oberonjs/browser/jslibs/codemirror/javascript.js

// TODO actually recognize syntax of TypeScript constructs

CodeMirror.defineMode("javascript", function(config, parserConfig) {
  var indentUnit = config.indentUnit;
  var statementIndent = parserConfig.statementIndent;
  var jsonMode = parserConfig.json;
  var isTS = parserConfig.typescript;

  // Tokenizer

  var keywords = function(){
    function kw(type) {return {type: type, style: "keyword"};}
    var A = kw("keyword a"), B = kw("keyword b"), C = kw("keyword c");
    var operator = kw("operator"), atom = {type: "atom", style: "atom"};

    var jsKeywords = {
      "if": kw("if"), "while": A, "with": A, "else": B, "do": B, "try": B, "finally": B,
      "return": C, "break": C, "continue": C, "new": C, "delete": C, "throw": C,
      "var": kw("var"), "const": kw("var"), "let": kw("var"),
      "function": kw("function"), "catch": kw("catch"),
      "for": kw("for"), "switch": kw("switch"), "case": kw("case"), "default": kw("default"),
      "in": operator, "typeof": operator, "instanceof": operator,
      "true": atom, "false": atom, "null": atom, "undefined": atom, "NaN": atom, "Infinity": atom,
      "this": kw("this")
    };

    // Extend the 'normal' keywords with the TypeScript language extensions
    if (isTS) {
      var type = {type: "variable", style: "variable-3"};
      var tsKeywords = {
        // object-like things
        "interface": kw("interface"),
        "class": kw("class"),
        "extends": kw("extends"),
        "constructor": kw("constructor"),

        // scope modifiers
        "public": kw("public"),
        "private": kw("private"),
        "protected": kw("protected"),
        "static": kw("static"),

        "super": kw("super"),

        // types
        "string": type, "number": type, "bool": type, "any": type
      };

      for (var attr in tsKeywords) {
        jsKeywords[attr] = tsKeywords[attr];
      }
    }

    return jsKeywords;
  }();

  var isOperatorChar = /[+\-*&%=<>!?|~^]/;

  function chain(stream, state, f) {
    state.tokenize = f;
    return f(stream, state);
  }

  function nextUntilUnescaped(stream, end) {
    var escaped = false, next;
    while ((next = stream.next()) != null) {
      if (next == end && !escaped)
        return false;
      escaped = !escaped && next == "\\";
    }
    return escaped;
  }

  // Used as scratch variables to communicate multiple values without
  // consing up tons of objects.
  var type, content;
  function ret(tp, style, cont) {
    type = tp; content = cont;
    return style;
  }
  function jsTokenBase(stream, state) {
    var ch = stream.next();
    if (ch == '"' || ch == "'")
      return chain(stream, state, jsTokenString(ch));
    else if (ch == "." && stream.match(/^\d+(?:[eE][+\-]?\d+)?/))
      return ret("number", "number");
    else if (/[\[\]{}\(\),;\:\.]/.test(ch))
      return ret(ch);
    else if (ch == "0" && stream.eat(/x/i)) {
      stream.eatWhile(/[\da-f]/i);
      return ret("number", "number");
    }
    else if (/\d/.test(ch)) {
      stream.match(/^\d*(?:\.\d*)?(?:[eE][+\-]?\d+)?/);
      return ret("number", "number");
    }
    else if (ch == "/") {
      if (stream.eat("*")) {
        return chain(stream, state, jsTokenComment);
      }
      else if (stream.eat("/")) {
        stream.skipToEnd();
        return ret("comment", "comment");
      }
      else if (state.lastType == "operator" || state.lastType == "keyword c" ||
               /^[\[{}\(,;:]$/.test(state.lastType)) {
        nextUntilUnescaped(stream, "/");
        stream.eatWhile(/[gimy]/); // 'y' is "sticky" option in Mozilla
        return ret("regexp", "string-2");
      }
      else {
        stream.eatWhile(isOperatorChar);
        return ret("operator", null, stream.current());
      }
    }
    else if (ch == "#") {
      stream.skipToEnd();
      return ret("error", "error");
    }
    else if (isOperatorChar.test(ch)) {
      stream.eatWhile(isOperatorChar);
      return ret("operator", null, stream.current());
    }
    else {
      stream.eatWhile(/[\w\$_]/);
      var word = stream.current(), known = keywords.propertyIsEnumerable(word) && keywords[word];
      return (known && state.lastType != ".") ? ret(known.type, known.style, word) :
                     ret("variable", "variable", word);
    }
  }

  function jsTokenString(quote) {
    return function(stream, state) {
      if (!nextUntilUnescaped(stream, quote))
        state.tokenize = jsTokenBase;
      return ret("string", "string");
    };
  }

  function jsTokenComment(stream, state) {
    var maybeEnd = false, ch;
    while (ch = stream.next()) {
      if (ch == "/" && maybeEnd) {
        state.tokenize = jsTokenBase;
        break;
      }
      maybeEnd = (ch == "*");
    }
    return ret("comment", "comment");
  }

