feat: linkExternalUrls to use GFM-style detection

Signed-off-by: Pedro Lamas <pedrolamas@gmail.com>

Author: pedrolamas

src/util/__tests__/link-external-urls.spec.ts

@@ -0,0 +1,47 @@
+import linkExternalUrls from '@/util/link-external-urls'
+
+const anchor = (url: string) => `<a target="_blank" rel="noopener noreferrer" href="${url}">${url}</a>`
+
+describe('linkExternalUrls', () => {
+  it('wraps a plain URL in an anchor', () => {
+    expect(linkExternalUrls('https://x.com'))
+      .toBe(anchor('https://x.com'))
+  })
+
+  it('keeps wrapping quotes outside the link', () => {
+    expect(linkExternalUrls('"https://x.com"'))
+      .toBe(`"${anchor('https://x.com')}"`)
+  })
+
+  it('keeps a sentence-ending period as plain text', () => {
+    expect(linkExternalUrls('see https://x.com.'))
+      .toBe(`see ${anchor('https://x.com')}.`)
+  })
+
+  it('keeps a trailing comma in mid-sentence as plain text', () => {
+    expect(linkExternalUrls('go to https://x.com, then stop'))
+      .toBe(`go to ${anchor('https://x.com')}, then stop`)
+  })
+
+  it('keeps wrapping parentheses outside the link', () => {
+    expect(linkExternalUrls('(https://x.com)'))
+      .toBe(`(${anchor('https://x.com')})`)
+  })
+
+  it('preserves balanced parentheses inside the URL', () => {
+    const url = 'https://en.wikipedia.org/wiki/Foo_(bar)'
+
+    expect(linkExternalUrls(url))
+      .toBe(anchor(url))
+  })
+
+  it('links multiple URLs in one string', () => {
+    expect(linkExternalUrls('a https://x.com b https://y.com'))
+      .toBe(`a ${anchor('https://x.com')} b ${anchor('https://y.com')}`)
+  })
+
+  it('returns text without a URL unchanged', () => {
+    expect(linkExternalUrls('no link here'))
+      .toBe('no link here')
+  })
+})

src/util/link-external-urls.ts

@@ -1,5 +1,27 @@
-const externalUrlRegExp = /(https?:\/\/\S+)/gi
+const externalUrlRegExp = /https?:\/\/[^\s<>"'`]+/gi
 
-const linkExternalUrls = (text: string) => text.replace(externalUrlRegExp, '<a target="_blank" href="$1">$1</a>')
+const trailingPunctuationRegExp = /[?!.,:*_~]+$/
+
+// Trim trailing punctuation and unmatched closing parens (GFM autolink rules).
+const trimUrl = (url: string): string => {
+  let result = url.replace(trailingPunctuationRegExp, '')
+
+  while (
+    result.endsWith(')') &&
+    (result.match(/\)/g)?.length ?? 0) > (result.match(/\(/g)?.length ?? 0)
+  ) {
+    result = result.slice(0, -1).replace(trailingPunctuationRegExp, '')
+  }
+
+  return result
+}
+
+const linkExternalUrls = (text: string) => text.replace(externalUrlRegExp, (match) => {
+  const url = trimUrl(match)
+
+  const trailing = match.slice(url.length)
+
+  return `<a target="_blank" rel="noopener noreferrer" href="${url}">${url}</a>${trailing}`
+})
 
 export default linkExternalUrls