fix(actions): don't swallow HTML entities into linkified URLs (#38239)

In the Actions log viewer, a double-quoted URL renders with a stray
extra `;` after it.

Reported in `gitea/runner#1046`

Remove the buggy AI slop `linkifyURLs` and use new approach to process
URLs in text

---------

Signed-off-by: wxiaoguang <wxiaoguang@gmail.com>
Co-authored-by: wxiaoguang <wxiaoguang@gmail.com>
This commit is contained in:
bircni
2026-06-28 19:37:16 +08:00
committed by GitHub
co-authored by wxiaoguang
parent 5b9251150c
commit ce8cf22af9
7 changed files with 104 additions and 117 deletions
+12 -29
View File
@@ -1,4 +1,15 @@
import {html, htmlRaw} from './html.ts';
/** Matches URLs, excluding characters that are never valid unencoded in URLs per RFC 3986. */
export const urlRawRegex = () => /\bhttps?:\/\/[^\s<>[\]]+/gi; // JS regexp has internal states, so always use a new instance
/** Strip trailing punctuation that is likely not part of the URL. */
export function trimUrlPunctuation(url: string): string {
url = url.replace(/[.,;:'"]+$/, '');
// Strip trailing closing parens only if unbalanced (not part of the URL like Wikipedia links)
while (url.endsWith(')') && (url.match(/\(/g) || []).length < (url.match(/\)/g) || []).length) {
url = url.slice(0, -1);
}
return url;
}
export function urlQueryEscape(s: string) {
// See "TestQueryEscape" in backend
@@ -31,31 +42,3 @@ export function pathEscapeSegments(s: string): string {
// The same as backend's PathEscapeSegments
return s.split('/').map(pathEscape).join('/');
}
// Match HTML tags (to skip) or URLs (to linkify) in HTML content
const urlLinkifyPattern = /(<([-\w]+)[^>]*>)|(<\/([-\w]+)[^>]*>)|(https?:\/\/[^\s<>"'`|(){}[\]]+)/gi;
const trailingPunctPattern = /[.,;:!?]+$/;
// Convert URLs to clickable links in HTML, preserving existing HTML tags
export function linkifyURLs(htmlString: string): string {
let inAnchor = false;
return htmlString.replace(urlLinkifyPattern, (match, _openTagFull, openTag, _closeTagFull, closeTag, url) => {
// skip URLs inside existing <a> tags
if (openTag === 'a') {
inAnchor = true;
return match;
} else if (closeTag === 'a') {
inAnchor = false;
return match;
}
if (inAnchor || !url) {
return match;
}
const trailingPunct = url.match(trailingPunctPattern);
const cleanUrl = trailingPunct ? url.slice(0, -trailingPunct[0].length) : url;
const trailing = trailingPunct ? trailingPunct[0] : '';
// safe because regexp only matches valid URLs (no quotes or angle brackets)
return html`<a href="${htmlRaw(cleanUrl)}" target="_blank">${htmlRaw(cleanUrl)}</a>${htmlRaw(trailing)}`;
});
}