Lex

Gem Version Build Status Code Climate Coverage Status Inline docs

Lex is an implementation of complier construction tool lex in Ruby. The goal is to stay close to the way the original tool works and combine it with the expressivness of Ruby.

Features

  • Very focused tool that mimics the basic lex functionality.
  • 100% Ruby implementation.
  • Provides comprehensive error reporting to assist in lexer construction.

Installation

Add this line to your application's Gemfile:

gem 'lex'

And then execute:

$ bundle

Or install it yourself as:

$ gem install lex

Contents

1. Overview

Lex is a library that processes character input streams. For example, suppose you have the following input string:

x = 5 + 44 * (s - t)

Lex then partitions the input string into tokens that match a series of regular expression rules. In this instance given the tokens definitions:

:ID, :EQUALS, :NUMBER, :PLUS, :TIMES, :LPAREN, :RPAREN, :MINUS

the output will contain the following tokens:

[:ID, 'x', 1, 1], [:EQUALS, '=', 1, 3], [:NUMBER, '5', 1, 5],
[:PLUS, '+', 1, 7], [:NUMBER, 44, 1, 9], [:TIMES, '*', 1, 12],
[:LPAREN, '(', 1, 14], [:ID, 's', 1, 15], [:MINUS, '-', 1, 17],
[:ID, 't', 1, 19], [:RPAREN, ')', 1, 20]

The Lex rules specified in the lexer will determine how the chunking of the input is performed. The following example demonstrates a high level overview of how this is done.

1.1 Example

Given an input:

input = "x = 5 + 44 * (s - t)"

and a simple tokenizer:

class MyLexer < Lex::Lexer
  tokens(
    :NUMBER,
    :PLUS,
    :MINUS,
    :TIMES,
    :DIVIDE,
    :LPAREN,
    :RPAREN,
    :EQUALS,
    :ID
  )

  # Regular expression rules for simple tokens
  rule(:PLUS,   /\+/)
  rule(:MINUS,  /\-/)
  rule(:TIMES,  /\*/)
  rule(:DIVIDE, /\//)
  rule(:LPAREN, /\(/)
  rule(:RPAREN, /\)/)
  rule(:ID,     /\A[_\$a-zA-Z][_\$0-9a-zA-Z]*/)

  # A regular expression rules with actions
  rule(:NUMBER, /[0-9]+/) do |lexer, token|
    token.value = token.value.to_i
    token
  end

  # Define a rule so we can track line numbers
  rule(:newline, /\n+/) do |lexer, token|
    lexer.advance_line(token.value.length)
  end

  # A string containing ignored characters (spaces and tabs)
  ignore " \t"

  error do |lexer, token|
    puts "Illegal character: #{value}"
  end
end

# build the lexer
my_lexer = MyLexer.new

To use the lexer you need to provide it some input using the lex method. After that, the method lex will either yield tokens to a given block or return an enumereator to allow you to retrieve tokens by repeatedly calling next method.

output = my_lexer.lex(input)
output.next  # =>  Lex::Token(ID,x,1,1)
output.next  # =>  Lex::Token(EQUALS,=,1,3)
output.next  # =>  Lex::Token(NUMBER,5,1,5)
...

The tokens returned by the lexer are instances of Lex::Token. This object has attributes such as name, value, line and column.

1.2 Tokens list

A lexer always requires a list of tokens that define all the possible token names that can be produced by the lexer. This list is used to perform validation checks.

The following list is an example of token names:

tokens(
  :NUMBER,
  :PLUS,
  :MINUS,
  :TIMES,
  :DIVIDE,
  :LPAREN,
  :RPAREN
)

1.3 Specifying rules

Each token is specified by writting a regular expression rule defined by by calling the rule method. For simple tokens you can just specify the name and regular expression:

rule(:PLUS, /\+/)

In this case, the first argument is the name of the token that needs to match exactly one of the names supplied in tokens. If you need to perform further processing on the matched token, the rule can be further expaned by adding an action inside a block. For instance, this rule matches numbers and converts the matched string into integer type:

token(:NUMBER, /\d+/) do |lexer, token|
  token.value = token.value.to_i
  token
end

The action block always takes two arguments, the first being the lexer itself and the second the token which is an instance of Lex::Token. This object has attributes of name which is the token name as string, value which is the actual text matched, line which is the current line indexed from 1, column which is the position of the token in relation to the current line. By default the name is set to the rule name. Inside the block you can modify the token object properties. However, when you change token properties, the token itself needs to be returned. If no value is returned by the action block, the token is simply discarded and lexer moves to another token.

The rules are processed in the same order as they appear in the lexer definition. Therefore, if you wanted to have a separate tokens for "=" and "==", you need to ensure that rule for matching "==" is checked first.

