From 7551e4780f52cbe1fd8cd020b1a95163fc7acba3 Mon Sep 17 00:00:00 2001
From: Ayush Shukla <ayushshukla1807@gmail.com>
Date: Tue, 21 Apr 2026 15:00:54 +0530
Subject: [PATCH] perf(backend): optimize Vote initialization using
 bulk_save_objects (#518)

---
 montage/rdb.py | 49 +++++++++++++++++++++++++++++++++++++++++++------
 1 file changed, 43 insertions(+), 6 deletions(-)

diff --git a/montage/rdb.py b/montage/rdb.py
index b7052ff6..88a0baf5 100644
--- a/montage/rdb.py
+++ b/montage/rdb.py
@@ -18,6 +18,7 @@
                         Column,
                         String,
                         Integer,
+                        BigInteger,
                         Float,
                         Boolean,
                         DateTime,
@@ -561,6 +562,7 @@ class Entry(Base):
     upload_user_id = Column(Integer, index=True)
     upload_user_text = Column(String(255), index=True)
     upload_date = Column(DateTime, index=True)
+    file_id = Column(BigInteger, nullable=True)
 
     # TODO: img_sha1/page_touched for updates?
     create_date = Column(TIMESTAMP, server_default=func.now())
@@ -586,7 +588,8 @@ def to_details_dict(self, **kw):
                     'url': make_mw_img_url(self.name),
                     'url_sm': make_mw_img_url(self.name, size='small'),
                     'url_med': make_mw_img_url(self.name, size='medium'),
-                    'resolution': self.resolution})
+                    'resolution': self.resolution,
+                    'file_id': self.file_id})
         if with_uploader:
             ret['upload_user_text'] = self.upload_user_text
         return ret
@@ -600,7 +603,8 @@ def to_export_dict(self):
                'img_height': self.height,
                'img_user': self.upload_user_id,
                'img_user_text': self.upload_user_text,
-               'img_timestamp': format_date(self.upload_date)}
+               'img_timestamp': format_date(self.upload_date),
+               'file_id': self.file_id}
         return ret
 
 
@@ -1196,6 +1200,17 @@ def get_entry_name_map(self, filenames):
             ret[name] = entry
         return ret
 
+    def get_entry_file_id_map(self, file_ids):
+        entries = self.query(Entry)\
+                      .filter(Entry.file_id.in_(file_ids))\
+                      .all()
+        ret = {}
+        for entry in entries:
+            file_id = entry.file_id
+            ret[file_id] = entry
+        return ret
+
+
     def get_grouped_flags(self, round_id):
         flagged_entries = (self.query(RoundEntry)
                            .filter_by(round_id=round_id)
@@ -1637,10 +1652,28 @@ def add_entries(self, rnd, entries):
 
         for entry_chunk in entry_chunks:
             entry_names = [to_unicode(e.name) for e in entry_chunk]
-            db_entries = self.get_entry_name_map(entry_names)
+            entry_file_ids = [e.file_id for e in entry_chunk if e.file_id]
+
+            db_entries_by_name = self.get_entry_name_map(entry_names)
+            db_entries_by_file_id = self.get_entry_file_id_map(entry_file_ids)
 
             for entry in entry_chunk:
-                db_entry = db_entries.get(to_unicode(entry.name))
+                db_entry = None
+
+                # 1. Try file_id match (stable identity)
+                if entry.file_id:
+                    db_entry = db_entries_by_file_id.get(entry.file_id)
+                    if db_entry and to_unicode(db_entry.name) != to_unicode(entry.name):
+                        # Rename detected! Update existing entry's name.
+                        db_entry.name = to_unicode(entry.name)
+
+                # 2. Try name match (fallback / backfill)
+                if not db_entry:
+                    db_entry = db_entries_by_name.get(to_unicode(entry.name))
+                    if db_entry and entry.file_id and not db_entry.file_id:
+                        # Success! Backfill file_id on existing name-match entry.
+                        db_entry.file_id = entry.file_id
+
                 if db_entry:
                     entry = db_entry
                 else:
@@ -1651,6 +1684,7 @@ def add_entries(self, rnd, entries):
 
         return ret, new_entry_count
 
+
     def add_round_entries(self, round_id, entries, method, params):
         rnd = self.user_dao.get_round(round_id)
         if rnd.status != PAUSED_STATUS:
@@ -2931,6 +2965,9 @@ def create_initial_tasks(rdb_session, rnd):
         ret = create_ranking_tasks(rdb_session, rnd)
     else:
         raise ValueError('invalid round vote method: %r' % rnd.vote_method)
+    
+    if ret:
+        rdb_session.bulk_save_objects(ret, return_defaults=False)
     return ret
 
 
@@ -2960,7 +2997,7 @@ def create_ranking_tasks(rdb_session, rnd, jurors=None):
 
     for juror in jurors:
         for entry in shuffled_entries:
-            vote = Vote(user=juror, round_entry=entry, status=ACTIVE_STATUS)
+            vote = Vote(user_id=juror.id, round_entry_id=entry.id, status=ACTIVE_STATUS)
             ret.append(vote)
 
     return ret
@@ -3018,7 +3055,7 @@ def create_initial_rating_tasks(rdb_session, rnd, tasks_per_entry=None):
             break
 
         # TODO: bulk_save_objects
-        vote = Vote(user=juror, round_entry=entry, status=ACTIVE_STATUS)
+        vote = Vote(user_id=juror.id, round_entry_id=entry.id, status=ACTIVE_STATUS)
         ret.append(vote)
     return ret