fix(ap): use slug-only OG image path /og/{slug}.png

The previous fix incorrectly generated /og/{year}-{month}-{day}-{slug}.png
but the Eleventy blog generates OG images at /og/{slug}.png (e.g. /og/2615b.png).
Remove the unnecessary date extraction and simplify to slug-only.

https://claude.ai/code/session_0124D41vdLYE3DkJxhPqYthX

README.md

@@ -148,7 +148,7 @@ Posts are converted from Indiekit's JF2 format to ActivityStreams 2.0 in two mod
 - Permalink appended to content body
 - Nested hashtags normalized: `on/art/music` → `#music` (Mastodon doesn't support path-style tags)
 - Sensitive posts flagged with `sensitive: true`; summary doubles as CW text for notes
-- Per-post OG image added to Note/Article objects (`/og/{year}-{month}-{day}-{slug}.png`) for fediverse preview cards
+- Per-post OG image added to Note/Article objects (`/og/{slug}.png`) for fediverse preview cards
 
 ### Express ↔ Fedify bridge
 
@@ -173,11 +173,7 @@ These patches are applied to `node_modules` via postinstall and at serve startup
 **`patch-ap-og-image.mjs`**
 The fork (both 842fc5af and 45f8ba9) attempts to derive the OG image path by matching a date-based URL pattern like `/articles/2024/01/15/slug/`. This blog uses flat URLs (`/articles/slug/`) with no date component, so the regex never matches and no `image` property is set on ActivityPub objects — Mastodon and other clients never show a preview card.
 
-The patch replaces the URL-pattern extraction with:
-1. Slug from the last URL path segment.
-2. Date from `properties.published` (ISO-8601 string).
-
-This produces the correct `/og/{year}-{month}-{day}-{slug}.png` filename that the Eleventy build generates for per-post OG images. Applied to both `jf2ToActivityStreams()` (plain JSON-LD) and `jf2ToAS2Activity()` (Fedify vocab objects).
+The patch replaces the broken date-from-URL regex with a simple last-path-segment extraction, producing `/og/{slug}.png` — the actual filename the Eleventy build generates (e.g. `/og/2615b.png`). Applied to both `jf2ToActivityStreams()` (plain JSON-LD) and `jf2ToAS2Activity()` (Fedify vocab objects).
 
 ### AP environment variables
 
@@ -658,7 +654,6 @@ Environment variables are loaded from `.env` via `dotenv`. See `indiekit.config.
 
 ## Changelog
 
-<<<<<<< HEAD
 ### 2026-03-21
 
 **chore(deps): merge upstream activitypub v3.7.1–v3.7.5 into fork** (`97a902b` in svemagie/indiekit-endpoint-activitypub)
@@ -700,7 +695,7 @@ New `patch-endpoint-github-contributions-log.mjs` suppresses the noisy per-contr
 ### 2026-03-20
 
 **fix(ap): fix OG image not included in ActivityPub activities**
-The fork's OG image code expected date-based URLs (`/articles/YYYY/MM/DD/slug/`) but this blog uses flat URLs (`/articles/slug/`). The regex never matched so no `image` property was set and Mastodon/fediverse clients showed no preview card. Added `patch-ap-og-image.mjs` which extracts the slug from the URL's last path segment and constructs the correct `/og/{slug}.png` filename.
+The fork's OG image code expected date-based URLs (`/articles/YYYY/MM/DD/slug/`) but this blog uses flat URLs (`/articles/slug/`). The regex never matched so no `image` property was set and Mastodon/fediverse clients showed no preview card. Added `patch-ap-og-image.mjs` which extracts the slug from the URL's last path segment and constructs `/og/{slug}.png` — the actual Eleventy OG filename format (e.g. `/og/2615b.png`).
 
