Jump To …

lexer.coffee

The CoffeeScript Lexer. Uses a series of token-matching regexes to attempt matches against the beginning of the source code. When a match is found, a token is produced, we consume the match, and start again. Tokens are in the form:

[tag, value, lineNumber]

Which is a format that can be fed directly into Jison.

{Rewriter} = require './rewriter'

Import the helpers we need.

{count, starts, compact, last} = require './helpers'

The Lexer Class

The Lexer class reads a stream of CoffeeScript and divvies it up into tagged tokens. Some potential ambiguity in the grammar has been avoided by pushing some extra smarts into the Lexer.

exports.Lexer = class Lexer

tokenize is the Lexer's main method. Scan by attempting to match tokens one at a time, using a regular expression anchored at the start of the remaining code, or a custom recursive token-matching method (for interpolations). When the next token has been recorded, we move forward within the code past the token, and begin again.

Each tokenizing method is responsible for returning the number of characters it has consumed.

Before returning the token stream, run it through the Rewriter unless explicitly asked not to.

  tokenize: (code, opts = {}) ->
    code     = "\n#{code}" if WHITESPACE.test code
    code     = code.replace(/\r/g, '').replace TRAILING_SPACES, ''

    @code    = code           # The remainder of the source code.
    @line    = opts.line or 0 # The current line.
    @indent  = 0              # The current indentation level.
    @indebt  = 0              # The over-indentation at the current level.
    @outdebt = 0              # The under-outdentation at the current level.
    @indents = []             # The stack of all current indentation levels.
    @tokens  = []             # Stream of parsed tokens in the form `['TYPE', value, line]`.

At every position, run through this list of attempted matches, short-circuiting if any of them succeed. Their order determines precedence: @literalToken is the fallback catch-all.

    i = 0
    while @chunk = code.slice i
      i += @identifierToken() or
           @commentToken()    or
           @whitespaceToken() or
           @lineToken()       or
           @heredocToken()    or
           @stringToken()     or
           @numberToken()     or
           @regexToken()      or
           @jsToken()         or
           @literalToken()

    @closeIndentation()
    return @tokens if opts.rewrite is off
    (new Rewriter).rewrite @tokens

Tokenizers

Matches identifying literals: variables, keywords, method names, etc. Check to ensure that JavaScript reserved words aren't being used as identifiers. Because CoffeeScript reserves a handful of keywords that are allowed in JavaScript, we're careful not to tag them as keywords when referenced as property names here, so you can still do jQuery.is() even though is means === otherwise.

  identifierToken: ->
    return 0 unless match = IDENTIFIER.exec @chunk
    [input, id, colon] = match

    if id is 'own' and @tag() is 'FOR'
      @token 'OWN', id
      return id.length
    forcedIdentifier = colon or
      (prev = last @tokens) and (prev[0] in ['.', '?.', '::'] or
      not prev.spaced and prev[0] is '@')
    tag = 'IDENTIFIER'

    if id in JS_KEYWORDS or
       not forcedIdentifier and id in COFFEE_KEYWORDS
      tag = id.toUpperCase()
      if tag is 'WHEN' and @tag() in LINE_BREAK
        tag = 'LEADING_WHEN'
      else if tag is 'FOR'
        @seenFor = yes
      else if tag is 'UNLESS'
        tag = 'IF'
      else if tag in UNARY
        tag = 'UNARY'
      else if tag in RELATION
        if tag isnt 'INSTANCEOF' and @seenFor
          tag = 'FOR' + tag
          @seenFor = no
        else
          tag = 'RELATION'
          if @value() is '!'
            @tokens.pop()
            id = '!' + id

    if id in JS_FORBIDDEN
      if forcedIdentifier
        tag = 'IDENTIFIER'
        id  = new String id
        id.reserved = yes
      else if id in RESERVED
        @identifierError id

    unless forcedIdentifier
      id  = COFFEE_ALIAS_MAP[id] if id in COFFEE_ALIASES
      tag = switch id
        when '!'                                  then 'UNARY'
        when '==', '!='                           then 'COMPARE'
        when '&&', '||'                           then 'LOGIC'
        when 'true', 'false', 'null', 'undefined' then 'BOOL'
        when 'break', 'continue', 'debugger'      then 'STATEMENT'
        else  tag

