"""Replace mojibake sequences in UTF-8 files with their original characters.

The repo had a few files saved by a tool that interpreted UTF-8 as cp1252 and
re-saved as UTF-8, producing classic mojibake (em-dash --> 'â€"', section sign
--> '§', right arrow --> 'â†'', etc.). This script rewrites those sequences
in-place without touching legitimately-encoded UTF-8 elsewhere in the file.
"""
import sys

PAIRS = [
    ('—', '—'),  # em-dash
    ('–', '–'),  # en-dash
    ('’', '’'),  # right single quote
    ('‘', '‘'),  # left single quote
    ('“', '“'),  # left double quote
    ('”', '”'),  # right double quote
    ('…', '…'),  # horizontal ellipsis
    ('•', '•'),  # bullet
    ('→', '→'),  # right arrow
    ('↑', '←'),  # left arrow
    ('↑', '↑'),  # up arrow (same prefix collision - rare)
    ('↓', '↓'),  # down arrow
    ('≤', '≤'),  # less-than-or-equal
    ('≥', '≥'),  # greater-than-or-equal
    ('×',       '×'),  # multiplication sign
    ('÷',       '÷'),  # division sign
    ('§',       '§'),  # section sign
    ('°',       '°'),  # degree
    ('½',       '½'),  # one-half
    ('«',       '«'),  # «
    ('»',       '»'),  # »
    ('©',       '©'),  # ©
    ('®',       '®'),  # ®
]


def fix(path: str) -> int:
    with open(path, 'rb') as f:
        raw = f.read()
    had_bom = raw.startswith(b'\xef\xbb\xbf')
    if had_bom:
        raw = raw[3:]
    text = raw.decode('utf-8')
    n = 0
    for bad, good in PAIRS:
        if bad in text:
            n += text.count(bad)
            text = text.replace(bad, good)
    if n == 0 and not had_bom:
        print(f'{path}: no mojibake found')
        return 0
    with open(path, 'w', encoding='utf-8', newline='\n') as f:
        f.write(text)
    note = f' (also stripped BOM)' if had_bom else ''
    print(f'{path}: replaced {n} sequence(s){note}')
    return n


if __name__ == '__main__':
    for p in sys.argv[1:]:
        fix(p)