Add extractUrls function

main.go

@@ -30,6 +30,33 @@ func main() {
 	html.Parse(strings.NewReader(mark.Render(string(bs))))
 }
 
+func extractUrls(n *html.Node) map[string]bool {
+	ss := make(map[string]bool)
+	ns := make([]*html.Node, 0, 1024)
+	ns = append(ns, n)
+
+	for len(ns) > 0 {
+		i := len(ns) - 1
+		n := ns[i]
+		ns = ns[:i]
+
+		if n.Type == html.ElementNode && n.Data == "a" {
+			for _, a := range n.Attr {
+				if a.Key == "href" {
+					ss[a.Val] = true
+					break
+				}
+			}
+		}
+
+		for n := n.FirstChild; n != nil; n = n.NextSibling {
+			ns = append(ns, n)
+		}
+	}
+
+	return ss
+}
+
 func getArgs() map[string]interface{} {
 	usage := `Link checker for Markdown and HTML
 

main_test.go

@@ -0,0 +1,41 @@
+package main
+
+import (
+	"strings"
+	"testing"
+
+	"github.com/stretchr/testify/assert"
+	"golang.org/x/net/html"
+)
+
+func TestExtractUrls(t *testing.T) {
+	for _, c := range []struct {
+		html    string
+		numUrls int
+	}{
+		{`<a href="https://google.com">Google</a>`, 1},
+		{
+			`
+				<div>
+					<a href="https://google.com">Google</a>
+					<a href="https://google.com">Google</a>
+				</div>
+			`,
+			1,
+		},
+		{
+			`
+				<div>
+					<a href="https://google.com">Google</a>
+					<a href="https://yahoo.com">Yahoo!</a>
+				</div>
+			`,
+			2,
+		},
+	} {
+		n, err := html.Parse(strings.NewReader(c.html))
+
+		assert.Equal(t, nil, err)
+		assert.Equal(t, c.numUrls, len(extractUrls(n)))
+	}
+}