Compare commits
No commits in common. "master" and "v0.0.1" have entirely different histories.
6 changed files with 40 additions and 82 deletions
3
.gitignore
vendored
3
.gitignore
vendored
|
|
@ -1,3 +0,0 @@
|
|||
*.swp
|
||||
*.gemspec
|
||||
pkg
|
||||
14
Rakefile
14
Rakefile
|
|
@ -6,12 +6,12 @@ begin
|
|||
Jeweler::Tasks.new do |gem|
|
||||
gem.name = "RegExpr"
|
||||
gem.summary = %Q{Regular Expression Creator}
|
||||
gem.description = %Q{Write Regular Expressions in a Hash and generates an optimized Regex}
|
||||
gem.description = %Q{Write Regular Expressions in a Hash and generats an optimized Regex}
|
||||
gem.email = "Denis.Knauf@gmail.com"
|
||||
gem.homepage = "http://github.com/DenisKnauf/RegExpr"
|
||||
gem.authors = ["Denis Knauf"]
|
||||
gem.files = %w[AUTHORS README.md VERSION lib/**/*.rb test/**/*.rb]
|
||||
gem.require_paths = %w[lib]
|
||||
gem.files = ["README.md", "VERSION", "lib/**/*.rb", "test/**/*.rb"]
|
||||
gem.require_paths = ["lib"]
|
||||
end
|
||||
Jeweler::GemcutterTasks.new
|
||||
rescue LoadError
|
||||
|
|
@ -38,11 +38,11 @@ rescue LoadError
|
|||
end
|
||||
end
|
||||
|
||||
#task :test => :check_dependencies
|
||||
task :test => :check_dependencies
|
||||
|
||||
#task :default => :test
|
||||
task :default => :test
|
||||
|
||||
require 'rdoc/task'
|
||||
require 'rake/rdoctask'
|
||||
Rake::RDocTask.new do |rdoc|
|
||||
if File.exist?('VERSION')
|
||||
version = File.read('VERSION')
|
||||
|
|
@ -51,7 +51,7 @@ Rake::RDocTask.new do |rdoc|
|
|||
end
|
||||
|
||||
rdoc.rdoc_dir = 'rdoc'
|
||||
rdoc.title = "RegExpr #{version}"
|
||||
rdoc.title = "sbdb #{version}"
|
||||
rdoc.rdoc_files.include('README*')
|
||||
rdoc.rdoc_files.include('lib/**/*.rb')
|
||||
end
|
||||
|
|
|
|||
2
VERSION
2
VERSION
|
|
@ -1 +1 @@
|
|||
0.0.2
|
||||
0.0.1
|
||||
|
|
|
|||
|
|
@ -2,7 +2,6 @@
|
|||
class RegExpr< Hash
|
||||
end
|
||||
|
||||
# any thing
|
||||
class RegExpr::Segment
|
||||
attr_accessor :value
|
||||
def initialize( val) self.value= val end
|
||||
|
|
@ -23,14 +22,12 @@ class RegExpr::Segment
|
|||
v
|
||||
end
|
||||
|
||||
# can't have any segment as value
|
||||
def deepest
|
||||
self.class_eval do
|
||||
def names() [] end
|
||||
end
|
||||
end
|
||||
|
||||
# can't have any value
|
||||
def novalue
|
||||
self.class_eval do
|
||||
def initialize() end
|
||||
|
|
@ -48,7 +45,6 @@ class RegExpr::Segment
|
|||
end
|
||||
end
|
||||
|
||||
# /(VALUE)/ or /(?:VALUE)/
|
||||
class RegExpr::Block< RegExpr::Segment
|
||||
attr_accessor :name, :hidden
|
||||
def hidden?() @hidden end
|
||||
|
|
@ -58,10 +54,6 @@ class RegExpr::Block< RegExpr::Segment
|
|||
def empty?() @value.empty? end
|
||||
def size() @value.size end
|
||||
|
||||
def inspect
|
||||
"#<#{self.class.name}: (#{hidden? ? '?:' : ''} #{value.map(&:inspect).join ' '} )>"
|
||||
end
|
||||
|
||||
def names
|
||||
names= @value.collect &:names
|
||||
names.push( name) unless self.hidden?
