hako/internal/archival/rules/url_analyzer_test.go

package rules

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

func TestAnalyzeURL(t *testing.T) {
	tests := []struct {
		name           string
		serverResponse func(w http.ResponseWriter)
		url            string
		wantErr        bool
		check          func(t *testing.T, metadata *URLMetadata)
	}{
		{
			name: "successful HEAD request",
			serverResponse: func(w http.ResponseWriter) {
				w.Header().Set("Content-Type", "text/html; charset=utf-8")
				w.WriteHeader(http.StatusOK)
			},
			url:     "", // Will be replaced with server URL
			wantErr: false,
			check: func(t *testing.T, metadata *URLMetadata) {
				// Domain will be from test server (127.0.0.1), so just check it's not empty
				if metadata.Domain == "" {
					t.Error("Domain is empty")
				}
				if metadata.MimeType != "text/html" {
					t.Errorf("MimeType = %q, want %q", metadata.MimeType, "text/html")
				}
			},
		},
		{
			name: "content type without charset",
			serverResponse: func(w http.ResponseWriter) {
				w.Header().Set("Content-Type", "application/pdf")
				w.WriteHeader(http.StatusOK)
			},
			url:     "http://example.com/file.pdf",
			wantErr: false,
			check: func(t *testing.T, metadata *URLMetadata) {
				if metadata.MimeType != "application/pdf" {
					t.Errorf("MimeType = %q, want %q", metadata.MimeType, "application/pdf")
				}
			},
		},
		{
			name: "domain with port",
			serverResponse: func(w http.ResponseWriter) {
				w.Header().Set("Content-Type", "text/html")
				w.WriteHeader(http.StatusOK)
			},
			url:     "", // Will be replaced with server URL
			wantErr: false,
			check: func(t *testing.T, metadata *URLMetadata) {
				// Domain will be from test server, so just check it's not empty
				if metadata.Domain == "" {
					t.Error("Domain is empty")
				}
			},
		},
		{
			name: "no content type header",
			serverResponse: func(w http.ResponseWriter) {
				w.WriteHeader(http.StatusOK)
			},
			url:     "http://example.com/page",
			wantErr: false,
			check: func(t *testing.T, metadata *URLMetadata) {
				if metadata.MimeType != "" {
					t.Errorf("MimeType = %q, want empty", metadata.MimeType)
				}
			},
		},
		{
			name: "invalid URL",
			serverResponse: func(w http.ResponseWriter) {
				w.WriteHeader(http.StatusOK)
			},
			url:     "not-a-valid-url",
			wantErr: true,
		},
		{
			name: "server error",
			serverResponse: func(w http.ResponseWriter) {
				w.WriteHeader(http.StatusInternalServerError)
			},
			url:     "http://example.com/page",
			wantErr: false, // HEAD request succeeds even with error status
			check: func(t *testing.T, metadata *URLMetadata) {
				// Should still get metadata even if server returns error
				if metadata == nil {
					t.Fatal("metadata is nil")
				}
			},
		},
	}

	for _, tt := range tests {
		t.Run(tt.name, func(t *testing.T) {
			// Create test server
			server := httptest.NewServer(http.HandlerFunc(func(w http.ResponseWriter, r *http.Request) {
				if r.Method != http.MethodHead {
					t.Errorf("Expected HEAD request, got %s", r.Method)
				}
				tt.serverResponse(w)
			}))
			defer server.Close()

			// Use server URL if url is not explicitly invalid
			testURL := tt.url
			if tt.url != "not-a-valid-url" {
				testURL = server.URL + "/test"
			}

			ctx := context.Background()
			metadata, err := AnalyzeURL(ctx, testURL)
			if (err != nil) != tt.wantErr {
				t.Errorf("AnalyzeURL() error = %v, wantErr %v", err, tt.wantErr)
				return
			}
			if tt.wantErr {
				return
			}
			if metadata == nil {
				t.Fatal("AnalyzeURL() returned nil metadata")
			}
			if tt.check != nil {
				tt.check(t, metadata)
			}
		})
	}
}

func TestAnalyzeURL_DomainExtraction(t *testing.T) {
	tests := []struct {
		name       string
		url        string
		wantDomain string
	}{
		{
			name:       "simple domain",
			url:        "http://example.com",
			wantDomain: "example.com",
		},
		{
			name:       "domain with port",
			url:        "http://example.com:8080",
			wantDomain: "example.com",
		},
		{
			name:       "subdomain",
			url:        "http://api.example.com",
			wantDomain: "api.example.com",
		},
		{
			name:       "https",
			url:        "https://example.com",
			wantDomain: "example.com",
		},
	}

	for _, tt := range tests {
		t.Run(tt.name, func(t *testing.T) {
			// Create test server
			server := httptest.NewServer(http.HandlerFunc(func(w http.ResponseWriter, r *http.Request) {
				w.WriteHeader(http.StatusOK)
			}))
			defer server.Close()

			// Replace server URL with our test URL
			ctx := context.Background()
			metadata, err := AnalyzeURL(ctx, tt.url)
			if err != nil {
				// If URL parsing fails, that's expected for some test cases
				// We'll test domain extraction separately
				return
			}
			if metadata.Domain != tt.wantDomain {
				t.Errorf("Domain = %q, want %q", metadata.Domain, tt.wantDomain)
			}
		})
	}
}