    @token tag, id
    @token ':', ':' if colon
    input.length

Matches numbers, including decimals, hex, and exponential notation. Be careful not to interfere with ranges-in-progress.

  numberToken: ->
    return 0 unless match = NUMBER.exec @chunk
    number = match[0]
    @token 'NUMBER', number
    number.length

Matches strings, including multi-line strings. Ensures that quotation marks are balanced within the string's contents, and within nested interpolations.

  stringToken: ->
    switch @chunk.charAt 0
      when "'"
        return 0 unless match = SIMPLESTR.exec @chunk
        @token 'STRING', (string = match[0]).replace MULTILINER, '\\\n'
      when '"'
        return 0 unless string = @balancedString @chunk, '"'
        if 0 < string.indexOf '#{', 1
          @interpolateString string.slice 1, -1
        else
          @token 'STRING', @escapeLines string
      else
        return 0
    @line += count string, '\n'
    string.length

Matches heredocs, adjusting indentation to the correct level, as heredocs preserve whitespace, but ignore indentation to the left.

  heredocToken: ->
    return 0 unless match = HEREDOC.exec @chunk
    heredoc = match[0]
    quote = heredoc.charAt 0
    doc = @sanitizeHeredoc match[2], quote: quote, indent: null
    if quote is '"' and 0 <= doc.indexOf '#{'
      @interpolateString doc, heredoc: yes
    else
      @token 'STRING', @makeString doc, quote, yes
    @line += count heredoc, '\n'
    heredoc.length

Matches and consumes comments.

  commentToken: ->
    return 0 unless match = @chunk.match COMMENT
    [comment, here] = match
    if here
      @token 'HERECOMMENT', @sanitizeHeredoc here,
        herecomment: true, indent: Array(@indent + 1).join(' ')
      @token 'TERMINATOR', '\n'
    @line += count comment, '\n'
    comment.length

Matches JavaScript interpolated directly into the source via backticks.

  jsToken: ->
    return 0 unless @chunk.charAt(0) is '`' and match = JSTOKEN.exec @chunk
    @token 'JS', (script = match[0]).slice 1, -1
    script.length

Matches regular expression literals. Lexing regular expressions is difficult to distinguish from division, so we borrow some basic heuristics from JavaScript and Ruby.

  regexToken: ->
    return 0 if @chunk.charAt(0) isnt '/'
    return @heregexToken match if match = HEREGEX.exec @chunk
    prev = last @tokens
    return 0 if prev and (prev[0] in (if prev.spaced then NOT_REGEX else NOT_SPACED_REGEX))
    return 0 unless match = REGEX.exec @chunk
    [regex] = match
    @token 'REGEX', if regex is '//' then '/(?:)/' else regex
    regex.length

Matches multiline extended regular expressions.

  heregexToken: (match) ->
    [heregex, body, flags] = match
    if 0 > body.indexOf '#{'
      re = body.replace(HEREGEX_OMIT, '').replace(/\//g, '\\/')
      @token 'REGEX', "/#{ re or '(?:)' }/#{flags}"
      return heregex.length
    @token 'IDENTIFIER', 'RegExp'
    @tokens.push ['CALL_START', '(']
    tokens = []
    for [tag, value] in @interpolateString(body, regex: yes)
      if tag is 'TOKENS'
        tokens.push value...
      else
        continue unless value = value.replace HEREGEX_OMIT, ''
        value = value.replace /\\/g, '\\\\'
        tokens.push ['STRING', @makeString(value, '"', yes)]
      tokens.push ['+', '+']
    tokens.pop()
    @tokens.push ['STRING', '""'], ['+', '+'] unless tokens[0]?[0] is 'STRING'
    @tokens.push tokens...
    @tokens.push [',', ','], ['STRING', '"' + flags + '"'] if flags
    @token ')', ')'
    heregex.length

Matches newlines, indents, and outdents, and determines which is which. If we can detect that the current line is continued onto the the next line, then the newline is suppressed:

elements
  .each( ... )
  .map( ... )

