[Python-checkins] gh-67790: Support float-style formatting for Fraction instances (#100161)
mdickinson
webhook-mailer at python.org
Sun Jan 22 13:44:58 EST 2023
https://github.com/python/cpython/commit/3e09f3152e518cdc8779b52943b86812114ce071
commit: 3e09f3152e518cdc8779b52943b86812114ce071
branch: main
author: Mark Dickinson <dickinsm at gmail.com>
committer: mdickinson <dickinsm at gmail.com>
date: 2023-01-22T18:44:49Z
summary:
gh-67790: Support float-style formatting for Fraction instances (#100161)
This PR adds support for float-style formatting for `Fraction` objects: it supports the `"e"`, `"E"`, `"f"`, `"F"`, `"g"`, `"G"` and `"%"` presentation types, and all the various bells and whistles of the formatting mini-language for those presentation types. The behaviour almost exactly matches that of `float`, but the implementation works with the exact `Fraction` value and does not do an intermediate conversion to `float`, and so avoids loss of precision or issues with numbers that are outside the dynamic range of the `float` type.
Note that the `"n"` presentation type is _not_ supported. That support could be added later if people have a need for it.
There's one corner-case where the behaviour differs from that of float: for the `float` type, if explicit alignment is specified with a fill character of `'0'` and alignment type `'='`, then thousands separators (if specified) are inserted into the padding string:
```python
>>> format(3.14, '0=11,.2f')
'0,000,003.14'
```
The exact same effect can be achieved by using the `'0'` flag:
```python
>>> format(3.14, '011,.2f')
'0,000,003.14'
```
For `Fraction`, only the `'0'` flag has the above behaviour with respect to thousands separators: there's no special-casing of the particular `'0='` fill-character/alignment combination. Instead, we treat the fill character `'0'` just like any other:
```python
>>> format(Fraction('3.14'), '0=11,.2f')
'00000003.14'
>>> format(Fraction('3.14'), '011,.2f')
'0,000,003.14'
```
The `Fraction` formatter is also stricter about combining these two things: it's not permitted to use both the `'0'` flag _and_ explicit alignment, on the basis that we should refuse the temptation to guess in the face of ambiguity. `float` is less picky:
```python
>>> format(3.14, '0<011,.2f')
'3.140000000'
>>> format(Fraction('3.14'), '0<011,.2f')
Traceback (most recent call last):
File "<stdin>", line 1, in <module>
File "/Users/mdickinson/Repositories/python/cpython/Lib/fractions.py", line 414, in __format__
raise ValueError(
ValueError: Invalid format specifier '0<011,.2f' for object of type 'Fraction'; can't use explicit alignment when zero-padding
```
files:
A Misc/NEWS.d/next/Library/2022-12-10-15-30-17.gh-issue-67790.P9YUZM.rst
M Doc/library/fractions.rst
M Doc/whatsnew/3.12.rst
M Lib/fractions.py
M Lib/test/test_fractions.py
diff --git a/Doc/library/fractions.rst b/Doc/library/fractions.rst
index dc9d5fc18215..06b0e038c89b 100644
--- a/Doc/library/fractions.rst
+++ b/Doc/library/fractions.rst
@@ -101,6 +101,11 @@ another rational number, or from a string.
.. versionchanged:: 3.12
Space is allowed around the slash for string inputs: ``Fraction('2 / 3')``.
+ .. versionchanged:: 3.12
+ :class:`Fraction` instances now support float-style formatting, with
+ presentation types ``"e"``, ``"E"``, ``"f"``, ``"F"``, ``"g"``, ``"G"``
+ and ``"%""``.
+
.. attribute:: numerator
Numerator of the Fraction in lowest term.
@@ -193,6 +198,29 @@ another rational number, or from a string.
``ndigits`` is negative), again rounding half toward even. This
method can also be accessed through the :func:`round` function.
+ .. method:: __format__(format_spec, /)
+
+ Provides support for float-style formatting of :class:`Fraction`
+ instances via the :meth:`str.format` method, the :func:`format` built-in
+ function, or :ref:`Formatted string literals <f-strings>`. The
+ presentation types ``"e"``, ``"E"``, ``"f"``, ``"F"``, ``"g"``, ``"G"``
+ and ``"%"`` are supported. For these presentation types, formatting for a
+ :class:`Fraction` object ``x`` follows the rules outlined for
+ the :class:`float` type in the :ref:`formatspec` section.
