'use strict';

const ZIP_ENCODING_UNKNOWN = 0;
const ZIP_ENCODING_ASCII = 1;
const ZIP_ENCODING_UTF8_KNOWN = 2;
const ZIP_ENCODING_UTF8_GUESSED = 3;
const ZIP_ENCODING_CP437 = 4;
const ZIP_ENCODING_ERROR = 5;

module.exports.zipEncodings = {
  unknown: ZIP_ENCODING_UNKNOWN,
  ascii: ZIP_ENCODING_ASCII,
  utf8known: ZIP_ENCODING_UTF8_KNOWN,
  utf8guessed: ZIP_ENCODING_UTF8_GUESSED,
  cp437: ZIP_ENCODING_CP437,
  error: ZIP_ENCODING_ERROR
}

const UTF_8_LEN_2_MASK     = 0xe0;
const UTF_8_LEN_2_MATCH    = 0xc0;
const UTF_8_LEN_3_MASK     = 0xf0;
const UTF_8_LEN_3_MATCH    = 0xe0;
const UTF_8_LEN_4_MASK     = 0xf8;
const UTF_8_LEN_4_MATCH    = 0xf0;
const UTF_8_CONTINUE_MASK  = 0xc0;
const UTF_8_CONTINUE_MATCH = 0x80;

module.exports.zipGuessEncoding = zipGuessEncoding;

function zipGuessEncoding (buf, expectedEncoding) {
  let enc = ZIP_ENCODING_ASCII;
  if (!buf) {
    return enc;
  }

  for (let i=0; i < buf.length; ++i) {
    if ((buf[i] > 31 && buf[i] < 128) || buf[i] == '\r' || buf[i] == '\n' || buf[i] == '\t') {
      continue;
    }

    let ulen;
    enc = ZIP_ENCODING_UTF8_GUESSED;
    if ((buf[i] & UTF_8_LEN_2_MASK) == UTF_8_LEN_2_MATCH)
      ulen = 1;
    else if ((buf[i] & UTF_8_LEN_3_MASK) == UTF_8_LEN_3_MATCH)
      ulen = 2;
    else if ((buf[i] & UTF_8_LEN_4_MASK) == UTF_8_LEN_4_MATCH)
      ulen = 3;
    else {
      enc = ZIP_ENCODING_CP437;
      break;
    }

    if (i + ulen >= buf.length) {
      enc = ZIP_ENCODING_CP437;
      break;
    }

    for (let j=1; j<=ulen; j++) {
      if ((buf[i+j] & UTF_8_CONTINUE_MASK) != UTF_8_CONTINUE_MATCH) {
        enc = ZIP_ENCODING_CP437;
        return done();
      }
    }
    i += ulen;
  }

  return done();

  function done () {
    const expected = expectedEncoding || ZIP_ENCODING_UNKNOWN;
    if (expected !== ZIP_ENCODING_UNKNOWN) {
      if (expected === ZIP_ENCODING_UTF8_KNOWN && enc === ZIP_ENCODING_UTF8_GUESSED) {
        enc = ZIP_ENCODING_UTF8_KNOWN;
      }
      if (expected !== enc && enc !== ZIP_ENCODING_ASCII) {
        return ZIP_ENCODING_ERROR;
      }
    }
    return enc;
  }
}