merge duplicate posts #348

app/models.py

@@ -1117,7 +1117,7 @@ class Post(db.Model):
     cross_posts = db.Column(MutableList.as_mutable(ARRAY(db.Integer)))
     tags = db.relationship('Tag', lazy='dynamic', secondary=post_tag, backref=db.backref('posts', lazy='dynamic'))
 
-    ap_id = db.Column(db.String(255), index=True)
+    ap_id = db.Column(db.String(255), index=True, unique=True)
     ap_create_id = db.Column(db.String(100))
     ap_announce_id = db.Column(db.String(100))
 

migrations/versions/299e0384c8f3_unique_post_ap_id.py

@@ -0,0 +1,72 @@
+"""unique post ap id
+
+Revision ID: 299e0384c8f3
+Revises: 7f7dfd4d4f1b
+Create Date: 2024-11-24 15:18:51.979847
+
+"""
+from alembic import op
+import sqlalchemy as sa
+from sqlalchemy import text
+
+# revision identifiers, used by Alembic.
+revision = '299e0384c8f3'
+down_revision = '7f7dfd4d4f1b'
+branch_labels = None
+depends_on = None
+
+
+def upgrade():
+    # Find duplicate posts by ap_id
+    dupes_query = text('''
+                SELECT ap_id FROM "post" 
+                GROUP BY ap_id 
+                HAVING COUNT(*) > 1
+            ''')
+
+    conn = op.get_bind()
+    duplicate_posts = conn.execute(dupes_query).scalars()
+    print('Cleaning up duplicate posts, this may take a while...')
+
+    for ap_id in duplicate_posts:
+        if ap_id is None:
+            continue
+        # Get all communities with the same ap_profile_id, ordered by ID
+        users_query = text('''
+                    SELECT id FROM "post" 
+                    WHERE ap_id = :ap_id 
+                    ORDER BY id
+                ''')
+        posts = conn.execute(users_query, {"ap_id": ap_id}).fetchall()
+
+        # Set the lowest ID as the new_id, and collect other IDs to update/delete
+        new_id = posts[0].id
+        old_ids = [post.id for post in posts[1:]]
+
+        print(ap_id)
+
+        if old_ids:
+            # Update tables with batch IN clause
+            conn.execute(text('UPDATE "post_reply" SET post_id = :new_id WHERE post_id IN :old_ids'), {"new_id": new_id, "old_ids": tuple(old_ids)})
+            conn.execute(text('DELETE FROM "post_tag" WHERE post_id IN :old_ids'), {"old_ids": tuple(old_ids)})
+            conn.execute(text('DELETE FROM "read_posts" WHERE read_post_id IN :old_ids'), {"old_ids": tuple(old_ids)})
+            conn.execute(text('DELETE FROM "post_vote" WHERE post_id IN :old_ids'), {"old_ids": tuple(old_ids)})
+            conn.execute(text('DELETE FROM "report" WHERE suspect_post_id IN :old_ids'), {"old_ids": tuple(old_ids)})
+
+            # Delete the duplicate posts
+            conn.execute(text('DELETE FROM "post" WHERE id IN :old_ids'), {"old_ids": tuple(old_ids)})
+    # ### commands auto generated by Alembic - please adjust! ###
+    with op.batch_alter_table('post', schema=None) as batch_op:
+        batch_op.drop_index('ix_post_ap_id')
+        batch_op.create_index(batch_op.f('ix_post_ap_id'), ['ap_id'], unique=True)
+
+    # ### end Alembic commands ###
+
+
+def downgrade():
+    # ### commands auto generated by Alembic - please adjust! ###
+    with op.batch_alter_table('post', schema=None) as batch_op:
+        batch_op.drop_index(batch_op.f('ix_post_ap_id'))
+        batch_op.create_index('ix_post_ap_id', ['ap_id'], unique=False)
+
+    # ### end Alembic commands ###