uniword 1.2.5 → 1.4.0

data/lib/uniword/validation/rules/font_table_signature_rule.rb

@@ -1,7 +1,5 @@
 # frozen_string_literal: true
 
-require_relative "base"
-
 module Uniword
   module Validation
     module Rules

data/lib/uniword/validation/rules/fonts_rule.rb

@@ -1,7 +1,5 @@
 # frozen_string_literal: true
 
-require_relative "base"
-
 module Uniword
   module Validation
     module Rules

data/lib/uniword/validation/rules/footnotes_rule.rb

@@ -1,7 +1,5 @@
 # frozen_string_literal: true
 
-require_relative "base"
-
 module Uniword
   module Validation
     module Rules

data/lib/uniword/validation/rules/headers_footers_rule.rb

@@ -1,7 +1,5 @@
 # frozen_string_literal: true
 
-require_relative "base"
-
 module Uniword
   module Validation
     module Rules

data/lib/uniword/validation/rules/images_rule.rb

@@ -1,7 +1,5 @@
 # frozen_string_literal: true
 
-require_relative "base"
-
 module Uniword
   module Validation
     module Rules

data/lib/uniword/validation/rules/mc_ignorable_namespace_rule.rb

@@ -1,7 +1,5 @@
 # frozen_string_literal: true
 
-require_relative "base"
-
 module Uniword
   module Validation
     module Rules

data/lib/uniword/validation/rules/model_context.rb

@@ -0,0 +1,32 @@
+# frozen_string_literal: true
+
+module Uniword
+  module Validation
+    module Rules
+      # In-memory validation context: wraps a document model.
+      #
+      # Counterpart to DocumentContext (which wraps an on-disk DOCX
+      # package). Model-level rules consume this context via the Engine.
+      #
+      # @example Validate a built document
+      #   context = ModelContext.new(document)
+      #   issues = Validation::Engine.run(context)
+      class ModelContext
+        # @return [Wordprocessingml::DocumentRoot] the document model
+        attr_reader :document
+
+        # @param document [Wordprocessingml::DocumentRoot] in-memory model
+        def initialize(document)
+          @document = document
+        end
+
+        # Context type used by the Engine to select rules.
+        #
+        # @return [Symbol] :model — this context wraps an in-memory model
+        def context_type
+          :model
+        end
+      end
+    end
+  end
+end

data/lib/uniword/validation/rules/model_rule.rb

@@ -0,0 +1,34 @@
+# frozen_string_literal: true
+
+module Uniword
+  module Validation
+    module Rules
+      # Base class for in-memory (model-level) validation rules.
+      #
+      # Model rules run against a Rules::ModelContext — the pre-save
+      # front-end of the validation engine — and inspect the document
+      # model directly instead of parsing package XML.
+      #
+      # @example Implement a model rule
+      #   class MyModelRule < ModelRule
+      #     def code = "DOC-900"
+      #
+      #     def check(context)
+      #       return [] if context.document.body
+      #
+      #       [issue("Document body is missing")]
+      #     end
+      #   end
+      #
+      #   Rules::Registry.register(MyModelRule)
+      class ModelRule < Base
+        # Model rules consume the in-memory ModelContext.
+        #
+        # @return [Symbol] :model
+        def context_type
+          :model
+        end
+      end
+    end
+  end
+end

data/lib/uniword/validation/rules/numbering_preservation_rule.rb

@@ -1,7 +1,5 @@
 # frozen_string_literal: true
 
-require_relative "base"
-
 module Uniword
   module Validation
     module Rules

data/lib/uniword/validation/rules/numbering_rule.rb

@@ -1,7 +1,5 @@
 # frozen_string_literal: true
 
-require_relative "base"
-
 module Uniword
   module Validation
     module Rules

data/lib/uniword/validation/rules/relationship_integrity_rule.rb

@@ -1,7 +1,6 @@
 # frozen_string_literal: true
 
 require "zip"
-require_relative "base"
 
 module Uniword
   module Validation

data/lib/uniword/validation/rules/rsid_rule.rb

@@ -1,7 +1,5 @@
 # frozen_string_literal: true
 
-require_relative "base"
-
 module Uniword
   module Validation
     module Rules

data/lib/uniword/validation/rules/section_properties_rule.rb

@@ -1,7 +1,5 @@
 # frozen_string_literal: true
 
-require_relative "base"
-
 module Uniword
   module Validation
     module Rules

