Merge pull request #15 from webexpress-framework/copilot/improve-reverse-html-renderer

Add missing factory mappings and comprehensive parser tests

Author: ReneSchwarzer

src/WebExpress.WebCore.Test/Html/Parser/UnitTestHtmlElementFactory.cs

@@ -113,5 +113,38 @@ public void NullTagName_Throws()
         {
             Assert.Throws<System.ArgumentNullException>(() => HtmlElementFactory.Create(null));
         }
+
+        /// <summary>
+        /// Both 'kbd' and 'kdb' map to the existing <see cref="HtmlElementTextSemanticsKdb"/>.
+        /// </summary>
+        [Fact]
+        public void KbdTag_MapsToKdbElement()
+        {
+            var element = HtmlElementFactory.Create("kbd");
+
+            Assert.IsType<HtmlElementTextSemanticsKdb>(element);
+        }
+
+        /// <summary>
+        /// The 'keygen' tag is resolved to <see cref="HtmlElementFormKeygen"/>.
+        /// </summary>
+        [Fact]
+        public void KnownTag_Keygen_ReturnsCorrectType()
+        {
+            var element = HtmlElementFactory.Create("keygen");
+
+            Assert.IsType<HtmlElementFormKeygen>(element);
+        }
+
+        /// <summary>
+        /// The 'command' tag is resolved to <see cref="HtmlElementInteractiveCommand"/>.
+        /// </summary>
+        [Fact]
+        public void KnownTag_Command_ReturnsCorrectType()
+        {
+            var element = HtmlElementFactory.Create("command");
+
+            Assert.IsType<HtmlElementInteractiveCommand>(element);
+        }
     }
 }