 **fix(ap): include commentary in repost ActivityPub activities** (`b53afe2e`)
 Reposts with a body were silently broken in two ways: (1) `jf2ToAS2Activity()` always emitted a bare `Announce` pointing at an external URL that doesn't serve AP JSON, so Mastodon dropped the activity from followers' timelines; (2) `jf2ToActivityStreams()` hard-coded Note content to `🔁 <url>`, ignoring `properties.content`. New `patch-ap-repost-commentary.mjs` (4 targeted replacements): skips the `Announce` early-return when commentary is present and falls through to `Create(Note)` instead; formats Note as `<commentary>\n\n🔁 <url>`; extracts commentary in the content-negotiation path. Pure reposts (no body) keep the `Announce` behaviour unchanged.

scripts/patch-ap-og-image.mjs

@@ -3,17 +3,15 @@
  *
  * Root cause:
  *   Both 842fc5af and 45f8ba9 versions of jf2-to-as2.js try to extract the
- *   post date from the URL using a regex that expects date-based URLs like
+ *   post slug from the URL using a regex that expects date-based URLs like
  *   /articles/2024/01/15/slug/ but this blog uses flat URLs like /articles/slug/.
  *   The regex never matches so the `image` property is never set — no OG image
  *   preview card reaches Mastodon or other fediverse servers.
  *
  * Fix:
- *   Replace the date-from-URL regex with an approach that:
- *     1. Extracts the slug from the last path segment of the post URL.
- *     2. Reads the date from properties.published (ISO-8601 string).
- *   Constructs /og/{year}-{month}-{day}-{slug}.png — the filename pattern that
- *   the Eleventy build generates for static OG preview images.
+ *   Replace the date-from-URL regex with a simple last-path-segment extraction.
+ *   Constructs /og/{slug}.png — the actual filename pattern the Eleventy build
+ *   generates for static OG preview images (e.g. /og/2615b.png).
  *
  *   Both jf2ToActivityStreams() (plain JSON-LD) and jf2ToAS2Activity() (Fedify
  *   vocab objects) are patched. Both 842fc5af and 45f8ba9 variants are handled
@@ -44,30 +42,27 @@ const AS2_BLOCK_RE =
   /  const ogMatchF = postUrl && postUrl\.match\([^\n]+\n  if \(ogMatchF\) \{[\s\S]*?\n  \}/;
 
 // ---------------------------------------------------------------------------
-// Replacement: extract slug from URL last segment, date from published ISO string.
-// Build /og/{year}-{month}-{day}-{slug}.png to match the Eleventy OG filenames.
+// Replacement: extract slug from last URL path segment.
+// Build /og/{slug}.png to match the Eleventy OG filenames (e.g. /og/2615b.png).
 //
-// Template literal note: backslashes in regex literals inside the injected code
-// are doubled here so they survive the template literal → string conversion:
+// Template literal note: backslashes inside the injected regex are doubled so
+// they survive the template literal → string conversion:
 //   \\\/ → \/ (escaped slash in regex)
-//   \\d  → \d (digit class)
 //   [\\\w-] → [\w-] (word char class)
 // ---------------------------------------------------------------------------
 const NEW_CN = `  const ogSlug = postUrl && postUrl.match(/\\/([\\\w-]+)\\/?$/)?.[1]; // og-image fix
-  const ogPub = properties.published && properties.published.match(/^(\\d{4})-(\\d{2})-(\\d{2})/); // og-image fix
-  if (ogSlug && ogPub) { // og-image fix
+  if (ogSlug) { // og-image fix
     object.image = {
       type: "Image",
-      url: \`\${publicationUrl.replace(/\\/$/, "")}/og/\${ogPub[1]}-\${ogPub[2]}-\${ogPub[3]}-\${ogSlug}.png\`, // og-image fix
+      url: \`\${publicationUrl.replace(/\\/$/, "")}/og/\${ogSlug}.png\`, // og-image fix
       mediaType: "image/png",
     };
   }`;
 
 const NEW_AS2 = `  const ogSlugF = postUrl && postUrl.match(/\\/([\\\w-]+)\\/?$/)?.[1]; // og-image fix
-  const ogPubF = properties.published && properties.published.match(/^(\\d{4})-(\\d{2})-(\\d{2})/); // og-image fix
-  if (ogSlugF && ogPubF) { // og-image fix
+  if (ogSlugF) { // og-image fix
     noteOptions.image = new Image({
-      url: new URL(\`\${publicationUrl.replace(/\\/$/, "")}/og/\${ogPubF[1]}-\${ogPubF[2]}-\${ogPubF[3]}-\${ogSlugF}.png\`), // og-image fix
+      url: new URL(\`\${publicationUrl.replace(/\\/$/, "")}/og/\${ogSlugF}.png\`), // og-image fix
       mediaType: "image/png",
     });
   }`;