REXML 3.1.9.1

git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/branches/ruby_2_6@67940 b2dd03c8-39d4-4d8f-98ff-823fe69b080e

7 files changed, 414 insertions, 20 deletions

diff --git a/test/rexml/parse/test_document_type_declaration.rb b/test/rexml/parse/test_document_type_declaration.rb
index 80f70888fb..55713909e7 100644
--- a/test/rexml/parse/test_document_type_declaration.rb
+++ b/test/rexml/parse/test_document_type_declaration.rb
@@ -5,17 +5,187 @@ require "rexml/document"
 module REXMLTests
   class TestParseDocumentTypeDeclaration < Test::Unit::TestCase
     private
-    def xml(internal_subset)
-      <<-XML
-<!DOCTYPE r SYSTEM "urn:x-rexml:test" [
-#{internal_subset}
-]>
+    def parse(doctype)
+      REXML::Document.new(<<-XML).doctype
+#{doctype}
 <r/>
       XML
     end
 
-    def parse(internal_subset)
-      REXML::Document.new(xml(internal_subset)).doctype
+    class TestName < self
+      def test_valid
+        doctype = parse(<<-DOCTYPE)
+<!DOCTYPE r>
+        DOCTYPE
+        assert_equal("r", doctype.name)
+      end
+
+      def test_garbage_plus_before_name_at_line_start
+        exception = assert_raise(REXML::ParseException) do
+          parse(<<-DOCTYPE)
+<!DOCTYPE +
+r SYSTEM "urn:x-rexml:test" [
+]>
+          DOCTYPE
+        end
+        assert_equal(<<-DETAIL.chomp, exception.to_s)
+Malformed DOCTYPE: invalid name
+Line: 5
+Position: 51
+Last 80 unconsumed characters:
++ r SYSTEM "urn:x-rexml:test" [ ]>  <r/> 
+        DETAIL
+      end
+    end
+
+    class TestExternalID < self
+      class TestSystem < self
+        def test_left_bracket_in_system_literal
+          doctype = parse(<<-DOCTYPE)
+<!DOCTYPE r SYSTEM "urn:x-rexml:[test" [
+]>
+          DOCTYPE
+          assert_equal([
+                         "r",
+                         "SYSTEM",
+                         nil,
+                         "urn:x-rexml:[test",
+                       ],
+                       [
+                         doctype.name,
+                         doctype.external_id,
+                         doctype.public,
+                         doctype.system,
+                       ])
+        end
+
+        def test_greater_than_in_system_literal
+          doctype = parse(<<-DOCTYPE)
+<!DOCTYPE r SYSTEM "urn:x-rexml:>test" [
+]>
+          DOCTYPE
+          assert_equal([
+                         "r",
+                         "SYSTEM",
+                         nil,
+                         "urn:x-rexml:>test",
+                       ],
+                       [
+                         doctype.name,
+                         doctype.external_id,
+                         doctype.public,
+                         doctype.system,
+                       ])
+        end
+
+        def test_no_literal
+          exception = assert_raise(REXML::ParseException) do
+            parse(<<-DOCTYPE)
+<!DOCTYPE r SYSTEM>
+            DOCTYPE
+          end
+          assert_equal(<<-DETAIL.chomp, exception.to_s)
