converter: allow base domain with url scheme

JohannesKaufmann · Nov 17, 2024 · 4d6b2ff · 4d6b2ff
1 parent 5af1444
commit 4d6b2ff
Show file tree

Hide file tree

Showing 3 changed files with 60 additions and 17 deletions.
diff --git a/cli/cmd/exec_test.go b/cli/cmd/exec_test.go
@@ -454,15 +454,14 @@ func TestExecute_General(t *testing.T) {
 
 			expectedStdout: []byte("![](http://example.com/image.png)\n"),
 		},
-		// TODO: with https domain
-		// {
-		// 	desc: "[domain] with https domain",
-		//
-		// 	inputStdin: []byte(`<img src="/image.png" />`),
-		// 	inputArgs:  []string{"html2markdown", "--domain", "https://example.com"},
-		//
-		// 	expectedStdout: []byte("![](https://example.com/image.png)\n"),
-		// },
+		{
+			desc: "[domain] with https domain",
+
+			inputStdin: []byte(`<img src="/image.png" />`),
+			inputArgs:  []string{"html2markdown", "--domain", "https://example.com"},
+
+			expectedStdout: []byte("![](https://example.com/image.png)\n"),
+		},
 
 		// - - - - - selectors - - - - - //
 		{

diff --git a/converter/url.go b/converter/url.go
@@ -16,6 +16,25 @@ var percentEncodingReplacer = strings.NewReplacer(
 	">", "%3E",
 )
 
+func parseBaseDomain(rawDomain string) *url.URL {
+	if rawDomain == "" {
+		return nil
+	}
+
+	u1, err := url.Parse(rawDomain)
+	if err == nil && u1.Host != "" {
+		// Yes, we got valid domain (probably with a http/https scheme)
+		return u1
+	}
+
+	u2, err := url.Parse("http://" + rawDomain)
+	if err == nil && u2.Host != "" {
+		// Yes, we got a valid domain (by choosing a fallback scheme)
+		return u2
+	}
+
+	return nil
+}
 func defaultAssembleAbsoluteURL(tagName string, rawURL string, domain string) string {
 	rawURL = strings.TrimSpace(rawURL)
 
@@ -51,13 +70,10 @@ func defaultAssembleAbsoluteURL(tagName string, rawURL string, domain string) st
 	// e.g. the email reading "Hi+Johannes" instead of "Hi Johannes"
 	u.RawQuery = strings.ReplaceAll(u.RawQuery, "+", "%20")
 
-	if domain != "" {
-		if u.Scheme == "" {
-			u.Scheme = "http"
-		}
-		if u.Host == "" {
-			u.Host = domain
-		}
+	if base := parseBaseDomain(domain); base != nil {
+		// If a "domain" is provided, we use that to convert relative links
+		// to absolute links.
+		u = base.ResolveReference(u)
 	}
 
 	return percentEncodingReplacer.Replace(u.String())

diff --git a/converter/url_test.go b/converter/url_test.go
@@ -66,6 +66,34 @@ func TestDefaultAssembleAbsoluteURL(t *testing.T) {
 
 			expected: "http://test.com/page.html?key=val#hash",
 		},
+		{
+			desc: "with http domain",
+
+			tagName: "a",
+			input:   "/page.html?key=val#hash",
+			domain:  "http://test.com",
+
+			expected: "http://test.com/page.html?key=val#hash",
+		},
+		{
+			desc: "with https domain",
+
+			tagName: "a",
+			input:   "/page.html?key=val#hash",
+			domain:  "https://test.com",
+
+			expected: "https://test.com/page.html?key=val#hash",
+		},
+		{
+			desc: "with domain that includes path",
+
+			tagName: "a",
+			input:   "/page.html?key=val#hash",
+			domain:  "https://test.com/random_stuff",
+
+			expected: "https://test.com/page.html?key=val#hash",
+		},
+
 		{
 			desc: "data uri",
 
@@ -223,7 +251,7 @@ func TestDefaultAssembleAbsoluteURL(t *testing.T) {
 	}
 }
 
-func TestParseAndEncode(t *testing.T) {
+func TestParseAndEncodeQuery(t *testing.T) {
 	runs := []struct {
 		desc string