deac/RegExpr
1
0
Fork 0
Code Issues Pull requests Releases Wiki Activity

optimizer broken. :/ (not compatible to ruby 1.9.1, changes needed)

Browse source
This commit is contained in:
Denis Knauf 2010-03-19 12:11:16 +01:00
parent 203f8103e5
commit 7f98fd7d4b
2 changed files with 161 additions and 157 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

2
lib/regexpr-uri.rb
View file

@ -45,7 +45,7 @@ class Uri
'userinfo' => 'username ( ":" password ) ?',
'domainlabel' => 'alphadigit ( ( alphadigit | "-" ) * alphadigit ) ?',
'hostname' => 'domainlabel ( "." | domainlabel ) *',
'ipv4' => 'ipv4digits ( "." ipv4digits ) {3}',
'ipv4' => 'ipv4digits ( "." ipv4digits ) {3,3}',
'request_uri' => '( path ? file ) ? ( "?" flags ) ?',
'serv' => 'host ( ":" port ? ) ?',
'pre' => '( ( scheme "://" ) ( auth "@" ) ? serv ) ?',

72
lib/regexpr.rb
View file

@ -17,8 +17,8 @@ class RegExpr::Segment
class <<self
def optimize v
v= v.optimize
v= nil if v && v. empty?
v= v. value[ 0] if v. instance_of?( RegExpr::Block) && v. hidden && v. size == 1
v= nil if v and v.empty?
v= v.value[ 0] if RegExpr::Block === v and v.hidden and v.size == 1
v
end
@ -55,13 +55,13 @@ class RegExpr::Block< RegExpr::Segment
def size() @value.size end
def names
names= @value. collect {|v| v. names }
names= @value.collect &:names
names.push( name) unless self.hidden?
names.flatten.compact
end
def initialize *val
val= val[ 0] if val. size == 1 && val[ 0]. instance_of?( Array)
val= val[ 0] if ::Array === val[ 0] and val.size == 1
super val
@hidden= true
end
@ -71,14 +71,14 @@ class RegExpr::Block< RegExpr::Segment
@value.each do |v|
v= self.class.optimize v
if v.instance_of? RegExpr::Or
if RegExpr::Or === v
list.push []
else list[ -1].push v
end
end
list.delete_if do |v|
if v. size == 1 && ( v[ 0]. instance_of?( RegExpr::Chars) || v[ 0]. instance_of?( RegExpr::Char) )
if (RegExpr::Chars === v[ 0] and v.size == 1 ) or RegExpr::Char === v[ 0]
chars+= v[ 0]
else false
end
@ -88,16 +88,14 @@ class RegExpr::Block< RegExpr::Segment
values= []
list.each do |v|
values.push RegExpr::Or.new
values+= if v. size == 1 &&
v[ 0]. instance_of?( RegExpr::Block) &&
v[ 0]. hidden
values+= if v.size == 1 and RegExpr::Block === v[ 0] and v[ 0].hidden
v[ 0].value
else
v.collect do |w|
if w. instance_of?( RegExpr::Block) && w. hidden
if RegExpr::Block === w and w.hidden
u= false
w.value.each do |i|
break unless u||= i. instance_of?( RegExpr::Or)
break unless u||= RegExpr::Or === i
end
u ? w : w.value
else w
@ -112,7 +110,7 @@ class RegExpr::Block< RegExpr::Segment
end
def to_r()
(@hidden ? '(?:%s)' : '(%s)')% @value. collect {|i| i.to_r }. join( '')
(@hidden ? '(?:%s)' : '(%s)')% @value.collect( &:to_r).join( '')
end
end
@ -178,7 +176,7 @@ class RegExpr::Chars< RegExpr::Segment
def not?() @not end
def empty?() @chars. empty? end
def size() @chars. size end
def value=( val) @chars= (@not= val[ 0] == ?^) ? val[ 1.. -1] : val end
def value=( val) @chars= (@not= val[ 0] == ?^) ? val[ 1.. -1] : val ; val end
def value() (self. not? ? '^' : '')+ (@chars) end
def not!() @not= !@not end
alias -@ not!
@ -186,12 +184,12 @@ class RegExpr::Chars< RegExpr::Segment
def split
chars= []
@chars. gsub( /\\-/) do |r|
chars. push ?-
chars. push ?-.ord
nil
end. gsub( /.-./) do |r|
chars+= ((r[ 0] .. r[ 2]). to_a)
chars+= (r[ 0].ord .. r[ 2].ord). to_a
nil
end. each_byte do |c|
end. bytes. each do |c|
chars. push c
end
chars
@ -199,16 +197,16 @@ class RegExpr::Chars< RegExpr::Segment
