[Ericsson AB]





The Erlang Token Scanner


This module contains functions for tokenizing characters into Erlang tokens.


category() = atom()
column() = integer() > 0
line() = integer()
location() = line() | {line(), column()}
reserved_word_fun() -> fun(atom()) -> bool()
symbol() = atom() | float() | integer() | string()
token() = {category(), token_data()} | {category(), token_data(), symbol()}
token_data() = line() | tuple()


string(String) -> Return
string(String, StartLocation) -> Return
string(String, StartLocation, Options) -> Return


String = string()
Return = {ok, Tokens, EndLocation} | Error
Tokens = [token()]
Error = {error, ErrorInfo, EndLocation}
StartLocation = EndLocation = location()
Options = Option | [Option]
Option = {reserved_word_fun,reserved_word_fun()} | return_comments | return_white_spaces | return

Takes the list of characters String and tries to scan (tokenize) them. Returns {ok, Tokens, EndLocation}, where Tokens are the Erlang tokens from String. EndLocation is the first location after the last token.

{error, ErrorInfo, EndLocation} is returned if an error occurs. EndLocation is the first location after the erroneous token.

string(String) is equivalent to string(String, 1).

StartLocation indicates the initial location when scanning starts. If StartLocation is a line, token_data() as well as EndLocation and ErrorLocation will be lines. If StartLocation is a pair of a line and a column, token_data() takes the form of a tuple, and EndLocation and ErrorLocation will be pairs of a line and a column. The token data tuple contains information about the column and the line where the token begins, as well as the text of the token, all of which can be accessed by calling token_info/1,2.

A token is a tuple containing information about syntactic category, the token data, and the actual terminal symbol. For punctuation characters (e.g. ;, |) and reserved words, the category and the symbol coincide, and the token is represented by a two-tuple. Three-tuples have one of the following forms: {atom, Info, atom()}, {char, Info, integer()}, {comment, Info, string()}, {float, Info, float()}, {integer, Info, integer()}, {var, Info, atom()}, and {white_space, Info, string()}.

The valid options are:

{reserved_word_fun, reserved_word_fun()}
A callback function that is called when the scanner has found an unquoted atom. If the function returns true, the unquoted atom itself will be the category of the token; if the function returns false, atom will be the category of the unquoted atom.
Return comment tokens.
Return white space tokens. By convention, if there is a newline character, it is always the first character of the text (there cannot be more than one newline in a white space token).
Short for [return_comments, return_white_spaces].

tokens(Continuation, CharSpec, StartLocation) -> Return
tokens(Continuation, CharSpec, StartLocation, Options) -> Return


Continuation = [] | Continuation1
Return = {done, Result, LeftOverChars} | {more, Continuation1}
LeftOverChars = CharSpec
CharSpec = string() | eof
Continuation1 = tuple()
Result = {ok, Tokens, EndLocation} | {eof, EndLocation} | Error
Tokens = [token()]
Error = {error, ErrorInfo, EndLocation}
StartLocation = EndLocation = location()
Options = Option | [Option]
Option = {reserved_word_fun,reserved_word_fun()} | return_comments | return_white_spaces | return

This is the re-entrant scanner which scans characters until a dot ('.' followed by a white space) or eof has been reached. It returns:

{done, Result, LeftOverChars}
This return indicates that there is sufficient input data to get a result. Result is:
{ok, Tokens, EndLocation}
The scanning was successful. Tokens is the list of tokens including dot.
{eof, EndLocation}
End of file was encountered before any more tokens.
{error, ErrorInfo, EndLocation}
An error occurred.
{more, Continuation1}
More data is required for building a term. Continuation1 must be passed in a new call to tokens/3,4 when more data is available.

The CharSpec eof signals end of file. LeftoverChars will then take the value eof as well.

See string/3 for a description of the various options.

reserved_word(Atom) -> bool()


Atom = atom()

Returns true if Atom is an Erlang reserved word, otherwise false.

token_info(Token) -> InfoResult


Token = token()
InfoResult = [InfoTuple]
InfoTuple = {Item, Info}
Item = atom()
Info = term()

Returns a list containing information about the Erlang token Token. The order of the InfoTuples is not defined. Currently the following Items are returned: category, column, length, line, symbol, and text. See token_info/2 for information about specific InfoTuples.

Note that if token_info(Token, Item) returns undefined for some Item in the list above, the item is not included in InfoResult.

token_info(Token, ItemSpec) -> InfoResult


Token = token()
ItemSpec = Item | [Item]
InfoResult = InfoTuple | undefined | [InfoTuple]
InfoTuple = {Item, Info}
Item = atom()
Info = term()

Returns a list containing information about the Erlang token Token. If ItemSpec is a single Item, the returned value is the corresponding InfoTuple, or undefined if the Item has no value. If ItemSpec is a list of ItemSpec, the result is a list of InfoTuple. The InfoTuples will appear with the corresponding Items in the same order as the Items appeared in the list of Items. Items with no value are not included in the list of InfoTuple.

Currently the following InfoTuples with corresponding Items are valid:

{category, category()}
The category of the token.
{column, column()}
The column where the token begins.
{length, integer() > 0}
The length of the token's text.
{line, line()}
The line where the token begins.
{location, location()}
The line and column where the token begins, or just the line if the column unknown.
{symbol, symbol()}
The token's symbol.
{text, string()}
The token's text. The text is the part of the input corresponding to the token.

The token_info/1,2 functions may change in the R13B release.

format_error(ErrorDescriptor) -> string()


ErrorDescriptor = errordesc()

Takes an ErrorDescriptor and returns a string which describes the error or warning. This function is usually called implicitly when processing an ErrorInfo structure (see below).

Error Information

The ErrorInfo mentioned above is the standard ErrorInfo structure which is returned from all IO modules. It has the following format:

{ErrorLocation, Module, ErrorDescriptor}

A string which describes the error is obtained with the following call:



The continuation of the first call to the re-entrant input functions must be []. Refer to Armstrong, Virding and Williams, 'Concurrent Programming in Erlang', Chapter 13, for a complete description of how the re-entrant input scheme works.

See Also

io(3), erl_parse(3)

stdlib 1.16
Copyright © 1991-2009 Ericsson AB