moxml 0.5.94 → 0.5.95

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: e54a93041f955c95affa2211704839bfaf47dafc0908cd73b58012946ffe99f1
-  data.tar.gz: 4d28d6376f61ae7441e13075ac5e9433bd18b6c393cf95afe363a639a4b8d251
+  metadata.gz: de8e16de76cef81c34b0a4e097d5aa16196fdbc03d9bfe2cbee1e765312f1599
+  data.tar.gz: ab5cc00b4c1a4a209065620c89b7541e107ed3d0a80e1e91e4455cb10c17fd98
 SHA512:
-  metadata.gz: 3e13a34f69a9e91b18d7fc9cef1e30c40ced24c8c73b322459ecba15af568439b002db8eec51097af2188571d026bfae7861646cc33077127c1fbd570b179863
-  data.tar.gz: a873c43e7ee3ccffc1997bf103263b7848d8ed17c61e25322a8d550bdf62a46d2f1e9e4a93f6e2832484a6df9d93350fd53d833a0b48ce04e1247583fc447d50
+  metadata.gz: bf21f0b903d142e6ebce40004d4e3bb657ec812396a2e8e6bd49c2eb06b6763fa6917f18c6918a134784d279360092c127fad56b2a75449f21d7e2dc94ae2e3c
+  data.tar.gz: 0bc9b8912c18c946101b76c88ce2c5c1d7f3ab2d8f643d06a8760d78332e1fd0d426844caff21efd8bb5241af9b35471478223fff9754b78405c3dda28d1990b

data/docs/_guides/advanced-features.adoc

@@ -80,6 +80,24 @@ adapter = element.context.config.adapter
 element = Moxml::Element.new(native_node, context)
 ----
 
+=== Parse bookkeeping opt-outs (leptris)
+
+Data-pipeline parses that never read diagnostics or source positions can
+skip that bookkeeping. Both are opt-in per call; the default parse is
+byte-identical either way.
+
+[source,ruby]
+----
+# Duplicate attributes admit silently: queries answer first-wins, but
+# the duplicate stays in the tree, so the serialized form can carry it.
+doc = Moxml.new.parse(xml, skip_dup_detection: true)
+
+# Source columns degrade; lines still resolve.
+doc = Moxml.new.parse(xml, skip_source_positions: true)
+----
+
+Without the flags, a duplicate attribute records a recover diagnostic in
+`doc.parse_diagnostics` and parses with the first value.
 
 See also:
 

data/lib/moxml/adapter/leptris.rb

@@ -856,6 +856,17 @@ module Moxml
         # Nokogiri semantics: blanks kept, no ATTLIST defaults).
         # noblanks forwards only once the engine flag is libxml2-safe
         # (see ENGINE_NOBLANKS_SAFE); otherwise it is moxml-side.
+        #
+        # The SKIP_ bookkeeping opt-outs (leptris-ruby#352, engine
+        # 1.9.272): duplicate attributes admit silently (first wins,
+        # no recover diagnostic) and source columns degrade (lines
+        # still resolve). Caller opt-in only — never defaults. The
+        # constant probe is the dependency-boundary check (the gem
+        # is zero-dep; user-resolved binding versions predate them).
+        NATIVE_PARSE_SKIPS =
+          ::Leptris::XML::ParseOptions.const_defined?(:SKIP_DUP_DETECTION) &&
+          ::Leptris::XML::ParseOptions.const_defined?(:SKIP_SOURCE_POSITIONS)
+
         def parse_flags(options, context: nil)
           flags = 0
           flags |= ::Leptris::XML::ParseOptions::DTDATTR if options[:dtdattr] == true
@@ -863,6 +874,11 @@ module Moxml
             (context&.config&.entity_mode == :keep ||
              options[:keep_entity_refs] == true)
           flags |= ::Leptris::XML::ParseOptions::NOBLANKS if options[:noblanks] == true && ENGINE_NOBLANKS_SAFE
+          if NATIVE_PARSE_SKIPS
+            po = ::Leptris::XML::ParseOptions
+            flags |= po::SKIP_DUP_DETECTION if options[:skip_dup_detection] == true
+            flags |= po::SKIP_SOURCE_POSITIONS if options[:skip_source_positions] == true
+          end
           flags.zero? ? nil : ::Leptris::XML::ParseOptions.new(flags)
         end
 

data/lib/moxml/adapter/nokogiri.rb

@@ -59,6 +59,7 @@ module Moxml
         # (Document/C14n), which is circular while THIS adapter file
         # is still loading.
         def native_inclusive10(native)
+          return nil if native.nil?
           return nil unless native_c14n_byte_safe?
 
