TwitchBot/node_modules/jade/lib/lexer.js

'use strict';

var utils = require('./utils');
var characterParser = require('character-parser');


/**
 * Initialize `Lexer` with the given `str`.
 *
 * @param {String} str
 * @param {String} filename
 * @api private
 */

var Lexer = module.exports = function Lexer(str, filename) {
  this.input = str.replace(/\r\n|\r/g, '\n');
  this.filename = filename;
  this.deferredTokens = [];
  this.lastIndents = 0;
  this.lineno = 1;
  this.stash = [];
  this.indentStack = [];
  this.indentRe = null;
  this.pipeless = false;
};


function assertExpression(exp) {
  //this verifies that a JavaScript expression is valid
  Function('', 'return (' + exp + ')');
}
function assertNestingCorrect(exp) {
  //this verifies that code is properly nested, but allows
  //invalid JavaScript such as the contents of `attributes`
  var res = characterParser(exp)
  if (res.isNesting()) {
    throw new Error('Nesting must match on expression `' + exp + '`')
  }
}

/**
 * Lexer prototype.
 */

Lexer.prototype = {

  /**
   * Construct a token with the given `type` and `val`.
   *
   * @param {String} type
   * @param {String} val
   * @return {Object}
   * @api private
   */

  tok: function(type, val){
    return {
        type: type
      , line: this.lineno
      , val: val
    }
  },

  /**
   * Consume the given `len` of input.
   *
   * @param {Number} len
   * @api private
   */

  consume: function(len){
    this.input = this.input.substr(len);
  },

  /**
   * Scan for `type` with the given `regexp`.
   *
   * @param {String} type
   * @param {RegExp} regexp
   * @return {Object}
   * @api private
   */

  scan: function(regexp, type){
    var captures;
    if (captures = regexp.exec(this.input)) {
      this.consume(captures[0].length);
      return this.tok(type, captures[1]);
    }
  },

  /**
   * Defer the given `tok`.
   *
   * @param {Object} tok
   * @api private
   */

  defer: function(tok){
    this.deferredTokens.push(tok);
  },

  /**
   * Lookahead `n` tokens.
   *
   * @param {Number} n
   * @return {Object}
   * @api private
   */

  lookahead: function(n){
    var fetch = n - this.stash.length;
    while (fetch-- > 0) this.stash.push(this.next());
    return this.stash[--n];
  },

  /**
   * Return the indexOf `(` or `{` or `[` / `)` or `}` or `]` delimiters.
   *
   * @return {Number}
   * @api private
   */

  bracketExpression: function(skip){
    skip = skip || 0;
    var start = this.input[skip];
    if (start != '(' && start != '{' && start != '[') throw new Error('unrecognized start character');
    var end = ({'(': ')', '{': '}', '[': ']'})[start];
    var range = characterParser.parseMax(this.input, {start: skip + 1});
    if (this.input[range.end] !== end) throw new Error('start character ' + start + ' does not match end character ' + this.input[range.end]);
    return range;
  },

  /**
   * Stashed token.
   */

  stashed: function() {
    return this.stash.length
      && this.stash.shift();
  },

  /**
   * Deferred token.
   */

  deferred: function() {
    return this.deferredTokens.length
      && this.deferredTokens.shift();
  },

  /**
   * end-of-source.
   */

  eos: function() {
    if (this.input.length) return;
    if (this.indentStack.length) {
      this.indentStack.shift();
      return this.tok('outdent');
    } else {
      return this.tok('eos');
    }
  },

  /**
   * Blank line.
   */

  blank: function() {
    var captures;
    if (captures = /^\n *\n/.exec(this.input)) {
      this.consume(captures[0].length - 1);
      ++this.lineno;
      if (this.pipeless) return this.tok('text', '');
      return this.next();
    }
  },

  /**
   * Comment.
   */

  comment: function() {
    var captures;
    if (captures = /^\/\/(-)?([^\n]*)/.exec(this.input)) {
      this.consume(captures[0].length);
      var tok = this.tok('comment', captures[2]);
      tok.buffer = '-' != captures[1];
      return tok;
    }
  },

