fix: only sync new YouTube likes, not existing ones

First sync snapshots all current liked video IDs into a
youtubeLikesSeen collection without creating posts. Subsequent
syncs only create posts for likes not in the seen set.

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

lib/likes-sync.js

@@ -1,13 +1,14 @@
 /**
  * YouTube Likes → Indiekit "like" posts sync
  *
- * Fetches the authenticated user's liked videos and creates
- * corresponding "like" posts via the Micropub posts collection.
+ * On first run after connecting, snapshots all current liked video IDs
+ * as "known" without creating posts. Subsequent syncs only create posts
+ * for newly liked videos (ones not in the known set and not already
+ * in the posts collection).
  */
 
 import { YouTubeClient } from "./youtube-client.js";
 import { getValidAccessToken } from "./oauth.js";
-import crypto from "node:crypto";
 
 /**
  * Generate a deterministic slug from a YouTube video ID.
@@ -18,11 +19,54 @@ function slugFromVideoId(videoId) {
   return `yt-like-${videoId}`;
 }
 
+/**
+ * Fetch all current liked video IDs (up to maxPages) and store them
+ * as the baseline. No posts are created.
+ * @returns {Promise<number>} number of IDs snapshotted
+ */
+async function snapshotExistingLikes(db, client, accessToken, maxPages) {
+  const collection = db.collection("youtubeLikesSeen");
+  await collection.createIndex({ videoId: 1 }, { unique: true });
+
+  let pageToken;
+  let count = 0;
+
+  for (let page = 0; page < maxPages; page++) {
+    const result = await client.getLikedVideos(accessToken, 50, pageToken);
+
+    const ops = result.videos.map((v) => ({
+      updateOne: {
+        filter: { videoId: v.id },
+        update: { $setOnInsert: { videoId: v.id, seenAt: new Date() } },
+        upsert: true,
+      },
+    }));
+
+    if (ops.length > 0) {
+      await collection.bulkWrite(ops, { ordered: false });
+      count += ops.length;
+    }
+
+    pageToken = result.nextPageToken;
+    if (!pageToken) break;
+  }
+
+  // Mark that the baseline snapshot is done
+  await db.collection("youtubeMeta").updateOne(
+    { key: "likes_baseline" },
+    { $set: { key: "likes_baseline", completedAt: new Date(), count } },
+    { upsert: true },
+  );
+
+  return count;
+}
+
 /**
  * Sync liked videos into the Indiekit posts collection.
  *
- * For each liked video that doesn't already have a corresponding post
- * in the database we insert a new "like" post document.
+ * First-run behaviour: snapshots all existing likes as "seen" without
+ * creating posts. Only likes that appear after the baseline are turned
+ * into posts.
  *
  * @param {object} opts
  * @param {import("mongodb").Db} opts.db
@@ -30,7 +74,7 @@ function slugFromVideoId(videoId) {
  * @param {object} opts.publication - Indiekit publication config
  * @param {import("mongodb").Collection} [opts.postsCollection]
  * @param {number} [opts.maxPages=3] - max pages to fetch (50 likes/page)
- * @returns {Promise<{synced: number, skipped: number, total: number, error?: string}>}
+ * @returns {Promise<{synced: number, skipped: number, total: number, baselined?: number, error?: string}>}
  */
 export async function syncLikes({ db, youtubeConfig, publication, postsCollection, maxPages = 3 }) {
   const { apiKey, oauth } = youtubeConfig;
@@ -38,7 +82,6 @@ export async function syncLikes({ db, youtubeConfig, publication, postsCollectio
     return { synced: 0, skipped: 0, total: 0, error: "OAuth not configured" };
   }
 
-  // Get a valid access token (auto‑refreshes if needed)
   let accessToken;
   try {
     accessToken = await getValidAccessToken(db, {
@@ -55,6 +98,17 @@ export async function syncLikes({ db, youtubeConfig, publication, postsCollectio
 
   const client = new YouTubeClient({ apiKey: apiKey || "unused", cacheTtl: 0 });
 
