moxml 0.5.82 → 0.5.84

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: 2945a0f85059a2979b59e3623e208fcc7bbd0195b2e00b581502adb5fa9f76ac
-  data.tar.gz: a5ca2152292ba794be5bc6edd766e36bf769b86fa63ce235a396645245b38bf7
+  metadata.gz: ff836e409dd2cf77368c7b26015e57de1b0af5e0678489e6e10452a6c70cc61a
+  data.tar.gz: a2a09b561505f90fc624089e925c3b41d4ab867668b00fd35f1132787e9dbfde
 SHA512:
-  metadata.gz: 9c57a2679fa6b1230ba4965f868c40de1eb98a8a4dac5dfe52a24a81f131d4ebb566707150b6d04e1f8b25d7e08bd284eae75d848b4488a0eed9c999b25662d5
-  data.tar.gz: 76107d50a3f3bfa76d2926931cbc54d49aa7e7e14d68f0aa932fb93513597d254f5063daf0341780bf9a2783472a884444b884fe50fa16e2248f08e0dcc68978
+  metadata.gz: decb847bc6cfd50e816c916d98c74b837d63e9e38a9b49a2d964333a141a8fafd9e55291f83a52739403ad99126884c2a37928567ac9f4f8fdaae3ca9df37692
+  data.tar.gz: 59be5497f4f1a4d439e86b4138be240b1bcabb370bc6d991d79b92aa64e9a40f0243b29f090d4171fd45146593ed27fc2da29ae2a2a72d0f542919e3e5aecf3b

data/lib/moxml/adapter/nokogiri.rb

@@ -30,6 +30,7 @@ module Moxml
           ::Nokogiri::XML::DocumentFragment.parse(processed) do |config|
             config.strict.nonet
             config.recover
+            config.huge
           end.children.to_a
         end
 
@@ -64,12 +65,18 @@ module Moxml
                 config.strict.nonet
                 config.recover unless options[:strict]
                 config.noblanks if options[:noblanks]
+                config.huge
               end
             else
               ::Nokogiri::XML(processed_xml, nil, "UTF-8") do |config|
                 config.strict.nonet
                 config.recover unless options[:strict]
                 config.noblanks if options[:noblanks]
+                # XML_PARSE_HUGE: lifts libxml2's arbitrary limits —
+                # text nodes > 10MB are silently TRUNCATED at 10MB in
+                # recover mode, so large documents lose data without
+                # it (users parse >10MB XML).
+                config.huge
               end
             end
           rescue ::Nokogiri::XML::SyntaxError => e
@@ -89,7 +96,7 @@ module Moxml
         # malformed input never raises.
         def parse_html(html, _options = {}, _context = nil)
           html_string = html.is_a?(IO) || html.is_a?(StringIO) ? html.read : html.to_s
-          native_doc = ::Nokogiri::HTML(html_string)
+          native_doc = ::Nokogiri::HTML(html_string, &:huge)
           Wrappers::Document.new(native_doc, _context || Context.new(:nokogiri))
         end
 

data/lib/moxml/version.rb

@@ -1,5 +1,5 @@
 # frozen_string_literal: true
 
 module Moxml
-  VERSION = "0.5.82"
+  VERSION = "0.5.84"
 end

data/spec/moxml/adapter/leptris_sax_dup_attr_spec.rb

@@ -0,0 +1,50 @@
+# frozen_string_literal: true
+
+require "spec_helper"
+
+# Duplicate attributes are a fatal XML 1.0 §3.1 error (moxml#281,
+# issue #130 banner semantics): the SAX lane must surface them via
+# on_error exactly like the DOM lane's parse_diagnostics. On
+# bindings whose drain claims dup-containing documents (engine <
+# 1.9.242 — leptris#1374), the drain swallows the error; those
+# bindings skip this spec until the ride lands.
+module SaxDupAttrHandlers
+  class ErrorCollector < Moxml::SAX::Handler
+    attr_reader :errors, :elements
+
+    def initialize
+      super
+      @errors = []
+      @elements = 0
+    end
+
+    def on_error(error)
+      @errors << error.message
+    end
+
+    def on_start_element(_name, _attrs = {}, _namespaces = {})
+      @elements += 1
+    end
+  end
+end
+
+RSpec.describe "SAX duplicate-attribute error surfacing" do
+  let(:context) { Moxml.new(:leptris) }
+  let(:dup_xml) do
+    '<body lang="en" xml:lang="en" xml:lang="en"><div>x</div></body>'
+  end
+
+  it "surfaces the redefined-attribute error through sax_parse" do
+    if defined?(Leptris::XML::SAX::Records) &&
+        !Leptris::XML::SAX::Records.open(dup_xml).nil?
+      skip "engine < 1.9.242 — the drain claims dup-attr documents " \
+           "(leptris#1374)"
+    end
+
+    handler = SaxDupAttrHandlers::ErrorCollector.new
+    context.sax_parse(dup_xml, handler)
+
+    expect(handler.errors.join).to include("xml:lang redefined")
+    expect(handler.elements).to eq(2)
+  end
+end

data/spec/moxml/adapter/nokogiri_spec.rb

@@ -36,4 +36,22 @@ RSpec.describe Moxml::Adapter::Nokogiri do
       expect(doc.parse_errors).to all(be_a(String))
     end
   end
+
+  # libxml2 truncates text nodes over XML_MAX_TEXT_LENGTH (10MB) —
+  # silently in recover mode — unless XML_PARSE_HUGE is set. The
+  # adapter always passes huge so large documents keep their data.
+  describe "XML_PARSE_HUGE (large documents)" do
+    let(:ctx) { Moxml.new(:nokogiri) }
+    let(:huge_xml) { "<r>#{'x' * 10_000_001}</r>" }
+
+    it "keeps text nodes over 10MB intact in recover mode" do
+      doc = ctx.parse(huge_xml, strict: false)
+      expect(doc.root.children.first.text.length).to eq(10_000_001)
+    end
+
+    it "keeps text nodes over 10MB intact in strict mode" do
+      doc = ctx.parse(huge_xml)
+      expect(doc.root.children.first.text.length).to eq(10_000_001)
+    end
+  end
 end

metadata

@@ -1,7 +1,7 @@
 --- !ruby/object:Gem::Specification
 name: moxml
 version: !ruby/object:Gem::Version
-  version: 0.5.82
+  version: 0.5.84
 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_sax_dup_attr_spec.rb
 - spec/moxml/adapter/leptris_sax_records_spec.rb
 - spec/moxml/adapter/leptris_spec.rb
 - spec/moxml/adapter/libxml_internals_spec.rb