+
+ Here are some examples::
+
+ >>> from fractions import Fraction
+ >>> format(Fraction(1, 7), '.40g')
+ '0.1428571428571428571428571428571428571429'
+ >>> format(Fraction('1234567.855'), '_.2f')
+ '1_234_567.86'
+ >>> f"{Fraction(355, 113):*>20.6e}"
+ '********3.141593e+00'
+ >>> old_price, new_price = 499, 672
+ >>> "{:.2%} price increase".format(Fraction(new_price, old_price) - 1)
+ '34.67% price increase'
+
.. seealso::
diff --git a/Doc/whatsnew/3.12.rst b/Doc/whatsnew/3.12.rst
index 851105f004ba..164cd89522d2 100644
--- a/Doc/whatsnew/3.12.rst
+++ b/Doc/whatsnew/3.12.rst
@@ -269,6 +269,12 @@ dis
:data:`~dis.hasarg` collection instead.
(Contributed by Irit Katriel in :gh:`94216`.)
+fractions
+---------
+
+* Objects of type :class:`fractions.Fraction` now support float-style
+ formatting. (Contributed by Mark Dickinson in :gh:`100161`.)
+
math
----
diff --git a/Lib/fractions.py b/Lib/fractions.py
index bdba6c339593..49a3f2841a2e 100644
--- a/Lib/fractions.py
+++ b/Lib/fractions.py
@@ -69,6 +69,96 @@ def _hash_algorithm(numerator, denominator):
""", re.VERBOSE | re.IGNORECASE)
+# Helpers for formatting
+
+def _round_to_exponent(n, d, exponent, no_neg_zero=False):
+ """Round a rational number to the nearest multiple of a given power of 10.
+
+ Rounds the rational number n/d to the nearest integer multiple of
+ 10**exponent, rounding to the nearest even integer multiple in the case of
+ a tie. Returns a pair (sign: bool, significand: int) representing the
+ rounded value (-1)**sign * significand * 10**exponent.
+
+ If no_neg_zero is true, then the returned sign will always be False when
+ the significand is zero. Otherwise, the sign reflects the sign of the
+ input.
+
+ d must be positive, but n and d need not be relatively prime.
+ """
+ if exponent >= 0:
+ d *= 10**exponent
+ else:
+ n *= 10**-exponent
+
+ # The divmod quotient is correct for round-ties-towards-positive-infinity;
+ # In the case of a tie, we zero out the least significant bit of q.
+ q, r = divmod(n + (d >> 1), d)
+ if r == 0 and d & 1 == 0:
+ q &= -2
+
+ sign = q < 0 if no_neg_zero else n < 0
+ return sign, abs(q)
+
+
+def _round_to_figures(n, d, figures):
+ """Round a rational number to a given number of significant figures.
+
+ Rounds the rational number n/d to the given number of significant figures
+ using the round-ties-to-even rule, and returns a triple
+ (sign: bool, significand: int, exponent: int) representing the rounded
+ value (-1)**sign * significand * 10**exponent.
+
+ In the special case where n = 0, returns a significand of zero and
+ an exponent of 1 - figures, for compatibility with formatting.
+ Otherwise, the returned significand satisfies
+ 10**(figures - 1) <= significand < 10**figures.
+
+ d must be positive, but n and d need not be relatively prime.
+ figures must be positive.
+ """
+ # Special case for n == 0.
+ if n == 0:
+ return False, 0, 1 - figures
+
+ # Find integer m satisfying 10**(m - 1) <= abs(n)/d <= 10**m. (If abs(n)/d
+ # is a power of 10, either of the two possible values for m is fine.)
+ str_n, str_d = str(abs(n)), str(d)
+ m = len(str_n) - len(str_d) + (str_d <= str_n)
+
+ # Round to a multiple of 10**(m - figures). The significand we get
+ # satisfies 10**(figures - 1) <= significand <= 10**figures.
+ exponent = m - figures
+ sign, significand = _round_to_exponent(n, d, exponent)
+
+ # Adjust in the case where significand == 10**figures, to ensure that
+ # 10**(figures - 1) <= significand < 10**figures.
+ if len(str(significand)) == figures + 1:
+ significand //= 10
+ exponent += 1
+
+ return sign, significand, exponent
+
+
+# Pattern for matching float-style format specifications;
+# supports 'e', 'E', 'f', 'F', 'g', 'G' and '%' presentation types.
+_FLOAT_FORMAT_SPECIFICATION_MATCHER = re.compile(r"""
+ (?:
+ (?P<fill>.)?
