X-Git-Url: https://wannabe.guru.org/gitweb/?a=blobdiff_plain;f=histogram.py;h=f85abea2c45f05ef65c70657bd1b2b6f53f04c04;hb=e8fbbb7306430478dec55d2c963eed116d8330cc;hp=87ef77340fd036b8ea29d5ddf3def5d902db3688;hpb=f41c2bbae228ea2305e7530148c4d2dca5131200;p=python_utils.git

diff --git a/histogram.py b/histogram.py
index 87ef773..f85abea 100644
--- a/histogram.py
+++ b/histogram.py
@@ -1,47 +1,54 @@
 #!/usr/bin/env python3
+# -*- coding: utf-8 -*-
+
+"""A text-based simple histogram helper class."""
 
 import math
-from numbers import Number
-from typing import Generic, Iterable, List, Optional, Tuple, TypeVar
+from typing import Dict, Generic, Iterable, List, Optional, Tuple, TypeVar
 
-T = TypeVar("T", bound=Number)
+T = TypeVar("T", int, float)
+Bound = int
+Count = int
 
 
 class SimpleHistogram(Generic[T]):
+    """A simple histogram."""
 
     # Useful in defining wide open bottom/top bucket bounds:
     POSITIVE_INFINITY = math.inf
     NEGATIVE_INFINITY = -math.inf
 
-    def __init__(self, buckets: List[Tuple[T, T]]):
+    def __init__(self, buckets: List[Tuple[Bound, Bound]]):
         from math_utils import RunningMedian
 
-        self.buckets = {}
+        self.buckets: Dict[Tuple[Bound, Bound], Count] = {}
         for start_end in buckets:
             if self._get_bucket(start_end[0]) is not None:
                 raise Exception("Buckets overlap?!")
             self.buckets[start_end] = 0
-        self.sigma = 0
-        self.median = RunningMedian()
-        self.maximum = None
-        self.minimum = None
-        self.count = 0
+        self.sigma: float = 0.0
+        self.median: RunningMedian = RunningMedian()
+        self.maximum: Optional[T] = None
+        self.minimum: Optional[T] = None
+        self.count: Count = 0
 
     @staticmethod
     def n_evenly_spaced_buckets(
         min_bound: T,
         max_bound: T,
         n: int,
-    ) -> List[Tuple[T, T]]:
-        ret = []
+    ) -> List[Tuple[int, int]]:
+        ret: List[Tuple[int, int]] = []
         stride = int((max_bound - min_bound) / n)
         if stride <= 0:
             raise Exception("Min must be < Max")
-        for bucket_start in range(min_bound, max_bound, stride):
+        imax = math.ceil(max_bound)
+        imin = math.floor(min_bound)
+        for bucket_start in range(imin, imax, stride):
             ret.append((bucket_start, bucket_start + stride))
         return ret
 
-    def _get_bucket(self, item: T) -> Optional[Tuple[T, T]]:
+    def _get_bucket(self, item: T) -> Optional[Tuple[int, int]]:
         for start_end in self.buckets:
             if start_end[0] <= item < start_end[1]:
                 return start_end
@@ -67,9 +74,10 @@ class SimpleHistogram(Generic[T]):
             all_true = all_true and self.add_item(item)
         return all_true
 
-    def __repr__(self, *, width: int = 80, label_formatter: str = None) -> str:
+    def __repr__(self, *, width: int = 80, label_formatter: str = '%d') -> str:
         from text_utils import bar_graph
 
+        txt = ""
         max_population: Optional[int] = None
         for bucket in self.buckets:
             pop = self.buckets[bucket]
@@ -77,14 +85,12 @@ class SimpleHistogram(Generic[T]):
                 last_bucket_start = bucket[0]  # beginning of range
             if max_population is None or pop > max_population:
                 max_population = pop  # bucket with max items
-
-        txt = ""
         if max_population is None:
             return txt
 
-        max_label_width = None
-        lowest_start = None
-        highest_end = None
+        max_label_width: Optional[int] = None
+        lowest_start: Optional[int] = None
+        highest_end: Optional[int] = None
         for bucket in sorted(self.buckets, key=lambda x: x[0]):
             start = bucket[0]
             if lowest_start is None:
@@ -92,19 +98,16 @@ class SimpleHistogram(Generic[T]):
             end = bucket[1]
             if highest_end is None or end > highest_end:
                 highest_end = end
-            if label_formatter is None:
-                if type(start) == int and type(end) == int:
-                    label_formatter = '%d'
-                elif type(start) == float and type(end) == float:
-                    label_formatter = '%.2f'
-                else:
-                    label_formatter = '%s'
             label = f'[{label_formatter}..{label_formatter}): ' % (start, end)
             label_width = len(label)
             if max_label_width is None or label_width > max_label_width:
                 max_label_width = label_width
             if start == last_bucket_start:
                 break
+        assert max_label_width is not None
+        assert lowest_start is not None
+        assert highest_end is not None
+
         sigma_label = f'[{label_formatter}..{label_formatter}): ' % (
             lowest_start,
             highest_end,