summaryrefslogtreecommitdiff
path: root/tool/transcode-tblgen.rb
diff options
context:
space:
mode:
authorakr <akr@b2dd03c8-39d4-4d8f-98ff-823fe69b080e>2010-03-14 17:01:06 +0000
committerakr <akr@b2dd03c8-39d4-4d8f-98ff-823fe69b080e>2010-03-14 17:01:06 +0000
commit1db141ed13454411acbf0e77cb91408514eacbe7 ()
tree862eb4e54c33382b7b13c5178fab1cd668e984c2 /tool/transcode-tblgen.rb
parentf5ce5551c8404d608815e95241dd3d91c74002f2 (diff)
* tool/transcode-tblgen.rb: refactored.
git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@26923 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
-rwxr-xr-xtool/transcode-tblgen.rb470
1 files changed, 245 insertions, 225 deletions
@@ -1,6 +1,23 @@
require 'optparse'
require 'erb'
require 'fileutils'
NUM_ELEM_BYTELOOKUP = 2
@@ -18,275 +35,277 @@ def c_esc(str)
'"' + str.gsub(C_ESC_PAT) { C_ESC[$&] } + '"'
end
-HEX2 = /[0-9A-Fa-f]{2}/
-
-class StrSet
- attr_reader :pat
- SINGLE_BYTE_RANGES = (0..255).map {|i| [i..i] }
- def self.parse(pattern)
- if /\A\s*((#{HEX2}|\{(#{HEX2}|#{HEX2}-#{HEX2})(,(#{HEX2}|#{HEX2}-#{HEX2}))*\})+(\s+|\z))*\z/o !~ pattern
- raise ArgumentError, "invalid pattern: #{pattern.inspect}"
- end
- result = []
- pattern.scan(/\S+/) {|seq|
- seq_result = []
- while !seq.empty?
- if /\A(#{HEX2})/o =~ seq
- byte = $1.to_i(16)
- seq_result << SINGLE_BYTE_RANGES[byte]
- seq = $'
- elsif /\A\{([^\}]+)\}/ =~ seq
- set = $1
- seq = $'
- set_result = []
- set.scan(/[^,]+/) {|range|
- if /\A(#{HEX2})-(#{HEX2})\z/o =~ range
- b = $1.to_i(16)
- e = $2.to_i(16)
- set_result << (b..e)
- elsif /\A(#{HEX2})\z/o =~ range
- byte = $1.to_i(16)
- set_result << (byte..byte)
- else
- raise "invalid range: #{range.inspect}"
- end
- }
- seq_result << set_result
- else
- raise "invalid sequence: #{seq.inspect}"
- end
- end
- result << seq_result
- }
- self.new(result)
end
- def initialize(pat)
- @pat = pat
end
- def hash
- return @hash if defined? @hash
- @hash = @pat.hash
end
- def eql?(other)
- self.class == other.class &&
- @pat == other.pat
end
- alias == eql?
- def to_s
- if @pat.empty?
- "(empset)"
- else
- @pat.map {|seq|
- if seq.empty?
- "(empstr)"
- else
- seq.map {|byteset|
- if byteset.length == 1 && byteset[0].begin == byteset[0].end
- "%02x" % byteset[0].begin
else
- "{" +
- byteset.map {|range|
- if range.begin == range.end
- "%02x" % range.begin
else
- "%02x-%02x" % [range.begin, range.end]
end
- }.join(',') +
- "}"
end
- }.join('')
- end
- }.join(' ')
- end
- end
-
- def inspect
- "\#<#{self.class}: #{self.to_s}>"
end
- def min_length
- if @pat.empty?
- nil
else
- @pat.map {|seq| seq.length }.min
end
end
- def max_length
- if @pat.empty?
- nil
- else
- @pat.map {|seq| seq.length }.max
- end
end
- def emptyable?
- @pat.any? {|seq|
- seq.empty?
- }
end
- def has_nonempty?
- @pat.any? {|seq|
- !seq.empty?
}
- end
- def first_bytes
- result = {}
- @pat.each {|seq|
- next if seq.empty?
- seq.first.each {|range|
- range.each {|byte|
- result[byte] = true
- }
}
}
- result.keys.sort
end
- def each_firstbyte
- h = {}
- @pat.each {|seq|
- next if seq.empty?
