format.py

import re
import string

""" Custom formatter mimicing str.format() with additional capabilities.

A format string can now additionally take the form:
    {literal|:format_spec}
This form will treat the text "literal" as the text to be formatted
instead of attempting to get the value from the passed-in arguments.

Furthermore, the built-in format_spec mini language has been expanded.
The following may now appear as the "type":
    plural[(int)] - converts the value to plural form
    num(int) - converts the value to plural form and prefixes with int
    list[(sep=", ")] - converts value to list separated by sep
    is - returns "is" if value is 1 or "are" otherwise (value must be an int)
    was - returns "was" if value is 1 or "were" otherwise
    an - prefixes value with "a" or "an"

If the # specifier is used with num or an, it will cause the prefix
to be bolded. If it is used with list, it will omit the "and" prior to
the last list item and will not special-case the two-item form.
"""

# Note: {0:is} is simply shorthand for {is|:plural({0})}, same with {0:was}

# https://docs.python.org/3/library/string.html#format-specification-mini-language
# (enhanced with the above additional types)
FORMAT = re.compile(
    r"^(?:(?P<fill>.?)(?P<align>[<>^=]))?(?P<sign>[+\- ]?)(?P<alt>#?)(?P<zero>0?)"
    r"(?P<width>[1-9]+|0+|0[oO][1-7]+|0[xX][0-9a-fA-F]+|0[bB][01]+)?(?P<comma>,?)"
    r"(?P<precision>\.(?:[1-9]+|0+|0[oO][1-7]+|0[xX][0-9a-fA-F]+|0[bB][01]+))?"
    r"(?P<type>[bcdeEfFgGnosxX%]|plural|num|list|is|was|an)(?:\((?P<arg>.*)\))?$"
    )

class CustomFormatter(string.Formatter):
    def get_value(self, key, args, kwargs):
        if not isinstance(key, int) and key[-1] == "|":
            return key[:-1]
        return super(CustomFormatter, self).get_value(key, args, kwargs)

    def format_field(self, value, format_spec):
        if format_spec:
            m = FORMAT.match(format_spec)
            if m:
                fill = m.group("fill") or ""
                align = m.group("align") or ""
                sign = m.group("sign") or ""
                alt = m.group("alt") == "#" # we don't pass alt through
                zero = m.group("zero") or ""
                width = m.group("width") or ""
                comma = m.group("comma") or ""
                precision = m.group("precision") or ""
                ty = m.group("type")
                arg = m.group("arg") # may be None
                if ty in ("plural", "num", "list", "is", "was", "an"):
                    format_spec = fill + align + sign + zero + width + comma + precision + "s"
                if ty == "plural":
                    arg = int(arg, 0) if arg else 2
                    value = self.plural(value, arg)
                elif ty == "num":
                    if arg is None:
                        raise ValueError("num() format requires an argument")
                    arg = int(arg, 0)
                    prefix = str(arg) if arg > 0 else "no"
                    if alt:
                        prefix = "\u0002" + prefix + "\u0002"
                    value = prefix + " " + self.plural(value, arg)
                elif ty == "an":
                    prefix = "an" if value[0] in ("a", "e", "i", "o", "u") else "a"
                    if alt:
                        prefix = "\u0002" + prefix + "\u0002"
                    value = prefix + " " + value
                elif ty == "is":
                    if not isinstance(value, int):
                        value = int(value, 0)
                    value = "is" if value == 1 else "are"
                elif ty == "was":
                    if not isinstance(value, int):
                        value = int(value, 0)
                    value = "was" if value == 1 else "were"
                elif ty == "list":
                    sep = arg or ", "
                    lv = len(value)
                    if alt:
                        value = sep.join(value)
                    elif lv == 0:
                        value = ""
                    elif lv == 1:
                        value = value[0]
                    elif lv == 2:
                        value = value[0] + " and " + value[1]
                    else:
                        value = sep.join(value[:-1]) + sep + "and " + value[-1]

        return super(CustomFormatter, self).format_field(value, format_spec)

    def plural(self, value, count=2):
        if count == 1:
            return value
        bits = value.split()
        if bits[-1][-2:] == "'s":
            bits[-1] = self.plural(bits[-1][:-2], count)
            bits[-1] += "'" if bits[-1][-1] == "s" else "'s"
        else:
            bits[-1] = {"person": "people",
                        "wolf": "wolves",
                        "has": "have",
                        "is": "are",
                        "was": "were",
                        "succubus": "succubi"}.get(bits[-1], bits[-1] + "s")
        return " ".join(bits)

# vim: set expandtab:sw=4:ts=4: