Type hint and validate Instance and Solution dictionaries

pyproject.toml

@@ -18,6 +18,7 @@ pytest = "^7.1.2"
 codecov = "^2.1.13"
 pytest-cov = "^4.0.0"
 pre-commit = "^2.19.0"
+mypy = "^1.2.0"
 
 [pytest]
 pythonpath = [".", "vrplib"]

tests/parse/test_parse_solution.py

@@ -8,15 +8,17 @@
     "text, data",
     [
         ("", {"routes": []}),  # empty solution
-        (
-            "Route #1: 1 2 3\n Route #2: 5 6\n COST: 10",
-            {"routes": [[1, 2, 3], [5, 6]], "cost": 10},
-        ),
         (
             "Route #1: 1 \n Route #2: 6\n comment: VRPLIB",
             {"routes": [[1], [6]], "comment": "VRPLIB"},
         ),
         (
+            # lower case cost
+            "Route #1: 1 2 3\n Route #2: 5 6\n COST: 10",
+            {"routes": [[1, 2, 3], [5, 6]], "cost": 10},
+        ),
+        (
+            # parse time on whitespace instead of colon
             "Route #1: 1 \n Route #2: 6\n time 180.23",
             {"routes": [[1], [6]], "time": 180.23},
         ),

vrplib/Instance.py

@@ -0,0 +1,23 @@
+from typing import TypedDict
+
+import numpy as np
+
+
+class Instance(TypedDict, total=False):
+    # Specifications
+    name: str
+    dimension: int  # number of depots + number of customers
+    edge_weight_format: str
+    edge_weight_type: str
+    display_type: str
+    comment: str
+    vehicles: int  # number of vehicles
+    capacity: int
+
+    # Data sections
+    node_coord: np.ndarray
+    demand: np.ndarray
+    depot: np.ndarray
+    service_time: np.ndarray
+    time_window: np.ndarray
+    edge_weight: np.ndarray

vrplib/Solution.py

@@ -0,0 +1,6 @@
+from typing import List, TypedDict
+
+
+class Solution(TypedDict, total=False):
+    routes: List[List[int]]
+    cost: float

vrplib/parse/parse_distances.py

@@ -1,16 +1,18 @@
 from itertools import combinations
-from typing import List, Optional, Union
+from typing import List, Optional
 
 import numpy as np
 
+from vrplib.Instance import Instance
+
 
 def parse_distances(
     data: List,
     edge_weight_type: str,
     edge_weight_format: Optional[str] = None,
     node_coord: Optional[np.ndarray] = None,
     comment: Optional[str] = None,
-    **kwargs: Union[float, str, np.ndarray],  # noqa
+    **kwargs: Instance,  # noqa
 ) -> np.ndarray:
     """
     Parses the distances. The specification "edge_weight_type" describes how
@@ -35,8 +37,7 @@ def parse_distances(
 
     Returns
     -------
-    np.ndarray
-        An n-by-n distances matrix.
+    An n-by-n distances matrix.
     """
     if "2D" in edge_weight_type:  # Euclidean distance on node coordinates
         if node_coord is None:
@@ -83,8 +84,7 @@ def pairwise_euclidean(coords: np.ndarray) -> np.ndarray:
 
     Returns
     -------
-    np.ndarray
-        An n-by-n Euclidean distances matrix.
+    An n-by-n Euclidean distances matrix.
 
     """
     diff = coords[:, np.newaxis, :] - coords
@@ -93,7 +93,7 @@ def pairwise_euclidean(coords: np.ndarray) -> np.ndarray:
     return np.sqrt(square_dist)
 
 
-def from_lower_row(triangular: np.ndarray) -> np.ndarray:
+def from_lower_row(triangular: List) -> np.ndarray:
     """
     Computes a full distances matrix from a lower row triangular matrix.
     The triangular matrix should not contain the diagonal.
@@ -106,8 +106,7 @@ def from_lower_row(triangular: np.ndarray) -> np.ndarray:
 
     Returns
     -------
-    np.ndarray
-        A n-by-n distances matrix.
+    A n-by-n distances matrix.
     """
     n = len(triangular) + 1
     distances = np.zeros((n, n))
@@ -118,7 +117,7 @@ def from_lower_row(triangular: np.ndarray) -> np.ndarray:
     return distances + distances.T
 