  /**
   * Interpolated tag.
   */

  interpolation: function() {
    if (/^#\{/.test(this.input)) {
      var match;
      try {
        match = this.bracketExpression(1);
      } catch (ex) {
        return;//not an interpolation expression, just an unmatched open interpolation
      }

      this.consume(match.end + 1);
      return this.tok('interpolation', match.src);
    }
  },

  /**
   * Tag.
   */

  tag: function() {
    var captures;
    if (captures = /^(\w[-:\w]*)(\/?)/.exec(this.input)) {
      this.consume(captures[0].length);
      var tok, name = captures[1];
      if (':' == name[name.length - 1]) {
        name = name.slice(0, -1);
        tok = this.tok('tag', name);
        this.defer(this.tok(':'));
        while (' ' == this.input[0]) this.input = this.input.substr(1);
      } else {
        tok = this.tok('tag', name);
      }
      tok.selfClosing = !! captures[2];
      return tok;
    }
  },

  /**
   * Filter.
   */

  filter: function() {
    return this.scan(/^:([\w\-]+)/, 'filter');
  },

  /**
   * Doctype.
   */

  doctype: function() {
    if (this.scan(/^!!! *([^\n]+)?/, 'doctype')) {
      throw new Error('`!!!` is deprecated, you must now use `doctype`');
    }
    var node = this.scan(/^(?:doctype) *([^\n]+)?/, 'doctype');
    if (node && node.val && node.val.trim() === '5') {
      throw new Error('`doctype 5` is deprecated, you must now use `doctype html`');
    }
    return node;
  },

  /**
   * Id.
   */

  id: function() {
    return this.scan(/^#([\w-]+)/, 'id');
  },

  /**
   * Class.
   */

  className: function() {
    return this.scan(/^\.([\w-]+)/, 'class');
  },

  /**
   * Text.
   */

  text: function() {
    return this.scan(/^(?:\| ?| )([^\n]+)/, 'text') || this.scan(/^(<[^\n]*)/, 'text');
  },

  textFail: function () {
    var tok;
    if (tok = this.scan(/^([^\.\n][^\n]+)/, 'text')) {
      console.warn('Warning: missing space before text for line ' + this.lineno + 
          ' of jade file "' + this.filename + '"');
      return tok;
    }
  },

  /**
   * Dot.
   */

  dot: function() {
    return this.scan(/^\./, 'dot');
  },

  /**
   * Extends.
   */

  "extends": function() {
    return this.scan(/^extends? +([^\n]+)/, 'extends');
  },

  /**
   * Block prepend.
   */

  prepend: function() {
    var captures;
    if (captures = /^prepend +([^\n]+)/.exec(this.input)) {
      this.consume(captures[0].length);
      var mode = 'prepend'
        , name = captures[1]
        , tok = this.tok('block', name);
      tok.mode = mode;
      return tok;
    }
  },

  /**
   * Block append.
   */

  append: function() {
    var captures;
    if (captures = /^append +([^\n]+)/.exec(this.input)) {
      this.consume(captures[0].length);
      var mode = 'append'
        , name = captures[1]
        , tok = this.tok('block', name);
      tok.mode = mode;
      return tok;
    }
  },

  /**
   * Block.
   */

  block: function() {
    var captures;
    if (captures = /^block\b *(?:(prepend|append) +)?([^\n]+)/.exec(this.input)) {
      this.consume(captures[0].length);
      var mode = captures[1] || 'replace'
        , name = captures[2]
        , tok = this.tok('block', name);

      tok.mode = mode;
      return tok;
    }
  },

  /**
   * Mixin Block.
   */

  mixinBlock: function() {
    var captures;
    if (captures = /^block\s*(\n|$)/.exec(this.input)) {
      this.consume(captures[0].length - 1);
      return this.tok('mixin-block');
    }
  },

