Add black formatter (#15)

* Add black formatter to dependencies

* Document contributing changes

Add formatter check in CI tests

* Run black formatter

* Limit line length to 79 characters

* Add flake8 to docs

.github/workflows/python-app.yml

@@ -37,6 +37,9 @@ jobs:
     #- name: Check type hints with mypy
       #run: |
         #poetry run mypy --ignore-missing-imports .
+    - name: Check whether black formatter was run
+      run: |
+        poetry run black --check .
     - name: Run tests and check coverage
       run: |
         poetry run pytest tests/

docs/quickstart.md

@@ -165,3 +165,43 @@ poetry run python -m loggibud.v1.eval.task1 \
     --instance tests/results/cvrp-instances/train/rj-0-cvrp-0.json \
     --solution results/rj-0-cvrp-0.json
 ```
+
+
+## Contributing
+
+First of all, thanks for the interest in the project. If you found a bug or have any question, feel free to open an issue.
+
+If you prefer to contribute with code or documentation, first fork the repository, make the appropriate changes and open a pull request to our `master` branch.
+
+Notice we use Python Poetry to manage dependencies. So if you need to add, remove or update any dependency make sure to use the proper [`poetry`](https://python-poetry.org/docs/) commands to write the changes in the `pyproject.toml` and `poetry.lock` files.
+
+Moreover, before opening a pull request, make sure the following were taken care:
+
+- The `black` formatter was run:
+```bash
+poetry run black .
+```
+
+- The code is conformant with `flake8`:
+```bash
+poetry run flake8 .
+```
+
+- The tests are still passing:
+```bash
+poetry run pytest tests/
+```
+
+### Note to Windows users
+
+In some cases, Windows uses CRLF as end of line instead of LF, which is the norm in Unix-based systems. This erroneously makes git thinks that a whole file was changed when saved in different operating systems.
+
+To alleviate this issue, we recommend Windows users to do one of the following:
+
+- When installing Git for Windows, choose the option "Checkout Windows-style, commit Unix-style line endings" [(see this StackOverflow answer)](https://stackoverflow.com/questions/1889559/git-diff-to-ignore-m)
+
+- If Git is already installed, write the following in the LoggiBUD repository before making any commit:
+
+```bash
+git config core.whitespace cr-at-eol
+```

loggibud/v1/baselines/run_task1.py

@@ -27,7 +27,9 @@
     # Load method from path.
     module = importlib.import_module(args.module)
     method = getattr(module, args.method)
-    params_class = getattr(module, args.params_class) if args.params_class else None
+    params_class = (
+        getattr(module, args.params_class) if args.params_class else None
+    )
 
     # Load instance and heuristic params.
     path = Path(args.instances)

loggibud/v1/baselines/shared/ortools.py

@@ -68,13 +68,17 @@ def solve(
     model = pywrapcp.RoutingModel(manager)
 
     # Unwrap the size index for every point.
-    sizes = np.array([0] + [d.size for d in instance.deliveries], dtype=np.int32)
+    sizes = np.array(
+        [0] + [d.size for d in instance.deliveries], dtype=np.int32
+    )
 
     def capacity_callback(src):
         src = manager.IndexToNode(src)
         return sizes[src]
 
-    capacity_callback_index = model.RegisterUnaryTransitCallback(capacity_callback)
+    capacity_callback_index = model.RegisterUnaryTransitCallback(
+        capacity_callback
+    )
     model.AddDimension(
         capacity_callback_index, 0, instance.vehicle_capacity, True, "Capacity"
     )
@@ -84,7 +88,9 @@ def capacity_callback(src):
 
     # Compute the distance matrix between points.
