breakpilot-lehrer/edu-search-service/internal/robots/robots_test.go

package robots

import (
	"context"
	"net/http"
	"net/http/httptest"
	"testing"
)

func TestNewChecker(t *testing.T) {
	checker := NewChecker("TestBot/1.0")
	if checker == nil {
		t.Fatal("Expected non-nil checker")
	}
}

func TestIsAllowed_NoRobots(t *testing.T) {
	// Server that returns 404 for robots.txt
	server := httptest.NewServer(http.HandlerFunc(func(w http.ResponseWriter, r *http.Request) {
		w.WriteHeader(http.StatusNotFound)
	}))
	defer server.Close()

	checker := NewChecker("TestBot/1.0")
	allowed, err := checker.IsAllowed(context.Background(), server.URL+"/some/page")

	if err != nil {
		t.Errorf("Unexpected error: %v", err)
	}
	if !allowed {
		t.Error("Should be allowed when robots.txt doesn't exist")
	}
}

func TestIsAllowed_AllowAll(t *testing.T) {
	robotsTxt := `User-agent: *
Allow: /
`
	server := httptest.NewServer(http.HandlerFunc(func(w http.ResponseWriter, r *http.Request) {
		if r.URL.Path == "/robots.txt" {
			w.Write([]byte(robotsTxt))
			return
		}
		w.WriteHeader(http.StatusOK)
	}))
	defer server.Close()

	checker := NewChecker("TestBot/1.0")
	allowed, _ := checker.IsAllowed(context.Background(), server.URL+"/any/path")

	if !allowed {
		t.Error("Should be allowed with Allow: /")
	}
}

func TestIsAllowed_DisallowPath(t *testing.T) {
	robotsTxt := `User-agent: *
Disallow: /private/
Disallow: /admin/
`
	server := httptest.NewServer(http.HandlerFunc(func(w http.ResponseWriter, r *http.Request) {
		if r.URL.Path == "/robots.txt" {
			w.Write([]byte(robotsTxt))
			return
		}
		w.WriteHeader(http.StatusOK)
	}))
	defer server.Close()

	checker := NewChecker("TestBot/1.0")

	// Should be disallowed
	allowed, _ := checker.IsAllowed(context.Background(), server.URL+"/private/secret")
	if allowed {
		t.Error("/private/secret should be disallowed")
	}

	allowed, _ = checker.IsAllowed(context.Background(), server.URL+"/admin/users")
	if allowed {
		t.Error("/admin/users should be disallowed")
	}

	// Should be allowed
	allowed, _ = checker.IsAllowed(context.Background(), server.URL+"/public/page")
	if !allowed {
		t.Error("/public/page should be allowed")
	}
}

func TestIsAllowed_AllowTakesPrecedence(t *testing.T) {
	robotsTxt := `User-agent: *
Disallow: /api/
Allow: /api/public/
`
	server := httptest.NewServer(http.HandlerFunc(func(w http.ResponseWriter, r *http.Request) {
		if r.URL.Path == "/robots.txt" {
			w.Write([]byte(robotsTxt))
			return
		}
		w.WriteHeader(http.StatusOK)
	}))
	defer server.Close()

	checker := NewChecker("TestBot/1.0")

	// Allow takes precedence
	allowed, _ := checker.IsAllowed(context.Background(), server.URL+"/api/public/docs")
	if !allowed {
		t.Error("/api/public/docs should be allowed (Allow takes precedence)")
	}

	// Still disallowed
	allowed, _ = checker.IsAllowed(context.Background(), server.URL+"/api/internal")
	if allowed {
		t.Error("/api/internal should be disallowed")
	}
}

func TestIsAllowed_SpecificUserAgent(t *testing.T) {
	robotsTxt := `User-agent: BadBot
Disallow: /

User-agent: *
Allow: /
`
	server := httptest.NewServer(http.HandlerFunc(func(w http.ResponseWriter, r *http.Request) {
		if r.URL.Path == "/robots.txt" {
			w.Write([]byte(robotsTxt))
			return
		}
		w.WriteHeader(http.StatusOK)
	}))
	defer server.Close()

	checker := NewChecker("GoodBot/1.0")
	allowed, _ := checker.IsAllowed(context.Background(), server.URL+"/page")

	if !allowed {
		t.Error("GoodBot should be allowed")
	}
}

func TestGetCrawlDelay(t *testing.T) {
	robotsTxt := `User-agent: *
Crawl-delay: 5
`
	server := httptest.NewServer(http.HandlerFunc(func(w http.ResponseWriter, r *http.Request) {
		if r.URL.Path == "/robots.txt" {
			w.Write([]byte(robotsTxt))
			return
		}
		w.WriteHeader(http.StatusOK)
	}))
	defer server.Close()

	checker := NewChecker("TestBot/1.0")
	delay, err := checker.GetCrawlDelay(context.Background(), server.URL+"/page")

