feat(vocab): add logic to aggregate and clean up duplicate SRS items based on answer patterns

backend/migrations-active/20260818000000-add-bisaya-rest-answer-alternative.cjs

@@ -95,6 +95,90 @@ module.exports = {
         transaction
       });
 
+      await queryInterface.sequelize.query(`
+        WITH candidates AS (
+          SELECT id, user_id,
+                 ENCODE(DIGEST(CONCAT_WS(
+                   '|',
+                   course_id::text,
+                   COALESCE(lesson_id::text, 'course'),
+                   UPPER(direction),
+                   :normalizedCombinedAnswer,
+                   :normalizedTarget
+                 ), 'sha1'), 'hex') AS new_item_key,
+                 correct_count, wrong_count, lapse_count, stage, interval_days,
+                 last_reviewed_at, next_due_at
+          FROM community.vocab_srs_item
+          WHERE learning = :shortAnswer
+            AND reference = :target
+        ), ranked AS (
+          SELECT *, ROW_NUMBER() OVER (PARTITION BY user_id, new_item_key ORDER BY id) AS position
+          FROM candidates
+        ), aggregates AS (
+          SELECT user_id, new_item_key, MIN(id) AS keeper_id,
+                 SUM(correct_count) AS correct_count,
+                 SUM(wrong_count) AS wrong_count,
+                 SUM(lapse_count) AS lapse_count,
+                 MAX(stage) AS stage,
+                 MAX(interval_days) AS interval_days,
+                 MAX(last_reviewed_at) AS last_reviewed_at,
+                 MIN(next_due_at) AS next_due_at
+          FROM ranked
+          GROUP BY user_id, new_item_key
+        )
+        UPDATE community.vocab_srs_item AS keeper
+        SET correct_count = aggregates.correct_count,
+            wrong_count = aggregates.wrong_count,
+            lapse_count = aggregates.lapse_count,
+            stage = aggregates.stage,
+            interval_days = aggregates.interval_days,
+            last_reviewed_at = aggregates.last_reviewed_at,
+            next_due_at = aggregates.next_due_at,
+            updated_at = NOW()
+        FROM aggregates
+        WHERE keeper.id = aggregates.keeper_id;
+      `, {
+        replacements: {
+          target: TARGET,
+          shortAnswer: SHORT_ANSWER,
+          normalizedCombinedAnswer: NORMALIZED_COMBINED_ANSWER,
+          normalizedTarget: NORMALIZED_TARGET
+        },
+        transaction
+      });
+
+      await queryInterface.sequelize.query(`
+        WITH candidates AS (
+          SELECT id, user_id,
+                 ENCODE(DIGEST(CONCAT_WS(
+                   '|',
+                   course_id::text,
+                   COALESCE(lesson_id::text, 'course'),
+                   UPPER(direction),
+                   :normalizedCombinedAnswer,
+                   :normalizedTarget
+                 ), 'sha1'), 'hex') AS new_item_key
+          FROM community.vocab_srs_item
+          WHERE learning = :shortAnswer
+            AND reference = :target
+        ), ranked AS (
+          SELECT id, ROW_NUMBER() OVER (PARTITION BY user_id, new_item_key ORDER BY id) AS position
+          FROM candidates
+        )
+        DELETE FROM community.vocab_srs_item AS duplicate
+        USING ranked
+        WHERE duplicate.id = ranked.id
+          AND ranked.position > 1;
+      `, {
+        replacements: {
+          target: TARGET,
+          shortAnswer: SHORT_ANSWER,
+          normalizedCombinedAnswer: NORMALIZED_COMBINED_ANSWER,
+          normalizedTarget: NORMALIZED_TARGET
+        },
+        transaction
+      });
+
       await queryInterface.sequelize.query(`
         UPDATE community.vocab_srs_item
         SET learning = :combinedAnswer,