+ (?P<align>[<>=^])
+ )?
+ (?P<sign>[-+ ]?)
+ (?P<no_neg_zero>z)?
+ (?P<alt>\#)?
+ # A '0' that's *not* followed by another digit is parsed as a minimum width
+ # rather than a zeropad flag.
+ (?P<zeropad>0(?=[0-9]))?
+ (?P<minimumwidth>0|[1-9][0-9]*)?
+ (?P<thousands_sep>[,_])?
+ (?:\.(?P<precision>0|[1-9][0-9]*))?
+ (?P<presentation_type>[eEfFgG%])
+""", re.DOTALL | re.VERBOSE).fullmatch
+
+
class Fraction(numbers.Rational):
"""This class implements rational numbers.
@@ -314,6 +404,122 @@ def __str__(self):
else:
return '%s/%s' % (self._numerator, self._denominator)
+ def __format__(self, format_spec, /):
+ """Format this fraction according to the given format specification."""
+
+ # Backwards compatiblility with existing formatting.
+ if not format_spec:
+ return str(self)
+
+ # Validate and parse the format specifier.
+ match = _FLOAT_FORMAT_SPECIFICATION_MATCHER(format_spec)
+ if match is None:
+ raise ValueError(
+ f"Invalid format specifier {format_spec!r} "
+ f"for object of type {type(self).__name__!r}"
+ )
+ elif match["align"] is not None and match["zeropad"] is not None:
+ # Avoid the temptation to guess.
+ raise ValueError(
+ f"Invalid format specifier {format_spec!r} "
+ f"for object of type {type(self).__name__!r}; "
+ "can't use explicit alignment when zero-padding"
+ )
+ fill = match["fill"] or " "
+ align = match["align"] or ">"
+ pos_sign = "" if match["sign"] == "-" else match["sign"]
+ no_neg_zero = bool(match["no_neg_zero"])
+ alternate_form = bool(match["alt"])
+ zeropad = bool(match["zeropad"])
+ minimumwidth = int(match["minimumwidth"] or "0")
+ thousands_sep = match["thousands_sep"]
+ precision = int(match["precision"] or "6")
+ presentation_type = match["presentation_type"]
+ trim_zeros = presentation_type in "gG" and not alternate_form
+ trim_point = not alternate_form
+ exponent_indicator = "E" if presentation_type in "EFG" else "e"
+
+ # Round to get the digits we need, figure out where to place the point,
+ # and decide whether to use scientific notation. 'point_pos' is the
+ # relative to the _end_ of the digit string: that is, it's the number
+ # of digits that should follow the point.
+ if presentation_type in "fF%":
+ exponent = -precision
+ if presentation_type == "%":
+ exponent -= 2
+ negative, significand = _round_to_exponent(
+ self._numerator, self._denominator, exponent, no_neg_zero)
+ scientific = False
+ point_pos = precision
+ else: # presentation_type in "eEgG"
+ figures = (
+ max(precision, 1)
+ if presentation_type in "gG"
+ else precision + 1
+ )
+ negative, significand, exponent = _round_to_figures(
+ self._numerator, self._denominator, figures)
+ scientific = (
+ presentation_type in "eE"
+ or exponent > 0
+ or exponent + figures <= -4
+ )
+ point_pos = figures - 1 if scientific else -exponent
+
+ # Get the suffix - the part following the digits, if any.
+ if presentation_type == "%":
+ suffix = "%"
+ elif scientific:
+ suffix = f"{exponent_indicator}{exponent + point_pos:+03d}"
+ else:
+ suffix = ""
+
+ # String of output digits, padded sufficiently with zeros on the left
+ # so that we'll have at least one digit before the decimal point.
+ digits = f"{significand:0{point_pos + 1}d}"
+
+ # Before padding, the output has the form f"{sign}{leading}{trailing}",
+ # where `leading` includes thousands separators if necessary and
+ # `trailing` includes the decimal separator where appropriate.
+ sign = "-" if negative else pos_sign
+ leading = digits[: len(digits) - point_pos]
+ frac_part = digits[len(digits) - point_pos :]
+ if trim_zeros:
+ frac_part = frac_part.rstrip("0")
+ separator = "" if trim_point and not frac_part else "."
+ trailing = separator + frac_part + suffix
+
+ # Do zero padding if required.
+ if zeropad:
+ min_leading = minimumwidth - len(sign) - len(trailing)
+ # When adding thousands separators, they'll be added to the
+ # zero-padded portion too, so we need to compensate.