	if err != nil {
		t.Errorf("Unexpected error: %v", err)
	}
	if delay != 5 {
		t.Errorf("Expected delay 5, got %d", delay)
	}
}

func TestMatchPattern_Simple(t *testing.T) {
	tests := []struct {
		pattern string
		path    string
		match   bool
	}{
		{"/private/", "/private/secret", true},
		{"/private/", "/public/", false},
		{"/", "/anything", true},
		{"", "/anything", false},
	}

	for _, tt := range tests {
		result := matchPattern(tt.pattern, tt.path)
		if result != tt.match {
			t.Errorf("Pattern '%s' vs Path '%s': expected %v, got %v",
				tt.pattern, tt.path, tt.match, result)
		}
	}
}

func TestMatchPattern_Wildcard(t *testing.T) {
	tests := []struct {
		pattern string
		path    string
		match   bool
	}{
		{"/*.pdf", "/document.pdf", true},
		{"/*.pdf", "/folder/doc.pdf", true},
		{"/*.pdf", "/document.html", false},
		{"/dir/*/page", "/dir/sub/page", true},
		{"/dir/*/page", "/dir/other/page", true},
	}

	for _, tt := range tests {
		result := matchPattern(tt.pattern, tt.path)
		if result != tt.match {
			t.Errorf("Pattern '%s' vs Path '%s': expected %v, got %v",
				tt.pattern, tt.path, tt.match, result)
		}
	}
}

func TestMatchPattern_EndAnchor(t *testing.T) {
	tests := []struct {
		pattern string
		path    string
		match   bool
	}{
		{"/exact$", "/exact", true},
		{"/exact$", "/exactmore", false},
		{"/exact$", "/exact/more", false},
	}

	for _, tt := range tests {
		result := matchPattern(tt.pattern, tt.path)
		if result != tt.match {
			t.Errorf("Pattern '%s' vs Path '%s': expected %v, got %v",
				tt.pattern, tt.path, tt.match, result)
		}
	}
}

func TestCacheStats(t *testing.T) {
	robotsTxt := `User-agent: *
Allow: /
`
	server := httptest.NewServer(http.HandlerFunc(func(w http.ResponseWriter, r *http.Request) {
		w.Write([]byte(robotsTxt))
	}))
	defer server.Close()

	checker := NewChecker("TestBot/1.0")

	// Initially empty
	count, _ := checker.CacheStats()
	if count != 0 {
		t.Errorf("Expected 0 cached entries, got %d", count)
	}

	// Fetch robots.txt
	checker.IsAllowed(context.Background(), server.URL+"/page")

	// Should have 1 entry
	count, hosts := checker.CacheStats()
	if count != 1 {
		t.Errorf("Expected 1 cached entry, got %d", count)
	}
	if len(hosts) != 1 {
		t.Errorf("Expected 1 host, got %v", hosts)
	}
}

func TestClearCache(t *testing.T) {
	robotsTxt := `User-agent: *
Allow: /
`
	server := httptest.NewServer(http.HandlerFunc(func(w http.ResponseWriter, r *http.Request) {
		w.Write([]byte(robotsTxt))
	}))
	defer server.Close()

	checker := NewChecker("TestBot/1.0")

	// Populate cache
	checker.IsAllowed(context.Background(), server.URL+"/page")

	count, _ := checker.CacheStats()
	if count != 1 {
		t.Errorf("Expected 1 cached entry, got %d", count)
	}

	// Clear cache
	checker.ClearCache()

	count, _ = checker.CacheStats()
	if count != 0 {
		t.Errorf("Expected 0 cached entries after clear, got %d", count)
	}
}

func TestParseRobotsTxt_Comments(t *testing.T) {
	robotsTxt := `# This is a comment
User-agent: *
# Another comment
Disallow: /private/  # inline comment
Allow: /public/
`
	server := httptest.NewServer(http.HandlerFunc(func(w http.ResponseWriter, r *http.Request) {
		if r.URL.Path == "/robots.txt" {
			w.Write([]byte(robotsTxt))
			return
		}
		w.WriteHeader(http.StatusOK)
	}))
	defer server.Close()

	checker := NewChecker("TestBot/1.0")

	allowed, _ := checker.IsAllowed(context.Background(), server.URL+"/public/page")
	if !allowed {
		t.Error("/public/page should be allowed")
	}

	allowed, _ = checker.IsAllowed(context.Background(), server.URL+"/private/page")
	if allowed {
		t.Error("/private/page should be disallowed")
	}
}

func TestIsAllowed_InvalidURL(t *testing.T) {
	checker := NewChecker("TestBot/1.0")

	_, err := checker.IsAllowed(context.Background(), "not a valid url ://")
	if err == nil {
		t.Error("Expected error for invalid URL")
	}
}