From a4f7427540840d31195ad37d7c68955ec7a36584 Mon Sep 17 00:00:00 2001
From: Ben Sima <ben@bsima.me>
Date: Sat, 15 Nov 2025 23:13:56 -0500
Subject: Implement episode deduplication in submission flow

- Check for existing episodes by URL hash before processing - Reuse
existing episodes when user submits duplicate URL - Add episode
to user's feed if they don't have it - Track 'added' metrics when
episode added to feed - Worker now creates user_episodes link and
tracks metrics - Show appropriate messages for already-in-feed vs
newly-added episodes
---
 Biz/PodcastItLater/Web.py    | 38 ++++++++++++++++++++++++++++++++++++--
 Biz/PodcastItLater/Worker.py | 12 ++++++++++++
 2 files changed, 48 insertions(+), 2 deletions(-)

(limited to 'Biz')

diff --git a/Biz/PodcastItLater/Web.py b/Biz/PodcastItLater/Web.py
index 1a94007..8caaf8c 100644
--- a/Biz/PodcastItLater/Web.py
+++ b/Biz/PodcastItLater/Web.py
@@ -1149,7 +1149,10 @@ def logout(request: Request) -> Response:
 
 
 @app.post("/submit")
-def submit_article(request: Request, data: FormData) -> html.div:  # noqa: PLR0911
+def submit_article(  # noqa: PLR0911, PLR0914
+    request: Request,
+    data: FormData,
+) -> html.div:
     """Handle manual form submission."""
     try:
         # Check if user is logged in
@@ -1208,7 +1211,38 @@ def submit_article(request: Request, data: FormData) -> html.div:  # noqa: PLR09
                 classes=["alert", "alert-warning"],
             )
 
-        # Extract Open Graph metadata
+        # Check if episode already exists for this URL
+        url_hash = Core.hash_url(url)
+        existing_episode = Core.Database.get_episode_by_url_hash(url_hash)
+
+        if existing_episode:
+            # Episode already processed - check if user has it
+            episode_id = existing_episode["id"]
+            if Core.Database.user_has_episode(user_id, episode_id):
+                return html.div(
+                    html.i(classes=["bi", "bi-info-circle", "me-2"]),
+                    "This episode is already in your feed.",
+                    classes=["alert", "alert-info"],
+                )
+            # Add existing episode to user's feed
+            Core.Database.add_episode_to_user(user_id, episode_id)
+            Core.Database.track_episode_event(
+                episode_id,
+                "added",
+                user_id,
+            )
+            return html.div(
+                html.i(classes=["bi", "bi-check-circle", "me-2"]),
+                "✓ Episode added to your feed! ",
+                html.a(
+                    "View episode",
+                    href=f"/episode/{encode_episode_id(episode_id)}",
+                    classes=["alert-link"],
+                ),
+                classes=["alert", "alert-success"],
+            )
+
+        # Episode doesn't exist yet - extract metadata and queue for processing
         title, author = extract_og_metadata(url)
 
         job_id = Core.Database.add_to_queue(
diff --git a/Biz/PodcastItLater/Worker.py b/Biz/PodcastItLater/Worker.py
index 202e512..75a111c 100644
--- a/Biz/PodcastItLater/Worker.py
+++ b/Biz/PodcastItLater/Worker.py
@@ -519,6 +519,7 @@ class ArticleProcessor:
             duration = ArticleProcessor.estimate_duration(audio_data)
 
             # Step 5: Create episode record
+            url_hash = Core.hash_url(url)
             episode_id = Core.Database.create_episode(
                 title=title,
                 audio_url=audio_url,
@@ -527,8 +528,19 @@ class ArticleProcessor:
                 user_id=job.get("user_id"),
                 author=job.get("author"),  # Pass author from job
                 original_url=url,  # Pass the original article URL
+                original_url_hash=url_hash,
             )
 
+            # Add episode to user's feed
+            user_id = job.get("user_id")
+            if user_id:
+                Core.Database.add_episode_to_user(user_id, episode_id)
+                Core.Database.track_episode_event(
+                    episode_id,
+                    "added",
+                    user_id,
+                )
+
             # Step 6: Mark job as complete
             Core.Database.update_job_status(
                 job_id,
-- 
cgit v1.2.3