"""
scripts/extract_hyphenation.py
Extrait les points de cesure depuis les logs de compilation TeX.

Deux formats geres, tous deux valides par execution reelle
(pdfTeX/LuaTeX, TeX Live 2023, 2026-09-02) :

  1. \\showhyphens{mot} (pdfTeX ou LuaTeX) -> une ligne par mot :
         [] \\tenrm mot-avec-ti-rets
     Fiable pour les tests mot-par-mot (test/words.tex).

  2. Callback LuaTeX post_linebreak_filter -> une ligne par ligne de
     paragraphe composee :
         LINE: Ny ma-ha-fi-na-ri-tra dia
     Necessaire pour les tests en contexte (test/sentences_lua.tex) --
     l'approche pdfTeX via les messages Overfull/Underfull a un angle
     mort confirme : les lignes "Loose" (badness finie, pas de
     depassement) n'impriment pas leur contenu meme avec \\hbadness=0.
     Voir docs/methodology.md section 3.
"""

from __future__ import annotations

import re
from pathlib import Path

SHOWHYPHENS_LINE = re.compile(r"^\[\]\s+\\\w+\s+(\S+)$", re.MULTILINE)
LUA_CALLBACK_LINE = re.compile(r"^LINE: (.+)$", re.MULTILINE)

PARAGRAPH_OVERFULL_LINE = re.compile(r"^\\\w+\s+(.+?)\s*\|$", re.MULTILINE)


def extract_hyphenated_words(log_content: str) -> list[str]:
    return [m.group(1) for m in SHOWHYPHENS_LINE.finditer(log_content)]


def extract_lua_lines(log_content: str) -> list[str]:
    results = []
    for m in LUA_CALLBACK_LINE.finditer(log_content):
        line = re.split(r"\s*\[\d+\{", m.group(1))[0]
        results.append(line.rstrip())
    return results


def extract_broken_lines_overfull(log_content: str) -> list[str]:
    return [m.group(1) for m in PARAGRAPH_OVERFULL_LINE.finditer(log_content)]


def main() -> None:
    import argparse
    parser = argparse.ArgumentParser()
    parser.add_argument("log_path", type=Path)
    parser.add_argument("output_path", type=Path)
    parser.add_argument(
        "--mode", choices=["showhyphens", "lua", "overfull"], default="showhyphens",
    )
    args = parser.parse_args()

    content = args.log_path.read_text(encoding="utf-8", errors="replace")
    extractors = {
        "showhyphens": extract_hyphenated_words,
        "lua": extract_lua_lines,
        "overfull": extract_broken_lines_overfull,
    }
    results = extractors[args.mode](content)
    args.output_path.write_text("\n".join(results) + "\n", encoding="utf-8")
    print(f"{len(results)} entree(s) extraite(s) -> {args.output_path}")


if __name__ == "__main__":
    main()