def optimize!
b2chr= lambda do |b|
"-[]".include?( b.chr) ? '\%c'% b : b. chr
b = b.chr
"-[]".include?( b) ? '\%c'% b : b
end
chars= self. chars. bytes. sort. uniq
$stderr.puts chars.inspect
chars= self. split. sort. uniq
@chars= ''
return self if chars.empty?
b= chars.shift
chars.each do |i|
if b+1 == i
unless @chars[ -1] == ?- && @chars[-2] != ?\\
unless @chars[ -1] == ?- and @chars[-2] != ?\\
@chars+= b2chr.call( b)+ '-'
end
else @chars+= b2chr.call b
@ -221,9 +219,10 @@ class RegExpr::Chars< RegExpr::Segment
def optimize
n= self.dup.optimize!
if (n. size == 1 || (n. size == 2 && n. value[ 0] == ?\\ )) && ! n. not?
RegExpr::Char. new n. chars[ -1]. chr
else n
if (n.size == 1 or (n.size == 2 and n.value[ 0] == ?\\ )) and not n.not?
RegExpr::Char.new n.chars[ -1]
else
n
end
end
@ -254,7 +253,8 @@ class RegExpr::Repeat< RegExpr::Segment
def optimize
super
min == 1 && max == 1 ? @value : self
r = (min == 1 and max == 1) ? @value : self
r
end
def initialize value, min= 1, max= min
@ -264,6 +264,7 @@ class RegExpr::Repeat< RegExpr::Segment
def to_r
t= '{%s,%s}'% [ @min||'', @max||'' ]
return '' if '{0,0}' == t
t= Hash[ *%w<{,1} ? {0,1} ? {0,} * {,} * {1,} +>+ ['{1,1}', ''] ][ t]|| t
@value.to_r+ t
end
@ -343,7 +344,8 @@ class RegExpr
end
def to_r exp= :main
r = self. to_re( exp). optimize
r = self.to_re( exp)
#r.optimize!
h, r = r.hidden?, r.to_r
r = r[ 1...-1] unless h
::Regexp.new r
@ -351,7 +353,7 @@ class RegExpr
def to_re exp= :main
u= RegExpr::Block.new
t, u. hidden= if exp. instance_of? Symbol
t, u.hidden= if Symbol === exp
u.name= exp.to_sym
if self[ exp]
[ self[ exp], false]
@ -360,7 +362,7 @@ class RegExpr
else [ exp.to_s, true]
end
until !t || t. empty?
until !t or t.empty?
v, t= self.to_r_next t
case v
when ')' then return u, t
@ -373,7 +375,7 @@ class RegExpr
def to_r_next exp
exp.strip!
/^/. match exp[ 1.. -1]
/^/ =~ exp[ 1.. -1]
t= case exp[ 0]
when ?^ then return RegExpr::Begin.new, exp[ 1.. -1]
when ?$ then return RegExpr::End.new, exp[ 1.. -1]
@ -387,7 +389,7 @@ class RegExpr
when ?0.. ?9
exp= 'XX'+ exp[ 1.. -1]
10
else Kernel. raise ArgumentError, 'Unknown form "%s"'% exp
else raise ArgumentError, 'Unknown form "%s"'% exp
end
i= exp[ 2.. -1].to_i h
return RegExpr::Char.new( i.chr), exp[ (i.to_s( h). size+ 2).. -1]
@ -406,7 +408,7 @@ class RegExpr
case exp
when %r<(\d+)..(\d+)> then RegExpr::Range.new $1.to_i, $2.to_i
when %r<^(\d+,\d+|,\d+|\d+,?)> then RegExpr::Repeat.new '', *$1.split( ',')
else Kernel. raise ArgumentError, 'Unknown form "%s"'% exp
else raise ArgumentError, 'Unknown form "%s"'% exp
end
end
@ -430,7 +432,7 @@ class RegExpr
when %r<^([a-z_][a-z_0-9]*\b)>i then self.to_re $1.to_sym
when %r<(\d+)..(\d+)> then RegExpr::Range.new $1.to_i, $2.to_i
when %r<^(\d+,\d+|,\d+|\d+,?)> then RegExpr::Repeat.new '', *$1.split( ',')
else Kernel. raise ArgumentError, 'Unknown form "%s"'% exp
else raise ArgumentError, 'Unknown form "%s"'% exp
end
end
[ t, $' ]
@ -440,8 +442,8 @@ class RegExpr
exp= [ :main ] if exp.empty?
exp.each do |e|
re= self.to_re e
names= re. names. collect {|n| '@%s'% n }. join ', '
re= ::Regexp. new '^%s$'% re. optimize. to_r
names= re.names.collect('@%s'.method(:%)).join ', '
re= ::Regexp.new '^%s$'% re.to_r
ev= <<-EOF
def #{e}= val
m= #{re.inspect}. match val
@ -454,5 +456,7 @@ class RegExpr
cl
end
def match( m, exp= :main) self. to_r( exp). match m end
def match( m, exp= :main) to_r( exp).match m end
def =~( x) to_r =~ x end
end