src/WebExpress.WebCore.Test/Html/Parser/UnitTestHtmlParser.cs

@@ -340,5 +340,161 @@ public void RoundTrip_Img_ProducesEquivalentHtml()
             // validation
             Assert.Equal(html, restoredHtml);
         }
+
+        // ------------------------------------------------------------------
+        // Additional tests
+        // ------------------------------------------------------------------
+
+        /// <summary>
+        /// ParseSingle returns the first node.
+        /// </summary>
+        [Fact]
+        public void ParseSingle_ReturnsFirstNode()
+        {
+            var node = Parser.ParseSingle("<div></div>");
+
+            Assert.IsType<HtmlElementTextContentDiv>(node);
+        }
+
+        /// <summary>
+        /// ParseSingle returns null for an empty input.
+        /// </summary>
+        [Fact]
+        public void ParseSingle_EmptyInput_ReturnsNull()
+        {
+            var node = Parser.ParseSingle("");
+
+            Assert.Null(node);
+        }
+
+        /// <summary>
+        /// An element with an inline style attribute retains its value.
+        /// </summary>
+        [Fact]
+        public void InlineStyleAttribute_IsPreserved()
+        {
+            var nodes = Parser.Parse("<div style=\"color: red;\"></div>");
+            var div = nodes.OfType<HtmlElementTextContentDiv>().Single();
+
+            Assert.Equal("color: red;", div.Style);
+        }
+
+        /// <summary>
+        /// A table structure with thead, tbody, and rows is correctly reconstructed.
+        /// </summary>
+        [Fact]
+        public void TableStructure_IsReconstructed()
+        {
+            var nodes = Parser.Parse("<table><thead><tr><th>Header</th></tr></thead><tbody><tr><td>Cell</td></tr></tbody></table>");
+            var table = nodes.OfType<HtmlElementTableTable>().Single();
+            var thead = table.Elements.OfType<HtmlElementTableThead>().Single();
+            var tbody = table.Elements.OfType<HtmlElementTableTbody>().Single();
+
+            Assert.NotNull(thead);
+            Assert.NotNull(tbody);
+        }
+
+        /// <summary>
+        /// Multiple top-level elements are all returned.
+        /// </summary>
+        [Fact]
+        public void MultipleRoots_AreAllReturned()
+        {
+            var nodes = Parser.Parse("<p>one</p><p>two</p>");
+
+            Assert.Equal(2, nodes.Count);
+            Assert.All(nodes, n => Assert.IsType<HtmlElementTextContentP>(n));
+        }
+
+        /// <summary>
+        /// A mismatched end tag is handled gracefully without throwing.
+        /// </summary>
+        [Fact]
+        public void MismatchedEndTag_IsHandledGracefully()
+        {
+            var nodes = Parser.Parse("<div><span>text</div>");
+
+            var div = nodes.OfType<HtmlElementTextContentDiv>().Single();
+            Assert.NotNull(div);
+        }
+
+        /// <summary>
+        /// Mixed text and element children are preserved in order.
+        /// </summary>
+        [Fact]
+        public void MixedContent_TextAndElements_ArePreserved()
+        {
+            var nodes = Parser.Parse("<p>Hello <strong>World</strong>!</p>");
+            var p = nodes.OfType<HtmlElementTextContentP>().Single();
+
+            Assert.Equal(3, p.Elements.Count());
+        }
+
+        /// <summary>
+        /// Roundtrip of a styled element preserves the style attribute.
+        /// </summary>
+        [Fact]
+        public void RoundTrip_StyleAttribute_IsPreserved()
+        {
+            // arrange
+            var original = new HtmlElementTextContentDiv();
+            original.Style = "color: red;";
+
+            // act
+            var html = original.ToString().Trim();
+            var parsed = Parser.Parse(html);
+
+            var restored = parsed.OfType<HtmlElementTextContentDiv>().Single();
+
+            // validation
+            Assert.Equal("color: red;", restored.Style);
+        }
+
+        /// <summary>
+        /// Roundtrip of an anchor element preserves href and text content.
+        /// </summary>
+        [Fact]
+        public void RoundTrip_Anchor_PreservesHrefAndText()
+        {
+            // arrange
+            var original = new HtmlElementTextSemanticsA(new HtmlText("click me"));
+            original.Href = "https://example.com";
+
+            // act
+            var html = original.ToString().Trim();
+            var parsed = Parser.Parse(html);
+
+            var a = parsed.OfType<HtmlElementTextSemanticsA>().Single();
+            var text = a.Elements.OfType<HtmlText>().Single();
+
+            // validation
+            Assert.Equal("https://example.com", a.Href);
+            Assert.Equal("click me", text.Value);
+        }
+
+        /// <summary>
+        /// A form with input fields is correctly reconstructed.
+        /// </summary>
+        [Fact]
+        public void FormWithInputs_IsReconstructed()
+        {
+            var nodes = Parser.Parse("<form action=\"/submit\"><input type=\"text\" name=\"q\"></form>");
+            var form = nodes.OfType<HtmlElementFormForm>().Single();
+            var input = form.Elements.OfType<HtmlElementFieldInput>().Single();
+
+            Assert.NotNull(input);
+        }
+
+        /// <summary>
+        /// The kbd tag (standard HTML) maps to HtmlElementTextSemanticsKdb.
+        /// </summary>
+        [Fact]
+        public void KbdTag_MapsToKdbElement()
+        {
+            var nodes = Parser.Parse("<kbd>Ctrl+C</kbd>");
+            var kbd = nodes.OfType<HtmlElementTextSemanticsKdb>().Single();
+
+            Assert.NotNull(kbd);
+        }
     }
 }