+ leading = leading.zfill(
+ 3 * min_leading // 4 + 1 if thousands_sep else min_leading
+ )
+
+ # Insert thousands separators if required.
+ if thousands_sep:
+ first_pos = 1 + (len(leading) - 1) % 3
+ leading = leading[:first_pos] + "".join(
+ thousands_sep + leading[pos : pos + 3]
+ for pos in range(first_pos, len(leading), 3)
+ )
+
+ # We now have a sign and a body. Pad with fill character if necessary
+ # and return.
+ body = leading + trailing
+ padding = fill * (minimumwidth - len(sign) - len(body))
+ if align == ">":
+ return padding + sign + body
+ elif align == "<":
+ return sign + body + padding
+ elif align == "^":
+ half = len(padding) // 2
+ return padding[:half] + sign + body + padding[half:]
+ else: # align == "="
+ return sign + padding + body
+
def _operator_fallbacks(monomorphic_operator, fallback_operator):
"""Generates forward and reverse operators given a purely-rational
operator and a function from the operator module.
diff --git a/Lib/test/test_fractions.py b/Lib/test/test_fractions.py
index 819d680fd88e..3bc6b409e05d 100644
--- a/Lib/test/test_fractions.py
+++ b/Lib/test/test_fractions.py
@@ -843,6 +843,382 @@ def denominator(self):
self.assertEqual(type(f.numerator), myint)
self.assertEqual(type(f.denominator), myint)
+ def test_format_no_presentation_type(self):
+ # Triples (fraction, specification, expected_result)
+ testcases = [
+ (F(1, 3), '', '1/3'),
+ (F(-1, 3), '', '-1/3'),
+ (F(3), '', '3'),
+ (F(-3), '', '-3'),
+ ]
+ for fraction, spec, expected in testcases:
+ with self.subTest(fraction=fraction, spec=spec):
+ self.assertEqual(format(fraction, spec), expected)
+
+ def test_format_e_presentation_type(self):
+ # Triples (fraction, specification, expected_result)
+ testcases = [
+ (F(2, 3), '.6e', '6.666667e-01'),
+ (F(3, 2), '.6e', '1.500000e+00'),
+ (F(2, 13), '.6e', '1.538462e-01'),
+ (F(2, 23), '.6e', '8.695652e-02'),
+ (F(2, 33), '.6e', '6.060606e-02'),
+ (F(13, 2), '.6e', '6.500000e+00'),
+ (F(20, 2), '.6e', '1.000000e+01'),
+ (F(23, 2), '.6e', '1.150000e+01'),
+ (F(33, 2), '.6e', '1.650000e+01'),
+ (F(2, 3), '.6e', '6.666667e-01'),
+ (F(3, 2), '.6e', '1.500000e+00'),
+ # Zero
+ (F(0), '.3e', '0.000e+00'),
+ # Powers of 10, to exercise the log10 boundary logic
+ (F(1, 1000), '.3e', '1.000e-03'),
+ (F(1, 100), '.3e', '1.000e-02'),
+ (F(1, 10), '.3e', '1.000e-01'),
+ (F(1, 1), '.3e', '1.000e+00'),
+ (F(10), '.3e', '1.000e+01'),
+ (F(100), '.3e', '1.000e+02'),
+ (F(1000), '.3e', '1.000e+03'),
+ # Boundary where we round up to the next power of 10
+ (F('99.999994999999'), '.6e', '9.999999e+01'),
+ (F('99.999995'), '.6e', '1.000000e+02'),
+ (F('99.999995000001'), '.6e', '1.000000e+02'),
+ # Negatives
+ (F(-2, 3), '.6e', '-6.666667e-01'),
+ (F(-3, 2), '.6e', '-1.500000e+00'),
+ (F(-100), '.6e', '-1.000000e+02'),
+ # Large and small
+ (F('1e1000'), '.3e', '1.000e+1000'),
+ (F('1e-1000'), '.3e', '1.000e-1000'),
+ # Using 'E' instead of 'e' should give us a capital 'E'
+ (F(2, 3), '.6E', '6.666667E-01'),
+ # Tiny precision
+ (F(2, 3), '.1e', '6.7e-01'),
+ (F('0.995'), '.0e', '1e+00'),
+ # Default precision is 6
+ (F(22, 7), 'e', '3.142857e+00'),
+ # Alternate form forces a decimal point
+ (F('0.995'), '#.0e', '1.e+00'),
+ # Check that padding takes the exponent into account.