  /**
   * Yield.
   */

  yield: function() {
    return this.scan(/^yield */, 'yield');
  },

  /**
   * Include.
   */

  include: function() {
    return this.scan(/^include +([^\n]+)/, 'include');
  },

  /**
   * Include with filter
   */

  includeFiltered: function() {
    var captures;
    if (captures = /^include:([\w\-]+) +([^\n]+)/.exec(this.input)) {
      this.consume(captures[0].length);
      var filter = captures[1];
      var path = captures[2];
      var tok = this.tok('include', path);
      tok.filter = filter;
      return tok;
    }
  },

  /**
   * Case.
   */

  "case": function() {
    return this.scan(/^case +([^\n]+)/, 'case');
  },

  /**
   * When.
   */

  when: function() {
    return this.scan(/^when +([^:\n]+)/, 'when');
  },

  /**
   * Default.
   */

  "default": function() {
    return this.scan(/^default */, 'default');
  },

  /**
   * Call mixin.
   */

  call: function(){

    var tok, captures;
    if (captures = /^\+(([-\w]+)|(#\{))/.exec(this.input)) {
      // try to consume simple or interpolated call
      if (captures[2]) {
        // simple call
        this.consume(captures[0].length);
        tok = this.tok('call', captures[2]);
      } else {
        // interpolated call
        var match;
        try {
          match = this.bracketExpression(2);
        } catch (ex) {
          return;//not an interpolation expression, just an unmatched open interpolation
        }
        this.consume(match.end + 1);
        assertExpression(match.src);
        tok = this.tok('call', '#{'+match.src+'}');
      }

      // Check for args (not attributes)
      if (captures = /^ *\(/.exec(this.input)) {
        try {
          var range = this.bracketExpression(captures[0].length - 1);
          if (!/^ *[-\w]+ *=/.test(range.src)) { // not attributes
            this.consume(range.end + 1);
            tok.args = range.src;
          }
        } catch (ex) {
          //not a bracket expcetion, just unmatched open parens
        }
      }

      return tok;
    }
  },

  /**
   * Mixin.
   */

  mixin: function(){
    var captures;
    if (captures = /^mixin +([-\w]+)(?: *\((.*)\))? */.exec(this.input)) {
      this.consume(captures[0].length);
      var tok = this.tok('mixin', captures[1]);
      tok.args = captures[2];
      return tok;
    }
  },

  /**
   * Conditional.
   */

  conditional: function() {
    var captures;
    if (captures = /^(if|unless|else if|else)\b([^\n]*)/.exec(this.input)) {
      this.consume(captures[0].length);
      var type = captures[1]
        , js = captures[2];

      switch (type) {
        case 'if':
          assertExpression(js)
          js = 'if (' + js + ')';
          break;
        case 'unless':
          assertExpression(js)
          js = 'if (!(' + js + '))';
          break;
        case 'else if':
          assertExpression(js)
          js = 'else if (' + js + ')';
          break;
        case 'else':
          if (js && js.trim()) {
            throw new Error('`else` cannot have a condition, perhaps you meant `else if`');
          }
          js = 'else';
          break;
      }

      return this.tok('code', js);
    }
  },

  /**
   * While.
   */

  "while": function() {
    var captures;
    if (captures = /^while +([^\n]+)/.exec(this.input)) {
      this.consume(captures[0].length);
      assertExpression(captures[1])
      return this.tok('code', 'while (' + captures[1] + ')');
    }
  },

  /**
   * Each.
   */

  each: function() {
    var captures;
    if (captures = /^(?:- *)?(?:each|for) +([a-zA-Z_$][\w$]*)(?: *, *([a-zA-Z_$][\w$]*))? * in *([^\n]+)/.exec(this.input)) {
      this.consume(captures[0].length);
      var tok = this.tok('each', captures[1]);
      tok.key = captures[2] || '$index';
      assertExpression(captures[3])
      tok.code = captures[3];
      return tok;
    }
  },

