Merge pull request #362 from systemli/Include-Fragment-to-ExtractURL

internal/util/url.go

@@ -4,6 +4,6 @@ import "regexp"
 
 // ExtractURLs extracts URLs from a text.
 func ExtractURLs(text string) []string {
-	urlRegex := regexp.MustCompile(`(http[s]?://(?:[a-zA-Z]|[0-9]|[$-_@.&+]|[!*\\(\\),]|(?:%[0-9a-fA-F][0-9a-fA-F]))+)`)
+	urlRegex := regexp.MustCompile(`(https?://(?:[a-zA-Z]|[0-9]|[$-_@.&+]|[!*\\(),]|%[0-9a-fA-F][0-9a-fA-F]|#)+)`)
 	return urlRegex.FindAllString(text, -1)
 }

internal/util/url_test.go

@@ -7,33 +7,41 @@ import (
 )
 
 func TestExtractURL(t *testing.T) {
-	text := "This is a text with a URL https://example.com"
-	urls := ExtractURLs(text)
-
-	assert.Equal(t, 1, len(urls))
-	assert.Equal(t, "https://example.com", urls[0])
-
-	text = "This is a text with a URL https://example.com and another URL http://example.org"
-	urls = ExtractURLs(text)
-
-	assert.Equal(t, 2, len(urls))
-	assert.Equal(t, "https://example.com", urls[0])
-	assert.Equal(t, "http://example.org", urls[1])
-
-	text = "This is a text without a URL"
-	urls = ExtractURLs(text)
-
-	assert.Equal(t, 0, len(urls))
-
-	text = "This is a text with a URL https://www.systemli.org/en/contact/"
-	urls = ExtractURLs(text)
-
-	assert.Equal(t, 1, len(urls))
-	assert.Equal(t, "https://www.systemli.org/en/contact/", urls[0])
-
-	text = "This is a text with a URL https://www.systemli.org/en/contact/?key=value"
-	urls = ExtractURLs(text)
-
-	assert.Equal(t, 1, len(urls))
-	assert.Equal(t, "https://www.systemli.org/en/contact/?key=value", urls[0])
+	testCases := []struct {
+		text     string
+		expected []string
+	}{
+		{
+			"This is a text with a URL https://example.com",
+			[]string{"https://example.com"},
+		},
+		{
+			"This is a text with a URL https://example.com and another URL http://example.org",
+			[]string{"https://example.com", "http://example.org"},
+		},
+		{
+			"This is a text without a URL",
+			[]string{},
+		},
+		{
+			"This is a text with a URL https://www.systemli.org/en/contact/",
+			[]string{"https://www.systemli.org/en/contact/"},
+		},
+		{
+			"This is a text with a URL https://www.systemli.org/en/contact/?key=value",
+			[]string{"https://www.systemli.org/en/contact/?key=value"},
+		},
+		{
+			"This is a text with a URL https://www.systemli.org/en/contact/?key=value#fragment",
+			[]string{"https://www.systemli.org/en/contact/?key=value#fragment"},
+		},
+	}
+
+	for _, tc := range testCases {
+		urls := ExtractURLs(tc.text)
+		assert.Equal(t, len(tc.expected), len(urls))
+		for i, url := range tc.expected {
+			assert.Equal(t, url, urls[i])
+		}
+	}
 }