|
||||
|
|
@ -84,11 +76,9 @@ class RegExpr::Block< RegExpr::Segment
|
|||
else list[ -1].push v
|
||||
end
|
||||
end
|
||||
#return self
|
||||
|
||||
# [A,C,A,C,Cs,As,C,Cs,C] => [A,A,As], chars = [C,C|Cs]++[C|Cs]++[C]
|
||||
list.delete_if do |v|
|
||||
if (1 == v.size and RegExpr::Chars === v[ 0]) or RegExpr::Char === v[ 0]
|
||||
if (RegExpr::Chars === v[ 0] and v.size == 1 ) or RegExpr::Char === v[ 0]
|
||||
chars+= v[ 0]
|
||||
else false
|
||||
end
|
||||
|
|
@ -110,7 +100,7 @@ class RegExpr::Block< RegExpr::Segment
|
|||
u ? w : w.value
|
||||
else w
|
||||
end
|
||||
end.flatten.compact
|
||||
end.flatten
|
||||
end
|
||||
end
|
||||
values.push RegExpr::Or.new, chars if chars.size > 0
|
||||
|
|
@ -124,7 +114,6 @@ class RegExpr::Block< RegExpr::Segment
|
|||
end
|
||||
end
|
||||
|
||||
# /(?!VALUE)/
|
||||
class RegExpr::Not< RegExpr::Segment
|
||||
deepest
|
||||
novalue
|
||||
|
|
@ -137,7 +126,6 @@ class RegExpr::Not< RegExpr::Segment
|
|||
end
|
||||
end
|
||||
|
||||
# eg: 1..99 => /[1-9]|[1-9][1-9]/
|
||||
class RegExpr::Range< RegExpr::Segment
|
||||
novalue
|
||||
attr_accessor :v1, :v2
|
||||
|
|
@ -155,9 +143,9 @@ class RegExpr::Range< RegExpr::Segment
|
|||
bf= b == 0 ? 1.0 : b.to_f
|
||||
1.upto( b.to_s.length- 1) do |i|
|
||||
pot= 10** i
|
||||
num= (af/ pot).ceil * pot # next higher number with i zeros
|
||||
num= (af/ pot).ceil* pot # next higher number with i zeros
|
||||
arr.insert i, num if num < @v2
|
||||
num= (bf/ pot).floor * pot # next lower number with i zeros
|
||||
num= (bf/ pot).floor* pot # next lower number with i zeros
|
||||
arr.insert -i, num
|
||||
end
|
||||
arr.uniq!
|
||||
|
|
@ -165,13 +153,15 @@ class RegExpr::Range< RegExpr::Segment
|
|||
|
||||
result= RegExpr::Block.new
|
||||
0.upto( arr. length- 2) do |i|
|
||||
first, second= arr[ i].to_s, (arr[ i+ 1]- 1).to_s
|
||||
first= arr[ i].to_s
|
||||
second= (arr[ i+ 1]- 1).to_s
|
||||
result.push RegExpr::Or.new
|
||||
0.upto( first.length- 1) do |j|
|
||||
fst, sec= first[ j], second[ j]
|
||||
result.push fst == sec ?
|
||||
RegExpr::Char.new( fst.chr) :
|
||||
RegExpr::Chars.new( '%c-%c'% [ fst, sec ])
|
||||
result.push( if first[ j] == second[ j]
|
||||
RegExpr::Char.new first[ j].chr
|
||||
else
|
||||
RegExpr::Chars.new '%c-%c'% [ first[ j], second[ j] ]
|
||||
end)
|
||||
end
|
||||
end
|
||||
result. value. shift
|
||||
|
|
@ -179,7 +169,6 @@ class RegExpr::Range< RegExpr::Segment
|
|||
end
|
||||
end
|
||||
|
||||
# /[CHARS]/ or /[^CHARS]/
|
||||
class RegExpr::Chars< RegExpr::Segment
|
||||
deepest
|
||||
attr_reader :chars, :not
|
||||
|
|
@ -192,10 +181,6 @@ class RegExpr::Chars< RegExpr::Segment
|
|||
def not!() @not= !@not end
|
||||
alias -@ not!
|
||||
|
||||
def inspect
|
||||
"#<#{self.class.name}: [#{value}]>"
|
||||
end
|
||||
|
||||
def split
|
||||
chars= []
|
||||
@chars. gsub( /\\-/) do |r|
|
||||
|
|
@ -256,9 +241,7 @@ class RegExpr::Chars< RegExpr::Segment
|
|||
end
|
||||
end
|
||||
|
||||
# /VALUE{MIN,MAX}/
|
||||
class RegExpr::Repeat< RegExpr::Segment
|
||||
SimpleChar= Hash[ *%w<{,1} ? {0,1} ? {0,} * {,} * {1,} +> + ['{1,1}', ''] ]
|
||||
attr_reader :min, :max
|
||||
|
||||
def minandmax x
|
||||
|
|
@ -282,7 +265,8 @@ class RegExpr::Repeat< RegExpr::Segment
|
|||
def to_r
|
||||
t= '{%s,%s}'% [ @min||'', @max||'' ]
|
||||
return '' if '{0,0}' == t
|
||||
@value.to_r+ (SimpleChar[ t] || t)
|
||||
t= Hash[ *%w<{,1} ? {0,1} ? {0,} * {,} * {1,} +>+ ['{1,1}', ''] ][ t]|| t
|
||||
@value.to_r+ t
|
||||
end
|
||||
end
|
||||
|
||||
|
|
@ -359,16 +343,15 @@ class RegExpr
|
|||
end
|
||||
end
|
||||
|
||||
def to_r exp= nil
|
||||
r = self.to_re exp
|
||||
r.optimize!