  /**
   * Code.
   */

  code: function() {
    var captures;
    if (captures = /^(!?=|-)[ \t]*([^\n]+)/.exec(this.input)) {
      this.consume(captures[0].length);
      var flags = captures[1];
      captures[1] = captures[2];
      var tok = this.tok('code', captures[1]);
      tok.escape = flags.charAt(0) === '=';
      tok.buffer = flags.charAt(0) === '=' || flags.charAt(1) === '=';
      if (tok.buffer) assertExpression(captures[1])
      return tok;
    }
  },

  /**
   * Attributes.
   */

  attrs: function() {
    if ('(' == this.input.charAt(0)) {
      var index = this.bracketExpression().end
        , str = this.input.substr(1, index-1)
        , tok = this.tok('attrs');

      assertNestingCorrect(str);

      var quote = '';
      var interpolate = function (attr) {
        return attr.replace(/(\\)?#\{(.+)/g, function(_, escape, expr){
          if (escape) return _;
          try {
            var range = characterParser.parseMax(expr);
            if (expr[range.end] !== '}') return _.substr(0, 2) + interpolate(_.substr(2));
            assertExpression(range.src)
            return quote + " + (" + range.src + ") + " + quote + interpolate(expr.substr(range.end + 1));
          } catch (ex) {
            return _.substr(0, 2) + interpolate(_.substr(2));
          }
        });
      }

      this.consume(index + 1);
      tok.attrs = [];

      var escapedAttr = true
      var key = '';
      var val = '';
      var interpolatable = '';
      var state = characterParser.defaultState();
      var loc = 'key';
      var isEndOfAttribute = function (i) {
        if (key.trim() === '') return false;
        if (i === str.length) return true;
        if (loc === 'key') {
          if (str[i] === ' ' || str[i] === '\n') {
            for (var x = i; x < str.length; x++) {
              if (str[x] != ' ' && str[x] != '\n') {
                if (str[x] === '=' || str[x] === '!' || str[x] === ',') return false;
                else return true;
              }
            }
          }
          return str[i] === ','
        } else if (loc === 'value' && !state.isNesting()) {
          try {
            Function('', 'return (' + val + ');');
            if (str[i] === ' ' || str[i] === '\n') {
              for (var x = i; x < str.length; x++) {
                if (str[x] != ' ' && str[x] != '\n') {
                  if (characterParser.isPunctuator(str[x]) && str[x] != '"' && str[x] != "'") return false;
                  else return true;
                }
              }
            }
            return str[i] === ',';
          } catch (ex) {
            return false;
          }
        }
      }

      this.lineno += str.split("\n").length - 1;

      for (var i = 0; i <= str.length; i++) {
        if (isEndOfAttribute(i)) {
          val = val.trim();
          if (val) assertExpression(val)
          key = key.trim();
          key = key.replace(/^['"]|['"]$/g, '');
          tok.attrs.push({
            name: key,
            val: '' == val ? true : val,
            escaped: escapedAttr
          });
          key = val = '';
          loc = 'key';
          escapedAttr = false;
        } else {
          switch (loc) {
            case 'key-char':
              if (str[i] === quote) {
                loc = 'key';
                if (i + 1 < str.length && [' ', ',', '!', '=', '\n'].indexOf(str[i + 1]) === -1)
                  throw new Error('Unexpected character ' + str[i + 1] + ' expected ` `, `\\n`, `,`, `!` or `=`');
              } else if (loc === 'key-char') {
                key += str[i];
              }
              break;
            case 'key':
              if (key === '' && (str[i] === '"' || str[i] === "'")) {
                loc = 'key-char';
                quote = str[i];
              } else if (str[i] === '!' || str[i] === '=') {
                escapedAttr = str[i] !== '!';
                if (str[i] === '!') i++;
                if (str[i] !== '=') throw new Error('Unexpected character ' + str[i] + ' expected `=`');
                loc = 'value';
                state = characterParser.defaultState();
              } else {
                key += str[i]
              }
              break;
            case 'value':
              state = characterParser.parseChar(str[i], state);
              if (state.isString()) {
                loc = 'string';
                quote = str[i];
                interpolatable = str[i];
              } else {
                val += str[i];
              }
              break;
            case 'string':
              state = characterParser.parseChar(str[i], state);
              interpolatable += str[i];
              if (!state.isString()) {
                loc = 'value';
                val += interpolate(interpolatable);
              }
              break;
          }
        }
      }

