canon 0.3.94 → 0.3.95

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: 5ab5a312bdd441c58ceda79bce75ccfab2b20a1c90c616bf763a90a2aafa78d1
-  data.tar.gz: c84b14c1296931956a10b090dd997d9e1b0987bc398469d58c5746dcf48b4c7f
+  metadata.gz: 5f4fab3f22311e8b17dff88e02d8ed2fd51e9d88822236a174dec1f60dcc1cef
+  data.tar.gz: e952e4fe4c278eb614c42a30a104c950d4dd4ecc871ea5387c4d483c8e292a7a
 SHA512:
-  metadata.gz: a76fc0c73ced250fda7897c1f53fe37283c97028d906e245982b44b810143bca8502da09b8550072e2fdb216009aa4c2338de4c138a06e5214df12a5e0ddd960
-  data.tar.gz: c125a5fe4d6fbd4755ce08cced6e34423ba23c76dcfd7c0bcba587bbf0dba24160421603d2b3b5b26564125eb156b8a6b31168888d3a53df63f460ca4d4e8494
+  metadata.gz: c3ecc4eeeb85076a6a1b3d9f75c2779b05baba59dd6f590f0dea301ac77cf4292278dcd6bef68f66ef97211fbb0c4a31cc3858f99055c0954d1bfcc57c0be125
+  data.tar.gz: 42bd8379d6117adc0b28d321a2e01be4851666d1348e39b42f41884254cf291449bb019f5ec1174970b3fa0ce7d77130065f6419a18a002e075666571b26314e

data/lib/canon/version.rb

@@ -1,5 +1,5 @@
 # frozen_string_literal: true
 
 module Canon
-  VERSION = "0.3.94"
+  VERSION = "0.3.95"
 end

data/lib/canon/xml/data_model.rb

@@ -227,12 +227,15 @@ inherited_namespaces: nil)
         # strict: false keeps the Nokogiri-engine contract — malformed
         # input yields a recovered document plus diagnostics instead of
         # an exception (moxml 0.5.15 defaults strict to true; issue
-        # #147 rides recover errors on Document#parse_errors).
+        # #147 rides recover errors on Document#parse_errors, with
+        # #271's parse_diagnostics as the second surface — leptris
+        # 1.9.362 records recovery exclusively there).
         doc = Canon::XmlParsing.moxml_context.parse(xml_string,
                                                     readonly: true,
                                                     strict: false)
         result = build_from_moxml(doc, preserve_whitespace: preserve_whitespace)
-        result.parse_errors = doc.parse_errors if doc.parse_errors.any?
+        records = Canon::XmlParsing.recover_records(doc)
+        result.parse_errors = records if records.any?
         # Canon owns this document's full lifecycle: the canon tree holds
         # no engine references once built, so release the native tree now
         # instead of waiting for the GC finalizer (moxml#134; no-op on

data/lib/canon/xml_parsing.rb

@@ -44,6 +44,17 @@ module Canon
         end
       end
 
+      # Recovery records across both engine surfaces: `parse_errors`
+      # (the legacy surface) and `parse_diagnostics` (moxml#271).
+      # leptris 1.9.362 moved recovery recording entirely to the
+      # diagnostics surface — `parse_errors` comes back empty on
+      # recovered parses — so the loudness contract and the banner
+      # carry-through must read both. Deduped by message: stacks that
+      # record on both surfaces would otherwise double-report.
+      def recover_records(doc)
+        (doc.parse_errors + doc.parse_diagnostics.map { |d| d[:message].to_s }).uniq
+      end
+
       # leptris#1564: HTML producers' output (void elements, unclosed
       # tags, inline JS with bare <) is not well-formed XML, but the
       # comparison pretty-printer still needs a tree. HTML-SHAPED
@@ -60,7 +71,7 @@ module Canon
         # contract keys off the RECORD, not the raise: non-HTML
         # malformed input stays loud; HTML-shaped input falls back to
         # the HTML parser. Engine-independent by construction.
-        recorded = doc.parse_errors
+        recorded = recover_records(doc)
         if recorded.any?
           raise Moxml::ParseError, recorded.join("; ") unless html_shaped?(xml_string)
 

metadata

@@ -1,14 +1,14 @@
 --- !ruby/object:Gem::Specification
 name: canon
 version: !ruby/object:Gem::Version
-  version: 0.3.94
+  version: 0.3.95
 platform: ruby
 authors:
 - Ribose Inc.
 autorequire:
 bindir: exe
 cert_chain: []
-date: 2026-10-15 00:00:00.000000000 Z
+date: 2026-10-16 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: diff-lcs