data/lib/uniword/validation/rules/settings_rule.rb

@@ -1,7 +1,5 @@
 # frozen_string_literal: true
 
-require_relative "base"
-
 module Uniword
   module Validation
     module Rules

data/lib/uniword/validation/rules/settings_values_rule.rb

@@ -1,7 +1,5 @@
 # frozen_string_literal: true
 
-require_relative "base"
-
 module Uniword
   module Validation
     module Rules

data/lib/uniword/validation/rules/style_references_rule.rb

@@ -1,7 +1,5 @@
 # frozen_string_literal: true
 
-require_relative "base"
-
 module Uniword
   module Validation
     module Rules

data/lib/uniword/validation/rules/table_grid_rule.rb

@@ -0,0 +1,27 @@
+# frozen_string_literal: true
+
+module Uniword
+  module Validation
+    module Rules
+      # DOC-204: every table must have a tblGrid.
+      #
+      # wml.xsd CT_Tbl requires exactly one tblGrid child; without it
+      # Word cannot lay out column widths and repairs the document.
+      class TableGridRule < ModelRule
+        def code = "DOC-204"
+        def category = :tables
+        def severity = "error"
+
+        def check(context)
+          context.document.tables.each_with_index.filter_map do |table, idx|
+            next if table.grid
+
+            issue("Table #{idx + 1} is missing the required tblGrid element",
+                  part: "word/document.xml",
+                  suggestion: "Add a w:tblGrid with one w:gridCol per column.")
+          end
+        end
+      end
+    end
+  end
+end

data/lib/uniword/validation/rules/table_properties_rule.rb

@@ -0,0 +1,27 @@
+# frozen_string_literal: true
+
+module Uniword
+  module Validation
+    module Rules
+      # DOC-205: every table must have table properties (tblPr).
+      #
+      # wml.xsd CT_Tbl requires exactly one tblPr child as the table's
+      # first property element.
+      class TablePropertiesRule < ModelRule
+        def code = "DOC-205"
+        def category = :tables
+        def severity = "error"
+
+        def check(context)
+          context.document.tables.each_with_index.filter_map do |table, idx|
+            next if table.properties
+
+            issue("Table #{idx + 1} is missing the required tblPr element",
+                  part: "word/document.xml",
+                  suggestion: "Add a w:tblPr element to the table.")
+          end
+        end
+      end
+    end
+  end
+end

data/lib/uniword/validation/rules/tables_rule.rb

@@ -1,7 +1,5 @@
 # frozen_string_literal: true
 
-require_relative "base"
-
 module Uniword
   module Validation
     module Rules

data/lib/uniword/validation/rules/theme_completeness_rule.rb

@@ -1,7 +1,5 @@
 # frozen_string_literal: true
 
-require_relative "base"
-
 module Uniword
   module Validation
     module Rules

data/lib/uniword/validation/rules/theme_rule.rb

@@ -1,7 +1,5 @@
 # frozen_string_literal: true
 
-require_relative "base"
-
 module Uniword
   module Validation
     module Rules

