lutaml-xsd 1.2.80 → 1.2.81

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: 0aed69846e615ba0095fd9030ca3c7d870fe590c855694797393e669a57085ec
-  data.tar.gz: abcb15e8a7482e4589d08bbb886529f6aff6462db5c56e4626c9610dcb25777f
+  metadata.gz: 467b4c2343707af1d200289a1666d12e3ab984c438f0a29ef601d92143914562
+  data.tar.gz: d249de5b32c7fb201c8025cf56af088bde07acb6e1f76d2cd3ae1f15bd39cf29
 SHA512:
-  metadata.gz: 928180700fa58c7965b7f57729b681a1580f18c5da136c6b7729a013426ca5bfeaad44b686ce0e28c2eb1247dd3e73ca1a1e34c81b61c37030165bf4f297ae62
-  data.tar.gz: 50441ce2cda522fed7e3df0acd8843ef3824a72b685bf663e21ade1d9c5a2366fbb2af244b4f83d839dbdb4e2c41faf3cf1eb14f3376bef8c0ef15f03de9fb98
+  metadata.gz: 0f9c9c9c71bf3e064dd93ab6e007df3f9d90aaad2478bf77d8f7b7915804b5e8218a6d8adb97be02293560dc261ed957a6fe09af5404a30b18ee7a0d5d1a941e
+  data.tar.gz: 539a1503358510cde20356dc3d6a0a1a2c0b9daaebf77bda9ef45bc54f45222a0eaf2d97477b05f1347b1d35dcc22742963b0fc388823217bc6cd45ceb900bdf

data/TODO.conformance.md

@@ -120,7 +120,7 @@ pinned commit) is measured via `rake conformance:w3c`:
 
 | mode | pass-rate |
 |---|---|
-| pipeline | 4563/4696 (97.17%) |
+| pipeline | 4573/4696 (97.38%) |
 
 (2026-11-02, bundled-stack measurement with the escape,
 hint-following, pattern-alternative, identity-crash, engine-set,
@@ -142,7 +142,7 @@ lockfile refresh) → 4411 (93.9%) → 4430 (94.3%) → 4435 (94.4%) →
 4436 (94.5%) → 4439 (94.5%) → 4454 (94.9%) → 4495 (95.7%) → 4524
 (96.3%) → 4534 (96.6%) → 4540 (96.7%) → 4542 (96.7%) → 4546 (96.8%)
 → 4557 (97.0%) → 4565 (97.2%) → 4569 (97.3%) → 4562 (drift raw) →
-4563 (97.17%, ref-namespace round), +602 net over the 1.2.43 baseline. The escapes round (PR #197 — \d→\p{Nd},
+4563 (97.17%, ref-namespace round) → 4573 (97.38%, group-content round), +612 net over the 1.2.43 baseline. The escapes round (PR #197 — \d→\p{Nd},
 \w/\W/\s to their XSD value spaces) collapses Regex 120→69;
 same-step pattern facets now behave as alternatives and
 whitespace-only text reaches the facets intact (PR #200), Regex
@@ -422,6 +422,34 @@ tuple keys, node-set field violations and absent-field rows handled, cross-eleme
   4569→4562 raw; 4563 with this round. NET -6 documented, not
   chased — the fixed-empty leg reverses the moment leptris-rs#25
   lands.
+- GROUP-CONTENT ROUND (2026-11-02): +13 (Group 8→0 — the family is
+  CLEAN at 126/126; Particles Db001/Db006 +2, Ea004/011/018 +3; suite
+  4563→4573 = 97.38% on the drifted stack, rspec green). FOUR Ruby
+  defects: (a) resolve_group scanned only top-level schemas — a group
+  declared in an IMPORTED document (the extension-body shape)
+  resolved to nothing (groupB009/010); nested resolved schemas are
+  now searched depth-guarded. (b) The NFA wiring missed the
+  lutaml-model container classes: a group whose declaration is xs:all
+  compiled ORDERED (groupO004), and with_occurrence gave min=0
+  particles one phantom extra copy (min=0/max=1 compiled capacity 2 —
+  masked for bare elements by the occurrence rule, visible wherever
+  nested group bounds lean on this verdict: groupJ003/J006/J008/L008
+  accepted forbidden repeats); a max=0 particle compiles to the empty
+  fragment (groupH008). (c) type_particles had no complexContent
+  restriction branch — a restriction body compiled EMPTY and children
+  passed unchecked (groupH008); a restriction replaces the base
+  content model. (d) The ref owner-namespace probe resolved ANY
+  declaration — including inline ones — so a qualified local namesake
+  hijacked unprefixed refs; the probe is GLOBAL-only (resolve_global),
+  the unprefixed ref canonically names a no-namespace declaration,
+  several distinct prefixed forms of one local name mean deliberate
+  disambiguation (no owner guess), and an unresolvable ref matches by
+  local name alone. KNOWN GAP left open: particlesQ032 — a
+  no-namespace document imported under an EXPLICIT namespace: its
+  globals must adopt the import namespace (MS-lenient chameleon
+  import; our index keeps them no-namespace so the four e2 refs
+  mis-resolve). Zero regressions; the drift entries above remain
+  (fixed-empty 9, reU6).
 - REGEX SEMANTICS ROUND (2026-10-31, PR #208): \w/\W now carry their
   Appendix F complements (\w = [^\p{P}\p{Z}\p{C}], class interiors
   via the exact positive partition L/M/N/S), \I/\C complements added