Keeps track of the level of indentation, because a single outdent token can close multiple indents, so we need to know how far in we happen to be.

  lineToken: ->
    return 0 unless match = MULTI_DENT.exec @chunk
    indent = match[0]
    @line += count indent, '\n'
    prev = last @tokens, 1
    size = indent.length - 1 - indent.lastIndexOf '\n'
    noNewlines = @unfinished()
    if size - @indebt is @indent
      if noNewlines then @suppressNewlines() else @newlineToken()
      return indent.length
    if size > @indent
      if noNewlines
        @indebt = size - @indent
        @suppressNewlines()
        return indent.length
      diff = size - @indent + @outdebt
      @token 'INDENT', diff
      @indents.push diff
      @outdebt = @indebt = 0
    else
      @indebt = 0
      @outdentToken @indent - size, noNewlines
    @indent = size
    indent.length

Record an outdent token or multiple tokens, if we happen to be moving back inwards past several recorded indents.

  outdentToken: (moveOut, noNewlines, close) ->
    while moveOut > 0
      len = @indents.length - 1
      if @indents[len] is undefined
        moveOut = 0
      else if @indents[len] is @outdebt
        moveOut -= @outdebt
        @outdebt = 0
      else if @indents[len] < @outdebt
        @outdebt -= @indents[len]
        moveOut  -= @indents[len]
      else
        dent = @indents.pop() - @outdebt
        moveOut -= dent
        @outdebt = 0
        @token 'OUTDENT', dent
    @outdebt -= moveOut if dent
    @token 'TERMINATOR', '\n' unless @tag() is 'TERMINATOR' or noNewlines
    this

Matches and consumes non-meaningful whitespace. Tag the previous token as being "spaced", because there are some cases where it makes a difference.

  whitespaceToken: ->
    return 0 unless (match = WHITESPACE.exec @chunk) or
                    (nline = @chunk.charAt(0) is '\n')
    prev = last @tokens
    prev[if match then 'spaced' else 'newLine'] = true if prev
    if match then match[0].length else 0

Generate a newline token. Consecutive newlines get merged together.

  newlineToken: ->
    @token 'TERMINATOR', '\n' unless @tag() is 'TERMINATOR'
    this

Use a \ at a line-ending to suppress the newline. The slash is removed here once its job is done.

  suppressNewlines: ->
    @tokens.pop() if @value() is '\\'
    this

We treat all other single characters as a token. E.g.: ( ) , . ! Multi-character operators are also literal tokens, so that Jison can assign the proper order of operations. There are some symbols that we tag specially here. ; and newlines are both treated as a TERMINATOR, we distinguish parentheses that indicate a method call from regular parentheses, and so on.

  literalToken: ->
    if match = OPERATOR.exec @chunk
      [value] = match
      @tagParameters() if CODE.test value
    else
      value = @chunk.charAt 0
    tag  = value
    prev = last @tokens
    if value is '=' and prev
      @assignmentError() if not prev[1].reserved and prev[1] in JS_FORBIDDEN
      if prev[1] in ['||', '&&']
        prev[0] = 'COMPOUND_ASSIGN'
        prev[1] += '='
        return value.length
    if      value is ';'             then tag = 'TERMINATOR'
    else if value in MATH            then tag = 'MATH'
    else if value in COMPARE         then tag = 'COMPARE'
    else if value in COMPOUND_ASSIGN then tag = 'COMPOUND_ASSIGN'
    else if value in UNARY           then tag = 'UNARY'
    else if value in SHIFT           then tag = 'SHIFT'
    else if value in LOGIC or value is '?' and prev?.spaced then tag = 'LOGIC'
    else if prev and not prev.spaced
      if value is '(' and prev[0] in CALLABLE
        prev[0] = 'FUNC_EXIST' if prev[0] is '?'
        tag = 'CALL_START'
      else if value is '[' and prev[0] in INDEXABLE
        tag = 'INDEX_START'
        switch prev[0]
          when '?'  then prev[0] = 'INDEX_SOAK'
          when '::' then prev[0] = 'INDEX_PROTO'
    @token tag, value
    value.length