data/lib/uniword/validation/rules.rb

@@ -1,30 +1,68 @@
 # frozen_string_literal: true
 
-require_relative "rules/base"
-require_relative "rules/registry"
-require_relative "rules/document_context"
-require_relative "rules/style_references_rule"
-require_relative "rules/numbering_rule"
-require_relative "rules/footnotes_rule"
-require_relative "rules/headers_footers_rule"
-require_relative "rules/bookmarks_rule"
-require_relative "rules/images_rule"
-require_relative "rules/tables_rule"
-require_relative "rules/fonts_rule"
-require_relative "rules/theme_rule"
-require_relative "rules/settings_rule"
-require_relative "rules/mc_ignorable_namespace_rule"
-require_relative "rules/settings_values_rule"
-require_relative "rules/theme_completeness_rule"
-require_relative "rules/numbering_preservation_rule"
-require_relative "rules/section_properties_rule"
-require_relative "rules/core_properties_namespace_rule"
-require_relative "rules/content_types_coverage_rule"
-require_relative "rules/font_table_signature_rule"
-require_relative "rules/relationship_integrity_rule"
-require_relative "rules/rsid_rule"
+module Uniword
+  module Validation
+    # Document semantic validation rules and their registry.
+    module Rules
+      autoload :Base, "#{__dir__}/rules/base"
+      autoload :Registry, "#{__dir__}/rules/registry"
+      autoload :DocumentContext, "#{__dir__}/rules/document_context"
+      autoload :ModelContext, "#{__dir__}/rules/model_context"
+      autoload :ModelRule, "#{__dir__}/rules/model_rule"
+      autoload :DocumentBodyRule, "#{__dir__}/rules/document_body_rule"
+      autoload :BookmarkPairingRule,
+               "#{__dir__}/rules/bookmark_pairing_rule"
+      autoload :BookmarkUniquenessRule,
+               "#{__dir__}/rules/bookmark_uniqueness_rule"
+      autoload :EmptyParagraphsRule,
+               "#{__dir__}/rules/empty_paragraphs_rule"
+      autoload :TableGridRule, "#{__dir__}/rules/table_grid_rule"
+      autoload :TablePropertiesRule,
+               "#{__dir__}/rules/table_properties_rule"
+      autoload :StyleReferencesRule,
+               "#{__dir__}/rules/style_references_rule"
+      autoload :NumberingRule, "#{__dir__}/rules/numbering_rule"
+      autoload :FootnotesRule, "#{__dir__}/rules/footnotes_rule"
+      autoload :HeadersFootersRule,
+               "#{__dir__}/rules/headers_footers_rule"
+      autoload :BookmarksRule, "#{__dir__}/rules/bookmarks_rule"
+      autoload :ImagesRule, "#{__dir__}/rules/images_rule"
+      autoload :TablesRule, "#{__dir__}/rules/tables_rule"
+      autoload :FontsRule, "#{__dir__}/rules/fonts_rule"
+      autoload :ThemeRule, "#{__dir__}/rules/theme_rule"
+      autoload :SettingsRule, "#{__dir__}/rules/settings_rule"
+      autoload :McIgnorableNamespaceRule,
+               "#{__dir__}/rules/mc_ignorable_namespace_rule"
+      autoload :SettingsValuesRule,
+               "#{__dir__}/rules/settings_values_rule"
+      autoload :ThemeCompletenessRule,
+               "#{__dir__}/rules/theme_completeness_rule"
+      autoload :NumberingPreservationRule,
+               "#{__dir__}/rules/numbering_preservation_rule"
+      autoload :SectionPropertiesRule,
+               "#{__dir__}/rules/section_properties_rule"
+      autoload :CorePropertiesNamespaceRule,
+               "#{__dir__}/rules/core_properties_namespace_rule"
+      autoload :ContentTypesCoverageRule,
+               "#{__dir__}/rules/content_types_coverage_rule"
+      autoload :FontTableSignatureRule,
+               "#{__dir__}/rules/font_table_signature_rule"
+      autoload :RelationshipIntegrityRule,
+               "#{__dir__}/rules/relationship_integrity_rule"
+      autoload :RsidRule, "#{__dir__}/rules/rsid_rule"
+    end
+  end
+end
 
 # Register all built-in validation rules
+# Model-level (in-memory) rules
+Uniword::Validation::Rules::Registry.register(Uniword::Validation::Rules::DocumentBodyRule)
+Uniword::Validation::Rules::Registry.register(Uniword::Validation::Rules::BookmarkPairingRule)
+Uniword::Validation::Rules::Registry.register(Uniword::Validation::Rules::BookmarkUniquenessRule)
+Uniword::Validation::Rules::Registry.register(Uniword::Validation::Rules::EmptyParagraphsRule)
+Uniword::Validation::Rules::Registry.register(Uniword::Validation::Rules::TableGridRule)
+Uniword::Validation::Rules::Registry.register(Uniword::Validation::Rules::TablePropertiesRule)
+# Package-level (on-disk) rules
 Uniword::Validation::Rules::Registry.register(Uniword::Validation::Rules::StyleReferencesRule)
 Uniword::Validation::Rules::Registry.register(Uniword::Validation::Rules::NumberingRule)
 Uniword::Validation::Rules::Registry.register(Uniword::Validation::Rules::FootnotesRule)

data/lib/uniword/validation/schema_registry.rb

@@ -57,6 +57,7 @@ module Uniword
         word/document.xml
         word/styles.xml
         word/settings.xml
+        word/webSettings.xml
         word/fontTable.xml
         word/numbering.xml
         word/footnotes.xml
@@ -64,6 +65,13 @@ module Uniword
         word/comments.xml
       ].freeze
 
