Clean up Typey Types progress output

cleanup_typey-type_progress.py

import json
import string
from collections import Counter
from datetime import datetime
from pathlib import Path
from typing import Dict, List

PROGRESS_FILE = (
    Path("~/.config/plover/typey-type_progress.json").expanduser().resolve(strict=True)
)


entries: Dict[str, int] = json.loads(PROGRESS_FILE.read_text())

# break into counts
words: List[str] = list(Counter(entries).elements())

# strip leading and trailing spaces
words = [word.strip() for word in words]

# strip punctuation and internal whitespace
CHARACTER_EXCEPTIONS = {"-"}

characters_to_remove = "".join(
    set(string.punctuation + string.whitespace) - CHARACTER_EXCEPTIONS
)
translation_table = str.maketrans("", "", characters_to_remove)
words = [word.translate(translation_table) for word in words]

# normalize capitalization
words = [word.casefold() for word in words]

CAPITALIZATION_EXCEPTIONS = {
    "paul": "Paul",
    "french": "French",
}

words = [CAPITALIZATION_EXCEPTIONS.get(word, word) for word in words]

# make a backup
old_progress_file = PROGRESS_FILE.with_name(
    PROGRESS_FILE.name + f"{datetime.now().isoformat(timespec='seconds')}.bak"
)
assert (
    not old_progress_file.exists()
), f"'{old_progress_file}' already exists; please rename it"
PROGRESS_FILE.rename(old_progress_file)

# write out cleaned progress
counts = Counter(words)
cleaned_progress = {key: counts[key] for key in counts}
PROGRESS_FILE.write_text(json.dumps(cleaned_progress))