#!/usr/bin/env python3
"""Classify native RS field scripts and movements by their first source macro."""
import argparse
import re
from pathlib import Path

ROOT = Path(__file__).resolve().parents[1]


def generate(pret, out):
    event = (pret / "include/macros/event.inc").read_text()
    movement = (pret / "include/macros/movement.inc").read_text()
    scripts = set(re.findall(r"^\s*\.macro\s+(\w+)", event, re.M))
    moves = set(re.findall(r"^\s*create_movement_action\s+(\w+)", movement, re.M))
    moves.update(re.findall(r"^\s*\.macro\s+(\w+)", movement, re.M))
    classified = {"scripts": set(), "movements": set(), "data": set()}
    paths = set()
    def include(path):
        if path in paths:
            return
        paths.add(path)
        for rel in re.findall(r'^\s*\.include\s+"(data/[^\"]+)"', path.read_text(), re.M):
            include(pret / rel)
    include(pret / "data/event_scripts.s")
    for path in sorted(paths):
        pending = []
        for raw in path.read_text(errors="replace").splitlines():
            line = raw.split("@", 1)[0].strip()
            if not line or line.startswith("#"):
                continue
            label = re.fullmatch(r"(\w+)::?", line)
            if label:
                pending.append(label[1])
                continue
            token = line.split()[0]
            if token in (".align", ".balign"):
                continue
            kind = ("data" if token in ("brailleformat", ".byte", ".2byte", ".4byte", ".string", ".incbin")
                    else "scripts" if token in scripts else "movements" if token in moves else None)
            if kind:
                classified[kind].update(pending)
            pending = []
    lines = ["-- Generated by tools/gen_rs_script_catalog.py from pokeruby source macros.", "return {"]
    for kind, names in classified.items():
        lines.append("  " + kind + " = {")
        lines.extend('    "' + name + '",' for name in sorted(names))
        lines.append("  },")
    lines.append("}")
    out.parent.mkdir(parents=True, exist_ok=True)
    out.write_text("\n".join(lines) + "\n")
    print(f"Classified {len(classified['scripts'])} script labels and {len(classified['movements'])} movement labels")


if __name__ == "__main__":
    parser = argparse.ArgumentParser(description=__doc__)
    parser.add_argument("--pret", type=Path, default=ROOT.parent / "pokeruby")
    parser.add_argument("--out", type=Path, default=ROOT / "src/import/gba/versions_scripts_rs.lua")
    args = parser.parse_args()
    generate(args.pret, args.out)