+      # Document property parts and their primary schemas.
+      DOCPROPS_SCHEMAS = {
+        "docProps/core.xml" => "ecma/opc-coreProperties.xsd",
+        "docProps/app.xml" => "iso/shared-documentPropertiesExtended.xsd",
+        "docProps/custom.xml" => "iso/shared-documentPropertiesCustom.xsd",
+      }.freeze
+
       # Pattern for header/footer parts
       HEADER_FOOTER_PATTERN = %r{\Aword/(header|footer)\d*\.xml\z}
 
@@ -115,6 +123,7 @@ module Uniword
       # For Word parts (document.xml, styles.xml, etc.), returns wml-2010.xsd
       # which imports the base wml.xsd and all extension schemas.
       # For relationship and content type parts, returns the appropriate schema.
+      # For docProps parts, returns the OPC/extended/custom properties schema.
       #
       # @param part_name [String] Path within ZIP (e.g., "word/document.xml")
       # @return [String, nil] XSD path relative to schemas_dir
@@ -124,6 +133,8 @@ module Uniword
           "ecma/opc-contentTypes.xsd"
         when "_rels/.rels", ->(n) { n.match?(RELS_PATTERN) }
           "ecma/opc-relationships.xsd"
+        when *DOCPROPS_SCHEMAS.keys
+          DOCPROPS_SCHEMAS[part_name]
         when *WORDML_PARTS, ->(n) { n.match?(HEADER_FOOTER_PATTERN) }
           "microsoft/wml-2010.xsd"
         when ->(n) { n.match?(THEME_PATTERN) }

data/lib/uniword/validation/validators/document_semantics_validator.rb

@@ -2,8 +2,6 @@
 
 require "zip"
 require "nokogiri"
-require_relative "../rules"
-require_relative "../report/validation_issue"
 # LayerValidator autoloaded via lib/uniword/validation.rb
 
 module Uniword
@@ -29,15 +27,9 @@ module Uniword
 
         def validate(path)
           issues = []
-
           context = Rules::DocumentContext.new(path)
           begin
-            rules = Rules::Registry.all
-            rules.each do |rule|
-              next unless rule.applicable?(context)
-
-              issues.concat(rule.check(context))
-            end
+            issues = Engine.run(context)
           ensure
             context.close
           end

data/lib/uniword/validation/validators/xml_schema_validator.rb

@@ -2,8 +2,6 @@
 
 require "zip"
 require "nokogiri"
-require_relative "../schema_registry"
-require_relative "../report/validation_issue"
 # LayerValidator autoloaded via lib/uniword/validation.rb
 
 module Uniword
@@ -22,6 +20,11 @@ module Uniword
       #   validator = XmlSchemaValidator.new("xml_schema" => { "xsd_validation" => true })
       #   result = validator.validate('/path/to/document.docx')
       class XmlSchemaValidator < LayerValidator
+        # Markup Compatibility namespace — the mc:Ignorable attribute
+        # itself lives here and is stripped during MCE preprocessing.
+        MCE_NAMESPACE_URI =
+          "http://schemas.openxmlformats.org/markup-compatibility/2006"
+
         def layer_name
           "XSD Schema"
         end
@@ -85,6 +88,7 @@ module Uniword
               content = entry.get_input_stream.read
               schema = registry.load_schema(schema_path)
               doc = Nokogiri::XML(content)
+              preprocess_mce(doc)
 
               schema.validate(doc).each do |error|
                 issues << Report::ValidationIssue.new(
@@ -116,6 +120,46 @@ module Uniword
           end
         end
 