     logger.info("Computing distance matrix.")
-    distance_matrix = (calculate_distance_matrix_m(locations) * 10).astype(np.int32)
+    distance_matrix = (calculate_distance_matrix_m(locations) * 10).astype(
+        np.int32
+    )
 
     def distance_callback(src, dst):
         x = manager.IndexToNode(src)
@@ -97,7 +103,9 @@ def distance_callback(src, dst):
     search_parameters = pywrapcp.DefaultRoutingSearchParameters()
     search_parameters.first_solution_strategy = params.first_solution_strategy
 
-    search_parameters.local_search_metaheuristic = params.local_search_metaheuristic
+    search_parameters.local_search_metaheuristic = (
+        params.local_search_metaheuristic
+    )
 
     if params.solution_limit:
         search_parameters.solution_limit = params.solution_limit

loggibud/v1/baselines/task1/kmeans_aggregation_ortools.py

@@ -21,7 +21,12 @@
 import numpy as np
 from sklearn.cluster import MiniBatchKMeans
 
-from loggibud.v1.types import CVRPInstance, CVRPSolution, CVRPSolutionVehicle, Delivery
+from loggibud.v1.types import (
+    CVRPInstance,
+    CVRPSolution,
+    CVRPSolutionVehicle,
+    Delivery,
+)
 from ..shared.ortools import solve_cvrp as ortools_solve, ORToolsParams
 
 
@@ -64,7 +69,9 @@ def solve(
     logger.info(f"Clustering instance into {num_clusters} subinstances")
     clustering = MiniBatchKMeans(num_clusters, random_state=params.seed)
 
-    points = np.array([[d.point.lng, d.point.lat] for d in instance.deliveries])
+    points = np.array(
+        [[d.point.lng, d.point.lat] for d in instance.deliveries]
+    )
     clusters = clustering.fit_predict(points)
 
     delivery_array = np.array(instance.deliveries)
@@ -123,7 +130,11 @@ def aggregate_deliveries(idx, deliveries):
             deliveries=[
                 d
                 for v in solve_cluster(
-                    [d for groups in v.deliveries for d in subsolutions[int(groups.id)]]
+                    [
+                        d
+                        for groups in v.deliveries
+                        for d in subsolutions[int(groups.id)]
+                    ]
                 )
                 for d in v
             ],

loggibud/v1/baselines/task1/kmeans_partition_ortools.py

@@ -59,13 +59,17 @@ def solve(
     num_deliveries = len(instance.deliveries)
     num_clusters = int(
         params.fixed_num_clusters
-        or np.ceil(num_deliveries / (params.variable_num_clusters or num_deliveries))
+        or np.ceil(
+            num_deliveries / (params.variable_num_clusters or num_deliveries)
+        )
     )
 
     logger.info(f"Clustering instance into {num_clusters} subinstances")
     clustering = KMeans(num_clusters, random_state=params.seed)
 
-    points = np.array([[d.point.lng, d.point.lat] for d in instance.deliveries])
+    points = np.array(
+        [[d.point.lng, d.point.lat] for d in instance.deliveries]
+    )
     clusters = clustering.fit_predict(points)
 
     delivery_array = np.array(instance.deliveries)

loggibud/v1/baselines/task1/lkh_3.py

@@ -16,7 +16,10 @@
 import numpy as np
 
 from loggibud.v1.types import (
-    CVRPInstance, CVRPSolution, CVRPSolutionVehicle, JSONDataclassMixin
+    CVRPInstance,
+    CVRPSolution,
+    CVRPSolutionVehicle,
+    JSONDataclassMixin,
 )
 from loggibud.v1.data_conversion import to_tsplib
 

loggibud/v1/baselines/task2/kmeans_greedy.py

@@ -16,7 +16,12 @@
 from sklearn.cluster import KMeans
 from tqdm import tqdm
 
-from loggibud.v1.types import Delivery, CVRPInstance, CVRPSolution, CVRPSolutionVehicle
+from loggibud.v1.types import (
+    Delivery,
+    CVRPInstance,
