Fix handling of orphaned UserWords (no bookmarks)

tools/migrations/26-02-10-a--add_monthly_activity_stats_cache.sql

@@ -0,0 +1,13 @@
+-- Cache table for monthly activity statistics by type
+-- Historical months are cached permanently, current month refreshed periodically
+
+CREATE TABLE IF NOT EXISTS monthly_activity_stats_cache (
+    id INT AUTO_INCREMENT PRIMARY KEY,
+    `year_month` VARCHAR(7) NOT NULL UNIQUE COMMENT 'Format: YYYY-MM',
+    `exercise_minutes` INT NOT NULL DEFAULT 0,
+    `reading_minutes` INT NOT NULL DEFAULT 0,
+    `browsing_minutes` INT NOT NULL DEFAULT 0,
+    `audio_minutes` INT NOT NULL DEFAULT 0,
+    `computed_at` DATETIME NOT NULL,
+    INDEX idx_year_month (`year_month`)
+) ENGINE=InnoDB DEFAULT CHARSET=utf8mb4 COLLATE=utf8mb4_bin;

zeeguu/api/endpoints/article.py

@@ -627,19 +627,50 @@ def clear_article_cache(article_id):
     )
 
     bookmark_count = len(bookmarks)
+    bookmark_ids_to_delete = {b.id for b in bookmarks}
 
-    # First pass: clear all preferred_bookmark_id references
+    # Track UserWords that need cleanup
+    user_words_to_check = {}
     for bookmark in bookmarks:
         user_word = bookmark.user_word
-        if user_word and user_word.preferred_bookmark_id == bookmark.id:
-            user_word.preferred_bookmark_id = None
+        if user_word:
+            user_words_to_check[user_word.id] = user_word
+            # Clear preferred_bookmark if it's one we're deleting
+            if user_word.preferred_bookmark_id == bookmark.id:
+                user_word.preferred_bookmark_id = None
+
     db_session.flush()  # Commit the nullifications before deleting
 
-    # Second pass: delete bookmarks
+    # Delete bookmarks
     for bookmark in bookmarks:
         db_session.delete(bookmark)
+    db_session.flush()
+
+    # Handle UserWords that are now orphaned or need new preferred_bookmark
+    user_words_marked_unfit = 0
+    for user_word in user_words_to_check.values():
+        # Check for remaining bookmarks not in our delete set
+        remaining_bookmarks = (
+            Bookmark.query
+            .filter(Bookmark.user_word_id == user_word.id)
+            .filter(Bookmark.id.notin_(bookmark_ids_to_delete))
+            .all()
+        )
+
+        if remaining_bookmarks:
+            # Set a new preferred bookmark if needed
+            if user_word.preferred_bookmark is None:
+                user_word.preferred_bookmark = remaining_bookmarks[0]
+        else:
+            # No bookmarks left - keep UserWord for history but mark unfit for study
+            user_word.set_unfit_for_study(db_session)
+            user_words_marked_unfit += 1
+
     db_session.commit()
 
+    if user_words_marked_unfit > 0:
+        log(f"[DEV] Marked {user_words_marked_unfit} UserWords as unfit for study (no bookmarks)")
+
     log(f"[DEV] Cleared cache and {bookmark_count} bookmarks for article {article_id}")
 
     return json_result({

zeeguu/api/endpoints/bookmarks_and_words.py

@@ -241,7 +241,8 @@ def delete_bookmark(bookmark_id):
                     # in the future we can generate an example for this user word with the help of the robots!
                     user_word.set_unfit_for_study(db_session)
             else:
-                # No other bookmarks exist - ALWAYS keep the user_word for historical data
+                # No other bookmarks exist - keep UserWord for historical data
+                # but mark as unfit for study (won't appear in exercises)
                 user_word.set_unfit_for_study(db_session)
 
         # Delete any ExampleSentenceContext records that reference this bookmark

zeeguu/api/endpoints/exercises.py

@@ -309,7 +309,6 @@ def _user_words_as_json_result(user_words):
             log(f"Failed to get tokenized context for user_word {uw.id}: {e}")
 
     dicts = []
-    words_to_delete = []
 
     for user_word in user_words:
         try:
@@ -320,28 +319,9 @@ def _user_words_as_json_result(user_words):
                 schedule=schedule,
                 pre_tokenized_context=tokenized_context
             ))
-        except ValueError as e:
-            # This means validate_data_integrity() couldn't repair the issue
-            # (i.e., UserWord has no bookmarks at all)
-            log(f"UserWord {user_word.id} failed validation and cannot be repaired: {str(e)}")
-            words_to_delete.append(user_word)
         except Exception as e:
-            # Log any other unexpected errors and skip
+            # Log unexpected errors and skip (orphaned UserWords are handled gracefully)
             log(f"Unexpected error processing UserWord {user_word.id}: {str(e)}")
             continue
 
-    # Delete UserWords that couldn't be repaired
-    if words_to_delete:
-        for word in words_to_delete:
-            try:
-                db.session.delete(word)
-                log(f"Deleted UserWord {word.id} due to unrepairable data integrity issues")
-            except:
-                log(f"Failed to delete UserWord {word.id}")
-        try:
-            db.session.commit()
-        except:
-            db.session.rollback()
-            log("Failed to commit UserWord deletions")
-
     return json_result(dicts)