feat: Enhanced CE schema and Notion sync (Pains/Gains)

company-explorer/backend/database.py

@@ -18,21 +18,35 @@ class Company(Base):
 
     id = Column(Integer, primary_key=True, index=True)
     
-    # Core Identity
+    # Core Identity (Golden Record - from Research)
     name = Column(String, index=True)
     website = Column(String, index=True) # Normalized Domain preferred
     crm_id = Column(String, unique=True, index=True, nullable=True) # Link to D365
     
+    # CRM Original Data (Source of Truth for Import)
+    crm_name = Column(String, nullable=True)
+    crm_website = Column(String, nullable=True)
+    crm_address = Column(String, nullable=True) # Full address string or JSON
+    crm_vat = Column(String, nullable=True)
+    
     # Classification
     industry_crm = Column(String, nullable=True) # The "allowed" industry
     industry_ai = Column(String, nullable=True)  # The AI suggested industry
     
-    # Location
+    # Location (Golden Record)
     city = Column(String, nullable=True)
     country = Column(String, default="DE")
     
     # Workflow Status
-    status = Column(String, default="NEW", index=True)
+    status = Column(String, default="NEW", index=True) # NEW, TO_ENRICH, ENRICHED, QUALIFIED, DISQUALIFIED
+    
+    # Quality & Confidence
+    confidence_score = Column(Float, default=0.0) # Overall confidence
+    data_mismatch_score = Column(Float, default=0.0) # 0.0=Match, 1.0=Mismatch
+    
+    # Scraping Status Flags
+    website_scrape_status = Column(String, default="PENDING") # PENDING, SUCCESS, FAILED, BLOCKED
+    wiki_search_status = Column(String, default="PENDING")    # PENDING, FOUND, NOT_FOUND
 
     # Granular Process Tracking (Timestamps)
     created_at = Column(DateTime, default=datetime.utcnow)
@@ -106,6 +120,13 @@ class Industry(Base):
     status_notion = Column(String, nullable=True) # e.g. "P1 Focus Industry"
     is_focus = Column(Boolean, default=False) # Derived from status_notion
     
+    # Enhanced Fields (v3.1 - Pains/Gains/Priority)
+    pains = Column(Text, nullable=True)
+    gains = Column(Text, nullable=True)
+    notes = Column(Text, nullable=True)
+    priority = Column(String, nullable=True) # Replaces old status concept ("Freigegeben")
+    ops_focus_secondary = Column(Boolean, default=False)
+
     # NEW SCHEMA FIELDS (from MIGRATION_PLAN)
     metric_type = Column(String, nullable=True) # Unit_Count, Area_in, Area_out
     min_requirement = Column(Float, nullable=True)
@@ -117,6 +138,7 @@ class Industry(Base):
     
     # Optional link to a Robotics Category (the "product" relevant for this industry)
     primary_category_id = Column(Integer, ForeignKey("robotics_categories.id"), nullable=True)
+    secondary_category_id = Column(Integer, ForeignKey("robotics_categories.id"), nullable=True)
     
     created_at = Column(DateTime, default=datetime.utcnow)