// Generates engine/chartable.zena (the byte-class constants) and engine/chartable.wat (the table itself, a
// data segment scripts/build.sh merges in) for engine/lex.zena's byte classification. Run:
// node scripts/gen/chartable.mjs
import fs from "node:fs";
import path from "node:path";
import { CLASS_BASE, LEXC_BASE, dataModule } from "./layout.mjs";

// Bit flags, one 256-byte table instead of lex.zena's isIdStart/isIdChar/isDigit range-check chains and
// its trivia if-chain (space/tab/VT/FF, CR/LF). ID_CHAR is precomputed as ID_START | DIGIT so the hot
// identifier/number continuation loops stay a single mask test.
const ID_START = 1;
const DIGIT = 2;
const ID_CHAR = 4;
const WS = 8;
const NL = 16;

const table = Buffer.alloc(256);
for (let c = 0; c < 256; c++) {
  let b = 0;
  const isStart = (c >= 97 && c <= 122) || (c >= 65 && c <= 90) || c === 95 || c === 36;
  const isDigit = c >= 48 && c <= 57;
  if (isStart) b |= ID_START;
  if (isDigit) b |= DIGIT;
  if (isStart || isDigit) b |= ID_CHAR;
  if (c === 32 || c === 9 || c === 11 || c === 12) b |= WS;
  if (c === 10 || c === 13) b |= NL;
  table[c] = b;
}

const out = `// GENERATED by scripts/gen/chartable.mjs. Do not edit by hand.
// Byte-class table for engine/lex.zena: one getU8(CLASS + c) load classifies a byte, replacing range-check
// chains for isIdStart/isIdChar/isDigit and the trivia if-chain.
export let CLASS: i32 = ${CLASS_BASE};
export let ID_START: i32 = ${ID_START};
export let DIGIT: i32 = ${DIGIT};
export let ID_CHAR: i32 = ${ID_CHAR}; // ID_START | DIGIT, so identifier/number continuation is one mask test
export let WS: i32 = ${WS};
export let NL: i32 = ${NL};
`;

const dest = path.resolve(import.meta.dirname, "../../engine/chartable.zena");
fs.writeFileSync(dest, out);
const wat = path.resolve(import.meta.dirname, "../../engine/chartable.wat");
fs.writeFileSync(wat, dataModule("chartable.mjs", CLASS_BASE, table, LEXC_BASE));
console.log(`wrote ${dest} and ${wat}: 256 bytes of table`);