Fix doubled negative sign in fractional() for negative mixed numbers

fractional(-1.3) returned '-1 -3/10' instead of '-1 3/10', and
fractional(-2.5) returned '-2 -1/2'. The whole-number part already
carries the sign (int(-1.3) == -1), so the numerator should not repeat
it. This was inconsistent with the pure-fraction case, which is already
correct (fractional(-0.3) == '-3/10').

Use abs(numerator) in the mixed-fraction output. Add regression cases
(-0.3, -1.3, -2.5) to the parametrized test_fractional, which previously
had no negative inputs.

Author: patchwright

src/humanize/number.py

@@ -367,7 +367,7 @@ def fractional(value: NumberOrString) -> str:
     if not whole_number:
         return f"{numerator:.0f}/{denominator:.0f}"
 
-    return f"{whole_number:.0f} {numerator:.0f}/{denominator:.0f}"
+    return f"{whole_number:.0f} {abs(numerator):.0f}/{denominator:.0f}"
 
 
 def scientific(value: NumberOrString, precision: int = 2) -> str:

tests/test_number.py

@@ -178,6 +178,9 @@ def test_apnumber(test_input: int | str, expected: str) -> None:
         (1.5, "1 1/2"),
         (0.3, "3/10"),
         (0.333, "333/1000"),
+        (-0.3, "-3/10"),
+        (-1.3, "-1 3/10"),
+        (-2.5, "-2 1/2"),
         (math.nan, "NaN"),
         (math.inf, "+Inf"),
         (-math.inf, "-Inf"),