Files
bakery-ia/services/distribution/app/services/routing_optimizer.py
2025-11-30 09:12:40 +01:00

457 lines
19 KiB
Python

"""
Routing optimizer for the distribution service using Google OR-Tools VRP
"""
import logging
from typing import List, Dict, Any, Optional, Tuple
from datetime import datetime, timedelta
import time
# Google OR-Tools - Vehicle Routing Problem
try:
from ortools.constraint_solver import routing_enums_pb2
from ortools.constraint_solver import pywrapcp
HAS_ORTOOLS = True
except ImportError:
print("Warning: OR-Tools not installed. Using fallback routing algorithm.")
HAS_ORTOOLS = False
logger = logging.getLogger(__name__)
class RoutingOptimizer:
"""
Vehicle Routing Problem optimizer using Google OR-Tools
"""
def __init__(self):
self.has_ortools = HAS_ORTOOLS
async def optimize_daily_routes(
self,
deliveries: List[Dict[str, Any]],
depot_location: Tuple[float, float],
vehicle_capacity_kg: Optional[float] = 1000.0,
time_limit_seconds: float = 30.0
) -> Dict[str, Any]:
"""
Optimize daily delivery routes using VRP
Args:
deliveries: List of delivery dictionaries with keys:
- id: str - delivery ID
- location: Tuple[float, float] - (lat, lng)
- weight_kg: float - weight of delivery
- time_window: Optional[Tuple[str, str]] - delivery time window
depot_location: Tuple[float, float] - depot location (lat, lng)
vehicle_capacity_kg: Maximum weight capacity per vehicle
time_limit_seconds: Time limit for optimization (timeout)
Returns:
Dict with optimized route sequences and metadata
"""
if not self.has_ortools:
logger.warning("OR-Tools not available, using fallback sequential routing")
return self._fallback_sequential_routing(deliveries, depot_location)
start_time = time.time()
try:
# Prepare data for VRP
locations = [depot_location] # Depot is first location (index 0)
demands = [0] # Depot has no demand
time_windows = [(0, 24*60)] # Depot available all day (in minutes from midnight)
delivery_mapping = {}
for i, delivery in enumerate(deliveries, 1):
locations.append(delivery['location'])
# Ensure demands are integers for OR-Tools compatibility
weight_kg = delivery.get('weight_kg', 0)
demands.append(int(weight_kg) if isinstance(weight_kg, (int, float)) else 0)
# Convert time windows to minutes from midnight
time_window = delivery.get('time_window', None)
if time_window:
start_time_str, end_time_str = time_window
start_minutes = self._time_to_minutes(start_time_str)
end_minutes = self._time_to_minutes(end_time_str)
time_windows.append((int(start_minutes), int(end_minutes)))
else:
time_windows.append((0, 24*60)) # Default to all day if no time window
delivery_mapping[i] = delivery['id']
# Check if we have no deliveries (only depot), return early with empty route
if len(locations) <= 1: # Only depot, no deliveries
logger.info("No deliveries to optimize, returning empty route")
return {
'routes': [],
'total_distance_km': 0,
'optimization_time_seconds': time.time() - start_time,
'algorithm_used': 'ortools_vrp',
'status': 'success'
}
# Calculate total demand first before checking it
total_demand = sum(demands)
# Check if total demand is 0 but we have deliveries - handle this case too
if total_demand == 0 and len(locations) > 1:
logger.info("Total demand is 0 but deliveries exist, returning simple route")
# Create simple route with all deliveries but no capacity constraints
simple_route = {
'route_number': 1,
'route_sequence': [delivery_mapping[i] for i in range(1, len(locations))],
'stops': [{
'stop_number': i,
'delivery_id': delivery_mapping.get(i, f"delivery_{i}"),
'sequence': i - 1
} for i in range(1, len(locations))],
'total_weight_kg': 0
}
return {
'routes': [simple_route],
'total_distance_km': 0,
'optimization_time_seconds': time.time() - start_time,
'algorithm_used': 'ortools_vrp_zero_demand',
'status': 'success'
}
# Calculate distance matrix using haversine formula
distance_matrix = self._calculate_distance_matrix(locations)
# Create VRP model
# Calculate required vehicles (total_demand already calculated above)
# Ensure at least 1 vehicle, and enough to cover demand plus buffer
min_vehicles = max(1, int(total_demand / vehicle_capacity_kg) + 1)
# Add a buffer vehicle just in case
num_vehicles = int(min_vehicles + 1)
logger.info(f"VRP Optimization: Demand={total_demand}kg, Capacity={vehicle_capacity_kg}kg, Vehicles={num_vehicles}")
# Create VRP model
manager = pywrapcp.RoutingIndexManager(
len(distance_matrix), # number of locations
num_vehicles, # number of vehicles
[0] * num_vehicles, # depot index for starts
[0] * num_vehicles # depot index for ends
)
routing = pywrapcp.RoutingModel(manager)
def distance_callback(from_index, to_index):
"""Returns the distance between the two nodes."""
