More refactoring

src/alpine_bits_python/db.py

@@ -12,6 +12,7 @@ from sqlalchemy import (
     DateTime,
     Double,
     ForeignKey,
+    ForeignKeyConstraint,
     Integer,
     String,
 )
@@ -368,19 +369,17 @@ class ConversionGuest(Base):
     """Guest information from hotel PMS conversions, with hashed fields for privacy.
 
     Stores both unhashed (for reference during transition) and hashed (SHA256 per Meta API)
-    versions of guest PII. Multiple conversions can reference the same guest if they have
-    the same hotel_id and guest_id (PMS guest identifier).
+    versions of guest PII. Uses composite primary key (hotel_id, guest_id) from the PMS.
 
     When multiple conversions for the same guest arrive with different guest info,
-    the most recent (by creation_time) data is kept as the canonical version.
+    the most recent (by last_seen) data is kept as the canonical version.
     """
 
     __tablename__ = "conversion_guests"
-    id = Column(Integer, primary_key=True)
 
-    # Natural keys from PMS (composite unique constraint)
-    hotel_id = Column(String, nullable=False, index=True)
-    guest_id = Column(String, index=True)  # PMS guest ID (nullable for unidentified guests)
+    # Natural keys from PMS - composite primary key
+    hotel_id = Column(String, nullable=False, primary_key=True, index=True)
+    guest_id = Column(String, nullable=False, primary_key=True, index=True)
 
     # Unhashed guest information (for reference/transition period)
     guest_first_name = Column(String)
@@ -396,6 +395,9 @@ class ConversionGuest(Base):
     hashed_country_code = Column(String(64))
     hashed_birth_date = Column(String(64))
 
+    # Guest classification
+    is_regular = Column(Boolean, default=False)  # True if guest has many prior stays before appearing in our reservations
+
     # Metadata
     first_seen = Column(DateTime(timezone=True))
     last_seen = Column(DateTime(timezone=True))
@@ -424,6 +426,7 @@ class ConversionGuest(Base):
         guest_country_code: str | None,
         guest_birth_date: Date | None,
         now: DateTime,
+        is_regular: bool = False,
     ):
         """Create a ConversionGuest from conversion guest data."""
         return cls(
@@ -441,6 +444,7 @@ class ConversionGuest(Base):
             hashed_birth_date=cls._normalize_and_hash(
                 guest_birth_date.isoformat() if guest_birth_date else None
             ),
+            is_regular=is_regular,
             first_seen=now,
             last_seen=now,
         )
@@ -544,6 +548,12 @@ class Conversion(Base):
     The tracking data transferered by the PMS is however somewhat shorter.
     We therefore also need to match on guest name/email and other metadata.
 
+    Attribution flags:
+    - directly_attributable: True if matched by ID (reservation_id is set), meaning
+      this conversion is directly responsible for this reservation
+    - guest_matched: True if matched only by guest details (customer_id/hashed_customer_id set),
+      meaning the same person made this request but the reservation may not be directly attributable
+
     """
 
     __tablename__ = "conversions"
@@ -557,12 +567,10 @@ class Conversion(Base):
     hashed_customer_id = Column(
         Integer, ForeignKey("hashed_customers.id"), nullable=True, index=True
     )
-    conversion_guest_id = Column(
-        Integer, ForeignKey("conversion_guests.id"), nullable=True, index=True
-    )
 
     # Reservation metadata from XML
     hotel_id = Column(String, index=True)  # hotelID attribute
+    guest_id = Column(String, nullable=True, index=True)  # PMS guest ID, FK to conversion_guests
     pms_reservation_id = Column(String, index=True)  # id attribute from reservation
     reservation_number = Column(String)  # number attribute
     reservation_date = Column(Date)  # date attribute (when reservation was made)
@@ -570,13 +578,8 @@ class Conversion(Base):
     reservation_type = Column(String)  # type attribute (e.g., "reservation")
     booking_channel = Column(String)  # bookingChannel attribute
 
-    # Guest information from reservation XML - used for matching
-    guest_first_name = Column(String, index=True)  # firstName from guest element
-    guest_last_name = Column(String, index=True)  # lastName from guest element
-    guest_email = Column(String, index=True)  # email from guest element
-    guest_country_code = Column(String)  # countryCode from guest element
-    guest_birth_date = Column(Date)  # birthDate from guest element
-    guest_id = Column(String)  # id from guest element
+
+
 
     # Advertising/tracking data - used for matching to existing reservations
     advertising_medium = Column(
@@ -589,10 +592,23 @@ class Conversion(Base):
         String, index=True
     )  # advertisingCampagne (contains fbclid/gclid)
 
+    # Attribution flags - track how this conversion was matched
+    directly_attributable = Column(Boolean, default=False)  # Matched by ID (high confidence)
+    guest_matched = Column(Boolean, default=False)  # Matched by guest details only
+
     # Metadata
     created_at = Column(DateTime(timezone=True))  # When this record was imported
     updated_at = Column(DateTime(timezone=True))  # When this record was last updated
 
+    # Composite foreign key constraint for ConversionGuest (hotel_id, guest_id)
+    __table_args__ = (
+        ForeignKeyConstraint(
+            ["hotel_id", "guest_id"],
+            ["conversion_guests.hotel_id", "conversion_guests.guest_id"],
+            ondelete="SET NULL",
+        ),
+    )
+
     # Relationships
     reservation = relationship("Reservation", backref="conversions")
     customer = relationship("Customer", backref="conversions")