1.4 Handling keywords

In order to handle keywords, you should write a single rule to match an identifier and then do a name lookup like so:

def self.keywords
  {
    if: :IF,
    then: :THEN,
    else: :ELSE,
    while: WHILE,
    ...
  }
end

tokens(:ID, *keywords.values)

rule(:ID, /\w[\w\d]*/) do |lexer, token|
  token.name = lexer.class.keywords.fetch(token.value.to_sym, :ID)
  token
end

1.5 Token values

By default token value is the text that was matched by the rule. However, the token value can be changed to any object. For example, when processing identifiers you may wish to return both identifier name and actual value.

rule(:ID, /\w[\w\d]*/) do |lexer, token|
  token.value = [token.value, lexer.class.keywords[token.value]]
  token
end

1.6 Discarded tokens

To discard a token, such as comment, define a rule that returns no token. For instance:

rule(:COMMENT, /\#.*/) do |lexer, token|
  # No return value. Token is discarded.
end

1.7 Line numbers

By default Lex knows nothing about line numbers since it doesn't understand what a "line" is. To provide this information you need to add a special rule called :newline:

rule(:newline, /\n+/) do |lexer, token|
  lexer.advance_line(token.value.length)
end

Calling the advance_line method the current_line is updated for the underlying lexer. Only the line is updated and since no token is returned the value is discarded.

Lex performs automatic column tracking for each token. This information is available by calling column on a Lex::Token instance.

1.8 Ignored characters

For any character that should be completely ignored in the input stream use the ignore rule. Usually this is used to skip over whitespace and other non-essential characters. For example:

ignore = " \t" # => Ignore whitespace and tabs

You could create a rule to achieve similar behaviour, however you are encourage to use this method as it has increased performance over the rule regular expression matching.

1.9 Literal characters

Not implemented yet!

1.10 Error handling

In order to handle lexing error conditions use the error method. In this case thetoken value attribute contains the offending string. For example:

error do |lexer, token|
  puts "Illegal character #{token.value}"
end

The lexer automatically skips the offending character and increments the column count.

When performing conditional lexing, you can handle errors per state like so:

error :foo do |lexer, token|
  puts "Illegal character #{token.value}"
end

1.11 Building the lexer

require 'lex'

class MyLexer < Lex::Lexer
  # required list of tokens
  tokens(
    :NUMBER,
  )
  ...
end

You can also provide lexer definition by using block:

my_lexer = Lex::Lexer.new do
  # required list of tokens
  tokens(
    :NUMBER,
  )
end

1.12 Maintaining state

In your lexer you may have a need to store state information.

1.13 Conditional lexing

A lexer can maintain internal lexing state. When lexer's state changes, the corresponding tokens for that state are only considered. The start condition is called :initial, similar to GNU flex.

To define a new lexical state, it must first be declared. This can be achieved by using a states declaration:

states(
  foo: :exclusive,
  bar: :inclusive
)

The above definition declares two states :foo and :bar. State may be of two types :exclusive and :inclusive. In an :exclusive state lexer contains no rules, which means that Lex will only return tokens and apply rules defined specifically for that state. On the other hand, an :inclusive state adds additional tokens and rules to the default set of rules. Thus, lex method will return both the tokens defined by default in addition to those defined specificially for the :inclusive state.

Once state has been declared, tokens and rules are declared by including the state name in token or rule definition. For example:

rule(:foo_NUMBER, /\d+/)
rule(:bar_ID, /[a-z][a-z0-9]+/)

The above rules define :NUMBER token in state :foo and :ID token in state :bar.

A token can be specified in multiple states by prefixing token name by state names like so:

rule(:foo_bar_NUMBER, /\d+/)

If no state information is provided, the lexer is assumed to be in :initial state. For example, the following declarations are equivalent:

rule(:NUMBER, /\d+/)
rule(:initial_NUMBER, /\d+/)

By default, lexing operates in :initial state. All the normally defined tokens are included in this state. During lexing if you wish to change the lexing state use the begin method. For example:

rule(:begin_foo, /start_foo/) do |lexer, token|
  lexer.begin(:foo)
end

To get out of state you can use begin like so:

rule(:foo_end, /end_foo/) do |lexer, token|
  lexer.begin(:initial)
end

For more complex scenarios with states you can use push_state and pop_state methods. For example:

rule(:begin_foo, /start_foo/) do |lexer, token|
  lexer.push_state(:foo)
end

rule(:foo_end, /end_foo/) do |lexer, token|
  lexer.pop_state(:foo)
end

Assume you are parsing HTML and you want to ignore anything inside comment. Here is how you may use lexer states to do this:

class MyLexer < Lex::Lexer
  tokens(
    :TAG,
    :ATTRIBUTE
  )

  # Declare the states
  states( htmlcomment: :exclusive )

  # Enter html comment
  rule(:begin_htmlcomment, /<!--/) do |lexer, token|
    lexer.begin(:htmlcomment)
  end

  # Leave html comment
  rule(:htmlcomment, /-->/) do |lexer, token|
    lexer.begin(:initial)
  end

  error :htmlcomment do |lexer, token|
    lexer.logger.info "Ignoring character #{token.value}"
  end

  ignore :htmlcomment, " \t\n"

  ignore " \t"
end

1.14 Debugging

In order to run lexer in debug mode pass in :debug flag set to true.

MyLexer.new(debug: true)

Contributing

  1. Fork it ( https://github.com/piotrmurach/lex/fork )
  2. Create your feature branch (git checkout -b my-new-feature)
  3. Commit your changes (git commit -am 'Add some feature')
  4. Push to the branch (git push origin my-new-feature)
  5. Create a new Pull Request

Copyright

Copyright (c) 2015 Piotr Murach. See LICENSE for further details.