  // Parser

  var atomicTypes = {"atom": true, "number": true, "variable": true, "string": true, "regexp": true, "this": true};

  function JSLexical(indented, column, type, align, prev, info) {
    this.indented = indented;
    this.column = column;
    this.type = type;
    this.prev = prev;
    this.info = info;
    if (align != null) this.align = align;
  }

  function inScope(state, varname) {
    for (var v = state.localVars; v; v = v.next)
      if (v.name == varname) return true;
  }

  function parseJS(state, style, type, content, stream) {
    var cc = state.cc;
    // Communicate our context to the combinators.
    // (Less wasteful than consing up a hundred closures on every call.)
    cx.state = state; cx.stream = stream; cx.marked = null, cx.cc = cc;

    if (!state.lexical.hasOwnProperty("align"))
      state.lexical.align = true;

    while(true) {
      var combinator = cc.length ? cc.pop() : jsonMode ? expression : statement;
      if (combinator(type, content)) {
        while(cc.length && cc[cc.length - 1].lex)
          cc.pop()();
        if (cx.marked) return cx.marked;
        if (type == "variable" && inScope(state, content)) return "variable-2";
        return style;
      }
    }
  }

  // Combinator utils

  var cx = {state: null, column: null, marked: null, cc: null};
  function pass() {
    for (var i = arguments.length - 1; i >= 0; i--) cx.cc.push(arguments[i]);
  }
  function cont() {
    pass.apply(null, arguments);
    return true;
  }
  function register(varname) {
    function inList(list) {
      for (var v = list; v; v = v.next)
        if (v.name == varname) return true;
      return false;
    }
    var state = cx.state;
    if (state.context) {
      cx.marked = "def";
      if (inList(state.localVars)) return;
      state.localVars = {name: varname, next: state.localVars};
    } else {
      if (inList(state.globalVars)) return;
      state.globalVars = {name: varname, next: state.globalVars};
    }
  }

  // Combinators

  var defaultVars = {name: "this", next: {name: "arguments"}};
  function pushcontext() {
    cx.state.context = {prev: cx.state.context, vars: cx.state.localVars};
    cx.state.localVars = defaultVars;
  }
  function popcontext() {
    cx.state.localVars = cx.state.context.vars;
    cx.state.context = cx.state.context.prev;
  }
  function pushlex(type, info) {
    var result = function() {
      var state = cx.state, indent = state.indented;
      if (state.lexical.type == "stat") indent = state.lexical.indented;
      state.lexical = new JSLexical(indent, cx.stream.column(), type, null, state.lexical, info);
    };
    result.lex = true;
    return result;
  }
  function poplex() {
    var state = cx.state;
    if (state.lexical.prev) {
      if (state.lexical.type == ")")
        state.indented = state.lexical.indented;
      state.lexical = state.lexical.prev;
    }
  }
  poplex.lex = true;

  function expect(wanted) {
    return function(type) {
      if (type == wanted) return cont();
      else if (wanted == ";") return pass();
      else return cont(arguments.callee);
    };
  }

