class SexpProcessor

SexpProcessor provides a uniform interface to process Sexps.

In order to create your own SexpProcessor subclass you’ll need to call super in the initialize method, then set any of the Sexp flags you want to be different from the defaults.

SexpProcessor uses a Sexp’s type to determine which process method to call in the subclass. For Sexp s(:lit, 1) SexpProcessor will call process_lit, if it is defined.

You can also specify a default method to call for any Sexp types without a process_<type> method or use the default processor provided to skip over them.

Here is a simple example:

class MyProcessor < SexpProcessor
  def initialize
    super
    self.strict = false
  end

  def process_lit(exp)
    val = exp.shift
    return val
  end
end

Constants

VERSION

duh

Attributes

auto_shift_type[RW]

Automatically shifts off the Sexp type before handing the Sexp to process_<type>

context[R]

Return a stack of contexts. Most recent node is first.

debug[RW]

A Hash of Sexp types and Regexp.

Print a debug message if the Sexp type matches the Hash key and the Sexp’s inspect output matches the Regexp.

default_method[RW]

A default method to call if a process_<type> method is not found for the Sexp type.

env[R]

A scoped environment to make you happy.

expected[RW]

Expected result class

require_empty[RW]

Raise an exception if the Sexp is not empty after processing

strict[RW]

Raise an exception if no process_<type> method is found for a Sexp.

unsupported[RW]

An array that specifies node types that are unsupported by this processor. SexpProcessor will raise UnsupportedNodeError if you try to process one of those node types.

warn_on_default[RW]

Emit a warning when the method in default_method is called.

Public Class Methods

expand_dirs_to_files(*dirs) click to toggle source

Expand an array of directories into a flattened array of paths, eg:

MyProcessor.run MyProcessor.expand_dirs_to_files ARGV
# File lib/sexp_processor.rb, line 101
def self.expand_dirs_to_files *dirs
  extensions = %w[rb rake]

  dirs.flatten.map { |p|
    if File.directory? p then
      Dir[File.join(p, "**", "*.{#{extensions.join ","}}")]
    else
      p
    end
  }.flatten.sort
end
new() click to toggle source

Creates a new SexpProcessor. Use super to invoke this initializer from SexpProcessor subclasses, then use the attributes above to customize the functionality of the SexpProcessor

# File lib/sexp_processor.rb, line 133
def initialize
  @default_method      = nil
  @warn_on_default     = true
  @auto_shift_type     = false
  @strict              = false
  @unsupported         = [:alloca, :cfunc, :cref, :ifunc, :last, :memo,
                          :newline, :opt_n, :method]
  @unsupported_checked = false
  @debug               = {}
  @expected            = Sexp
  @require_empty       = true
  @exceptions          = {}

  # we do this on an instance basis so we can subclass it for
  # different processors.
  @processors = self.class.processors
  @rewriters  = self.class.rewriters
  @context    = []

  if @processors.empty?
    public_methods.each do |name|
      case name
      when /^process_(.*)/ then
        @processors[$1.to_sym] = name.to_sym
      when /^rewrite_(.*)/ then
        @rewriters[$1.to_sym]  = name.to_sym
      end
    end
  end
end
processors() click to toggle source

Cache processor methods per class.

# File lib/sexp_processor.rb, line 116
def self.processors
  @processors ||= {}
end
rewriters() click to toggle source

Cache rewiter methods per class.

# File lib/sexp_processor.rb, line 123
def self.rewriters
  @rewriters ||= {}
end

Public Instance Methods

assert_empty(meth, exp, exp_orig) click to toggle source

Raise if exp is not empty.

# File lib/sexp_processor.rb, line 167
def assert_empty meth, exp, exp_orig
  unless exp.empty? then
    msg = "exp not empty after #{self.class}.#{meth} on #{exp.inspect}"
    msg += " from #{exp_orig.inspect}" if $DEBUG
    raise NotEmptyError, msg
  end
end
assert_type(list, typ) click to toggle source

Raises unless the Sexp type for list matches typ

# File lib/sexp_processor.rb, line 318
def assert_type list, typ
  raise SexpTypeError, "Expected type #{typ.inspect} in #{list.inspect}" if
    not Array === list or list.sexp_type != typ
end
in_context(type) { || ... } click to toggle source

Track a stack of contexts that the processor is in, pushing on type yielding, and then removing the context from the stack.

# File lib/sexp_processor.rb, line 385
def in_context type
  self.context.unshift type

  yield

ensure
  self.context.shift
end
on_error_in(node_type, &block) click to toggle source

Registers an error handler for node

# File lib/sexp_processor.rb, line 337
def on_error_in node_type, &block
  @exceptions[node_type] = block
end
process(exp) click to toggle source

Default Sexp processor. Invokes process_<type> methods matching the Sexp type given. Performs additional checks as specified by the initializer.