Token Manipulators

Sanitize a heredoc or herecomment by erasing all external indentation on the left-hand side.

  sanitizeHeredoc: (doc, options) ->
    {indent, herecomment} = options
    if herecomment
      if HEREDOC_ILLEGAL.test doc
        throw new Error "block comment cannot contain \"*/\", starting on line #{@line + 1}"
      return doc if doc.indexOf('\n') <= 0
    else
      while match = HEREDOC_INDENT.exec doc
        attempt = match[1]
        indent = attempt if indent is null or 0 < attempt.length < indent.length
    doc = doc.replace /// \n #{indent} ///g, '\n' if indent
    doc = doc.replace /^\n/, '' unless herecomment
    doc

A source of ambiguity in our grammar used to be parameter lists in function definitions versus argument lists in function calls. Walk backwards, tagging parameters specially in order to make things easier for the parser.

  tagParameters: ->
    return this if @tag() isnt ')'
    stack = []
    {tokens} = this
    i = tokens.length
    tokens[--i][0] = 'PARAM_END'
    while tok = tokens[--i]
      switch tok[0]
        when ')'
          stack.push tok
        when '(', 'CALL_START'
          if stack.length then stack.pop()
          else if tok[0] is '('
            tok[0] = 'PARAM_START'
            return this
    this

Close up all remaining open blocks at the end of the file.

  closeIndentation: ->
    @outdentToken @indent

The error for when you try to use a forbidden word in JavaScript as an identifier.

  identifierError: (word) ->
    throw SyntaxError "Reserved word \"#{word}\" on line #{@line + 1}"

The error for when you try to assign to a reserved word in JavaScript, like "function" or "default".

  assignmentError: ->
    throw SyntaxError "Reserved word \"#{@value()}\" on line #{@line + 1} can't be assigned"

Matches a balanced group such as a single or double-quoted string. Pass in a series of delimiters, all of which must be nested correctly within the contents of the string. This method allows us to have strings within interpolations within strings, ad infinitum.

  balancedString: (str, end) ->
    stack = [end]
    for i in [1...str.length]
      switch letter = str.charAt i
        when '\\'
          i++
          continue
        when end
          stack.pop()
          unless stack.length
            return str.slice 0, i + 1
          end = stack[stack.length - 1]
          continue
      if end is '}' and letter in ['"', "'"]
        stack.push end = letter
      else if end is '}' and letter is '{'
        stack.push end = '}'
      else if end is '"' and prev is '#' and letter is '{'
        stack.push end = '}'
      prev = letter
    throw new Error "missing #{ stack.pop() }, starting on line #{ @line + 1 }"

Expand variables and expressions inside double-quoted strings using Ruby-like notation for substitution of arbitrary expressions.

"Hello #{name.capitalize()}."

If it encounters an interpolation, this method will recursively create a new Lexer, tokenize the interpolated contents, and merge them into the token stream.

  interpolateString: (str, options = {}) ->
    {heredoc, regex} = options
    tokens = []
    pi = 0
    i  = -1
    while letter = str.charAt i += 1
      if letter is '\\'
        i += 1
        continue
      unless letter is '#' and str.charAt(i+1) is '{' and
             (expr = @balancedString str.slice(i + 1), '}')
        continue
      tokens.push ['NEOSTRING', str.slice(pi, i)] if pi < i
      inner = expr.slice(1, -1)
      if inner.length
        nested = new Lexer().tokenize inner, line: @line, rewrite: off
        nested.pop()
        nested.shift() if nested[0]?[0] is 'TERMINATOR'
        if len = nested.length
          if len > 1
            nested.unshift ['(', '(']
            nested.push    [')', ')']
          tokens.push ['TOKENS', nested]
      i += expr.length
      pi = i + 1
    tokens.push ['NEOSTRING', str.slice pi] if i > pi < str.length
    return tokens if regex
    return @token 'STRING', '""' unless tokens.length
    tokens.unshift ['', ''] unless tokens[0][0] is 'NEOSTRING'
    @token '(', '(' if interpolated = tokens.length > 1
    for [tag, value], i in tokens
      @token '+', '+' if i
      if tag is 'TOKENS'
        @tokens.push value...
      else
        @token 'STRING', @makeString value, '"', heredoc
    @token ')', ')' if interpolated
    tokens