 
-def from_eilon(edge_weights: np.ndarray) -> np.ndarray:
+def from_eilon(edge_weights: List) -> np.ndarray:
     """
     Computes a full distances matrix from the Eilon instances with "LOWER_ROW"
     edge weight format. The specification is incorrect, instead the edge weight

vrplib/parse/parse_solomon.py

@@ -1,12 +1,12 @@
-from typing import Dict, List, Union
+from typing import List
 
 import numpy as np
 
+from vrplib.Instance import Instance
+
 from .parse_distances import pairwise_euclidean
 from .parse_utils import text2lines
 
-Instance = Dict[str, Union[str, float, np.ndarray]]
-
 
 def parse_solomon(text: str, compute_edge_weights: bool = True) -> Instance:
     """

vrplib/parse/parse_solution.py

@@ -1,9 +1,7 @@
-from typing import Dict, List, Union
+from vrplib.Solution import Solution
 
 from .parse_utils import infer_type, text2lines
 
-Solution = Dict[str, Union[float, str, List]]
-
 
 def parse_solution(text: str) -> Solution:
     """
@@ -17,19 +15,18 @@ def parse_solution(text: str) -> Solution:
 
     Returns
     -------
-    dict
-        The soluion data.
+    The soluion data.
     """
     solution: Solution = {"routes": []}
 
     for line in text2lines(text):
         if "Route" in line:
             route = [int(idx) for idx in line.split(":")[1].split(" ") if idx]
-            solution["routes"].append(route)  # type: ignore
+            solution["routes"].append(route)
         elif ":" in line or " " in line:  # Split at first colon or whitespace
             split_at = ":" if ":" in line else " "
             k, v = [word.strip() for word in line.split(split_at, 1)]
-            solution[k.lower()] = infer_type(v)
+            solution[k.lower()] = infer_type(v)  # type: ignore
         else:  # Ignore lines without keyword-value pairs
             continue
 

vrplib/parse/parse_vrplib.py

@@ -1,13 +1,13 @@
 import re
-from typing import Dict, List, Tuple, Union
+from typing import List, Tuple, Union
 
 import numpy as np
 
+from vrplib.Instance import Instance
+
 from .parse_distances import parse_distances
 from .parse_utils import infer_type, text2lines
 
-Instance = Dict[str, Union[str, float, np.ndarray]]
-
 
 def parse_vrplib(text: str, compute_edge_weights: bool = True) -> Instance:
     """
@@ -33,17 +33,17 @@ def parse_vrplib(text: str, compute_edge_weights: bool = True) -> Instance:
     dict
         The instance data.
     """
-    instance = {}
+    instance = Instance()
 
     specs, sections = group_specifications_and_sections(text2lines(text))
 
     for spec in specs:
         key, value = parse_specification(spec)
-        instance[key] = value
+        instance[key] = value  # type: ignore
 
     for section in sections:
         section, data = parse_section(section, instance)
-        instance[section] = data
+        instance[section] = data  # type: ignore
 
     if instance and compute_edge_weights and "edge_weight" not in instance:
         # Compute edge weights if there was no explicit edge weight section
@@ -97,9 +97,18 @@ def parse_specification(line: str) -> Tuple[str, Union[float, str]]:
     return k.lower(), infer_type(v)
 
 
-def parse_section(lines: List, instance: Dict) -> np.ndarray:
+def parse_section(
+    lines: List[str], instance: Instance
+) -> Tuple[str, np.ndarray]:
     """
-    Parses the data section into numpy arrays.
+    Parses the lines corresponding to a section and returns the data.
+
+    Parameters
+    ----------
+    lines
+        The lines corresponding to a section.
+    instance
+        The instance data parsed so far.
     """
     section = _remove_suffix(lines[0].strip(), "_SECTION").lower()
     data_ = [[infer_type(n) for n in line.split()] for line in lines[1:]]
@@ -124,5 +133,5 @@ def parse_section(lines: List, instance: Dict) -> np.ndarray:
     return section, data
 
 
-def _remove_suffix(name: str, suffix: str):
+def _remove_suffix(name: str, suffix: str) -> str:
     return name[: -len(suffix)] if name.endswith(suffix) else name