           native.canonicalize(::Nokogiri::XML::XML_C14N_1_0)

data/lib/moxml/version.rb

@@ -1,5 +1,5 @@
 # frozen_string_literal: true
 
 module Moxml
-  VERSION = "0.5.94"
+  VERSION = "0.5.95"
 end

data/spec/moxml/adapter/leptris_parse_skips_spec.rb

@@ -0,0 +1,58 @@
+# frozen_string_literal: true
+
+require "spec_helper"
+
+DUP_XML = %(<r><e id="1">text</e><e id="2" id="3"/></r>)
+POS_XML = %(<r><e id="1">text</e></r>)
+
+RSpec.describe "Moxml leptris parse bookkeeping opt-outs" do
+  # leptris-ruby#352 (binding 1.9.273.1, engine 1.9.272 Door A):
+  # SKIP_DUP_DETECTION admits duplicate attributes silently (first
+  # wins, no recover diagnostic); SKIP_SOURCE_POSITIONS degrades
+  # source columns (lines still resolve). Caller opt-in only.
+  let(:ctx) { Moxml.new(:leptris) }
+  let(:adapter) { ctx.config.adapter }
+
+  before do
+    skip "binding without the SKIP_ parse flags" unless
+      Leptris::XML::ParseOptions.const_defined?(:SKIP_DUP_DETECTION)
+  end
+
+  it "drops the duplicate-attribute diagnostic when requested" do
+    doc = ctx.parse(DUP_XML, skip_dup_detection: true)
+    expect(adapter.parse_diagnostics(doc.native)).to eq([])
+  end
+
+  it "keeps the duplicate-attribute diagnostic by default" do
+    doc = ctx.parse(DUP_XML)
+    expect(adapter.parse_diagnostics(doc.native).map { |d| d[:kind] }).to include(:recover)
+  end
+
+  it "first attribute wins either way" do
+    expect(ctx.parse(DUP_XML, skip_dup_detection: true).root.children[1][:id])
+      .to eq(ctx.parse(DUP_XML).root.children[1][:id])
+  end
+
+  it "keeps lines but degrades columns for skipped positions" do
+    full = ctx.parse(POS_XML).root.children[0].source_position
+    skipped = ctx.parse(POS_XML, skip_source_positions: true)
+      .root.children[0].source_position
+    expect(skipped[:line]).to eq(full[:line])
+    expect(skipped[:col_start]).to be < full[:col_start]
+  end
+
+  it "keeps the duplicate in the tree — queries still answer first-wins" do
+    # The engine admits the duplicate silently: reads see the first
+    # value, but the second attribute stays in the tree, so the
+    # serialized form can carry it. Callers opting in trade the
+    # well-formed-output guarantee for the bookkeeping speed.
+    fast = ctx.parse(DUP_XML, skip_dup_detection: true)
+    expect(fast.root.children[1][:id]).to eq("2")
+    expect(fast.to_xml).to include('id="2" id="3"')
+  end
+
+  it "still raises on fatal errors with skips on" do
+    expect { ctx.parse("<r><e></r>", skip_dup_detection: true) }
+      .to raise_error(Moxml::ParseError)
+  end
+end

data/spec/moxml/struct_plan_typed_spec.rb

@@ -2,6 +2,10 @@
 
 require "spec_helper"
 
+# The per-item target class for the typed plan: Struct gives the
+# accessors StructPlan assigns the cast slots to.
+TypedItem = Struct.new(:price, :qty, :ok, :label)
+
 # Typed plan scalars (leptris plan ABI, #1269a consumer face):
 # [slot, type] declarations cast in C — Integer/Float/TrueClass/
 # FalseClass members with no Ruby String materialized; unparseable

metadata

@@ -1,7 +1,7 @@
 --- !ruby/object:Gem::Specification
 name: moxml
 version: !ruby/object:Gem::Version
-  version: 0.5.94
+  version: 0.5.95
 platform: ruby
 authors:
 - Ribose Inc.
@@ -438,6 +438,7 @@ files:
 - spec/moxml/adapter/entity_restoration_spec.rb
 - spec/moxml/adapter/headed_ox_spec.rb
 - spec/moxml/adapter/leptris_doc_parts_spec.rb
+- spec/moxml/adapter/leptris_parse_skips_spec.rb
 - spec/moxml/adapter/leptris_record_driver_spec.rb
 - spec/moxml/adapter/leptris_sax_dup_attr_spec.rb
 - spec/moxml/adapter/leptris_sax_records_spec.rb