+    CVRPSolution,
+    CVRPSolutionVehicle,
+)
 from loggibud.v1.baselines.shared.ortools import (
     solve as ortools_solve,
     ORToolsParams,
@@ -67,7 +72,9 @@ def pretrain(
     num_deliveries = len(points)
     num_clusters = int(
         params.fixed_num_clusters
-        or np.ceil(num_deliveries / (params.variable_num_clusters or num_deliveries))
+        or np.ceil(
+            num_deliveries / (params.variable_num_clusters or num_deliveries)
+        )
     )
 
     logger.info(f"Clustering instance into {num_clusters} subinstances")
@@ -80,13 +87,17 @@ def pretrain(
     )
 
 
-def finetune(model: KMeansGreedyModel, instance: CVRPInstance) -> KMeansGreedyModel:
+def finetune(
+    model: KMeansGreedyModel, instance: CVRPInstance
+) -> KMeansGreedyModel:
     """Prepare the model for one particular instance."""
 
     return KMeansGreedyModel(
         params=model.params,
         clustering=model.clustering,
-        cluster_subsolutions={i: [] for i in range(model.clustering.n_clusters)},
+        cluster_subsolutions={
+            i: [] for i in range(model.clustering.n_clusters)
+        },
         # Just fill some random instance.
         subinstance=instance,
     )
@@ -95,12 +106,17 @@ def finetune(model: KMeansGreedyModel, instance: CVRPInstance) -> KMeansGreedyMo
 def route(model: KMeansGreedyModel, delivery: Delivery) -> KMeansGreedyModel:
     """Route a single delivery using the model instance."""
 
-    cluster = model.clustering.predict([[delivery.point.lng, delivery.point.lat]])[0]
+    cluster = model.clustering.predict(
+        [[delivery.point.lng, delivery.point.lat]]
+    )[0]
 
     subsolution = model.cluster_subsolutions[cluster]
 
     def is_feasible(route):
-        return route.occupation + delivery.size < model.subinstance.vehicle_capacity
+        return (
+            route.occupation + delivery.size
+            < model.subinstance.vehicle_capacity
+        )
 
     # TODO: We could make this method faster by using a route size table, but seems a bit
     # overkill since it's not a bottleneck.
@@ -114,7 +130,9 @@ def is_feasible(route):
         route_idx, route = max(feasible_routes, key=lambda v: v[1].occupation)
 
     else:
-        route = CVRPSolutionVehicle(origin=model.subinstance.origin, deliveries=[])
+        route = CVRPSolutionVehicle(
+            origin=model.subinstance.origin, deliveries=[]
+        )
         subsolution.append(route)
         route_idx = len(subsolution) - 1
 
@@ -146,7 +164,9 @@ def finish(instance: CVRPInstance, model: KMeansGreedyModel) -> CVRPSolution:
 
     return CVRPSolution(
         name=instance.name,
-        vehicles=[v for subsolution in subsolutions for v in subsolution.vehicles],
+        vehicles=[
+            v for subsolution in subsolutions for v in subsolution.vehicles
+        ],
     )
 
 
@@ -179,11 +199,15 @@ def solve_instance(
     # Load instance and heuristic params.
     eval_path = Path(args.eval_instances)
     eval_path_dir = eval_path if eval_path.is_dir() else eval_path.parent
-    eval_files = [eval_path] if eval_path.is_file() else list(eval_path.iterdir())
+    eval_files = (
+        [eval_path] if eval_path.is_file() else list(eval_path.iterdir())
+    )
 
     train_path = Path(args.train_instances)
     train_path_dir = train_path if train_path.is_dir() else train_path.parent
-    train_files = [train_path] if train_path.is_file() else list(train_path.iterdir())
+    train_files = (
+        [train_path] if train_path.is_file() else list(train_path.iterdir())
+    )
 
     # params = params_class.from_file(args.params) if args.params else None
 