data/lib/lutaml/xsd/validation/content_model_matcher.rb

@@ -134,13 +134,13 @@ module Lutaml
 
         def compile_particle(particle)
           case particle
-          when Lutaml::Xsd::Sequence
+          when Lutaml::Xsd::Sequence, Lutaml::Xml::Schema::Xsd::Sequence
             compile_bounded_container(particle)
-          when Lutaml::Xsd::Choice
+          when Lutaml::Xsd::Choice, Lutaml::Xml::Schema::Xsd::Choice
             with_occurrence(particle) do
               compile_choice_particles(@container_particles.call(particle))
             end
-          when Lutaml::Xsd::All
+          when Lutaml::Xsd::All, Lutaml::Xml::Schema::Xsd::All
             # The all's own bounds wrap the member bitmask: an optional
             # all accepts the empty sequence even with mandatory
             # members (addB31).
@@ -181,10 +181,17 @@ module Lutaml
         end
 
         # A group ref expands to its declaration's container, whose own
-        # kind (sequence vs choice) decides the wiring.
+        # kind (sequence vs choice vs all) decides the wiring. Imported
+        # documents parse to the lutaml-model container classes —
+        # groupO004's xs:all compiled ORDERED before both All classes
+        # were wired to the bitmask.
         def compile_container_body(container)
-          if container.is_a?(Lutaml::Xsd::Choice)
+          if container.is_a?(Lutaml::Xsd::Choice) ||
+              container.is_a?(Lutaml::Xml::Schema::Xsd::Choice)
             compile_choice_particles(@container_particles.call(container))
+          elsif container.is_a?(Lutaml::Xsd::All) ||
+              container.is_a?(Lutaml::Xml::Schema::Xsd::All)
+            with_occurrence(container) { compile_all(container) }
           else
             compile_particles(@container_particles.call(container))
           end
@@ -235,14 +242,20 @@ module Lutaml
 
         # Wraps a body fragment with the particle's occurrence range:
         # min mandatory copies, then an unbounded loop on one extra copy
-        # or (max - min) optional copies.
+        # or (max - min) optional copies. The first body copy counts
+        # toward the bound: with min=0 it IS the first optional copy, so
+        # only max-1 more are wired (min=0/max=1 used to compile with
+        # capacity 2 — masked for bare elements by the occurrence rule,
+        # visible wherever nested group bounds lean on this verdict).
         def with_occurrence(particle)
           min = parse_occurs(particle.min_occurs, 1)
           max = parse_max_occurs(particle.max_occurs, 1)
-          eff_min = [min, MAX_UNROLL].min
+          return empty_fragment if max.is_a?(Integer) && max.zero?
+
           entry, exit_state = yield
           cursor = exit_state
           last_entry = entry
+          eff_min = [min, MAX_UNROLL].min
           (eff_min - 1).times do
             c_entry, c_exit = yield
             @epsilons[cursor] << c_entry
@@ -259,7 +272,8 @@ module Lutaml
             @epsilons[exit_state] << final
           else
             @epsilons[cursor] << final
-            (max - min).times do
+            optional = max - min - (min.zero? && max.is_a?(Integer) ? 1 : 0)
+            optional.times do
               c_entry, c_exit = yield
               @epsilons[cursor] << c_entry
               @epsilons[c_exit] << final

