(file, options: SAFE_PARSE_OPTIONS)
+        end
         @opt.remove_namespaces!
 
         uid = build_uid

data/lib/openehr/parser/xml_archetype_parser.rb

@@ -30,7 +30,9 @@ module OpenEHR
 
       def doc
         @doc ||= begin
-          parsed = Nokogiri::XML::Document.parse(File.open(@filename, 'rb:bom|utf-8'))
+          parsed = File.open(@filename, 'rb:bom|utf-8') do |file|
+            Nokogiri::XML::Document.parse(file, options: SAFE_PARSE_OPTIONS)
+          end
           parsed.remove_namespaces!
           parsed
         end

data/lib/openehr/parser/xml_constraint_parsing.rb

@@ -65,7 +65,18 @@ module OpenEHR
           child_node = Node.new(node)
           child_node.path = node.path
           child_node.id = node.id
-          send child.attributes['type'].text.downcase, child, child_node
+          type_name = child.attributes['type'].text
+          handler = type_name.downcase
+          if respond_to?(handler, true)
+            send handler, child, child_node
+          else
+            # children is open to vendor/newer-model constraint extensions;
+            # the other xsi:type dispatch sites are schema-closed. Preserve
+            # the C_OBJECT core here, while warning that type-specific
+            # constraints are dropped and validation becomes more permissive.
+            warn "openehr parser: unknown constraint type \"#{type_name}\" at #{child_node.path}; treating as C_COMPLEX_OBJECT (type-specific constraints dropped)"
+            c_complex_object(child, child_node)
+          end
         end
       end
 

data/lib/openehr/parser/xml_domain_type_parsing.rb

@@ -11,6 +11,20 @@ module OpenEHR
       private
 
       def c_code_phrase(attr_xml, node)
+        OpenEHR::AM::OpenEHRProfile::DataTypes::Text::CCodePhrase.new(
+          code_phrase_constraint_args(attr_xml, node)
+        )
+      end
+
+      def c_code_reference(attr_xml, node)
+        args = code_phrase_constraint_args(attr_xml, node)
+        args[:code_list] = nil if args[:code_list] && args[:code_list].empty?
+        uri = attr_xml.at('referenceSetUri')&.text&.strip
+        args[:reference_set_uri] = uri unless uri.nil? || uri.empty?
+        OpenEHR::AM::OpenEHRProfile::DataTypes::Text::CCodeReference.new(args)
+      end
+
+      def code_phrase_constraint_args(attr_xml, node)
         terminology_id_node = attr_xml.at('terminology_id/value')
         terminology_id = terminology_id_node ? OpenEHR::RM::Support::Identification::TerminologyID.new(value: terminology_id_node.text.strip) : nil
 
@@ -21,13 +35,13 @@ module OpenEHR
         occurrences_node = attr_xml.at('occurrences')
         occurrences_obj = occurrences_node ? occurrences(occurrences_node) : nil
 
-        OpenEHR::AM::OpenEHRProfile::DataTypes::Text::CCodePhrase.new(
+        {
           terminology_id: terminology_id,
           code_list: code_list,
           path: node.path,
           occurrences: occurrences_obj,
           rm_type_name: 'CODE_PHRASE'
-        )
+        }
       end
 
       # The <property> element is optional in real templates; return nil rather

data/lib/openehr/parser.rb

@@ -1,6 +1,34 @@
+require 'nokogiri'
+
 module OpenEHR
   module Parser
     class Base
+      # Explicit, safe Nokogiri ParseOptions for untrusted OPT/archetype XML.
+      # Deliberately lists RECOVER|NONET|BIG_LINES by name rather than
+      # referencing Nokogiri::XML::ParseOptions::DEFAULT_XML - the whole
+      # point is to not silently follow that constant if a future Nokogiri
+      # release (or a downstream app pinning an older one) changes it.
+      #
+      # - RECOVER: parse malformed-but-well-intentioned XML leniently
+      #   (matches today's implicit default; see
+      #   docs/design/xxe-safe-parse-options-plan.md for real-fixture
+      #   evidence that at least one downstream consumer OPT currently
+      #   depends on this).
+      # - NONET: forbid network access during parsing. Nokogiri's own docs:
+      #   "UNSAFE to unset this option" for untrusted input.
+      # - BIG_LINES: line numbers as `long int`, unrelated to safety.
+      # - NOENT (entity substitution) and DTDLOAD (external DTD subset
+      #   loading) are deliberately NOT set - both default off in Nokogiri
+      #   and both documented "UNSAFE to set...for untrusted documents".
+      #   This is what actually gates XXE (see the plan's attack-proof
+      #   section: enabling either makes local-file and external-DTD
+      #   entity resolution succeed).
+      SAFE_PARSE_OPTIONS = Nokogiri::XML::ParseOptions.new(
+        Nokogiri::XML::ParseOptions::RECOVER |
+        Nokogiri::XML::ParseOptions::NONET |
+        Nokogiri::XML::ParseOptions::BIG_LINES
+      )
+
       attr_reader :filename
 
       def initialize(filename)

data/lib/openehr/version.rb

@@ -1,3 +1,3 @@
 module OpenEHR
-  VERSION = "2.3.0"
+  VERSION = "2.3.2"
 end

metadata

@@ -1,7 +1,7 @@
 --- !ruby/object:Gem::Specification
 name: openehr
 version: !ruby/object:Gem::Version
-  version: 2.3.0
+  version: 2.3.2
 platform: ruby
 authors:
 - Shinji KOBAYASHI