loggibud/v1/baselines/task2/qrp_sweep.py

@@ -49,7 +49,10 @@
 from tqdm import tqdm
 
 from loggibud.v1.types import (
-    Delivery, CVRPInstance, CVRPSolution, CVRPSolutionVehicle
+    Delivery,
+    CVRPInstance,
+    CVRPSolution,
+    CVRPSolutionVehicle,
 )
 from loggibud.v1.baselines.shared.ortools import (
     solve as ortools_solve,
@@ -99,8 +102,7 @@ def predict(self, delivery: Delivery) -> int:
         """
 
         point_translated = (
-            np.array([delivery.point.lng, delivery.point.lat])
-            - self.center
+            np.array([delivery.point.lng, delivery.point.lat]) - self.center
         )
         angle = np.arctan2(point_translated[1], point_translated[0])
 

loggibud/v1/data_conversion.py

@@ -53,36 +53,30 @@ def to_tsplib(
     tspfile += "\n"
 
     # Demand section
-    tspfile += (
-        "DEMAND_SECTION\n"
-        "1 0\n"
-    )
+    tspfile += "DEMAND_SECTION\n" "1 0\n"
     tspfile += "\n".join(
         f"{i} {delivery.size}"
         for i, delivery in enumerate(instance.deliveries, start=2)
     )
     tspfile += "\n"
 
     # Depot section: ensure node 1 is the depot (-1 to terminate the list)
-    tspfile += (
-        "DEPOT_SECTION\n"
-        "1\n"
-        "-1\n"
-    )
+    tspfile += "DEPOT_SECTION\n" "1\n" "-1\n"
 
     # Edge section:
     # Compute distance matrix
     locations = [instance.origin] + [
         delivery.point for delivery in instance.deliveries
     ]
-    distance_matrix = (
-        calculate_distance_matrix_m(locations) * 10
-    ).astype(np.int32)
+    distance_matrix = (calculate_distance_matrix_m(locations) * 10).astype(
+        np.int32
+    )
 
     tspfile += "EDGE_WEIGHT_SECTION\n"
 
     def print_row(row):
         return " ".join(str(el) for el in row)
+
     tspfile += "\n".join(print_row(row) for row in distance_matrix)
 
     if not file_name:

loggibud/v1/distances.py

@@ -24,7 +24,9 @@ def calculate_distance_matrix_m(
     if len(points) < 2:
         return 0
 
-    coords_uri = ";".join(["{},{}".format(point.lng, point.lat) for point in points])
+    coords_uri = ";".join(
+        ["{},{}".format(point.lng, point.lat) for point in points]
+    )
 
     response = requests.get(
         f"{config.host}/table/v1/driving/{coords_uri}?annotations=distance",
@@ -44,7 +46,9 @@ def calculate_route_distance_m(
     if len(points) < 2:
         return 0
 
-    coords_uri = ";".join("{},{}".format(point.lng, point.lat) for point in points)
+    coords_uri = ";".join(
+        "{},{}".format(point.lng, point.lat) for point in points
+    )
 
     response = requests.get(
         f"{config.host}/route/v1/driving/{coords_uri}?annotations=distance&continue_straight=false",
@@ -57,7 +61,7 @@ def calculate_route_distance_m(
 
 
 def calculate_distance_matrix_great_circle_m(
-    points: Iterable[Point]
+    points: Iterable[Point],
 ) -> np.ndarray:
     """Distance matrix using the Great Circle distance
     This is an Euclidean-like distance but on spheres [1]. In this case it is
@@ -88,16 +92,17 @@ def calculate_distance_matrix_great_circle_m(
 
     delta_sigma = np.arctan2(
         np.sqrt(
-            (np.cos(phi2) * np.sin(delta_lambda))**2
+            (np.cos(phi2) * np.sin(delta_lambda)) ** 2
             + (
                 np.cos(phi1) * np.sin(phi2)
                 - np.sin(phi1) * np.cos(phi2) * np.cos(delta_lambda)
-            )**2
+            )
+            ** 2
         ),
         (
             np.sin(phi1) * np.sin(phi2)
             + np.cos(phi1) * np.cos(phi2) * np.cos(delta_lambda)
-        )
+        ),
     )
 
     return EARTH_RADIUS_METERS * delta_sigma