+Malformed DOCTYPE: system literal is missing
+Line: 3
+Position: 26
+Last 80 unconsumed characters:
+ SYSTEM>  <r/> 
+          DETAIL
+        end
+
+        def test_garbage_after_literal
+          exception = assert_raise(REXML::ParseException) do
+            parse(<<-DOCTYPE)
+<!DOCTYPE r SYSTEM 'r.dtd'x'>
+            DOCTYPE
+          end
+          assert_equal(<<-DETAIL.chomp, exception.to_s)
+Malformed DOCTYPE: garbage after external ID
+Line: 3
+Position: 36
+Last 80 unconsumed characters:
+x'>  <r/> 
+          DETAIL
+        end
+
+        def test_single_quote
+          doctype = parse(<<-DOCTYPE)
+<!DOCTYPE r SYSTEM 'r".dtd'>
+          DOCTYPE
+          assert_equal("r\".dtd", doctype.system)
+        end
+
+        def test_double_quote
+          doctype = parse(<<-DOCTYPE)
+<!DOCTYPE r SYSTEM "r'.dtd">
+          DOCTYPE
+          assert_equal("r'.dtd", doctype.system)
+        end
+      end
+
+      class TestPublic < self
+        class TestPublicIDLiteral < self
+          def test_content_double_quote
+            exception = assert_raise(REXML::ParseException) do
+              parse(<<-DOCTYPE)
+<!DOCTYPE r PUBLIC 'double quote " is invalid' "r.dtd">
+              DOCTYPE
+            end
+            assert_equal(<<-DETAIL.chomp, exception.to_s)
+Malformed DOCTYPE: invalid public ID literal
+Line: 3
+Position: 62
+Last 80 unconsumed characters:
+ PUBLIC 'double quote " is invalid' "r.dtd">  <r/> 
+            DETAIL
+          end
+
+          def test_single_quote
+            doctype = parse(<<-DOCTYPE)
+<!DOCTYPE r PUBLIC 'public-id-literal' "r.dtd">
+            DOCTYPE
+            assert_equal("public-id-literal", doctype.public)
+          end
+
+          def test_double_quote
+            doctype = parse(<<-DOCTYPE)
+<!DOCTYPE r PUBLIC "public'-id-literal" "r.dtd">
+            DOCTYPE
+            assert_equal("public'-id-literal", doctype.public)
+          end
+        end
+
+        class TestSystemLiteral < self
+          def test_garbage_after_literal
+            exception = assert_raise(REXML::ParseException) do
+              parse(<<-DOCTYPE)
+<!DOCTYPE r PUBLIC 'public-id-literal' 'system-literal'x'>
+              DOCTYPE
+            end
+            assert_equal(<<-DETAIL.chomp, exception.to_s)
+Malformed DOCTYPE: garbage after external ID
+Line: 3
+Position: 65
+Last 80 unconsumed characters:
+x'>  <r/> 
+           DETAIL
+          end
+
+          def test_single_quote
+            doctype = parse(<<-DOCTYPE)
+<!DOCTYPE r PUBLIC "public-id-literal" 'system"-literal'>
+            DOCTYPE
+            assert_equal("system\"-literal", doctype.system)
+          end
+
+          def test_double_quote
+            doctype = parse(<<-DOCTYPE)
+<!DOCTYPE r PUBLIC "public-id-literal" "system'-literal">
+            DOCTYPE
+            assert_equal("system'-literal", doctype.system)
+          end
+        end
+      end
     end
 
