From 2ff34369e0db350e17f3070fe9bc94e2742111c5 Mon Sep 17 00:00:00 2001
From: Paulo Castellano <paulo@castellanos.llc>
Date: Sat, 13 Jun 2026 20:02:54 -0300
Subject: [PATCH] Stop URL facets swallowing trailing punctuation

The URL regex captured trailing sentence punctuation and an unmatched closing
paren ("see https://x.com).") into the link facet, so Bluesky rendered the
punctuation as part of a malformed link. Trim it like the official atproto
tokenizer. Add tests for the trim and for byte-offset correctness after
multibyte characters (emoji/accents).
---
 app/Services/Social/BlueskyPublisher.php      | 20 +++++++-
 .../Services/Social/BlueskyPublisherTest.php  | 46 +++++++++++++++++++
 2 files changed, 65 insertions(+), 1 deletion(-)

diff --git a/app/Services/Social/BlueskyPublisher.php b/app/Services/Social/BlueskyPublisher.php
index a47115e1..b117ac62 100644
--- a/app/Services/Social/BlueskyPublisher.php
+++ b/app/Services/Social/BlueskyPublisher.php
@@ -180,7 +180,7 @@ private function parseFacets(string $text): array
         );
 
         foreach ($urlMatches[0] as $match) {
-            $url = $match[0];
+            $url = $this->trimTrailingUrlPunctuation($match[0]);
             $start = $this->getUtf8ByteOffset($text, (int) $match[1]);
             $end = $start + strlen($url);
 
@@ -299,6 +299,24 @@ private function getUtf8ByteOffset(string $text, int $charOffset): int
         return strlen(substr($text, 0, $charOffset));
     }
 
+    /**
+     * Trailing sentence punctuation and an unmatched closing paren are almost
+     * never part of a URL (e.g. "see https://x.com)."). Mirrors the official
+     * atproto link tokenizer so the link facet doesn't over-extend past the URL.
+     */
+    private function trimTrailingUrlPunctuation(string $url): string
+    {
+        if (preg_match('/[.,;:!?]$/', $url)) {
+            $url = substr($url, 0, -1);
+        }
+
+        if (str_ends_with($url, ')') && ! str_contains($url, '(')) {
+            $url = substr($url, 0, -1);
+        }
+
+        return $url;
+    }
+
     private function buildPostUrl(string $handle, string $postId): string
     {
         return "https://bsky.app/profile/{$handle}/post/{$postId}";
diff --git a/tests/Feature/Services/Social/BlueskyPublisherTest.php b/tests/Feature/Services/Social/BlueskyPublisherTest.php
index 8df42b21..400a0117 100644
--- a/tests/Feature/Services/Social/BlueskyPublisherTest.php
+++ b/tests/Feature/Services/Social/BlueskyPublisherTest.php
@@ -101,6 +101,52 @@
     });
 });
 
+test('bluesky publisher strips trailing punctuation from URL facets', function () {
+    $this->post->update(['content' => 'see https://example.com).']);
+
+    Http::fake([
+        config('trypost.platforms.bluesky.default_service').'/xrpc/com.atproto.repo.createRecord' => Http::response([
+            'uri' => 'at://did:plc:testuser123/app.bsky.feed.post/3abc123xyz',
+            'cid' => 'bafyreiabc123',
+        ], 200),
+    ]);
+
+    $this->publisher->publish($this->postPlatform);
+
+    Http::assertSent(function ($request) {
+        $link = collect($request['record']['facets'] ?? [])
+            ->first(fn ($facet) => $facet['features'][0]['$type'] === 'app.bsky.richtext.facet#link');
+
+        return $link
+            && $link['features'][0]['uri'] === 'https://example.com'
+            && $link['index']['byteEnd'] === $link['index']['byteStart'] + strlen('https://example.com');
+    });
+});
+
+test('bluesky publisher computes byte offsets after multibyte characters', function () {
+    $this->post->update(['content' => 'Olá 🎉 #café']);
+
+    Http::fake([
+        config('trypost.platforms.bluesky.default_service').'/xrpc/com.atproto.repo.createRecord' => Http::response([
+            'uri' => 'at://did:plc:testuser123/app.bsky.feed.post/3abc123xyz',
+            'cid' => 'bafyreiabc123',
+        ], 200),
+    ]);
+
+    $this->publisher->publish($this->postPlatform);
+
+    Http::assertSent(function ($request) {
+        $text = $request['record']['text'];
+        $tag = collect($request['record']['facets'] ?? [])
+            ->first(fn ($facet) => $facet['features'][0]['$type'] === 'app.bsky.richtext.facet#tag');
+
+        // byteStart must be the UTF-8 byte position of '#café', not its character index.
+        return $tag
+            && $tag['index']['byteStart'] === strpos($text, '#café')
+            && $tag['index']['byteEnd'] === strpos($text, '#café') + strlen('#café');
+    });
+});
+
 test('bluesky publisher resolves mentions to DIDs as facets', function () {
     $this->post->update(['content' => 'Shout out to @friend.bsky.social']);