From f1e6b9db9b08b42645b286743caabb079513d9c3 Mon Sep 17 00:00:00 2001
From: qananasikq <qananasikq@gitea.millionmiles.ru>
Date: Fri, 17 Apr 2026 16:07:19 +0300
Subject: [PATCH] fix photo merge

---
 encar_scraper/encar.py | 38 +++++++++++++++++++++++++++++++++-----
 1 file changed, 33 insertions(+), 5 deletions(-)

diff --git a/encar_scraper/encar.py b/encar_scraper/encar.py
index e2694d0..cceaf75 100644
--- a/encar_scraper/encar.py
+++ b/encar_scraper/encar.py
@@ -651,6 +651,29 @@ class EncarScraper:
         record.slug = self.car_mapper._build_slug(record.brand, record.model, canonical_vehicle_id)
         return True
 
+    @staticmethod
+    def _merge_record_images(primary: CarRecord, candidate: CarRecord) -> bool:
+        merged: dict[str, ImageRecord] = {}
+
+        def _put(images: list[ImageRecord]) -> None:
+            for image in images:
+                key = image.fullres_image
+                existing = merged.get(key)
+                if existing is None or image.order_index < existing.order_index:
+                    merged[key] = image
+
+        _put(primary.images)
+        _put(candidate.images)
+
+        merged_images = sorted(merged.values(), key=lambda img: (img.order_index, img.fullres_image))
+        if len(merged_images) != len(primary.images) or any(
+            left.fullres_image != right.fullres_image or left.order_index != right.order_index
+            for left, right in zip(primary.images, merged_images)
+        ):
+            primary.images = merged_images
+            return True
+        return False
+
     def collect_listing(
         self,
         limit: int | None = None,
@@ -1215,22 +1238,27 @@ class EncarScraper:
                     canonicalized_records += 1
 
             deduped_records: list[CarRecord] = []
-            seen_origin_ids: set[str] = set()
+            deduped_by_origin_id: dict[str, CarRecord] = {}
             collapsed_duplicates = 0
+            merged_image_sets = 0
             for record in records:
-                if record.origin_id in seen_origin_ids:
+                existing = deduped_by_origin_id.get(record.origin_id)
+                if existing is not None:
                     collapsed_duplicates += 1
+                    if self._merge_record_images(existing, record):
+                        merged_image_sets += 1
                     continue
-                seen_origin_ids.add(record.origin_id)
+                deduped_by_origin_id[record.origin_id] = record
                 deduped_records.append(record)
                 all_origin_ids.add(record.origin_id)
             records = deduped_records
 
-            if canonicalized_records or collapsed_duplicates:
+            if canonicalized_records or collapsed_duplicates or merged_image_sets:
                 logger.info(
-                    "Canonicalized %d Encar records by photo vehicle id and collapsed %d duplicates in batch",
+                    "Canonicalized %d Encar records by photo vehicle id collapsed %d duplicates and merged %d image sets in batch",
                     canonicalized_records,
                     collapsed_duplicates,
+                    merged_image_sets,
                 )
 
         # Фаза 3: upsert в БД