  function statement(type) {
    if (type == "var") return cont(pushlex("vardef"), vardef1, expect(";"), poplex);
    if (type == "keyword a") return cont(pushlex("form"), expression, statement, poplex);
    if (type == "keyword b") return cont(pushlex("form"), statement, poplex);
    if (type == "{") return cont(pushlex("}"), block, poplex);
    if (type == ";") return cont();
    if (type == "if") return cont(pushlex("form"), expression, statement, poplex, maybeelse);
    if (type == "function") return cont(functiondef);
    if (type == "for") return cont(pushlex("form"), expect("("), pushlex(")"), forspec1, expect(")"),
                                   poplex, statement, poplex);
    if (type == "variable") return cont(pushlex("stat"), maybelabel);
    if (type == "switch") return cont(pushlex("form"), expression, pushlex("}", "switch"), expect("{"),
                                      block, poplex, poplex);
    if (type == "case") return cont(expression, expect(":"));
    if (type == "default") return cont(expect(":"));
    if (type == "catch") return cont(pushlex("form"), pushcontext, expect("("), funarg, expect(")"),
                                     statement, poplex, popcontext);
    return pass(pushlex("stat"), expression, expect(";"), poplex);
  }
  function expression(type) {
    return expressionInner(type, false);
  }
  function expressionNoComma(type) {
    return expressionInner(type, true);
  }
  function expressionInner(type, noComma) {
    var maybeop = noComma ? maybeoperatorNoComma : maybeoperatorComma;
    if (atomicTypes.hasOwnProperty(type)) return cont(maybeop);
    if (type == "function") return cont(functiondef);
    if (type == "keyword c") return cont(noComma ? maybeexpressionNoComma : maybeexpression);
    if (type == "(") return cont(pushlex(")"), maybeexpression, expect(")"), poplex, maybeop);
    if (type == "operator") return cont(noComma ? expressionNoComma : expression);
    if (type == "[") return cont(pushlex("]"), commasep(expressionNoComma, "]"), poplex, maybeop);
    if (type == "{") return cont(pushlex("}"), commasep(objprop, "}"), poplex, maybeop);
    return cont();
  }
  function maybeexpression(type) {
    if (type.match(/[;\}\)\],]/)) return pass();
    return pass(expression);
  }
  function maybeexpressionNoComma(type) {
    if (type.match(/[;\}\)\],]/)) return pass();
    return pass(expressionNoComma);
  }

  function maybeoperatorComma(type, value) {
    if (type == ",") return cont(expression);
    return maybeoperatorNoComma(type, value, false);
  }
  function maybeoperatorNoComma(type, value, noComma) {
    var me = noComma == false ? maybeoperatorComma : maybeoperatorNoComma;
    var expr = noComma == false ? expression : expressionNoComma;
    if (type == "operator") {
      if (/\+\+|--/.test(value)) return cont(me);
      if (value == "?") return cont(expression, expect(":"), expr);
      return cont(expr);
    }
    if (type == ";") return;
    if (type == "(") return cont(pushlex(")", "call"), commasep(expressionNoComma, ")"), poplex, me);
    if (type == ".") return cont(property, me);
    if (type == "[") return cont(pushlex("]"), maybeexpression, expect("]"), poplex, me);
  }
  function maybelabel(type) {
    if (type == ":") return cont(poplex, statement);
    return pass(maybeoperatorComma, expect(";"), poplex);
  }
  function property(type) {
    if (type == "variable") {cx.marked = "property"; return cont();}
  }
  function objprop(type, value) {
    if (type == "variable") {
      cx.marked = "property";
      if (value == "get" || value == "set") return cont(getterSetter);
    } else if (type == "number" || type == "string") {
      cx.marked = type + " property";
    }
    if (atomicTypes.hasOwnProperty(type)) return cont(expect(":"), expressionNoComma);
  }
  function getterSetter(type) {
    if (type == ":") return cont(expression);
    if (type != "variable") return cont(expect(":"), expression);
    cx.marked = "property";
    return cont(functiondef);
  }
  function commasep(what, end) {
    function proceed(type) {
      if (type == ",") {
        var lex = cx.state.lexical;
        if (lex.info == "call") lex.pos = (lex.pos || 0) + 1;
        return cont(what, proceed);
      }
      if (type == end) return cont();
      return cont(expect(end));
    }
    return function(type) {
      if (type == end) return cont();
      else return pass(what, proceed);
    };
  }
  function block(type) {
    if (type == "}") return cont();
    return pass(statement, block);
  }
  function maybetype(type) {
    if (type == ":") return cont(typedef);
    return pass();
  }
  function typedef(type) {
    if (type == "variable"){cx.marked = "variable-3"; return cont();}
    return pass();
  }
  function vardef1(type, value) {
    if (type == "variable") {
      register(value);
      return isTS ? cont(maybetype, vardef2) : cont(vardef2);
    }
    return pass();
  }
  function vardef2(type, value) {
    if (value == "=") return cont(expressionNoComma, vardef2);
    if (type == ",") return cont(vardef1);
  }
  function maybeelse(type, value) {
    if (type == "keyword b" && value == "else") return cont(pushlex("form"), statement, poplex);
  }
  function forspec1(type) {
    if (type == "var") return cont(vardef1, expect(";"), forspec2);
    if (type == ";") return cont(forspec2);
    if (type == "variable") return cont(formaybein);
    return pass(expression, expect(";"), forspec2);
  }
  function formaybein(_type, value) {
    if (value == "in") return cont(expression);
    return cont(maybeoperatorComma, forspec2);
  }
  function forspec2(type, value) {
    if (type == ";") return cont(forspec3);
    if (value == "in") return cont(expression);
    return pass(expression, expect(";"), forspec3);
  }
  function forspec3(type) {
    if (type != ")") cont(expression);
  }
  function functiondef(type, value) {
    if (type == "variable") {register(value); return cont(functiondef);}
    if (type == "(") return cont(pushlex(")"), pushcontext, commasep(funarg, ")"), poplex, statement, popcontext);
  }
  function funarg(type, value) {
    if (type == "variable") {register(value); return isTS ? cont(maybetype) : cont();}
  }

