"""Reader-only vocabulary projection; calculation packets and CLI stay untouched."""

from __future__ import annotations

import json
from pathlib import Path
import re


# Both runtimes consume this file. ASCII regex semantics deliberately match JS:
# Python's default Unicode \b would leave Chinese page references untranslated.
_RULES = json.loads(Path(__file__).with_suffix(".rules.json").read_text(encoding="utf-8"))
_CHART_FENCE = re.compile(_RULES["chartFencePattern"], re.MULTILINE | re.ASCII)
_REPLACEMENTS = [
    (re.compile(rule["pattern"], re.ASCII | (re.IGNORECASE if rule.get("ignoreCase") else 0)), rule)
    for rule in _RULES["rules"]
]


def _replace(match: re.Match[str], rule: dict) -> str:
    if "replacement" in rule:
        return rule["replacement"]
    return (
        rule.get("prefix", "")
        + "".join(match[group] for group in rule.get("beforeGroups", []))
        + rule["labels"][match[rule["labelGroup"]]]
        + "".join(match[group] for group in rule.get("afterGroups", []))
    )


def clean_reader_appendix_markdown(markdown: str) -> str:
    """Translate copy without dropping lines, field names, or numeric cells.

    Ambiguous English words are changed only as complete pipe-table cells.
    Chart fences are immutable: their JSON is a separate renderer contract.
    """
    parts = _CHART_FENCE.split(markdown)
    for index in range(0, len(parts), 2):
        value = parts[index]
        for pattern, rule in _REPLACEMENTS:
            value = pattern.sub(lambda match: _replace(match, rule), value)
        parts[index] = value
    return "".join(parts)