Alignment marker is not optional for numbers when it should

[mraspaud]

While working on #35 I noticed that the alignment character is not optional for number as it should.

In [6]: parse("{foo:05d}", "00123")
Out[6]: {'foo': 123}

In [8]: parse("{foo:_>5d}", "__123")
Out[8]: {'foo': 123}

# but

In [7]: parse("{foo:_5d}", "__123")
---------------------------------------------------------------------------
ValueError                                Traceback (most recent call last)
Input In [7], in <module>
----> 1 parse("{foo:_5d}", "__123")

File ~/usr/src/trollsift/trollsift/parser.py:457, in parse(fmt, stri, full_match)
    447 """Parse keys and corresponding values from *stri* using format described in *fmt* string.
    448 
    449 Args:
   (...)
    454 
    455 """
    456 convdef = get_convert_dict(fmt)
--> 457 keyvals = regex_formatter.extract_values(fmt, stri, full_match=full_match)
    458 for key in convdef.keys():
    459     keyvals[key] = _convert(convdef[key], keyvals[key])

File ~/usr/src/trollsift/trollsift/parser.py:368, in RegexFormatter.extract_values(self, fmt, stri, full_match)
    366 match = re.match(regex, stri)
    367 if match is None:
--> 368     raise ValueError("String does not match pattern.")
    369 return match.groupdict()

ValueError: String does not match pattern.

The format specification here: https://docs.python.org/3/library/string.html#format-specification-mini-language says > is the default for numbers, and it's already working when the fill is 0, so maybe this could be fixed? I had a quick look, but got lost in regexes...

[djhoese]

I don't think _5d is a valid format specifier:

In [1]: a = 123

In [2]: f"{a:_5d}"
---------------------------------------------------------------------------
ValueError                                Traceback (most recent call last)
Input In [2], in <cell line: 1>()
----> 1 f"{a:_5d}"

ValueError: Invalid format specifier

In [3]: f"{a:_>5d}"
Out[3]: '__123'

What am I doing wrong?

[djhoese]

The alignment is only optional if fill isn't specified it seems. I'm not sure how 0 is working.

In [17]: parse("{foo:<5d}", "  123")
Out[17]: {'foo': 123}

In [18]: parse("{foo:<5d}", "123  ")
Out[18]: {'foo': 123}

In [19]: parse("{foo:=5d}", "123  ")
Out[19]: {'foo': 123}

In [20]: parse("{foo:=5d}", "  123")
Out[20]: {'foo': 123}

In [21]: parse("{foo:5d}", "  123")
Out[21]: {'foo': 123}

[djhoese]

I noticed in the docs:

When no explicit alignment is given, preceding the width field by a zero ('0') character enables sign-aware zero-padding for numeric types. This is equivalent to a fill character of '0' with an alignment type of '='.

Changed in version 3.10: Preceding the width field by '0' no longer affects the default alignment for strings.

This to me sounds like Python 3.10+ changes the behavior of the 0 fill as you're using it. So I think that means that what trollsift is doing is as expected and that your example is using an invalid format specifier (a fill without an alignment).

[mraspaud]

ok, sorry about this then, I misunderstood the format doc. We're good here, closing.