     class TestMixed < self
@@ -45,6 +215,15 @@ module REXMLTests
         assert_equal([REXML::NotationDecl, REXML::AttlistDecl],
                      doctype.children.collect(&:class))
       end
+
+      private
+      def parse(internal_subset)
+        super(<<-DOCTYPE)
+<!DOCTYPE r SYSTEM "urn:x-rexml:test" [
+#{internal_subset}
+]>
+        DOCTYPE
+      end
     end
   end
 end
diff --git a/test/rexml/parse/test_element.rb b/test/rexml/parse/test_element.rb
index aad915fe7b..7206fe595a 100644
--- a/test/rexml/parse/test_element.rb
+++ b/test/rexml/parse/test_element.rb
@@ -33,6 +33,32 @@ Last 80 unconsumed characters:
 
         DETAIL
       end
+
+      def test_garbage_less_than_before_root_element_at_line_start
+        exception = assert_raise(REXML::ParseException) do
+          parse("<\n<x/>")
+        end
+        assert_equal(<<-DETAIL.chomp, exception.to_s)
+malformed XML: missing tag start
+Line: 2
+Position: 6
+Last 80 unconsumed characters:
+< <x/>
+        DETAIL
+      end
+
+      def test_garbage_less_than_slash_before_end_tag_at_line_start
+        exception = assert_raise(REXML::ParseException) do
+          parse("<x></\n</x>")
+        end
+        assert_equal(<<-DETAIL.chomp, exception.to_s)
+Missing end tag for 'x'
+Line: 2
+Position: 10
+Last 80 unconsumed characters:
+</ </x>
+        DETAIL
+      end
     end
   end
 end
diff --git a/test/rexml/parse/test_notation_declaration.rb b/test/rexml/parse/test_notation_declaration.rb
index 0d29f0d81f..19a0536d0a 100644
--- a/test/rexml/parse/test_notation_declaration.rb
+++ b/test/rexml/parse/test_notation_declaration.rb
@@ -23,10 +23,100 @@ module REXMLTests
         doctype = parse("<!NOTATION name PUBLIC 'urn:public-id'>")
         assert_equal("name", doctype.notation("name").name)
       end
+
+      def test_no_name
+        exception = assert_raise(REXML::ParseException) do
+          parse(<<-INTERNAL_SUBSET)
+<!NOTATION>
+          INTERNAL_SUBSET
+        end
+        assert_equal(<<-DETAIL.chomp, exception.to_s)
+Malformed notation declaration: name is missing
+Line: 5
+Position: 72
+Last 80 unconsumed characters:
+ <!NOTATION>  ]> <r/> 
+        DETAIL
+      end
+
+      def test_invalid_name
+        exception = assert_raise(REXML::ParseException) do
+          parse(<<-INTERNAL_SUBSET)
+<!NOTATION '>
+          INTERNAL_SUBSET
+        end
+        assert_equal(<<-DETAIL.chomp, exception.to_s)
+Malformed notation declaration: invalid name
+Line: 5
+Position: 74
+Last 80 unconsumed characters:
+'>  ]> <r/> 
+        DETAIL
+      end
+
+      def test_no_id_type
+        exception = assert_raise(REXML::ParseException) do
+          parse(<<-INTERNAL_SUBSET)
+<!NOTATION name>
+          INTERNAL_SUBSET
+        end
+        assert_equal(<<-DETAIL.chomp, exception.to_s)
+Malformed notation declaration: invalid ID type
+Line: 5
+Position: 77
+Last 80 unconsumed characters:
+>  ]> <r/> 
+        DETAIL
+      end
+
+      def test_invalid_id_type
+        exception = assert_raise(REXML::ParseException) do
+          parse(<<-INTERNAL_SUBSET)
+<!NOTATION name INVALID>
+          INTERNAL_SUBSET
+        end
+        assert_equal(<<-DETAIL.chomp, exception.to_s)
+Malformed notation declaration: invalid ID type
+Line: 5
+Position: 85
+Last 80 unconsumed characters:
+ INVALID>  ]> <r/> 
+        DETAIL
+      end
     end
 
     class TestExternalID < self
       class TestSystem < self
+        def test_no_literal
+          exception = assert_raise(REXML::ParseException) do
+            parse(<<-INTERNAL_SUBSET)
+<!NOTATION name SYSTEM>
+            INTERNAL_SUBSET
+          end
+          assert_equal(<<-DETAIL.chomp, exception.to_s)
+Malformed notation declaration: system literal is missing
+Line: 5
+Position: 84
+Last 80 unconsumed characters:
+ SYSTEM>  ]> <r/> 
+          DETAIL
+        end
+
+        def test_garbage_after_literal
+          exception = assert_raise(REXML::ParseException) do
+            parse(<<-INTERNAL_SUBSET)
+<!NOTATION name SYSTEM 'system-literal'x'>
+            INTERNAL_SUBSET
+          end
+          assert_equal(<<-DETAIL.chomp, exception.to_s)
+Malformed notation declaration: garbage before end >
+Line: 5
+Position: 103
+Last 80 unconsumed characters:
+x'>  ]> <r/> 
+          DETAIL
+        end
+
         def test_single_quote
           doctype = parse(<<-INTERNAL_SUBSET)
 <!NOTATION name SYSTEM 'system-literal'>
@@ -44,6 +134,21 @@ module REXMLTests
 
