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 +1,115 @@
+package javadocs
+
+import (
+	"strings"
+	"testing"
+
+	"github.com/PuerkitoBio/goquery"
+)
+
+const testPackagePageHTML = `
+<!DOCTYPE html>
+<html>
+<body>
+<h1>Package java.util</h1>
+<div class="block">Contains the collections framework, legacy collection classes, event model, date and time facilities.</div>
+
+<table class="type-summary">
+<tr>
+<td><a href="ArrayList.html">ArrayList</a></td>
+<td>Resizable-array implementation of the List interface.</td>
+</tr>
+<tr>
+<td><a href="HashMap.html">HashMap</a></td>
+<td>Hash table based implementation of the Map interface.</td>
+</tr>
+</table>
+
+<table class="interface-summary">
+<tr>
+<td><a href="List.html">List</a></td>
+<td>An ordered collection (also known as a sequence).</td>
+</tr>
+</table>
+
+<table class="exception-summary">
+<tr>
+<td><a href="ConcurrentModificationException.html">ConcurrentModificationException</a></td>
+<td>This exception may be thrown by methods that detect concurrent modification.</td>
+</tr>
+</table>
+</body>
+</html>
+`
+
+func TestParsePackagePage(t *testing.T) {
+	parser := NewParser()
+	pkg, err := parser.ParsePackagePage(testPackagePageHTML, "https://docs.oracle.com/en/java/javase/17/docs/api/java.base/java/util/package-summary.html")
+	if err != nil {
+		t.Fatalf("ParsePackagePage failed: %v", err)
+	}
+
+	if pkg.Name == "" {
+		t.Error("Expected non-empty package name")
+	}
+
+	if pkg.Doc == "" {
+		t.Error("Expected non-empty doc")
+	}
+
+	if len(pkg.Classes) == 0 {
+		t.Error("Expected at least one class")
+	}
+
+	if len(pkg.Interfaces) == 0 {
+		t.Error("Expected at least one interface")
+	}
+
+	if len(pkg.Exceptions) == 0 {
+		t.Error("Expected at least one exception")
+	}
+}
+
+func TestExtractClasses(t *testing.T) {
+	parser := NewParser()
+
+	doc, err := goquery.NewDocumentFromReader(strings.NewReader(testPackagePageHTML))
+	if err != nil {
+		t.Fatalf("Failed to parse HTML: %v", err)
+	}
+
+	classes := parser.extractClasses(doc, "java.util", "https://docs.oracle.com/test")
+
+	if len(classes) == 0 {
+		t.Fatal("Expected at least one class")
+	}
+
+	first := classes[0]
+	if first.Name == "" {
+		t.Error("Expected non-empty class name")
+	}
+
+	if first.Package != "java.util" {
+		t.Errorf("Expected package 'java.util', got %q", first.Package)
+	}
+}
+
+func TestResolveURL(t *testing.T) {
+	tests := []struct {
+		base     string
+		href     string
+		expected string
+	}{
+		{"https://docs.oracle.com", "/api/ArrayList.html", "https://docs.oracle.com/api/ArrayList.html"},
+		{"https://docs.oracle.com", "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)
+			}
+		})
+	}
+}