from_node = manager.IndexToNode(from_index)
to_node = manager.IndexToNode(to_index)
return distance_matrix[from_node][to_node]
transit_callback_index = routing.RegisterTransitCallback(distance_callback)
routing.SetArcCostEvaluatorOfAllVehicles(transit_callback_index)
# Add capacity constraint
def demand_callback(index):
"""Returns the demand of the node."""
node = manager.IndexToNode(index)
return int(demands[node]) # Ensure demands are integers
demand_callback_index = routing.RegisterUnaryTransitCallback(demand_callback)
routing.AddDimensionWithVehicleCapacity(
demand_callback_index,
0, # null capacity slack
[int(vehicle_capacity_kg)] * num_vehicles, # vehicle maximum capacities (as integers)
True, # start cumul to zero
'Capacity'
)
# Add time window constraint
def time_callback(from_index, to_index):
"""Returns the travel time between the two nodes."""
from_node = manager.IndexToNode(from_index)
to_node = manager.IndexToNode(to_index)
# Calculate travel time based on distance (meters) and assumed speed (km/h)
distance_m = distance_matrix[from_node][to_node]
distance_km = distance_m / 1000.0 # Convert meters to km
# Assume 30 km/h average speed for city deliveries
travel_time_minutes = (distance_km / 30.0) * 60.0
return int(travel_time_minutes)
time_callback_index = routing.RegisterTransitCallback(time_callback)
routing.AddDimension(
time_callback_index,
60 * 24, # Allow waiting time (24 hours in minutes)
60 * 24, # Maximum time per vehicle (24 hours in minutes)
False, # Don't force start cumul to zero
'Time'
)
time_dimension = routing.GetDimensionOrDie('Time')
# Add time window constraints for each location
for location_idx in range(len(locations)):
index = manager.NodeToIndex(location_idx)
if index != -1: # Valid index
min_time, max_time = time_windows[location_idx]
time_dimension.CumulVar(index).SetRange(int(min_time), int(max_time))
# Setting first solution heuristic
search_parameters = pywrapcp.DefaultRoutingSearchParameters()
search_parameters.first_solution_strategy = (
routing_enums_pb2.FirstSolutionStrategy.PATH_CHEAPEST_ARC
)
search_parameters.time_limit.FromSeconds(time_limit_seconds)
# Solve the problem
solution = routing.SolveWithParameters(search_parameters)
# Check if solution was found
if solution:
optimized_routes = self._extract_routes(routing, manager, solution, delivery_mapping)
# Calculate total distance and duration
total_distance = 0
total_duration = 0
for route in optimized_routes:
route_distance = 0
for stop in route['stops']:
route_distance += stop.get('distance_to_next', 0)
route['total_distance_km'] = route_distance
total_distance += route_distance
logger.info(f"VRP optimization completed in {time.time() - start_time:.2f}s")
return {
'routes': optimized_routes,
'total_distance_km': total_distance,
'optimization_time_seconds': time.time() - start_time,
'algorithm_used': 'ortools_vrp',
'status': 'success'
}
else:
logger.warning("OR-Tools failed to find solution, using fallback routing")
return self._fallback_sequential_routing(deliveries, depot_location)
except Exception as e:
logger.error(f"Error in VRP optimization: {e}")
# Fallback to simple sequential routing
return self._fallback_sequential_routing(deliveries, depot_location)
def _calculate_distance_matrix(self, locations: List[Tuple[float, float]]) -> List[List[int]]:
"""
Calculate distance matrix using haversine formula (in meters)
"""
import math
def haversine_distance(lat1, lon1, lat2, lon2):
"""Calculate distance between two lat/lon points in meters"""
R = 6371000 # Earth's radius in meters
lat1, lon1, lat2, lon2 = map(math.radians, [lat1, lon1, lat2, lon2])
dlat = lat2 - lat1
dlon = lon2 - lon1
a = math.sin(dlat/2)**2 + math.cos(lat1) * math.cos(lat2) * math.sin(dlon/2)**2
c = 2 * math.asin(math.sqrt(a))
return R * c # Distance in meters
n = len(locations)
matrix = [[0] * n for _ in range(n)]
for i in range(n):
for j in range(n):
if i != j:
lat1, lon1 = locations[i]
lat2, lon2 = locations[j]
dist_m = haversine_distance(lat1, lon1, lat2, lon2)
matrix[i][j] = int(dist_m)
return matrix
def _extract_routes(self, routing, manager, solution, delivery_mapping) -> List[Dict[str, Any]]:
"""
Extract routes from OR-Tools solution
"""
routes = []
for vehicle_id in range(manager.GetNumberOfVehicles()):
index = routing.Start(vehicle_id)
# Skip if vehicle is not used (Start -> End directly)
if routing.IsEnd(solution.Value(routing.NextVar(index))):
continue
current_route = {
'route_number': vehicle_id + 1,
'stops': [],
'total_weight_kg': 0
}
# Initialize route sequence to store the delivery IDs in visit order
route_sequence = []