src/WebExpress.WebCore.Test/Html/Parser/UnitTestHtmlTokenizer.cs

@@ -246,5 +246,76 @@ public void TagNameNormalisation_IsLowerCase()
             Assert.Equal("div", tokens[0].TagName);
             Assert.Equal("class", tokens[0].Attributes[0].Name);
         }
+
+        // ------------------------------------------------------------------
+        // Whitespace and edge cases
+        // ------------------------------------------------------------------
+
+        /// <summary>
+        /// Whitespace-only text between tags is preserved as a text token.
+        /// </summary>
+        [Fact]
+        public void WhitespaceText_IsPreservedAsTextToken()
+        {
+            var tokens = Tokenize("<div> </div>");
+
+            Assert.Equal(HtmlTokenType.StartTag, tokens[0].Type);
+            Assert.Equal(HtmlTokenType.Text, tokens[1].Type);
+            Assert.Equal(" ", tokens[1].Value);
+            Assert.Equal(HtmlTokenType.EndTag, tokens[2].Type);
+        }
+
+        /// <summary>
+        /// An unquoted attribute value is read until whitespace or closing bracket.
+        /// </summary>
+        [Fact]
+        public void UnquotedAttributeValue_IsExtracted()
+        {
+            var tokens = Tokenize("<div class=foo>");
+
+            var attr = tokens[0].Attributes.Single();
+            Assert.Equal("class", attr.Name);
+            Assert.Equal("foo", attr.Value);
+        }
+
+        /// <summary>
+        /// An inline style attribute is preserved in its entirety.
+        /// </summary>
+        [Fact]
+        public void InlineStyleAttribute_IsPreserved()
+        {
+            var tokens = Tokenize("<div style=\"color: red; font-size: 14px;\">");
+
+            var attr = tokens[0].Attributes.Single();
+            Assert.Equal("style", attr.Name);
+            Assert.Equal("color: red; font-size: 14px;", attr.Value);
+        }
+
+        /// <summary>
+        /// A stray less-than character is emitted as text.
+        /// </summary>
+        [Fact]
+        public void StrayLessThan_IsEmittedAsText()
+        {
+            var tokens = Tokenize("a < b");
+
+            Assert.Equal(HtmlTokenType.Text, tokens[0].Type);
+            Assert.Equal("a ", tokens[0].Value);
+            Assert.Equal(HtmlTokenType.Text, tokens[1].Type);
+            Assert.Equal("<", tokens[1].Value);
+            Assert.Equal(HtmlTokenType.Text, tokens[2].Type);
+        }
+
+        /// <summary>
+        /// A keygen void element without slash is emitted as self-closing.
+        /// </summary>
+        [Fact]
+        public void KeygenVoidElement_ReturnsSelfClosing()
+        {
+            var tokens = Tokenize("<keygen>");
+
+            Assert.Equal(HtmlTokenType.SelfClosingTag, tokens[0].Type);
+            Assert.Equal("keygen", tokens[0].TagName);
+        }
     }
 }

src/WebExpress.WebCore/WebHtml/Parser/HtmlElementFactory.cs

@@ -77,6 +77,10 @@ public class HtmlElementFactory
                 ["dfn"] = () => new HtmlElementTextSemanticsDfn(),
                 ["em"] = () => new HtmlElementTextSemanticsEm(),
                 ["i"] = () => new HtmlElementTextSemanticsI(),
+                // The standard HTML element is <kbd>, but the existing class uses "kdb" as
+                // its element name.  Both spellings are mapped so that the parser handles
+                // real-world HTML (<kbd>) as well as the project's own renderer output (<kdb>).
+                ["kbd"] = () => new HtmlElementTextSemanticsKdb(),
                 ["kdb"] = () => new HtmlElementTextSemanticsKdb(),
                 // 'kbd' is the correct HTML tag name; 'kdb' mirrors the existing class typo.
                 ["kbd"] = () => new HtmlElementTextSemanticsKdb(),
@@ -140,6 +144,7 @@ public class HtmlElementFactory
                 ["datalist"] = () => new HtmlElementFormDatalist(),
                 ["fieldset"] = () => new HtmlElementFormFieldset(),
                 ["form"] = () => new HtmlElementFormForm(),
+                ["keygen"] = () => new HtmlElementFormKeygen(),
                 ["meter"] = () => new HtmlElementFormMeter(),
                 ["optgroup"] = () => new HtmlElementFormOptgroup(),
                 ["option"] = () => new HtmlElementFormOption(),
@@ -148,6 +153,7 @@ public class HtmlElementFactory
                 ["textarea"] = () => new HtmlElementFormTextarea(),
 
                 // Interactive
+                ["command"] = () => new HtmlElementInteractiveCommand(),
                 ["details"] = () => new HtmlElementInteractiveDetails(),
                 ["menu"] = () => new HtmlElementInteractiveMenu(),
                 ["summary"] = () => new HtmlElementInteractiveSummary(),

src/WebExpress.WebCore/WebHtml/Parser/HtmlTokenizer.cs

@@ -28,7 +28,7 @@ public class HtmlTokenizer
             new(StringComparer.OrdinalIgnoreCase)
             {
                 "area", "base", "br", "col", "embed", "hr", "img", "input",
-                "link", "meta", "param", "source", "track", "wbr"
+                "keygen", "link", "meta", "param", "source", "track", "wbr"
             };
 
         /// <summary>