A String converter for a pseudo Middle English style, c. AD 1150.

olden.py

"""Small Module providing text "conversion" into a pseudo-Middle-English style.
    Based mostly, but loosely, on the orthography of Middle English c. AD 1150.
"""

from functools import partial
from re import compile
from typing import Any, Callable, Dict, Match, Union

__all__ = ("antiquate", "SubMap")


DEBUG: bool = False
SubMap = Dict[str, Union[str, Dict[str, Union[str, Any]]]]
table_chars: SubMap = {
    # Ash:
    "A[Ee]": "Æ",
    "a[Ee]": "æ",
    # OE Ligature:
    "O[Ee]": "Œ",
    "o[Ee]": "œ",
    # Eth:
    "D[Hh]": "Ð",
    "d[Hh]": "ð",
    # Thorn:
    "T[Hh]": "Þ",
    "t[Hh]": "þ",
    # Wynn:
    "W[Hh]?": "Ƿ",
    "w[Hh]?": "ƿ",
    # Yogh:
    "G[Hh]": "Ȝ",
    "g[Hh]": "ȝ",
    # Z and Soft C to S:
    "Z|C(?=[EeIi])": "S",
    "z|c(?=[EeIi])": "s",
    # (Probably) Soft G:
    "(?<=[AaEeIiOoUu])G(?=[EeIi])": "J",
    "(?<=[AaEeIiOoUu])g(?=[EeIi])": "j",
    # Long Vowels:
    r"(?i)(?<!\b)[aio]\we[dnsþ]?\b": {
        r"(?i)((?<!s)e(!þ)(?=\w$))|(e$)|(e(?=[^dnsþ]$))": "",
        "A": "Æ",
        "a": "æ",
        "I": "Y",
        "i": "y",
        "O": "Œ",
        "o": "œ",
    },
    # U/V:
    "U": "V",
    "u": "v",
    r"(?<!\b)v": "u",
    # Middle English "-ing":
    r"(?<=\w{2}[CcGgLlYy])ING(?=S?\W)": "INGE",
    r"(?<=\w{3})ING(?=S?\W)": "YNGE",
    r"(?i)(?<=\w{2}[cgly])ing(?=s?\W)": "inge",
    r"(?i)(?<=\w{3})ing(?=s?\W)": "ynge",
    # Long S:
    r"s(?!\b)": "ſ",
}


def _prepare(d: SubMap, _db=False) -> Callable[[str], str]:
    calls = [
        partial(compile(k).sub, _prepare(v, DEBUG) if isinstance(v, dict) else v)
        for k, v in d.items()
    ]

    def run(text: Union[Match, str]) -> str:
        if isinstance(text, Match):
            text = text.group()

        pre = text

        for call in calls:
            text = call(text)

        if _db:
            print(pre, "->", text)

        return text

    return run


_main: Callable[[str], str] = _prepare(table_chars)


def antiquate(text: str) -> str:
    """Convert a String into pseudo Middle English."""
    return _main(text)


if __name__ == "__main__":
    from sys import stdin

    for line in stdin:
        print(antiquate(line.strip()))