      if ('/' == this.input.charAt(0)) {
        this.consume(1);
        tok.selfClosing = true;
      }

      return tok;
    }
  },

  /**
   * &attributes block
   */
  attributesBlock: function () {
    var captures;
    if (/^&attributes\b/.test(this.input)) {
      this.consume(11);
      var args = this.bracketExpression();
      this.consume(args.end + 1);
      return this.tok('&attributes', args.src);
    }
  },

  /**
   * Indent | Outdent | Newline.
   */

  indent: function() {
    var captures, re;

    // established regexp
    if (this.indentRe) {
      captures = this.indentRe.exec(this.input);
    // determine regexp
    } else {
      // tabs
      re = /^\n(\t*) */;
      captures = re.exec(this.input);

      // spaces
      if (captures && !captures[1].length) {
        re = /^\n( *)/;
        captures = re.exec(this.input);
      }

      // established
      if (captures && captures[1].length) this.indentRe = re;
    }

    if (captures) {
      var tok
        , indents = captures[1].length;

      ++this.lineno;
      this.consume(indents + 1);

      if (' ' == this.input[0] || '\t' == this.input[0]) {
        throw new Error('Invalid indentation, you can use tabs or spaces but not both');
      }

      // blank line
      if ('\n' == this.input[0]) return this.tok('newline');

      // outdent
      if (this.indentStack.length && indents < this.indentStack[0]) {
        while (this.indentStack.length && this.indentStack[0] > indents) {
          this.stash.push(this.tok('outdent'));
          this.indentStack.shift();
        }
        tok = this.stash.pop();
      // indent
      } else if (indents && indents != this.indentStack[0]) {
        this.indentStack.unshift(indents);
        tok = this.tok('indent', indents);
      // newline
      } else {
        tok = this.tok('newline');
      }

      return tok;
    }
  },

  /**
   * Pipe-less text consumed only when
   * pipeless is true;
   */

  pipelessText: function() {
    if (this.pipeless) {
      if ('\n' == this.input[0]) return;
      var i = this.input.indexOf('\n');
      if (-1 == i) i = this.input.length;
      var str = this.input.substr(0, i);
      this.consume(str.length);
      return this.tok('text', str);
    }
  },

  /**
   * ':'
   */

  colon: function() {
    return this.scan(/^: */, ':');
  },

  fail: function () {
    if (/^ ($|\n)/.test(this.input)) {
      this.consume(1);
      return this.next();
    }
    throw new Error('unexpected text ' + this.input.substr(0, 5));
  },

  /**
   * Return the next token object, or those
   * previously stashed by lookahead.
   *
   * @return {Object}
   * @api private
   */

  advance: function(){
    return this.stashed()
      || this.next();
  },

  /**
   * Return the next token object.
   *
   * @return {Object}
   * @api private
   */

  next: function() {
    return this.deferred()
      || this.blank()
      || this.eos()
      || this.pipelessText()
      || this.yield()
      || this.doctype()
      || this.interpolation()
      || this["case"]()
      || this.when()
      || this["default"]()
      || this["extends"]()
      || this.append()
      || this.prepend()
      || this.block()
      || this.mixinBlock()
      || this.include()
      || this.includeFiltered()
      || this.mixin()
      || this.call()
      || this.conditional()
      || this.each()
      || this["while"]()
      || this.tag()
      || this.filter()
      || this.code()
      || this.id()
      || this.className()
      || this.attrs()
      || this.attributesBlock()
      || this.indent()
      || this.text()
      || this.comment()
      || this.colon()
      || this.dot()
      || this.textFail()
      || this.fail();
  }
};