+ (F(22, 7), '11.6e', '3.142857e+00'),
+ (F(22, 7), '12.6e', '3.142857e+00'),
+ (F(22, 7), '13.6e', ' 3.142857e+00'),
+ # Thousands separators
+ (F('1234567.123456'), ',.5e', '1.23457e+06'),
+ (F('123.123456'), '012_.2e', '0_001.23e+02'),
+ # z flag is legal, but never makes a difference to the output
+ (F(-1, 7**100), 'z.6e', '-3.091690e-85'),
+ ]
+ for fraction, spec, expected in testcases:
+ with self.subTest(fraction=fraction, spec=spec):
+ self.assertEqual(format(fraction, spec), expected)
+
+ def test_format_f_presentation_type(self):
+ # Triples (fraction, specification, expected_result)
+ testcases = [
+ # Simple .f formatting
+ (F(0, 1), '.2f', '0.00'),
+ (F(1, 3), '.2f', '0.33'),
+ (F(2, 3), '.2f', '0.67'),
+ (F(4, 3), '.2f', '1.33'),
+ (F(1, 8), '.2f', '0.12'),
+ (F(3, 8), '.2f', '0.38'),
+ (F(1, 13), '.2f', '0.08'),
+ (F(1, 199), '.2f', '0.01'),
+ (F(1, 200), '.2f', '0.00'),
+ (F(22, 7), '.5f', '3.14286'),
+ (F('399024789'), '.2f', '399024789.00'),
+ # Large precision (more than float can provide)
+ (F(104348, 33215), '.50f',
+ '3.14159265392142104470871594159265392142104470871594'),
+ # Precision defaults to 6 if not given
+ (F(22, 7), 'f', '3.142857'),
+ (F(0), 'f', '0.000000'),
+ (F(-22, 7), 'f', '-3.142857'),
+ # Round-ties-to-even checks
+ (F('1.225'), '.2f', '1.22'),
+ (F('1.2250000001'), '.2f', '1.23'),
+ (F('1.2349999999'), '.2f', '1.23'),
+ (F('1.235'), '.2f', '1.24'),
+ (F('1.245'), '.2f', '1.24'),
+ (F('1.2450000001'), '.2f', '1.25'),
+ (F('1.2549999999'), '.2f', '1.25'),
+ (F('1.255'), '.2f', '1.26'),
+ (F('-1.225'), '.2f', '-1.22'),
+ (F('-1.2250000001'), '.2f', '-1.23'),
+ (F('-1.2349999999'), '.2f', '-1.23'),
+ (F('-1.235'), '.2f', '-1.24'),
+ (F('-1.245'), '.2f', '-1.24'),
+ (F('-1.2450000001'), '.2f', '-1.25'),
+ (F('-1.2549999999'), '.2f', '-1.25'),
+ (F('-1.255'), '.2f', '-1.26'),
+ # Negatives and sign handling
+ (F(2, 3), '.2f', '0.67'),
+ (F(2, 3), '-.2f', '0.67'),
+ (F(2, 3), '+.2f', '+0.67'),
+ (F(2, 3), ' .2f', ' 0.67'),
+ (F(-2, 3), '.2f', '-0.67'),
+ (F(-2, 3), '-.2f', '-0.67'),
+ (F(-2, 3), '+.2f', '-0.67'),
+ (F(-2, 3), ' .2f', '-0.67'),
+ # Formatting to zero places
+ (F(1, 2), '.0f', '0'),
+ (F(-1, 2), '.0f', '-0'),
+ (F(22, 7), '.0f', '3'),
+ (F(-22, 7), '.0f', '-3'),
+ # Formatting to zero places, alternate form
+ (F(1, 2), '#.0f', '0.'),
+ (F(-1, 2), '#.0f', '-0.'),
+ (F(22, 7), '#.0f', '3.'),
+ (F(-22, 7), '#.0f', '-3.'),
+ # z flag for suppressing negative zeros
+ (F('-0.001'), 'z.2f', '0.00'),
+ (F('-0.001'), '-z.2f', '0.00'),
+ (F('-0.001'), '+z.2f', '+0.00'),
+ (F('-0.001'), ' z.2f', ' 0.00'),
+ (F('0.001'), 'z.2f', '0.00'),
+ (F('0.001'), '-z.2f', '0.00'),
+ (F('0.001'), '+z.2f', '+0.00'),
+ (F('0.001'), ' z.2f', ' 0.00'),
+ # Specifying a minimum width
+ (F(2, 3), '6.2f', ' 0.67'),
+ (F(12345), '6.2f', '12345.00'),
+ (F(12345), '12f', '12345.000000'),
+ # Fill and alignment
+ (F(2, 3), '>6.2f', ' 0.67'),
+ (F(2, 3), '<6.2f', '0.67 '),
+ (F(2, 3), '^3.2f', '0.67'),
+ (F(2, 3), '^4.2f', '0.67'),
+ (F(2, 3), '^5.2f', '0.67 '),
+ (F(2, 3), '^6.2f', ' 0.67 '),
+ (F(2, 3), '^7.2f', ' 0.67 '),