- seq.first.each {|range|
- range.each {|byte|
- (h[byte] ||= []) << seq[1..-1]
- }
}
}
- h.keys.sort.each {|byte|
- yield byte, StrSet.new(h[byte])
}
- end
-end
-
-class ArrayCode
- def initialize(type, name)
- @type = type
- @name = name
- @len = 0;
- @content = ''
- end
-
- def length
- @len
- end
-
- def insert_at_last(num, str)
- newnum = self.length + num
- @content << str
- @len += num
- end
-
- def to_s
- <<"End"
-static const #{@type}
-#{@name}[#{@len}] = {
-#{@content}};
-End
- end
-end
-
-class ActionMap
- def self.parse(hash)
- h = {}
- hash.each {|pat, action|
- h[StrSet.parse(pat)] = action
}
- self.new(h)
end
- def initialize(h)
- @map = h
end
def hash
return @hash if defined? @hash
- hash = 0
- @map.each {|k,v|
- hash ^= k.hash ^ v.hash
- }
- @hash = hash
end
def eql?(other)
self.class == other.class &&
- @map == other.instance_eval { @map }
end
alias == eql?
def inspect
"\#<#{self.class}:" +
- @map.map {|k, v| " [" + k.to_s + "]=>" + v.inspect }.join('') +
">"
end
- def max_input_length
- @map.keys.map {|k| k.max_length }.max
end
- def check_conflict
- has_empty = false
- has_nonempty = false
- @map.each {|ss, action|
- has_empty = true if ss.emptyable?
- has_nonempty = true if ss.has_nonempty?
- }
- if has_empty && has_nonempty
- raise "conflict between empty and nonempty sequence"
- end
end
def empty_action
- @map.each {|ss, action|
- return action if ss.emptyable?
- }
- nil
end
- def each_firstbyte(valid_encoding=nil)
- h = {}
- @map.each {|ss, action|
- if ss.emptyable?
- raise "emptyable pattern"
- else
- ss.each_firstbyte {|byte, rest|
- h[byte] ||= {}
- if h[byte][rest].nil?
- elsif action == :nomap0
- next
- elsif h[byte][rest] != :nomap0
- raise "ambiguous %s or %s (%02X/%s)" % [h[byte][rest], action, byte, rest]
- end
- h[byte][rest] = action
- }
- end
- }
- if valid_encoding
- valid_encoding.each_firstbyte {|byte, rest|
- if h[byte]
- am = ActionMap.new(h[byte])
- yield byte, am, rest
- else
- am = ActionMap.new(rest => :undef)
- yield byte, am, nil
- end
- }
- else
- h.keys.sort.each {|byte|
- am = ActionMap.new(h[byte])
- yield byte, am, nil
}
- end
end
OffsetsMemo = {}
@@ -451,25 +470,24 @@ End
PostMemo = {}
NextName = "a"
- def generate_node(bytes_code, words_code, name_hint=nil, valid_encoding=nil)
- if n = PreMemo[[self,valid_encoding]]
return n
end
table = Array.new(0x100, :invalid)
- each_firstbyte(valid_encoding) {|byte, rest, rest_valid_encoding|
- rest.check_conflict
if a = rest.empty_action
table[byte] = a
else
name_hint2 = nil
name_hint2 = "#{name_hint}_#{'%02X' % byte}" if name_hint
- table[byte] = "/*BYTE_LOOKUP*/" + rest.gennode(bytes_code, words_code, name_hint2, rest_valid_encoding)
end
}
if n = PostMemo[table]
- return PreMemo[[self,valid_encoding]] = n
end
if !name_hint
@@ -477,16 +495,16 @@ End
NextName.succ!
end
- PreMemo[[self,valid_encoding]] = PostMemo[table] = name_hint
generate_lookup_node(bytes_code, words_code, name_hint, table)
name_hint
end
- def gennode(bytes_code, words_code, name_hint=nil, valid_encoding=nil)
@bytes_code = bytes_code
@words_code = words_code
- name = generate_node(bytes_code, words_code, name_hint, valid_encoding)
@bytes_code = nil
@words_code = nil
return name
@@ -627,18 +645,20 @@ def transcode_compile_tree(name, from, map)
map.each {|k, v|
h[k] = v unless h[k] # use first mapping
}
- am = ActionMap.parse(h)
-
- max_input = am.max_input_length
-
- if ValidEncoding[from]
- valid_encoding = StrSet.parse(ValidEncoding[from])
- max_input = [max_input, valid_encoding.max_length].max
else
- valid_encoding = nil
end
- defined_name = am.gennode(TRANSCODE_GENERATED_BYTES_CODE, TRANSCODE_GENERATED_WORDS_CODE, name, valid_encoding)
return defined_name, max_input
end
generated by cgit v1.2.3 (git 2.25.1) at 2025-06-24 11:25:09 +0000