data/lib/lutaml/xsd/validation/rules/content_model_validation_rule.rb

@@ -299,6 +299,7 @@ module Lutaml
                 type.complex_content
               content = type.complex_content
               extension = Array(content.extension).first
+              restriction = Array(content.restriction).first
               if extension
                 base_local = extension.base.to_s.split(":").last
                 base_type = base_local && repository&.resolve_named_type(base_local)
@@ -311,6 +312,14 @@ module Lutaml
                     particles << particle
                   end
                 end
+              elsif restriction
+                # A restriction REPLACES the base content model: only
+                # the restriction body's particles apply (groupH008's
+                # prohibited group ref was previously compiled as an
+                # EMPTY model and the children passed unchecked).
+                particles = COMPLEX_PARTICLE_KINDS.flat_map do |kind|
+                  Array(restriction.public_send(kind)).compact
+                end
               end
             end
 
@@ -508,12 +517,16 @@ module Lutaml
 
           # A qualified ref names a declaration in ONE namespace.
           # Global names duplicated across namespaces (import + main)
-          # resolve against the child's namespace today; when the owner
-          # schema's target namespace declares the same local name, a
-          # foreign child is an unexpected element (the engine — fed
-          # the documents one at a time — admits the substitute, so
-          # this runs on the engine-accept path). Wildcard territory
-          # is exempt.
+          # resolve against the child's namespace today; a foreign
+          # child is an unexpected element (the engine — fed the
+          # documents one at a time — admits the substitute, so this
+          # runs on the engine-accept path). Wildcard territory is
+          # exempt. The PREFIX is the signal: an unprefixed ref
+          # canonically names a NO-NAMESPACE declaration (Q016 — its
+          # qualified local namesake is irrelevant), a prefixed one
+          # names some namespace: a child in the opposite territory,
+          # or — for the duplicated-name shape — in any namespace but
+          # the owner's, cannot match.
           #
           # @return [void]
           def scan_cross_namespace_refs(xml_element, particles, collector,
@@ -526,12 +539,35 @@ module Lutaml
             end
             return if refs.empty?
 
+            # Several DISTINCT prefixed forms of one local name mean the
+            # schema disambiguates by prefix on purpose (particlesQ032:
+            # foo:e2 / x:e2 / bar:e2) — the owner-namespace guess is
+            # untenable there.
+            single_prefixed_form = refs.select do |r|
+              r.ref.to_s.include?(":")
+            end.group_by { |r| r.ref.to_s.split(":").last }
+              .select { |_, group| group.map { |r| r.ref.to_s.split(":").first }.uniq.size == 1 }
+              .map(&:first)
+
             xml_element.children.each do |child|
-              next if child.namespace_uri.to_s == owner_ns
-              next unless refs.any? do |r|
-                r.ref.to_s.split(":").last == child.name
+              child_ns = child.namespace_uri.to_s
+              next if child_ns == owner_ns
+
+              hit = refs.find do |r|
+                ref = r.ref.to_s
+                next false unless ref.split(":").last == child.name
+
+                if ref.include?(":")
+                  next false unless single_prefixed_form.include?(child.name)
+
+                  child_ns.empty? ||
+                    (child_ns != owner_ns &&
+                     repository.resolve_global(child.name, owner_ns))
+                else
+                  child_ns.match?(/\S/)
+                end
               end
-              next unless repository.resolve(child.name, owner_ns)
+              next unless hit
               next if wildcard_covers_child?(xml_element, particles, child,
                                              repository)
 
@@ -540,7 +576,7 @@ module Lutaml
                 code: "unexpected_element_in_sequence",
                 message: "Unexpected element '#{child.qualified_name}' " \
                          "in sequence — the reference names the same " \
-                         "declaration in #{owner_ns}",
+                         "declaration in another namespace",
                 location: child.xpath,
                 context: { element: child.qualified_name },
                 suggestion: "Use the declaration's namespace or rename " \
@@ -563,7 +599,7 @@ module Lutaml
             return nil unless repository
 
             local = particle.ref.to_s.split(":").last
-            (owner_ns && repository.resolve(local, owner_ns)) ||
+            (owner_ns && repository.resolve_global(local, owner_ns)) ||
               repository.resolve(local, child.namespace_uri) ||
               repository.resolve(local, nil)
           end

data/lib/lutaml/xsd/validation/schema_definition_resolver.rb

@@ -156,6 +156,27 @@ module Lutaml
           candidates.length == 1 ? candidates.first.first : nil
         end
 
