web-crawler/url_test.go

package main

import (
	"slices"
	"testing"
)

func TestNormaliseURL(t *testing.T) {
	t.Parallel()

	wantNormalisedURL := "blog.boot.dev/path"

	cases := []struct {
		name     string
		inputURL string
	}{
		{
			name:     "remove HTTPS scheme",
			inputURL: "https://blog.boot.dev/path",
		},
		{
			name:     "remove HTTP scheme",
			inputURL: "http://blog.boot.dev/path",
		},
		{
			name:     "remove HTTPS scheme with a trailing slash",
			inputURL: "https://blog.boot.dev/path/",
		},
		{
			name:     "remove HTTP scheme with a trailing slash",
			inputURL: "http://blog.boot.dev/path/",
		},
		{
			name:     "remove HTTPS scheme with port 443",
			inputURL: "https://blog.boot.dev:443/path",
		},
		{
			name:     "remove HTTP scheme with port 80",
			inputURL: "http://blog.boot.dev:80/path",
		},
		{
			name:     "normalised URL",
			inputURL: "blog.boot.dev/path",
		},
	}

	for ind, tc := range slices.All(cases) {
		t.Run(tc.name, func(t *testing.T) {
			t.Parallel()

			got, err := normaliseURL(tc.inputURL)
			if err != nil {
				t.Fatalf(
					"Test %d - '%s' FAILED: unexpected error: %v",
					ind,
					tc.name,
					err,
				)
			}

			if got != wantNormalisedURL {
				t.Errorf(
					"Test %d - %s FAILED: unexpected normalised URL returned: want %s, got %s",
					ind,
					tc.name,
					wantNormalisedURL,
					got,
				)
			} else {
				t.Logf(
					"Test %d - %s PASSED: expected normalised URL returned: got %s",
					ind,
					tc.name,
					got,
				)
			}
		})
	}
}

func TestEqualDomains(t *testing.T) {
	t.Parallel()

	cases := []struct {
		name string
		urlA string
		urlB string
		want bool
	}{
		{
			name: "Same domain, different paths",
			urlA: "https://example.com/news",
			urlB: "https://example.com/about/contact",
			want: true,
		},
		{
			name: "Different domains, same path",
			urlA: "http://example.com/blog",
			urlB: "http://example.org/blog",
			want: false,
		},
		{
			name: "Same domain, different protocols",
			urlA: "http://code.person.me.uk/projects/orion",
			urlB: "https://code.person.me.uk/user/person/README.md",
			want: true,
		},
	}

	for ind, tc := range slices.All(cases) {
		t.Run(tc.name, testEqualDomains(ind+1, tc.name, tc.urlA, tc.urlB, tc.want))
	}
}

func testEqualDomains(testNum int, testName, urlA, urlB string, want bool) func(t *testing.T) {
	return func(t *testing.T) {
		t.Parallel()

		got, err := equalDomains(urlA, urlB)
		if err != nil {
			t.Fatalf(
				"Test %d - '%s' FAILED: unexpected error: %v",
				testNum,
				testName,
				err,
			)
		}

		if got != want {
			t.Errorf(
				"Test %d - '%s' FAILED: unexpected domain comparison received: want %t, got %t",
				testNum,
				testName,
				want,
				got,
			)
		} else {
			t.Logf(
				"Test %d - '%s' PASSED: expected domain comparison received: got %t",
				testNum,
				testName,
				got,
			)
		}
	}
}