+ (F(2, 3), '^8.2f', ' 0.67 '),
+ # '=' alignment
+ (F(-2, 3), '=+8.2f', '- 0.67'),
+ (F(2, 3), '=+8.2f', '+ 0.67'),
+ # Fill character
+ (F(-2, 3), 'X>3.2f', '-0.67'),
+ (F(-2, 3), 'X>7.2f', 'XX-0.67'),
+ (F(-2, 3), 'X<7.2f', '-0.67XX'),
+ (F(-2, 3), 'X^7.2f', 'X-0.67X'),
+ (F(-2, 3), 'X=7.2f', '-XX0.67'),
+ (F(-2, 3), ' >7.2f', ' -0.67'),
+ # Corner cases: weird fill characters
+ (F(-2, 3), '\x00>7.2f', '\x00\x00-0.67'),
+ (F(-2, 3), '\n>7.2f', '\n\n-0.67'),
+ (F(-2, 3), '\t>7.2f', '\t\t-0.67'),
+ (F(-2, 3), '>>7.2f', '>>-0.67'),
+ (F(-2, 3), '<>7.2f', '<<-0.67'),
+ (F(-2, 3), '→>7.2f', '→→-0.67'),
+ # Zero-padding
+ (F(-2, 3), '07.2f', '-000.67'),
+ (F(-2, 3), '-07.2f', '-000.67'),
+ (F(2, 3), '+07.2f', '+000.67'),
+ (F(2, 3), ' 07.2f', ' 000.67'),
+ # An isolated zero is a minimum width, not a zero-pad flag.
+ # So unlike zero-padding, it's legal in combination with alignment.
+ (F(2, 3), '0.2f', '0.67'),
+ (F(2, 3), '>0.2f', '0.67'),
+ (F(2, 3), '<0.2f', '0.67'),
+ (F(2, 3), '^0.2f', '0.67'),
+ (F(2, 3), '=0.2f', '0.67'),
+ # Corner case: zero-padding _and_ a zero minimum width.
+ (F(2, 3), '00.2f', '0.67'),
+ # Thousands separator (only affects portion before the point)
+ (F(2, 3), ',.2f', '0.67'),
+ (F(2, 3), ',.7f', '0.6666667'),
+ (F('123456.789'), ',.2f', '123,456.79'),
+ (F('1234567'), ',.2f', '1,234,567.00'),
+ (F('12345678'), ',.2f', '12,345,678.00'),
+ (F('12345678'), ',f', '12,345,678.000000'),
+ # Underscore as thousands separator
+ (F(2, 3), '_.2f', '0.67'),
+ (F(2, 3), '_.7f', '0.6666667'),
+ (F('123456.789'), '_.2f', '123_456.79'),
+ (F('1234567'), '_.2f', '1_234_567.00'),
+ (F('12345678'), '_.2f', '12_345_678.00'),
+ # Thousands and zero-padding
+ (F('1234.5678'), '07,.2f', '1,234.57'),
+ (F('1234.5678'), '08,.2f', '1,234.57'),
+ (F('1234.5678'), '09,.2f', '01,234.57'),
+ (F('1234.5678'), '010,.2f', '001,234.57'),
+ (F('1234.5678'), '011,.2f', '0,001,234.57'),
+ (F('1234.5678'), '012,.2f', '0,001,234.57'),
+ (F('1234.5678'), '013,.2f', '00,001,234.57'),
+ (F('1234.5678'), '014,.2f', '000,001,234.57'),
+ (F('1234.5678'), '015,.2f', '0,000,001,234.57'),
+ (F('1234.5678'), '016,.2f', '0,000,001,234.57'),
+ (F('-1234.5678'), '07,.2f', '-1,234.57'),
+ (F('-1234.5678'), '08,.2f', '-1,234.57'),
+ (F('-1234.5678'), '09,.2f', '-1,234.57'),
+ (F('-1234.5678'), '010,.2f', '-01,234.57'),
+ (F('-1234.5678'), '011,.2f', '-001,234.57'),
+ (F('-1234.5678'), '012,.2f', '-0,001,234.57'),
+ (F('-1234.5678'), '013,.2f', '-0,001,234.57'),
+ (F('-1234.5678'), '014,.2f', '-00,001,234.57'),
+ (F('-1234.5678'), '015,.2f', '-000,001,234.57'),
+ (F('-1234.5678'), '016,.2f', '-0,000,001,234.57'),
+ # Corner case: no decimal point
+ (F('-1234.5678'), '06,.0f', '-1,235'),
+ (F('-1234.5678'), '07,.0f', '-01,235'),
+ (F('-1234.5678'), '08,.0f', '-001,235'),
+ (F('-1234.5678'), '09,.0f', '-0,001,235'),
+ # Corner-case - zero-padding specified through fill and align
+ # instead of the zero-pad character - in this case, treat '0' as a
+ # regular fill character and don't attempt to insert commas into
+ # the filled portion. This differs from the int and float
+ # behaviour.
