first commit

2026-06-04 04:23:02 +00:00 · 2026-02-22 10:42:17 +01:00
commit 55885a0e8f
239 changed files with 103690 additions and 0 deletions
@@ -0,0 +1,265 @@
+package rustdocs
+
+import (
+	"strings"
+	"testing"
+
+	"github.com/PuerkitoBio/goquery"
+)
+
+const testStructPageHTML = `
+<!DOCTYPE html>
+<html>
+<body>
+<main>
+<div class="main-heading">
+  <div class="rustdoc-breadcrumbs"><a href="../index.html">std</a>::<wbr><a href="index.html">simd</a></div>
+  <h1>Struct <span class="struct">Mask</span></h1>
+  <span class="sub-heading"><span class="since" title="Stable since Rust version 1.0.0">1.0.0</span></span>
+</div>
+<pre class="rust item-decl"><code>pub struct Mask&lt;T, const N: usize&gt;(<span class="comment">/* private fields */</span>)</code></pre>
+<details class="toggle top-doc" open="">
+  <div class="docblock"><p>A SIMD vector mask for N elements.</p></div>
+</details>
+<h2 id="implementations">Implementations</h2>
+<details class="toggle implementors-toggle" open="">
+  <details class="toggle method-toggle" open="">
+    <section id="method.test" class="method">
+      <h4 class="code-header">pub fn <a href="#method.test" class="fn">test</a>(&amp;self, index: usize) -&gt; bool</h4>
+    </section>
+    <div class="docblock"><p>Tests the value of the specified element.</p></div>
+  </details>
+</details>
+</main>
+</body>
+</html>
+`
+
+const testSearchHTML = `
+<div id="results">
+<ul class="search-results active">
+<a class="result-method" href="../std/simd/struct.Mask.html#method.test">
+  <span class="result-name">
+    <span class="typename">method</span>
+    <div class="path"><span>std::</span><span>simd::</span><span class="method">Mask::</span><span class="fn">test</span></div>
+  </span>
+  <div class="desc">Tests the value of the specified element.</div>
+</a>
+<a class="result-struct" href="../std/vec/struct.Vec.html">
+  <span class="result-name">
+    <span class="typename">struct</span>
+    <div class="path"><span>std::</span><span>vec::</span><span class="struct">Vec</span></div>
+  </span>
+  <div class="desc">A contiguous growable array type.</div>
+</a>
+<a class="result-fn" href="../std/io/fn.stdout.html">
+  <span class="result-name">
+    <span class="typename">fn</span>
+    <div class="path"><span>std::</span><span>io::</span><span class="fn">stdout</span></div>
+  </span>
+  <div class="desc">Constructs a new handle to the standard output.</div>
+</a>
+</ul>
+</div>
+`
+
+const testCratePageHTML = `
+<!DOCTYPE html>
+<html>
+<body>
+<main>
+<div class="main-heading">
+  <h1>Crate <span>serde</span></h1>
+  <span class="sub-heading"><span class="since">1.0.0</span></span>
+</div>
+<details class="toggle top-doc">
+  <div class="docblock"><p>A framework for serializing and deserializing Rust data structures.</p></div>
+</details>
+<h2 id="modules">Modules</h2>
+<div class="item-table">
+  <div class="module-item"><a class="mod" href="de/index.html">de</a><div class="desc">Deserialize implementation.</div></div>
+</div>
+<h2 id="structs">Structs</h2>
+<div class="item-table">
+  <div class="struct"><a class="struct" href="struct.Serializer.html">Serializer</a><div class="desc">A structure for serializing Rust values.</div></div>
+</div>
+<h2 id="enums">Enums</h2>
+<div class="item-table">
+  <div class="enum"><a class="enum" href="enum.Error.html">Error</a><div class="desc">Errors during serialization.</div></div>
+</div>
+</main>
+</body>
+</html>
+`
+
+func TestParseItemPage(t *testing.T) {
+	parser := NewParser()
+	symbol, err := parser.ParseItemPage(testStructPageHTML, "https://docs.rs/std/simd/struct.Mask.html")
+	if err != nil {
+		t.Fatalf("ParseItemPage failed: %v", err)
+	}
+
+	if symbol.Name != "Mask" {
+		t.Errorf("Expected name 'Mask', got '%s'", symbol.Name)
+	}
+
+	if symbol.Kind != ItemKindStruct {
+		t.Errorf("Expected kind 'struct', got '%s'", symbol.Kind)
+	}
+
+	if symbol.Doc == "" {
+		t.Error("Expected non-empty doc")