+        # Strip MCE-ignorable ATTRIBUTES per the part's own mc:Ignorable
+        # declaration (w14:paraId/textId and the mc:Ignorable attribute
+        # itself), so XSD validation is not drowned in false errors for
+        # content every real document carries.
+        #
+        # Extension ELEMENTS are deliberately kept: removing them changes
+        # libxml2's content-model evaluation and can expose pre-existing
+        # Word-vs-schema quirks (e.g. rsids placed late, tolerated via the
+        # sequence's trailing xsd:any) as false baseline errors.
+        # Operates in place on the parsed document; never touches the
+        # packaged bytes.
+        #
+        # @param doc [Nokogiri::XML::Document] parsed part to clean
+        # @return [void]
+        def preprocess_mce(doc)
+          root = doc.root
+          return unless root
+
+          ignorable = root["Ignorable"] || root["mc:Ignorable"]
+          return if ignorable.nil? || ignorable.empty?
+
+          uris = ignorable.split(/\s+/).filter_map do |prefix|
+            root.namespaces["xmlns:#{prefix}"]
+          end
+          # The mc:Ignorable attribute itself is MCE machinery and is
+          # undeclared on most root elements in the transitional XSDs.
+          strip_mce_attributes(root, uris + [MCE_NAMESPACE_URI])
+        end
+
+        # Remove attributes in ignorable namespaces from every element.
+        def strip_mce_attributes(doc, uris)
+          doc.traverse do |node|
+            next unless node.element?
+
+            node.attribute_nodes.each do |attr|
+              attr.remove if uris.include?(attr.namespace&.href)
+            end
+          end
+        end
+
         def check_unknown_namespaces(registry, content, part_name, issues)
           ns_uris = registry.detect_namespaces(content)
           unknown = registry.unknown_namespaces(ns_uris)

data/lib/uniword/validation/validators.rb

@@ -3,16 +3,7 @@
 module Uniword
   module Validation
     module Validators
-      autoload :FileStructureValidator,
-               "#{__dir__}/validators/file_structure_validator"
-      autoload :ZipIntegrityValidator,
-               "#{__dir__}/validators/zip_integrity_validator"
-      autoload :OoxmlPartValidator, "#{__dir__}/validators/ooxml_part_validator"
       autoload :XmlSchemaValidator, "#{__dir__}/validators/xml_schema_validator"
-      autoload :RelationshipValidator,
-               "#{__dir__}/validators/relationship_validator"
-      autoload :ContentTypeValidator,
-               "#{__dir__}/validators/content_type_validator"
       autoload :DocumentSemanticsValidator,
                "#{__dir__}/validators/document_semantics_validator"
     end

data/lib/uniword/validation/verify_orchestrator.rb

@@ -1,13 +1,6 @@
 # frozen_string_literal: true
 
 require "time"
-require_relative "opc_validator"
-require_relative "schema_registry"
-require_relative "report/verification_report"
-require_relative "report/layer_result"
-require_relative "report/terminal_formatter"
-require_relative "validators/xml_schema_validator"
-require_relative "validators/document_semantics_validator"
 
 module Uniword
   module Validation

data/lib/uniword/validation.rb

@@ -9,9 +9,14 @@ module Uniword
     autoload :LayerValidator, "#{__dir__}/validation/layer_validator"
     autoload :LayerValidationResult,
              "#{__dir__}/validation/layer_validation_result"
-    autoload :DocumentValidator, "#{__dir__}/validation/document_validator"
     autoload :Checkers, "#{__dir__}/validation/checkers"
     autoload :Validators, "#{__dir__}/validation/validators"
-    autoload :StructuralValidator, "#{__dir__}/validation/structural_validator"
+    autoload :Engine, "#{__dir__}/validation/engine"
+    autoload :OpcValidator, "#{__dir__}/validation/opc_validator"
+    autoload :SchemaRegistry, "#{__dir__}/validation/schema_registry"
+    autoload :VerifyOrchestrator,
+             "#{__dir__}/validation/verify_orchestrator"
+    autoload :Rules, "#{__dir__}/validation/rules"
+    autoload :Report, "#{__dir__}/validation/report"
   end
 end

data/lib/uniword/version.rb

@@ -1,5 +1,5 @@
 # frozen_string_literal: true
 
 module Uniword
-  VERSION = "1.2.5"
+  VERSION = "1.4.0"
 end

data/lib/uniword/vml/imagedata.rb

@@ -1,7 +1,6 @@
 # frozen_string_literal: true
 
 require "lutaml/model"
-require "uniword/ooxml/types/relationship_id"
 
 module Uniword
   module Vml

data/lib/uniword/wordprocessingml/attached_template.rb

@@ -1,7 +1,6 @@
 # frozen_string_literal: true
 
 require "lutaml/model"
-require_relative "../properties/relationship_id"
 
 module Uniword
   module Wordprocessingml

data/lib/uniword/wordprocessingml/deleted_text.rb

@@ -16,8 +16,8 @@ module Uniword
       xml do
         element "delText"
         namespace Uniword::Ooxml::Namespaces::WordProcessingML
-        mixed_content
 
+        map_content to: :content
         map_attribute "xml:space", to: :xml_space, render_nil: false
       end