  // Interface

  return {
    startState: function(basecolumn) {
      return {
        tokenize: jsTokenBase,
        lastType: null,
        cc: [],
        lexical: new JSLexical((basecolumn || 0) - indentUnit, 0, "block", false),
        localVars: parserConfig.localVars,
        globalVars: parserConfig.globalVars,
        context: parserConfig.localVars && {vars: parserConfig.localVars},
        indented: 0
      };
    },

    token: function(stream, state) {
      if (stream.sol()) {
        if (!state.lexical.hasOwnProperty("align"))
          state.lexical.align = false;
        state.indented = stream.indentation();
      }
      if (state.tokenize != jsTokenComment && stream.eatSpace()) return null;
      var style = state.tokenize(stream, state);
      if (type == "comment") return style;
      state.lastType = type == "operator" && (content == "++" || content == "--") ? "incdec" : type;
      return parseJS(state, style, type, content, stream);
    },

    indent: function(state, textAfter) {
      if (state.tokenize == jsTokenComment) return CodeMirror.Pass;
      if (state.tokenize != jsTokenBase) return 0;
      var firstChar = textAfter && textAfter.charAt(0), lexical = state.lexical;
      // Kludge to prevent 'maybelse' from blocking lexical scope pops
      for (var i = state.cc.length - 1; i >= 0; --i) {
        var c = state.cc[i];
        if (c == poplex) lexical = lexical.prev;
        else if (c != maybeelse || /^else\b/.test(textAfter)) break;
      }
      if (lexical.type == "stat" && firstChar == "}") lexical = lexical.prev;
      if (statementIndent && lexical.type == ")" && lexical.prev.type == "stat")
        lexical = lexical.prev;
      var type = lexical.type, closing = firstChar == type;

      if (type == "vardef") return lexical.indented + (state.lastType == "operator" || state.lastType == "," ? 4 : 0);
      else if (type == "form" && firstChar == "{") return lexical.indented;
      else if (type == "form") return lexical.indented + indentUnit;
      else if (type == "stat")
        return lexical.indented + (state.lastType == "operator" || state.lastType == "," ? statementIndent || indentUnit : 0);
      else if (lexical.info == "switch" && !closing && parserConfig.doubleIndentSwitch != false)
        return lexical.indented + (/^(?:case|default)\b/.test(textAfter) ? indentUnit : 2 * indentUnit);
      else if (lexical.align) return lexical.column + (closing ? 0 : 1);
      else return lexical.indented + (closing ? 0 : indentUnit);
    },

    electricChars: ":{}",
    blockCommentStart: jsonMode ? null : "/*",
    blockCommentEnd: jsonMode ? null : "*/",
    lineComment: jsonMode ? null : "//",
    fold: "brace",

    helperType: jsonMode ? "json" : "javascript",
    jsonMode: jsonMode
  };
});

CodeMirror.defineMIME("text/javascript", "javascript");
CodeMirror.defineMIME("text/ecmascript", "javascript");
CodeMirror.defineMIME("application/javascript", "javascript");
CodeMirror.defineMIME("application/ecmascript", "javascript");
CodeMirror.defineMIME("application/json", {name: "javascript", json: true});
CodeMirror.defineMIME("application/x-json", {name: "javascript", json: true});
CodeMirror.defineMIME("text/typescript", { name: "javascript", typescript: true });
CodeMirror.defineMIME("application/typescript", { name: "javascript", typescript: true });