Add statistics printing options

mesi/display.py

@@ -88,6 +88,39 @@ def comparison_progressbar(lst: List, **kwargs):
     return tqdm(lst, total=len(lst), unit="cmp", colour="blue", leave=False, **kwargs)
 
 
+def print_divider(title=""):
+    """Print a simple divider with an optional title."""
+    title = f" {title} " if title else ""
+    print(f"\n======{title}======\n")
+
+
+def print_average_distance(
+    distances: Dict[Tuple[Path, Path], float], distance_threshold: float
+):
+    """Print the average of the distances.
+
+    Args:
+        distances (Dict[Tuple[Path, Path], float]): A dictionary of comparisons to distances.
+        distance_threshold (float): The threshold below which to count values.
+    """
+    values = list(filter(lambda x: x <= distance_threshold, distances.values()))
+    avg = round(sum(values) / len(values), 2)
+
+    print(f"> Average distance: {avg}")
+
+
+def print_distance_distribution(
+    distances: Dict[Tuple[Path, Path], float], distance_threshold: float
+):
+    """Print the distribution of the distances.
+
+    Args:
+        distances (Dict[Tuple[Path, Path], float]): A dictionary of comparisons to distances.
+        distance_threshold (float): The threshold below which to count values.
+    """
+    print("> Distribution display is not yet supported")
+
+
 def print_distances(
     distances: Dict[Tuple[Path, Path], float],
     distance_threshold: float,

mesi/main.py

@@ -50,6 +50,12 @@ def main(
         help=f"String distance algorithm to use  [{', '.join(compare.DISTANCE_FUNCTIONS.keys())}]",
         metavar="ALGORITHM",
     ),
+    average: bool = typer.Option(
+        False, help="Display the average of the computed distances"
+    ),
+    distribution: bool = typer.Option(
+        False, help="Display the distribution of the computed distances"
+    ),
 ):
     """Calculate the similarity between all possible pairs among the given files.
 
@@ -74,6 +80,15 @@ def main(
             combinations, algorithm, ignore_whitespace
         )
         display.print_distances(distances, threshold, table_format, full_paths)
+
+        if average or distribution:
+            display.print_divider("Statistics")
+
+        if average:
+            display.print_average_distance(distances, threshold)
+
+        if distribution:
+            display.print_distance_distribution(distances, threshold)
     except ValueError as err:
         display.print_error(err)
         raise typer.Exit(1)