Helpers

Add a token to the results, taking note of the line number.

  token: (tag, value) ->
    @tokens.push [tag, value, @line]

Peek at a tag in the current token stream.

  tag: (index, tag) ->
    (tok = last @tokens, index) and if tag then tok[0] = tag else tok[0]

Peek at a value in the current token stream.

  value: (index, val) ->
    (tok = last @tokens, index) and if val then tok[1] = val else tok[1]

Are we in the midst of an unfinished expression?

  unfinished: ->
    LINE_CONTINUER.test(@chunk) or
    (prev = last @tokens, 1) and prev[0] isnt '.' and
      (value = @value()) and not value.reserved and
      NO_NEWLINE.test(value) and not CODE.test(value) and not ASSIGNED.test(@chunk)

Converts newlines for string literals.

  escapeLines: (str, heredoc) ->
    str.replace MULTILINER, if heredoc then '\\n' else ''

Constructs a string token by escaping quotes and newlines.

  makeString: (body, quote, heredoc) ->
    return quote + quote unless body
    body = body.replace /\\([\s\S])/g, (match, contents) ->
      if contents in ['\n', quote] then contents else match
    body = body.replace /// #{quote} ///g, '\\$&'
    quote + @escapeLines(body, heredoc) + quote

Constants

Keywords that CoffeeScript shares in common with JavaScript.

JS_KEYWORDS = [
  'true', 'false', 'null', 'this'
  'new', 'delete', 'typeof', 'in', 'instanceof'
  'return', 'throw', 'break', 'continue', 'debugger'
  'if', 'else', 'switch', 'for', 'while', 'do', 'try', 'catch', 'finally'
  'class', 'extends', 'super'
]

CoffeeScript-only keywords.

COFFEE_KEYWORDS = ['undefined', 'then', 'unless', 'until', 'loop', 'of', 'by', 'when']

COFFEE_ALIAS_MAP =
  and  : '&&'
  or   : '||'
  is   : '=='
  isnt : '!='
  not  : '!'
  yes  : 'true'
  no   : 'false'
  on   : 'true'
  off  : 'false'
  
COFFEE_ALIASES  = (key for key of COFFEE_ALIAS_MAP)
COFFEE_KEYWORDS = COFFEE_KEYWORDS.concat COFFEE_ALIASES

The list of keywords that are reserved by JavaScript, but not used, or are used by CoffeeScript internally. We throw an error when these are encountered, to avoid having a JavaScript error at runtime.

RESERVED = [
  'case', 'default', 'function', 'var', 'void', 'with'
  'const', 'let', 'enum', 'export', 'import', 'native'
  '__hasProp', '__extends', '__slice', '__bind', '__indexOf'
]

The superset of both JavaScript keywords and reserved words, none of which may be used as identifiers or properties.

JS_FORBIDDEN = JS_KEYWORDS.concat RESERVED

exports.RESERVED = RESERVED.concat(JS_KEYWORDS).concat(COFFEE_KEYWORDS)

Token matching regexes.

IDENTIFIER = /// ^
  ( [$A-Za-z_\x7f-\uffff][$\w\x7f-\uffff]* )
  ( [^\n\S]* : (?!:) )?  # Is this a property name?
///

NUMBER     = ///
  ^ 0x[\da-f]+ |                              # hex
  ^ (?: \d+(\.\d+)? | \.\d+ ) (?:e[+-]?\d+)?  # decimal
///i

HEREDOC    = /// ^ ("""|''') ([\s\S]*?) (?:\n[^\n\S]*)? \1 ///

OPERATOR   = /// ^ (
  ?: [-=]>             # function
   | [-+*/%<>&|^!?=]=  # compound assign / compare
   | >>>=?             # zero-fill right shift
   | ([-+:])\1         # doubles
   | ([&|<>])\2=?      # logic / shift
   | \?\.              # soak access
   | \.{2,3}           # range or splat
) ///

WHITESPACE = /^[^\n\S]+/

COMMENT    = /^###([^#][\s\S]*?)(?:###[^\n\S]*|(?:###)?$)|^(?:\s*#(?!##[^#]).*)+/

CODE       = /^[-=]>/

MULTI_DENT = /^(?:\n[^\n\S]*)+/

SIMPLESTR  = /^'[^\\']*(?:\\.[^\\']*)*'/

JSTOKEN    = /^`[^\\`]*(?:\\.[^\\`]*)*`/

Regex-matching-regexes.