       class TestPublic < self
         class TestPublicIDLiteral < self
+          def test_content_double_quote
+            exception = assert_raise(REXML::ParseException) do
+              parse(<<-INTERNAL_SUBSET)
+<!NOTATION name PUBLIC 'double quote " is invalid' "system-literal">
+              INTERNAL_SUBSET
+            end
+            assert_equal(<<-DETAIL.chomp, exception.to_s)
+Malformed notation declaration: invalid public ID literal
+Line: 5
+Position: 129
+Last 80 unconsumed characters:
+ PUBLIC 'double quote " is invalid' "system-literal">  ]> <r/> 
+            DETAIL
+          end
+
           def test_single_quote
             doctype = parse(<<-INTERNAL_SUBSET)
 <!NOTATION name PUBLIC 'public-id-literal' "system-literal">
@@ -60,6 +165,21 @@ module REXMLTests
         end
 
         class TestSystemLiteral < self
+          def test_garbage_after_literal
+            exception = assert_raise(REXML::ParseException) do
+              parse(<<-INTERNAL_SUBSET)
+<!NOTATION name PUBLIC 'public-id-literal' 'system-literal'x'>
+              INTERNAL_SUBSET
+            end
+            assert_equal(<<-DETAIL.chomp, exception.to_s)
+Malformed notation declaration: garbage before end >
+Line: 5
+Position: 123
+Last 80 unconsumed characters:
+x'>  ]> <r/> 
+           DETAIL
+          end
+
           def test_single_quote
             doctype = parse(<<-INTERNAL_SUBSET)
 <!NOTATION name PUBLIC "public-id-literal" 'system-literal'>
@@ -96,5 +216,66 @@ module REXMLTests
         end
       end
     end
+
+    class TestPublicID < self
+      def test_no_literal
+        exception = assert_raise(REXML::ParseException) do
+          parse(<<-INTERNAL_SUBSET)
+<!NOTATION name PUBLIC>
+          INTERNAL_SUBSET
+        end
+        assert_equal(<<-DETAIL.chomp, exception.to_s)
+Malformed notation declaration: public ID literal is missing
+Line: 5
+Position: 84
+Last 80 unconsumed characters:
+ PUBLIC>  ]> <r/> 
+        DETAIL
+      end
+
+      def test_literal_content_double_quote
+        exception = assert_raise(REXML::ParseException) do
+          parse(<<-INTERNAL_SUBSET)
+<!NOTATION name PUBLIC 'double quote " is invalid in PubidLiteral'>
+          INTERNAL_SUBSET
+        end
+        assert_equal(<<-DETAIL.chomp, exception.to_s)
+Malformed notation declaration: invalid public ID literal
+Line: 5
+Position: 128
+Last 80 unconsumed characters:
+ PUBLIC 'double quote \" is invalid in PubidLiteral'>  ]> <r/> 
+        DETAIL
+      end
+
+      def test_garbage_after_literal
+        exception = assert_raise(REXML::ParseException) do
+          parse(<<-INTERNAL_SUBSET)
+<!NOTATION name PUBLIC 'public-id-literal'x'>
+          INTERNAL_SUBSET
+        end
+        assert_equal(<<-DETAIL.chomp, exception.to_s)
+Malformed notation declaration: garbage before end >
+Line: 5
+Position: 106
+Last 80 unconsumed characters:
+x'>  ]> <r/> 
+        DETAIL
+      end
+
+      def test_literal_single_quote
+        doctype = parse(<<-INTERNAL_SUBSET)
+<!NOTATION name PUBLIC 'public-id-literal'>
+        INTERNAL_SUBSET
+        assert_equal("public-id-literal", doctype.notation("name").public)
+      end
+
+      def test_literal_double_quote
+        doctype = parse(<<-INTERNAL_SUBSET)
+<!NOTATION name PUBLIC "public-id-literal">
+        INTERNAL_SUBSET
+        assert_equal("public-id-literal", doctype.notation("name").public)
+      end
+    end
   end
 end
diff --git a/test/rexml/parse/test_processing_instruction.rb b/test/rexml/parse/test_processing_instruction.rb
index a23513fc6e..f0c0c24e67 100644
--- a/test/rexml/parse/test_processing_instruction.rb
+++ b/test/rexml/parse/test_processing_instruction.rb
@@ -20,6 +20,25 @@ Last 80 unconsumed characters:
 <??>
         DETAIL
       end
+
+      def test_garbage_text
+        # TODO: This should be parse error.
+        # Create test/parse/test_document.rb or something and move this to it.
+        doc = parse(<<-XML)
+x<?x y
+<!--?><?x -->?>
+<r/>
+        XML
+        pi = doc.children[1]
+        assert_equal([
+                       "x",
+                       "y\n<!--",
+                     ],
+                     [
+                       pi.target,
+                       pi.content,
+                     ])
+      end
     end
   end
 end
diff --git a/test/rexml/parser/test_ultra_light.rb b/test/rexml/parser/test_ultra_light.rb
index 8f4a3980d5..44fd1d1ec0 100644
--- a/test/rexml/parser/test_ultra_light.rb
+++ b/test/rexml/parser/test_ultra_light.rb
@@ -16,7 +16,6 @@ class TestUltraLightParser < Test::Unit::TestCase
                        nil,
                        [:entitydecl, "name", "value"]
                      ],
-                     [:text, "\n"],
                      [:start_element, :parent, "root", {}],
                      [:text, "\n"],
                    ],
diff --git a/test/rexml/test_core.rb b/test/rexml/test_core.rb
index 46036d7f12..ae528d75d3 100644
--- a/test/rexml/test_core.rb
+++ b/test/rexml/test_core.rb
@@ -1,4 +1,4 @@
-# coding: binary
+# coding: utf-8
 # frozen_string_literal: false
 