+ (F('1234.5678'), '0=12,.2f', '00001,234.57'),
+ # Corner case where it's not clear whether the '0' indicates zero
+ # padding or gives the minimum width, but there's still an obvious
+ # answer to give. We want this to work in case the minimum width
+ # is being inserted programmatically: spec = f'{width}.2f'.
+ (F('12.34'), '0.2f', '12.34'),
+ (F('12.34'), 'X>0.2f', '12.34'),
+ # 'F' should work identically to 'f'
+ (F(22, 7), '.5F', '3.14286'),
+ # %-specifier
+ (F(22, 7), '.2%', '314.29%'),
+ (F(1, 7), '.2%', '14.29%'),
+ (F(1, 70), '.2%', '1.43%'),
+ (F(1, 700), '.2%', '0.14%'),
+ (F(1, 7000), '.2%', '0.01%'),
+ (F(1, 70000), '.2%', '0.00%'),
+ (F(1, 7), '.0%', '14%'),
+ (F(1, 7), '#.0%', '14.%'),
+ (F(100, 7), ',.2%', '1,428.57%'),
+ (F(22, 7), '7.2%', '314.29%'),
+ (F(22, 7), '8.2%', ' 314.29%'),
+ (F(22, 7), '08.2%', '0314.29%'),
+ # Test cases from #67790 and discuss.python.org Ideas thread.
+ (F(1, 3), '.2f', '0.33'),
+ (F(1, 8), '.2f', '0.12'),
+ (F(3, 8), '.2f', '0.38'),
+ (F(2545, 1000), '.2f', '2.54'),
+ (F(2549, 1000), '.2f', '2.55'),
+ (F(2635, 1000), '.2f', '2.64'),
+ (F(1, 100), '.1f', '0.0'),
+ (F(49, 1000), '.1f', '0.0'),
+ (F(51, 1000), '.1f', '0.1'),
+ (F(149, 1000), '.1f', '0.1'),
+ (F(151, 1000), '.1f', '0.2'),
+ ]
+ for fraction, spec, expected in testcases:
+ with self.subTest(fraction=fraction, spec=spec):
+ self.assertEqual(format(fraction, spec), expected)
+
+ def test_format_g_presentation_type(self):
+ # Triples (fraction, specification, expected_result)
+ testcases = [
+ (F('0.000012345678'), '.6g', '1.23457e-05'),
+ (F('0.00012345678'), '.6g', '0.000123457'),
+ (F('0.0012345678'), '.6g', '0.00123457'),
+ (F('0.012345678'), '.6g', '0.0123457'),
+ (F('0.12345678'), '.6g', '0.123457'),
+ (F('1.2345678'), '.6g', '1.23457'),
+ (F('12.345678'), '.6g', '12.3457'),
+ (F('123.45678'), '.6g', '123.457'),
+ (F('1234.5678'), '.6g', '1234.57'),
+ (F('12345.678'), '.6g', '12345.7'),
+ (F('123456.78'), '.6g', '123457'),
+ (F('1234567.8'), '.6g', '1.23457e+06'),
+ # Rounding up cases
+ (F('9.99999e+2'), '.4g', '1000'),
+ (F('9.99999e-8'), '.4g', '1e-07'),
+ (F('9.99999e+8'), '.4g', '1e+09'),
+ # Check round-ties-to-even behaviour
+ (F('-0.115'), '.2g', '-0.12'),
+ (F('-0.125'), '.2g', '-0.12'),
+ (F('-0.135'), '.2g', '-0.14'),
+ (F('-0.145'), '.2g', '-0.14'),
+ (F('0.115'), '.2g', '0.12'),
+ (F('0.125'), '.2g', '0.12'),
+ (F('0.135'), '.2g', '0.14'),
+ (F('0.145'), '.2g', '0.14'),
+ # Trailing zeros and decimal point suppressed by default ...
