Implement indirect parser and fix some bugs.

This commit is contained in:
Jakob Rath 2013-12-11 14:37:36 +01:00 committed by Dan Hirsch
parent 8ff5e181ce
commit 76782bfa4a
3 changed files with 63 additions and 24 deletions

View file

@ -12,7 +12,7 @@ require 'hammer/parser_builder'
parser = Hammer::Parser.build do parser = Hammer::Parser.build do
token 'blah' token 'blah'
ch 'a' ch 'a'.ord
choice { choice {
sequence { sequence {
token 'abc' token 'abc'
@ -57,3 +57,17 @@ parser = h.token(s)
p parser.parse 'BLAH' # => false p parser.parse 'BLAH' # => false
s.upcase! s.upcase!
p parser.parse 'BLAH' # => false p parser.parse 'BLAH' # => false
x = nil
parser = Hammer::Parser.build {
token 'abc'
x = indirect
end_p
}
x.bind(h.token('abd'))
p parser.parse 'abcabdabd'
p parser.parse 'abcabd'
p parser.parse 'abdabd'
p parser.parse 'abd'

View file

@ -3,9 +3,22 @@ module Hammer
# Don't create new instances with Hammer::Parser.new, # Don't create new instances with Hammer::Parser.new,
# use the constructor methods instead (i.e. Hammer::Parser.int64 etc.) # use the constructor methods instead (i.e. Hammer::Parser.int64 etc.)
def initialize #
# name: Name of the parser. Should be a symbol.
# h_parser: The pointer to the parser as returned by hammer.
# dont_gc: Pass additional data that's used by the parser and needs to be saved from the garbage collector.
def initialize(name, h_parser, dont_gc)
@name = name
@h_parser = h_parser
@dont_gc = dont_gc
end end
attr_reader :name
attr_reader :h_parser
# Parse the given data. Returns true if successful, false otherwise.
#
# data: A string containing the data to parse.
def parse(data) def parse(data)
raise RuntimeError, '@h_parser is nil' if @h_parser.nil? raise RuntimeError, '@h_parser is nil' if @h_parser.nil?
raise ArgumentError, 'expecting a String' unless data.is_a? String # TODO: Not needed, FFI checks that. raise ArgumentError, 'expecting a String' unless data.is_a? String # TODO: Not needed, FFI checks that.
@ -16,24 +29,24 @@ module Hammer
!result.null? !result.null?
end end
# Binds an indirect parser.
def bind(other_parser)
raise RuntimeError, 'can only bind indirect parsers' unless self.name == :indirect
Hammer::Internal.h_bind_indirect(self.h_parser, other_parser.h_parser)
end
def self.token(string) def self.token(string)
h_string = string.dup h_string = string.dup
h_parser = Hammer::Internal.h_token(h_string, h_string.length) h_parser = Hammer::Internal.h_token(h_string, h_string.length)
parser = Hammer::Parser.new return Hammer::Parser.new(:token, h_parser, h_string)
parser.instance_variable_set :@h_parser, h_parser
# prevent string from getting garbage-collected
parser.instance_variable_set :@h_string, h_string
return parser
end end
def self.ch(num) def self.ch(num)
raise ArgumentError, 'expecting a Fixnum in 0..255', unless num.is_a?(Fixnum) and num.between?(0, 255) raise ArgumentError, 'expecting a Fixnum in 0..255' unless num.is_a?(Fixnum) and num.between?(0, 255)
h_parser = Hammer::Internal.h_ch(num) h_parser = Hammer::Internal.h_ch(num)
parser = Hammer::Parser.new return Hammer::Parser.new(:ch, h_parser, nil)
parser.instance_variable_set :@h_parser, h_parser
return parser
end end
# Defines a parser constructor with the given name. # Defines a parser constructor with the given name.
@ -41,7 +54,7 @@ module Hammer
# hammer_function: name of the hammer function to call (default: 'h_'+name) # hammer_function: name of the hammer function to call (default: 'h_'+name)
# varargs: Whether the function is taking a variable number of arguments (default: false) # varargs: Whether the function is taking a variable number of arguments (default: false)
def self.define_parser(name, options = {}) def self.define_parser(name, options = {})
hammer_function = options[:hammer_function] || ('h_' + name.to_s) hammer_function = options[:hammer_function] || ('h_' + name.to_s).to_sym
varargs = options[:varargs] || false varargs = options[:varargs] || false
# Define a new class method # Define a new class method
@ -54,10 +67,7 @@ module Hammer
end end
h_parser = Hammer::Internal.send hammer_function, *args h_parser = Hammer::Internal.send hammer_function, *args
parser = Hammer::Parser.new return Hammer::Parser.new(name, h_parser, parsers)
parser.instance_variable_set :@h_parser, h_parser
parser.instance_variable_set :@sub_parsers, parsers # store sub parsers to prevent them from being garbage-collected
return parser
end end
end end
private_class_method :define_parser private_class_method :define_parser
@ -77,8 +87,8 @@ module Hammer
define_parser :left define_parser :left
define_parser :right define_parser :right
define_parser :middle define_parser :middle
define_parser :end define_parser :end_p
define_parser :nothing define_parser :nothing_p
define_parser :butnot define_parser :butnot
define_parser :difference define_parser :difference
define_parser :xor define_parser :xor
@ -92,10 +102,7 @@ module Hammer
define_parser :length_value define_parser :length_value
define_parser :and define_parser :and
define_parser :not define_parser :not
# TODO: If indirect, add a bind method that calls h_bind_indirect
define_parser :indirect define_parser :indirect
attr_reader :h_parser
end end
end end

View file

@ -75,8 +75,8 @@ module Hammer
define_parser :left define_parser :left
define_parser :right define_parser :right
define_parser :middle define_parser :middle
define_parser :end define_parser :end_p
define_parser :nothing define_parser :nothing_p
define_parser :butnot define_parser :butnot
define_parser :difference define_parser :difference
define_parser :xor define_parser :xor
@ -90,7 +90,25 @@ module Hammer
define_parser :length_value define_parser :length_value
define_parser :and define_parser :and
define_parser :not define_parser :not
define_parser :indirect
# At least indirect must return the parser instead of the builder, so it can be stored in a variable.
# Other possible solution:
# Make indirect take a name parameter, and use the name to bind it later.
# Example:
# p = Hammer::Parser.build { indirect(:the_name) }
# p.bind(:the_name, inner_parser)
# (store names and parsers in hash in the builder,
# when building merge hashes from sub builders and store everything in the resulting sequence or choice.
# make Parser#bind take and optional symbol. if it is given, the name is looked up in the table.)
# TODO:
# Think about this more.
# Do we need to be able to build parsers by chaining function calls? DSL should be sufficient.
# If yes, the parser methods in this class should not return "self", but the Hammer::Parser object they create.
def indirect
parser = Hammer::Parser.indirect
@parsers << parser
return parser
end
end # class ParserBuilder end # class ParserBuilder