 require_relative "rexml_test_utils"
@@ -995,7 +995,7 @@ EOL
       document.write(s)
 
       ## XML Doctype
-      str = '<!DOCTYPE foo "bar">'
+      str = '<!DOCTYPE foo SYSTEM "bar">'
       source  = REXML::Source.new(str)
       doctype = REXML::DocType.new(source)
       document.add(doctype)
diff --git a/test/rexml/test_doctype.rb b/test/rexml/test_doctype.rb
index 7f42669170..915717de8e 100644
--- a/test/rexml/test_doctype.rb
+++ b/test/rexml/test_doctype.rb
@@ -18,12 +18,6 @@ module REXMLTests
       @doc_type_system = REXML::Document.new(xml_system).doctype
 
       @pubid = "TEST_ID"
-      xml_public = <<-XML
-      <!DOCTYPE root PUBLIC "#{@pubid}">
-      <root/>
-      XML
-      @doc_type_public = REXML::Document.new(xml_public).doctype
-
       xml_public_system = <<-XML
       <!DOCTYPE root PUBLIC "#{@pubid}" "#{@sysid}">
       <root/>
@@ -35,11 +29,9 @@ module REXMLTests
       assert_equal([
                      nil,
                      @pubid,
-                     @pubid,
                    ],
                    [
                      @doc_type_system.public,
-                     @doc_type_public.public,
                      @doc_type_public_system.public,
                    ])
     end
@@ -58,12 +50,10 @@ module REXMLTests
     def test_system
       assert_equal([
                      @sysid,
-                     nil,
                      @sysid,
                    ],
                    [
                      @doc_type_system.system,
-                     @doc_type_public.system,
                      @doc_type_public_system.system,
                    ])
     end