Merge pull request #2485 from afjcjsbx/fix/telegram-oauth-links

fix(telegram): preserve raw OAuth links in HTML rendering

pkg/channels/telegram/parser_markdown_to_html.go

@@ -2,9 +2,13 @@ package telegram
 
 import (
 	"fmt"
+	"html"
+	"regexp"
 	"strings"
 )
 
+var reRawURL = regexp.MustCompile(`https?://[^\s<]+`)
+
 func markdownToTelegramHTML(text string) string {
 	if text == "" {
 		return ""
@@ -19,6 +23,9 @@ func markdownToTelegramHTML(text string) string {
 	links := extractLinks(text)
 	text = links.text
 
+	rawURLs := extractRawURLs(text)
+	text = rawURLs.text
+
 	text = reHeading.ReplaceAllString(text, "$1")
 
 	text = reBlockquote.ReplaceAllString(text, "$1")
@@ -43,10 +50,19 @@ func markdownToTelegramHTML(text string) string {
 
 	for i, lnk := range links.links {
 		label := escapeHTML(lnk[0])
-		url := lnk[1]
+		url := escapeHTMLAttr(lnk[1])
 		text = strings.ReplaceAll(text, fmt.Sprintf("\x00LK%d\x00", i), fmt.Sprintf(`<a href="%s">%s</a>`, url, label))
 	}
 
+	for i, rawURL := range rawURLs.urls {
+		escaped := escapeHTML(rawURL)
+		text = strings.ReplaceAll(
+			text,
+			fmt.Sprintf("\x00RU%d\x00", i),
+			fmt.Sprintf(`<a href="%s">%s</a>`, escapeHTMLAttr(rawURL), escaped),
+		)
+	}
+
 	for i, code := range inlineCodes.codes {
 		escaped := escapeHTML(code)
 		text = strings.ReplaceAll(text, fmt.Sprintf("\x00IC%d\x00", i), fmt.Sprintf("<code>%s</code>", escaped))
@@ -92,6 +108,11 @@ type codeBlockMatch struct {
 	codes []string
 }
 
+type rawURLMatch struct {
+	text string
+	urls []string
+}
+
 func extractCodeBlocks(text string) codeBlockMatch {
 	matches := reCodeBlock.FindAllStringSubmatch(text, -1)
 
@@ -110,6 +131,24 @@ func extractCodeBlocks(text string) codeBlockMatch {
 	return codeBlockMatch{text: text, codes: codes}
 }
 
+func extractRawURLs(text string) rawURLMatch {
+	matches := reRawURL.FindAllString(text, -1)
+
+	urls := make([]string, 0, len(matches))
+	for _, match := range matches {
+		urls = append(urls, match)
+	}
+
+	i := 0
+	text = reRawURL.ReplaceAllStringFunc(text, func(string) string {
+		placeholder := fmt.Sprintf("\x00RU%d\x00", i)
+		i++
+		return placeholder
+	})
+
+	return rawURLMatch{text: text, urls: urls}
+}
+
 type inlineCodeMatch struct {
 	text  string
 	codes []string
@@ -139,3 +178,7 @@ func escapeHTML(text string) string {
 	text = strings.ReplaceAll(text, ">", "&gt;")
 	return text
 }
+
+func escapeHTMLAttr(text string) string {
+	return html.EscapeString(text)
+}

pkg/channels/telegram/parser_markdown_to_html_test.go

@@ -32,6 +32,11 @@ func Test_markdownToTelegramHTML(t *testing.T) {
 			input:    "[click here](https://example.com/path)",
 			expected: `<a href="https://example.com/path">click here</a>`,
 		},
+		{
+			name:     "raw oauth url with underscores survives",
+			input:    "Apri https://accounts.google.com/o/oauth2/auth?response_type=code&client_id=test-client&redirect_uri=http%3A%2F%2Flocalhost%3A8001%2Foauth2callback&code_challenge=abc_def&code_challenge_method=S256",
+			expected: `Apri <a href="https://accounts.google.com/o/oauth2/auth?response_type=code&amp;client_id=test-client&amp;redirect_uri=http%3A%2F%2Flocalhost%3A8001%2Foauth2callback&amp;code_challenge=abc_def&amp;code_challenge_method=S256">https://accounts.google.com/o/oauth2/auth?response_type=code&amp;client_id=test-client&amp;redirect_uri=http%3A%2F%2Flocalhost%3A8001%2Foauth2callback&amp;code_challenge=abc_def&amp;code_challenge_method=S256</a>`,
+		},
 		{
 			name: "link with underscores in URL is not corrupted by italic regex",
 			// Google Flights URLs use URL-safe base64 with underscores in the tfs param.
@@ -45,6 +50,11 @@ func Test_markdownToTelegramHTML(t *testing.T) {
 			input:    "[first](https://a.com/path_one) and [second](https://b.com/path_two_x)",
 			expected: `<a href="https://a.com/path_one">first</a> and <a href="https://b.com/path_two_x">second</a>`,
 		},
+		{
+			name:     "markdown link query params are escaped in href",
+			input:    "[oauth](https://example.com/cb?response_type=code&client_id=test-client)",
+			expected: `<a href="https://example.com/cb?response_type=code&amp;client_id=test-client">oauth</a>`,
+		},
 		{
 			name:     "link label with HTML special chars is escaped",
 			input:    "[a & b](https://example.com)",