+	}
+
+	if !strings.Contains(symbol.Signature, "struct Mask") {
+		t.Errorf("Expected signature to contain 'struct Mask', got '%s'", symbol.Signature)
+	}
+}
+
+func TestParseSearchResults(t *testing.T) {
+	parser := NewParser()
+	results, err := parser.ParseSearchResults(testSearchHTML)
+	if err != nil {
+		t.Fatalf("ParseSearchResults failed: %v", err)
+	}
+
+	if len(results) < 3 {
+		t.Fatalf("Expected at least 3 results, got %d", len(results))
+	}
+
+	method := results[0]
+	if method.Kind != "fn" {
+		t.Errorf("Expected kind 'fn' for method, got '%s'", method.Kind)
+	}
+	if method.Description == "" {
+		t.Error("Expected non-empty description")
+	}
+
+	structResult := results[1]
+	if structResult.Kind != "struct" {
+		t.Errorf("Expected kind 'struct', got '%s'", structResult.Kind)
+	}
+
+	fnResult := results[2]
+	if fnResult.Kind != "fn" {
+		t.Errorf("Expected kind 'fn', got '%s'", fnResult.Kind)
+	}
+}
+
+func TestParseCratePage(t *testing.T) {
+	parser := NewParser()
+	crate, err := parser.ParseCratePage(testCratePageHTML, "https://docs.rs/serde")
+	if err != nil {
+		t.Fatalf("ParseCratePage failed: %v", err)
+	}
+
+	if crate.Name == "" {
+		t.Error("Expected non-empty name")
+	}
+
+	if crate.Description == "" {
+		t.Error("Expected non-empty description")
+	}
+}
+
+func TestExtractKindFromClasses(t *testing.T) {
+	tests := []struct {
+		classes  string
+		expected string
+	}{
+		{"result-struct", "struct"},
+		{"result-enum", "enum"},
+		{"result-trait", "trait"},
+		{"result-fn", "fn"},
+		{"result-macro", "macro"},
+		{"result-const", "const"},
+		{"result-static", "static"},
+		{"result-mod", "mod"},
+		{"result-method", "fn"},
+		{"result-externcrate", "mod"},
+		{"unknown-class", ""},
+	}
+
+	for _, tt := range tests {
+		t.Run(tt.classes, func(t *testing.T) {
+			got := extractKindFromClasses(tt.classes)
+			if got != tt.expected {
+				t.Errorf("extractKindFromClasses(%q) = %q, want %q", tt.classes, got, tt.expected)
+			}
+		})
+	}
+}
+
+func TestResolveURL(t *testing.T) {
+	tests := []struct {
+		base     string
+		href     string
+		expected string
+	}{
+		{"https://docs.rs", "/serde/struct.Serializer.html", "https://docs.rs/serde/struct.Serializer.html"},
+		{"https://docs.rs", "https://example.com/page", "https://example.com/page"},
+	}
+
+	for _, tt := range tests {
+		t.Run(tt.href, func(t *testing.T) {
+			got := resolveURL(tt.base, tt.href)
+			if got != tt.expected {
+				t.Errorf("resolveURL(%q, %q) = %q, want %q", tt.base, tt.href, got, tt.expected)
+			}
+		})
+	}
+}
+
+func TestCleanText(t *testing.T) {
+	tests := []struct {
+		input    string
+		expected string
+	}{
+		{"  hello   world  ", "hello world"},
+		{"single", "single"},
+		{"\n\ttabs\t\n", "tabs"},
+	}
+
+	for _, tt := range tests {
+		t.Run(tt.input, func(t *testing.T) {
+			if got := cleanText(tt.input); got != tt.expected {
+				t.Errorf("cleanText(%q) = %q, want %q", tt.input, got, tt.expected)
+			}
+		})
+	}
+}
+
+func TestExtractItemPath(t *testing.T) {
+	parser := NewParser()
+
+	html := `<div class="rustdoc-breadcrumbs"><a href="../index.html">std</a>::<a href="index.html">simd</a></div>`
+	doc, err := goquery.NewDocumentFromReader(strings.NewReader(html))
+	if err != nil {
+		t.Fatalf("Failed to parse HTML: %v", err)
+	}
+
+	got := parser.extractItemPath(doc, "https://docs.rs/std/simd/struct.Mask.html")
+
+	if !strings.Contains(got, "std") || !strings.Contains(got, "simd") {
+		t.Errorf("extractItemPath() = %q, expected to contain std and simd", got)
+	}
+}
+
+func TestExtractMethods(t *testing.T) {
+	parser := NewParser()
+
+	doc, err := goquery.NewDocumentFromReader(strings.NewReader(testStructPageHTML))
+	if err != nil {
+		t.Fatalf("Failed to parse HTML: %v", err)
+	}
+
+	methods := parser.ExtractMethods(doc)
+	if len(methods) == 0 {
+		t.Error("Expected at least one method")
+		return
+	}
+
+	if methods[0].Name == "" {
+		t.Error("Expected non-empty method name")
+	}
+}