Initial commit - WhyRating Engine (Google Reviews Scraper)

core/database.py

@@ -354,7 +354,11 @@ class DatabaseManager:
                     callback_status,
                     callback_attempts,
                     scraper_version,
-                    scraper_variant
+                    scraper_variant,
+                    business_name,
+                    business_category,
+                    business_address,
+                    business_rating
                 FROM jobs
                 WHERE job_id = $1
             """, job_id)
@@ -575,6 +579,69 @@ class DatabaseManager:
 
             log.debug(f"Updated session fingerprint for job {job_id}")
 
+    async def update_job_metadata(
+        self,
+        job_id: UUID,
+        metadata_updates: Dict[str, Any]
+    ):
+        """
+        Update specific fields in job metadata without overwriting existing data.
+
+        Args:
+            job_id: Job UUID
+            metadata_updates: Dictionary of metadata fields to update/add
+                - bot_detected: True if sort button was hidden (bot detection)
+                - initial_sort_used: Sort order used for scraping
+                - sort_orders_attempted: List of all sort orders tried
+                - multi_sort: Multi-sort completion info
+        """
+        async with self.pool.acquire() as conn:
+            # Merge new metadata with existing metadata using JSONB concatenation
+            await conn.execute("""
+                UPDATE jobs
+                SET
+                    metadata = COALESCE(metadata, '{}'::jsonb) || $2::jsonb,
+                    updated_at = NOW()
+                WHERE job_id = $1
+            """, job_id, json.dumps(metadata_updates))
+
+            log.debug(f"Updated job metadata for job {job_id}: {list(metadata_updates.keys())}")
+
+    async def update_business_info(
+        self,
+        job_id: UUID,
+        business_name: Optional[str] = None,
+        business_category: Optional[str] = None,
+        business_address: Optional[str] = None,
+        business_rating: Optional[float] = None
+    ):
+        """
+        Update business info columns for a job.
+
+        These are dedicated columns (not JSONB) for queryable business data
+        captured from the Google Maps page during scraping.
+
+        Args:
+            job_id: Job UUID
+            business_name: Business name from Google Maps
+            business_category: Business category (e.g., "Restaurant", "Toy store")
+            business_address: Full address from Google Maps
+            business_rating: Aggregate rating at time of scrape (e.g., 4.5)
+        """
+        async with self.pool.acquire() as conn:
+            await conn.execute("""
+                UPDATE jobs
+                SET
+                    business_name = COALESCE($2, business_name),
+                    business_category = COALESCE($3, business_category),
+                    business_address = COALESCE($4, business_address),
+                    business_rating = COALESCE($5, business_rating),
+                    updated_at = NOW()
+                WHERE job_id = $1
+            """, job_id, business_name, business_category, business_address, business_rating)
+
+            log.debug(f"Updated business info for job {job_id}: name={business_name}, category={business_category}")
+
     async def mark_job_partial(
         self,
         job_id: UUID,
@@ -674,7 +741,11 @@ class DatabaseManager:
                     callback_status,
                     callback_attempts,
                     scraper_version,
-                    scraper_variant
+                    scraper_variant,
+                    business_name,
+                    business_category,
+                    business_address,
+                    business_rating
                 FROM jobs
                 {where_clause}
                 ORDER BY created_at DESC