Fix DOH labels in fx150

fix_doh_labels.py

#!/usr/bin/env python3
"""
Remove trailing punctuation from specific Fluent message attributes in
browser/browser/preferences/preferences.ftl across all locales.

- preferences-doh-radio-default/custom/off: fix .description
- preferences-doh-fallback-label: fix .label
"""

import os
import re
import unicodedata

ROOT = os.path.dirname(os.path.abspath(__file__))

# Map each message ID to the attribute that should have trailing punctuation removed
MESSAGE_ATTR = {
    "preferences-doh-radio-default": "description",
    "preferences-doh-radio-custom": "description",
    "preferences-doh-radio-off": "description",
    "preferences-doh-fallback-label": "label",
}

ATTR_RE = re.compile(r"^(\s+\.(label|description)\s*=\s*)(.+)$")


def strip_trailing_punctuation(text):
    """Strip all trailing Unicode punctuation characters."""
    v = text.rstrip()
    while v and unicodedata.category(v[-1]).startswith("P"):
        v = v[:-1]
    return v


def fix_file(filepath):
    with open(filepath, encoding="utf-8") as f:
        lines = f.readlines()

    new_lines = []
    current_target_attr = None  # attribute name to fix for the current message
    attr_consumed = False
    changed = False

    for line in lines:
        stripped = line.rstrip("\n")

        # Check if this line starts one of the target messages
        matched_msg = next(
            (msg for msg in MESSAGE_ATTR if stripped == f"{msg} =" or stripped.startswith(f"{msg} =")),
            None,
        )
        if matched_msg:
            current_target_attr = MESSAGE_ATTR[matched_msg]
            attr_consumed = False
            new_lines.append(line)
            continue

        if current_target_attr:
            # A non-indented, non-empty line means we've left the message block
            if stripped and not stripped[0] in (" ", "\t"):
                current_target_attr = None
                attr_consumed = False
                new_lines.append(line)
                continue

            # Check for the target attribute (only the first occurrence per message)
            if not attr_consumed:
                m = ATTR_RE.match(stripped)
                if m and m.group(2) == current_target_attr:
                    prefix = m.group(1)
                    value = m.group(3)
                    fixed = strip_trailing_punctuation(value)
                    if fixed != value.rstrip():
                        trailing = value[len(value.rstrip()):]
                        new_line = line.replace(prefix + value, prefix + fixed + trailing, 1)
                        print(f"  {os.path.relpath(filepath, ROOT)}: {value!r} -> {fixed!r}")
                        line = new_line
                        changed = True
                    attr_consumed = True

        new_lines.append(line)

    if changed:
        with open(filepath, "w", encoding="utf-8") as f:
            f.writelines(new_lines)

    return changed


def main():
    total_changed = 0
    for entry in sorted(os.listdir(ROOT)):
        if entry.startswith("."):
            continue
        locale_dir = os.path.join(ROOT, entry)
        if not os.path.isdir(locale_dir):
            continue

        ftl_path = os.path.join(
            locale_dir, "browser", "browser", "preferences", "preferences.ftl"
        )
        if not os.path.isfile(ftl_path):
            continue

        if fix_file(ftl_path):
            total_changed += 1

    print(f"\nDone. Modified files: {total_changed}")


if __name__ == "__main__":
    main()