matzo-web/js/language.js

const prefixRE = (words) => {
  return new RegExp("^(?:" + words.join("|") + ")", "i");
}
const wordRE = (words) => {
  return new RegExp("^(?:" + words.join("|") + ")$", "i");
}

const builtins = wordRE([
  // useful helpers
  "rep",

  // string manipulation
  "str/upper",
  "str/lower",
  "str/capitalize",
  "wd", "se",

  // arithmetic
  "add", "sub", "mul",

  // tuple stuff
  "tuple/len", "tuple/concat", "tuple/flatten", "tuple/join",
  "tuple/index", "tuple/rep", "tuple/replace", "tuple/fold",
  "tuple/map",
]);

const keywords = wordRE([
  "puts",
  "case",
  "let",
  "in",
  "fix",
  "record",
]);

const normal = (stream, state) => {
  let ch = stream.next();
  if (ch == "(") {
    if (stream.eat("*")) {
      comment(stream, state);
      return "comment";
    }
  }

  if (ch == "\"" || ch == "'")
    return (state.cur = string(ch))(stream, state);
  if (ch == "[" && /[\[=]/.test(stream.peek()))
    return (state.cur = bracketed(readBracket(stream), "string"))(stream, state);
  if (/\d/.test(ch)) {
    stream.eatWhile(/[\w.%]/);
    return "number";
  }
  if (/[\w_]/.test(ch)) {
    stream.eatWhile(/[\w\\\-_.]/);
    return "variable";
  }
  return null;
}

const comment = (stream, state) => {
  let ch;
  while (ch = stream.next()) {
    if (ch == "*" && stream.next() == ")") {
      state.cur = normal;
      return "comment";
    }
  }
  state.cur = comment;
};

const bracketed = (level, style) => {
  return (stream, state) => {
    let curlev = null, ch;
    while ((ch = stream.next()) != null) {
      if (curlev == null) {
        if (ch == "]") curlev = 0;
      } else if (ch == "=") {
        ++curlev;
      } else if (ch == "]" && curlev == level) {
        state.cur = normal;
        break;
      } else {
        curlev = null;
      }
    }
    return style;
  };
}

const string = (quote) => {
  return (stream, state) => {
    let escaped = false, ch;
    while ((ch = stream.next()) != null) {
      if (ch == quote && !escaped) break;
      escaped = !escaped && ch == "\\";
    }
    if (!escaped)
      state.cur = normal;
    return "string";
  };
}

export const matzoLang = {
  startState: (basecol) => {
    return {cur: normal};
  },

  token: (stream, state) => {
    if (stream.eatSpace())
      return null;

    let style = state.cur(stream, state);
    const word = stream.current();

    if (style == "variable") {
      if (keywords.test(word))
        style = "keyword";
      else if (builtins.test(word))
        style = "builtin";
    }

    return style;
  },

  languageData: {
    commentTokens: {block: {open: "(\*", close: "\*)"}}
  }
};