# Add depot as first stop
node_index = manager.IndexToNode(index)
delivery_id = delivery_mapping.get(node_index, f"depot_{node_index}")
current_route['stops'].append({
'stop_number': 1,
'delivery_id': delivery_id,
'location': 'depot',
'sequence': 0
})
stop_number = 1
while not routing.IsEnd(index):
index = solution.Value(routing.NextVar(index))
node_index = manager.IndexToNode(index)
if node_index != 0: # Not depot
stop_number += 1
delivery_id = delivery_mapping.get(node_index, f"delivery_{node_index}")
current_route['stops'].append({
'stop_number': stop_number,
'delivery_id': delivery_id,
'location_index': node_index,
'sequence': stop_number
})
# Add delivery ID to route sequence (excluding depot stops)
route_sequence.append(delivery_id)
else: # Back to depot
stop_number += 1
current_route['stops'].append({
'stop_number': stop_number,
'delivery_id': f"depot_end_{vehicle_id + 1}",
'location': 'depot',
'sequence': stop_number
})
break
# Add the route_sequence to the current route
current_route['route_sequence'] = route_sequence
routes.append(current_route)
return routes
def _time_to_minutes(self, time_str: str) -> int:
"""
Convert HH:MM string to minutes from midnight
"""
if ":" in time_str:
hour, minute = map(int, time_str.split(":"))
return hour * 60 + minute
else:
# If it's already in minutes, return as is
return int(time_str)
def _fallback_sequential_routing(self, deliveries: List[Dict[str, Any]], depot_location: Tuple[float, float]) -> Dict[str, Any]:
"""
Fallback routing algorithm that sequences deliveries sequentially
"""
import math
def haversine_distance(lat1, lon1, lat2, lon2):
"""Calculate distance between two lat/lon points in km"""
R = 6371 # Earth's radius in km
lat1, lon1, lat2, lon2 = map(math.radians, [lat1, lon1, lat2, lon2])
dlat = lat2 - lat1
dlon = lon2 - lon1
a = math.sin(dlat/2)**2 + math.cos(lat1) * math.cos(lat2) * math.sin(dlon/2)**2
c = 2 * math.asin(math.sqrt(a))
return R * c # Distance in km
# Calculate distances from depot to each delivery and between deliveries
deliveries_with_distance = []
for delivery in deliveries:
lat, lon = delivery['location']
depot_lat, depot_lon = depot_location
dist = haversine_distance(depot_lat, depot_lon, lat, lon)
deliveries_with_distance.append({
**delivery,
'distance_from_depot': dist
})
# Sort deliveries by distance from depot (nearest first)
deliveries_with_distance.sort(key=lambda x: x['distance_from_depot'])
# Create simple route
route_stops = []
total_distance = 0
# Start from depot
route_stops.append({
'stop_number': 1,
'delivery_id': 'depot_start',
'location': depot_location,
'sequence': 0,
'is_depot': True
})
# Add deliveries
for i, delivery in enumerate(deliveries_with_distance, 1):
route_stops.append({
'stop_number': i + 1,
'delivery_id': delivery['id'],
'location': delivery['location'],
'weight_kg': delivery.get('weight_kg', 0),
'sequence': i,
'is_depot': False
})
# Return to depot
route_stops.append({
'stop_number': len(deliveries_with_distance) + 2,
'delivery_id': 'depot_end',
'location': depot_location,
'sequence': len(deliveries_with_distance) + 1,
'is_depot': True
})
# Calculate total distance
for i in range(len(route_stops) - 1):
current_stop = route_stops[i]
next_stop = route_stops[i + 1]
if not current_stop['is_depot'] or not next_stop['is_depot']:
if not current_stop['is_depot'] and not next_stop['is_depot']:
# Between two deliveries
curr_lat, curr_lon = current_stop['location']
next_lat, next_lon = next_stop['location']
dist = haversine_distance(curr_lat, curr_lon, next_lat, next_lon)
elif current_stop['is_depot'] and not next_stop['is_depot']:
# From depot to delivery
depot_lat, depot_lon = current_stop['location']
del_lat, del_lon = next_stop['location']
dist = haversine_distance(depot_lat, depot_lon, del_lat, del_lon)
elif not current_stop['is_depot'] and next_stop['is_depot']:
# From delivery to depot
del_lat, del_lon = current_stop['location']
depot_lat, depot_lon = next_stop['location']
dist = haversine_distance(del_lat, del_lon, depot_lat, depot_lon)
else:
dist = 0 # depot to depot
total_distance += dist
route_stops[i]['distance_to_next'] = dist
# Create route sequence from delivery IDs in the order they appear
route_sequence = [stop['delivery_id'] for stop in route_stops if not stop.get('is_depot', False)]
return {
'routes': [{
'route_number': 1,
'stops': route_stops,
'route_sequence': route_sequence,
'total_distance_km': total_distance,
'total_weight_kg': sum(d.get('weight_kg', 0) for d in deliveries),
}],
'total_distance_km': total_distance,
'optimization_time_seconds': 0,
'algorithm_used': 'fallback_sequential',
'status': 'success'
}