+        # Global (top-level) element declaration only. A ref particle
+        # names a GLOBAL declaration; the index also holds inline ones,
+        # and a qualified local namesake must never satisfy the
+        # owner-namespace probe (particlesQ016: ref "foo" unprefixed vs
+        # the qualified local foo inside the restricted type).
+        #
+        # @param local_name [String]
+        # @param namespace_uri [String, nil]
+        # @return [Lutaml::Xml::Schema::Xsd::Element, nil]
+        def resolve_global(local_name, namespace_uri)
+          return nil if local_name.nil? || local_name.empty?
+
+          candidates = @elements[local_name]
+          return nil if candidates.empty?
+
+          exact = candidates.find do |_element, namespace, kind|
+            kind == :top && namespace_matches?(namespace, namespace_uri)
+          end
+          exact&.first
+        end
+
         # Named type for an element's `type="prefix:Name"` reference.
         #
         # @param schema_element [Lutaml::Xml::Schema::Xsd::Element, nil]
@@ -386,7 +407,10 @@ module Lutaml
         end
 
         # Named group declaration by local name (for xs:group reference
-        # expansion in the content and occurrence walks).
+        # expansion in the content and occurrence walks). Group refs
+        # reach across imports (groupB009: an extension body references
+        # a group declared in the imported document), so nested
+        # resolved schemas are searched too.
         #
         # @param local_name [String]
         # @return [Lutaml::Xml::Schema::Xsd::Group, nil]
@@ -396,6 +420,25 @@ module Lutaml
           schemas.each do |schema|
             found = Array(schema.group).find { |g| g.name == local_name }
             return found if found
+
+            found = find_nested_group(local_name, schema)
+            return found if found
+          end
+          nil
+        end
+
+        # @return [Object, nil]
+        def find_nested_group(local_name, schema, depth = 0)
+          return nil if depth > 8
+
+          (Array(schema.include) + Array(schema.import)).each do |nested|
+            next unless nested.is_a?(Lutaml::Xml::Schema::Xsd::Schema)
+
+            found = Array(nested.group).find { |g| g.name == local_name }
+            return found if found
+
+            found = find_nested_group(local_name, nested, depth + 1)
+            return found if found
           end
           nil
         end

data/lib/lutaml/xsd/validation/validation_rule.rb

@@ -252,12 +252,17 @@ module Lutaml
             # owner_ns nil means UNKNOWN (the caller has no owner
             # context), not "no namespace" — resolve(local, nil) only
             # matches no-namespace declarations, and a hint-loaded
-            # document can supply a foreign namesake.
+            # document can supply a foreign namesake. The owner probe
+            # is GLOBAL-only: a ref names a global declaration, and an
+            # inline namesake must not satisfy it.
             declaration =
-              (owner_ns && repository&.resolve(local, owner_ns)) ||
+              (owner_ns && repository&.resolve_global(local, owner_ns)) ||
               repository&.resolve(local, child.namespace_uri) ||
               repository&.resolve(local, nil)
-            return false unless declaration
+            # An unresolvable ref (particlesQ032: a no-namespace document
+            # imported under an explicit namespace — its globals index
+            # nowhere) is namespace-unconstrained: match by local name.
+            return child.name == local unless declaration
 
             child.name == declaration.name &&
               namespaces_match?(child.namespace_uri,

data/lib/lutaml/xsd/version.rb

@@ -2,6 +2,6 @@
 
 module Lutaml
   module Xsd
-    VERSION = "1.2.80"
+    VERSION = "1.2.81"
   end
 end

metadata

@@ -1,7 +1,7 @@
 --- !ruby/object:Gem::Specification
 name: lutaml-xsd
 version: !ruby/object:Gem::Version
-  version: 1.2.80
+  version: 1.2.81
 platform: ruby
 authors:
 - Ribose Inc.