From 86b2819ba6f77f478bedd94d871c18efd97f9b77 Mon Sep 17 00:00:00 2001
From: Greyson Parrelli <greyson@signal.org>
Date: Wed, 2 Sep 2026 21:18:50 +0000
Subject: [PATCH] Allow apostrophes within linkified URL paths.

---
 .../java/org/signal/core/util/Linkifier.kt    | 20 ++++++++++++++++---
 .../org/signal/core/util/LinkifierTest.kt     | 11 ++++++++++
 2 files changed, 28 insertions(+), 3 deletions(-)

diff --git a/core/util-jvm/src/main/java/org/signal/core/util/Linkifier.kt b/core/util-jvm/src/main/java/org/signal/core/util/Linkifier.kt
index b4a44a0ec9..a1c17d455d 100644
--- a/core/util-jvm/src/main/java/org/signal/core/util/Linkifier.kt
+++ b/core/util-jvm/src/main/java/org/signal/core/util/Linkifier.kt
@@ -27,13 +27,26 @@ object Linkifier {
   /** Closing brackets that are trimmed only if the URL has no matching opener inside */
   private val CLOSING_BRACKETS = mapOf(')' to '(', ']' to '[', '}' to '{')
 
+  /**
+   * Characters that always terminate a URL: whitespace (including the unicode variants), bidi/format
+   * controls, and punctuation that is only ever used to quote or wrap a URL.
+   */
+  private const val URL_TERMINATOR = "\\s\\u0085\\u00A0\\u1680\\u2000-\\u200D\\u2028\\u2029\\u202A-\\u202F\\u205F\\u2066-\\u2069\\u3000\\uFEFF<>\"`|\\\\"
+
   /**
    * Characters we treat as definitely-not-part-of-a-URL when extending past the host. Commas and
    * semicolons are allowed inside a URL, but not when they are acting as a separator before another
    * obvious URL.
    */
-  private const val URL_CHAR =
-    "(?:[^\\s\\u0085\\u00A0\\u1680\\u2000-\\u200D\\u2028\\u2029\\u202A-\\u202F\\u205F\\u2066-\\u2069\\u3000\\uFEFF<>\"'`,;|\\\\]|[,;](?!https?://|www\\.))"
+  private const val URL_CHAR = "(?:[^$URL_TERMINATOR,;]|[,;](?!https?://|www\\.))"
+
+  /**
+   * Like [URL_CHAR], but for the authority (host/port/userinfo) portion of a URL, so it stops at the
+   * first path/query/fragment delimiter. Apostrophes are excluded here so that a possessive like
+   * `https://signal.org's blog` doesn't drag the `'s` into the host, while still allowing them
+   * within a path (see [URL_CHAR]).
+   */
+  private const val AUTHORITY_CHAR = "(?:[^$URL_TERMINATOR,;'/?#]|[,;](?!https?://|www\\.))"
 
   /**
    * A single domain label: letter/digit, optional letter/digit/hyphen body. Used for intermediate
@@ -56,7 +69,8 @@ object Linkifier {
     "(?i)" +
       "(?:" +
       // Variant 1: explicit scheme (http or https). User intent is unambiguous; no TLD check.
-      "https?://" + URL_CHAR + "+" +
+      "https?://" + AUTHORITY_CHAR + "+" +
+      "(?:[/?#]" + URL_CHAR + "*)?" +
       "|" +
       // Variant 2: starts with www. — also unambiguous; no TLD check.
       "www\\." + DOMAIN_LABEL + "(?:\\." + DOMAIN_LABEL + ")+" +
diff --git a/core/util-jvm/src/test/java/org/signal/core/util/LinkifierTest.kt b/core/util-jvm/src/test/java/org/signal/core/util/LinkifierTest.kt
index c6c0d29c62..bbc2e3871d 100644
--- a/core/util-jvm/src/test/java/org/signal/core/util/LinkifierTest.kt
+++ b/core/util-jvm/src/test/java/org/signal/core/util/LinkifierTest.kt
@@ -108,6 +108,17 @@ class LinkifierTest(private val case: Case) {
       Case("semicolon in url path is preserved", "https://example.com/pricewatch/compare/2290176;2126182/", listOf(web("https://example.com/pricewatch/compare/2290176;2126182/"))),
       Case("semicolon in url query is preserved", "Go to https://example.com/search?a=1;b=2", listOf(web("https://example.com/search?a=1;b=2"))),
 
+      // ----- apostrophes -----
+      Case("apostrophe in url path is preserved", "https://en.wikipedia.org/wiki/Occam's_razor", listOf(web("https://en.wikipedia.org/wiki/Occam's_razor"))),
+      Case("apostrophe in url path is preserved mid-sentence", "see https://en.wikipedia.org/wiki/Occam's_razor for more", listOf(web("https://en.wikipedia.org/wiki/Occam's_razor"))),
+      Case("apostrophe in bare domain path is preserved", "en.wikipedia.org/wiki/Occam's_razor", listOf(web("en.wikipedia.org/wiki/Occam's_razor", url = "http://en.wikipedia.org/wiki/Occam's_razor"))),
+      Case("apostrophe in url query is preserved", "https://example.com/search?q=Occam's", listOf(web("https://example.com/search?q=Occam's"))),
+      Case("trailing apostrophe is trimmed", "https://example.com/foo' end", listOf(web("https://example.com/foo"))),
+      Case("single-quoted url does not include the quotes", "'https://signal.org'", listOf(web("https://signal.org"))),
+      Case("possessive after schemed host is not part of the url", "https://signal.org's blog is good", listOf(web("https://signal.org"))),
+      Case("possessive after bare domain is not part of the url", "signal.org's blog is good", listOf(web("signal.org", url = "http://signal.org"))),
+      Case("possessive after www domain is not part of the url", "www.signal.org's blog is good", listOf(web("www.signal.org", url = "http://www.signal.org"))),
+
       // ----- bracket / paren handling -----
       Case("trailing closing paren without opener is trimmed", "(see https://signal.org)", listOf(web("https://signal.org"))),
       Case("trailing closing paren with matching opener inside is preserved", "https://en.wikipedia.org/wiki/Foo_(bar) and more", listOf(web("https://en.wikipedia.org/wiki/Foo_(bar)"))),