+ (F(0), '.6g', '0'),
+ (F('123.400'), '.6g', '123.4'),
+ (F('123.000'), '.6g', '123'),
+ (F('120.000'), '.6g', '120'),
+ (F('12000000'), '.6g', '1.2e+07'),
+ # ... but not when alternate form is in effect
+ (F(0), '#.6g', '0.00000'),
+ (F('123.400'), '#.6g', '123.400'),
+ (F('123.000'), '#.6g', '123.000'),
+ (F('120.000'), '#.6g', '120.000'),
+ (F('12000000'), '#.6g', '1.20000e+07'),
+ # 'G' format (uses 'E' instead of 'e' for the exponent indicator)
+ (F('123.45678'), '.6G', '123.457'),
+ (F('1234567.8'), '.6G', '1.23457E+06'),
+ # Default precision is 6 significant figures
+ (F('3.1415926535'), 'g', '3.14159'),
+ # Precision 0 is treated the same as precision 1.
+ (F('0.000031415'), '.0g', '3e-05'),
+ (F('0.00031415'), '.0g', '0.0003'),
+ (F('0.31415'), '.0g', '0.3'),
+ (F('3.1415'), '.0g', '3'),
+ (F('3.1415'), '#.0g', '3.'),
+ (F('31.415'), '.0g', '3e+01'),
+ (F('31.415'), '#.0g', '3.e+01'),
+ (F('0.000031415'), '.1g', '3e-05'),
+ (F('0.00031415'), '.1g', '0.0003'),
+ (F('0.31415'), '.1g', '0.3'),
+ (F('3.1415'), '.1g', '3'),
+ (F('3.1415'), '#.1g', '3.'),
+ (F('31.415'), '.1g', '3e+01'),
+ # Thousands separator
+ (F(2**64), '_.25g', '18_446_744_073_709_551_616'),
+ # As with 'e' format, z flag is legal, but has no effect
+ (F(-1, 7**100), 'zg', '-3.09169e-85'),
+ ]
+ for fraction, spec, expected in testcases:
+ with self.subTest(fraction=fraction, spec=spec):
+ self.assertEqual(format(fraction, spec), expected)
+
+ def test_invalid_formats(self):
+ fraction = F(2, 3)
+ with self.assertRaises(TypeError):
+ format(fraction, None)
+
+ invalid_specs = [
+ 'Q6f', # regression test
+ # illegal to use fill or alignment when zero padding
+ 'X>010f',
+ 'X<010f',
+ 'X^010f',
+ 'X=010f',
+ '0>010f',
+ '0<010f',
+ '0^010f',
+ '0=010f',
+ '>010f',
+ '<010f',
+ '^010f',
+ '=010e',
+ '=010f',
+ '=010g',
+ '=010%',
+ '>00.2f',
+ '>00f',
+ # Too many zeros - minimum width should not have leading zeros
+ '006f',
+ # Leading zeros in precision
+ '.010f',
+ '.02f',
+ '.000f',
+ # Missing precision
+ '.e',
+ '.f',
+ '.g',
+ '.%',
+ # Z instead of z for negative zero suppression
+ 'Z.2f'
+ ]
+ for spec in invalid_specs:
+ with self.subTest(spec=spec):
+ with self.assertRaises(ValueError):
+ format(fraction, spec)
+
if __name__ == '__main__':
unittest.main()
diff --git a/Misc/NEWS.d/next/Library/2022-12-10-15-30-17.gh-issue-67790.P9YUZM.rst b/Misc/NEWS.d/next/Library/2022-12-10-15-30-17.gh-issue-67790.P9YUZM.rst
new file mode 100644
index 000000000000..ba0db774f8b3
--- /dev/null
+++ b/Misc/NEWS.d/next/Library/2022-12-10-15-30-17.gh-issue-67790.P9YUZM.rst
@@ -0,0 +1,2 @@
+Add float-style formatting support for :class:`fractions.Fraction`
+instances.
More information about the Python-checkins
mailing list