REGEX = /// ^
  / (?! [\s=] )       # disallow leading whitespace or equals signs
  [^ [ / \n \\ ]*  # every other thing
  (?:
    (?: \\[\s\S]   # anything escaped
      | \[         # character class
           [^ \] \n \\ ]*
           (?: \\[\s\S] [^ \] \n \\ ]* )*
         ]
    ) [^ [ / \n \\ ]*
  )*
  / [imgy]{0,4} (?!\w)
///

HEREGEX      = /// ^ /{3} ([\s\S]+?) /{3} ([imgy]{0,4}) (?!\w) ///

HEREGEX_OMIT = /\s+(?:#.*)?/g

Token cleaning regexes.

MULTILINER      = /\n/g

HEREDOC_INDENT  = /\n+([^\n\S]*)/g

HEREDOC_ILLEGAL = /\*\//

ASSIGNED        = /^\s*@?([$A-Za-z_][$\w\x7f-\uffff]*|['"].*['"])[^\n\S]*?[:=][^:=>]/

LINE_CONTINUER  = /// ^ \s* (?: , | \??\.(?![.\d]) | :: ) ///

TRAILING_SPACES = /\s+$/

NO_NEWLINE      = /// ^ (?:            # non-capturing group
  [-+*&|/%=<>!.\\][<>=&|]* |           # symbol operators
  and | or | is(?:nt)? | n(?:ot|ew) |  # word operators
  delete | typeof | instanceof
) $ ///

Compound assignment tokens.

COMPOUND_ASSIGN = [
  '-=', '+=', '/=', '*=', '%=', '||=', '&&=', '?=', '<<=', '>>=', '>>>=', '&=', '^=', '|='
]

Unary tokens.

UNARY   = ['!', '~', 'NEW', 'TYPEOF', 'DELETE', 'DO']

Logical tokens.

LOGIC   = ['&&', '||', '&', '|', '^']

Bit-shifting tokens.

SHIFT   = ['<<', '>>', '>>>']

Comparison tokens.

COMPARE = ['==', '!=', '<', '>', '<=', '>=']

Mathematical tokens.

MATH    = ['*', '/', '%']

Relational tokens that are negatable with not prefix.

RELATION = ['IN', 'OF', 'INSTANCEOF']

Boolean tokens.

BOOL = ['TRUE', 'FALSE', 'NULL', 'UNDEFINED']

Tokens which a regular expression will never immediately follow, but which a division operator might.

See: http://www.mozilla.org/js/language/js20-2002-04/rationale/syntax.html#regular-expressions

Our list is shorter, due to sans-parentheses method calls.

NOT_REGEX = ['NUMBER', 'REGEX', 'BOOL', '++', '--', ']']

If the previous token is not spaced, there are more preceding tokens that force a division parse:

NOT_SPACED_REGEX = NOT_REGEX.concat ')', '}', 'THIS', 'IDENTIFIER', 'STRING'

Tokens which could legitimately be invoked or indexed. A opening parentheses or bracket following these tokens will be recorded as the start of a function invocation or indexing operation.

CALLABLE  = ['IDENTIFIER', 'STRING', 'REGEX', ')', ']', '}', '?', '::', '@', 'THIS', 'SUPER']
INDEXABLE = CALLABLE.concat 'NUMBER', 'BOOL'

Tokens that, when immediately preceding a WHEN, indicate that the WHEN occurs at the start of a line. We disambiguate these from trailing whens to avoid an ambiguity in the grammar.

LINE_BREAK = ['INDENT', 'OUTDENT', 'TERMINATOR']