+  // --- First-run: snapshot existing likes, create zero posts ---
+  const baseline = await db.collection("youtubeMeta").findOne({ key: "likes_baseline" });
+  if (!baseline) {
+    console.log("[YouTube] First sync — snapshotting existing likes (no posts will be created)");
+    const count = await snapshotExistingLikes(db, client, accessToken, maxPages);
+    console.log(`[YouTube] Baselined ${count} existing liked videos`);
+    return { synced: 0, skipped: 0, total: count, baselined: count };
+  }
+
+  // --- Normal sync: only create posts for new likes ---
+  const seenCollection = db.collection("youtubeLikesSeen");
   let synced = 0;
   let skipped = 0;
   let total = 0;
@@ -68,10 +122,26 @@ export async function syncLikes({ db, youtubeConfig, publication, postsCollectio
     total = result.totalResults;
 
     for (const video of result.videos) {
-      const slug = slugFromVideoId(video.id);
-      const videoUrl = `https://www.youtube.com/watch?v=${video.id}`;
+      const videoId = video.id;
 
-      // Check if we already synced this like
+      // Already seen? Skip.
+      const seen = await seenCollection.findOne({ videoId });
+      if (seen) {
+        skipped++;
+        continue;
+      }
+
+      // Mark as seen immediately (even if post insert fails, don't retry)
+      await seenCollection.updateOne(
+        { videoId },
+        { $setOnInsert: { videoId, seenAt: new Date() } },
+        { upsert: true },
+      );
+
+      const slug = slugFromVideoId(videoId);
+      const videoUrl = `https://www.youtube.com/watch?v=${videoId}`;
+
+      // Also skip if a post already exists (belt-and-suspenders)
       if (postsCollection) {
         const existing = await postsCollection.findOne({
           $or: [
@@ -85,7 +155,6 @@ export async function syncLikes({ db, youtubeConfig, publication, postsCollectio
         }
       }
 
-      // Build the post path/url from the publication's like post type config
       const postPath = likePostType?.post?.path
         ? likePostType.post.path.replace("{slug}", slug)
         : `content/likes/${slug}.md`;
@@ -105,11 +174,11 @@ export async function syncLikes({ db, youtubeConfig, publication, postsCollectio
             text: `Liked "${video.title}" by ${video.channelTitle} on YouTube`,
             html: `Liked "<a href="${videoUrl}">${escapeHtml(video.title)}</a>" by ${escapeHtml(video.channelTitle)} on YouTube`,
           },
-          published: video.publishedAt || new Date().toISOString(),
+          published: new Date().toISOString(),
           url: postUrl,
           visibility: "public",
           "post-status": "published",
-          "youtube-video-id": video.id,
+          "youtube-video-id": videoId,
           "youtube-channel": video.channelTitle,
           "youtube-thumbnail": video.thumbnail || "",
         },
@@ -125,7 +194,6 @@ export async function syncLikes({ db, youtubeConfig, publication, postsCollectio
     if (!pageToken) break;
   }
 
-  // Update sync metadata
   await db.collection("youtubeMeta").updateOne(
     { key: "likes_sync" },
     {
@@ -158,7 +226,7 @@ export async function getLastSyncStatus(db) {
  * @param {object} options - endpoint options
  */
 export function startLikesSync(Indiekit, options) {
-  const interval = options.likes?.syncInterval || 3_600_000; // default 1 hour
+  const interval = options.likes?.syncInterval || 3_600_000;
 
   async function run() {
     const db = Indiekit.database;
@@ -178,15 +246,16 @@ export function startLikesSync(Indiekit, options) {
       if (result.synced > 0) {
         console.log(`[YouTube] Likes sync: ${result.synced} new, ${result.skipped} skipped`);
       }
+      if (result.baselined) {
+        console.log(`[YouTube] Baseline complete: ${result.baselined} existing likes recorded`);
+      }
     } catch (err) {
       console.error("[YouTube] Likes sync error:", err.message);
     }
   }
 
-  // First run after a short delay to let the DB connect
   setTimeout(() => {
     run().catch(() => {});
-    // Then repeat on interval
     setInterval(() => run().catch(() => {}), interval);
   }, 15_000);
 }