# File lib/sexp_processor.rb, line 217
def process exp
  return nil if exp.nil?

  unless Sexp === exp then
    raise SexpTypeError, "exp must be a Sexp, was #{exp.class}:#{exp.inspect}"
  end

  if self.context.empty? then
    p :rewriting unless debug.empty?
    exp = self.rewrite(exp)
    p :done_rewriting unless debug.empty?
  end

  unless @unsupported_checked then
    m = public_methods.grep(/^process_/) { |o| o.to_s.sub(/^process_/, "").to_sym }
    supported = m - (m - @unsupported)

    raise UnsupportedNodeError, "#{supported.inspect} shouldn't be in @unsupported" unless supported.empty?

    @unsupported_checked = true
  end

  result = self.expected.new

  type = exp.sexp_type
  raise "type should be a Symbol, not: #{exp.first.inspect}" unless
    Symbol === type

  in_context type do
    if @debug.key? type then
      str = exp.inspect
      puts "// DEBUG:(original ): #{str}" if str =~ @debug[type]
    end

    exp_orig = nil
    exp_orig = exp.deep_clone if $DEBUG or
      @debug.key? type or @exceptions.key?(type)

    raise UnsupportedNodeError, "'#{type}' is not a supported node type" if
      @unsupported.include? type

    # now do a pass with the real processor (or generic)
    meth = @processors[type] || @default_method
    if meth then

      if @warn_on_default and meth == @default_method then
        warn "WARNING: Using default method #{meth} for #{type}"
      end

      exp = exp.sexp_body if @auto_shift_type and meth != @default_method # HACK

      result = error_handler(type, exp_orig) {
        self.send meth, exp
      }

      if @debug.key? type then
        str = exp.inspect
        puts "// DEBUG (processed): #{str}" if str =~ @debug[type]
      end

      raise SexpTypeError, "Result of #{type} must be a #{@expected}, was #{result.class}:#{result.inspect}" unless
        @expected === result

      self.assert_empty(meth, exp, exp_orig) if @require_empty
    else
      unless @strict then
        until exp.empty? do
          sub_exp, *exp = exp # HACK
          sub_result = nil
          if Array === sub_exp then
            sub_result = error_handler(type, exp_orig) do
              process(sub_exp)
            end
            raise "Result is a bad type" unless Array === sub_exp
            raise "Result does not have a type in front: #{sub_exp.inspect}" unless
              Symbol === sub_exp.sexp_type unless
              sub_exp.empty?
          else
            sub_result = sub_exp
          end
          # result << sub_result
          result = result.class.new(*result, sub_result) # HACK
        end

        # NOTE: this is costly, but we are in the generic processor
        # so we shouldn't hit it too much with RubyToC stuff at least.
        result.c_type ||= exp.c_type if Sexp === exp and exp.respond_to?(:c_type)
      else
        msg = "Bug! Unknown node-type #{type.inspect} to #{self.class}"
        msg += " in #{exp_orig.inspect} from #{caller.inspect}" if $DEBUG
        raise UnknownNodeError, msg
      end
    end
  end

  result
end
process_dummy(exp) click to toggle source

A fairly generic processor for a dummy node. Dummy nodes are used when your processor is doing a complicated rewrite that replaces the current sexp with multiple sexps.

Bogus Example:

def process_something(exp)
  return s(:dummy, process(exp), s(:extra, 42))
end
# File lib/sexp_processor.rb, line 352
def process_dummy exp
  result = @expected.new(:dummy) rescue @expected.new
  result << self.process(exp.shift) until exp.empty?
  result
end
rewrite(exp) click to toggle source

Rewrite exp using rewrite_* method for exp‘s sexp_type, if one exists.

# File lib/sexp_processor.rb, line 179
def rewrite exp
  type = exp.sexp_type

  comments = exp.comments

  if @debug.key? type then
    str = exp.inspect
    puts "// DEBUG (original ): #{str}" if str =~ @debug[type]
  end

  in_context type do
    exp = exp.map { |sub| Array === sub ? rewrite(sub) : sub }
  end

  loop do
    meth = @rewriters[type]
    exp  = self.send(meth, exp) if meth
    break unless Sexp === exp

    if @debug.key? type then
      str = exp.inspect
      puts "// DEBUG (rewritten): #{str}" if str =~ @debug[type]
    end

    old_type, type = type, exp.sexp_type
    break if old_type == type
  end

  exp.comments = comments

  exp
end
scope(&block) click to toggle source

Add a scope level to the current env. Eg:

def process_defn exp
  name = exp.shift
  args = process(exp.shift)
  scope do
    body = process(exp.shift)
    # ...
  end
end

env[:x] = 42
scope do
  env[:x]       # => 42
  env[:y] = 24
end
env[:y]         # => nil
# File lib/sexp_processor.rb, line 377
def scope &block
  env.scope(&block)
end