|
||||
def to_r exp= :main
|
||||
r = self.to_re( exp)
|
||||
#r.optimize!
|
||||
h, r = r.hidden?, r.to_r
|
||||
r = r[ 1...-1] unless h
|
||||
::Regexp.new r
|
||||
end
|
||||
|
||||
def to_re exp= nil
|
||||
exp||= :main
|
||||
def to_re exp= :main
|
||||
u= RegExpr::Block.new
|
||||
t, u.hidden= if Symbol === exp
|
||||
u.name= exp.to_sym
|
||||
|
|
@ -378,11 +361,7 @@ class RegExpr
|
|||
end
|
||||
else [ exp.to_s, true]
|
||||
end
|
||||
parse t
|
||||
end
|
||||
|
||||
def parse t, u= nil
|
||||
u||= RegExpr::Block.new
|
||||
until !t or t.empty?
|
||||
v, t= self.to_r_next t
|
||||
case v
|
||||
|
|
@ -415,8 +394,7 @@ class RegExpr
|
|||
i= exp[ 2.. -1].to_i h
|
||||
return RegExpr::Char.new( i.chr), exp[ (i.to_s( h). size+ 2).. -1]
|
||||
|
||||
when ?.
|
||||
return RegExpr::WildCard.new( '.'), exp[ 1.. -1]
|
||||
when ?. then return RegExpr::WildCard.new( '.'), exp[ 1.. -1]
|
||||
|
||||
when ?0
|
||||
case exp[ 1]
|
||||
|
|
@ -428,17 +406,13 @@ class RegExpr
|
|||
return '', $1.to_i( 2).to_s+ $'
|
||||
else
|
||||
case exp
|
||||
when %r<(\d+)..(\d+)>
|
||||
RegExpr::Range.new $1.to_i, $2.to_i
|
||||
when %r<^(\d+,\d+|,\d+|\d+,?)>
|
||||
RegExpr::Repeat.new '', *$1.split( ',')
|
||||
else
|
||||
raise ArgumentError, 'Unknown form "%s"'% 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 raise ArgumentError, 'Unknown form "%s"'% exp
|
||||
end
|
||||
end
|
||||
|
||||
when ?(
|
||||
return parse( exp[ 1.. -1])
|
||||
when ?( then return self.to_re( exp[ 1.. -1])
|
||||
when ?) then ')'
|
||||
when ?| then RegExpr::Or.new
|
||||
|
||||
|
|
@ -446,26 +420,19 @@ class RegExpr
|
|||
when ?* then RegExpr::Repeat.new '', nil
|
||||
when ?? then RegExpr::Repeat.new '', 0, 1
|
||||
|
||||
when ?"
|
||||
RegExpr::Char.new %r<^"((?:[^"]|\\")*)">.match( exp)[ 1]
|
||||
when ?[
|
||||
RegExpr::Chars.new %r<^\[((?:[^\]]|\\\])*[^\\]|)\]>.match( exp)[ 1]
|
||||
when ?/
|
||||
_, re, f= %r<^/((?:[^/]|\\/)*)/(im?|mi)?>.match( exp)
|
||||
flg= $2=~ /i/ ? ::Regexp::IGNORECASE : 0
|
||||
flg+= $2=~ /m/ ? ::Regexp::MULTILINE : 0
|
||||
RegExpr::Regexp.new ::Regexp.new( re, flg)
|
||||
when ?" then RegExpr::Char.new %r<^"((?:[^"]|\\")*)">.match( exp)[ 1]
|
||||
when ?[ then RegExpr::Chars.new %r<^\[((?:[^\]]|\\\])*[^\\]|)\]>.match( exp)[ 1]
|
||||
when ?/ then exp =~ %r<^/((?:[^/]|\\/)*)/(im?|mi)?>
|
||||
RegExpr::Regexp.new ::Regexp.new( $1,
|
||||
($2 =~ /i/ ? ::Regexp::IGNORECASE : 0)+
|
||||
($2 =~ /m/ ? ::Regexp::MULTILINE : 0))
|
||||
|
||||
else
|
||||
case exp
|
||||
when %r<^([a-z_][a-z_0-9]*\b)>i
|
||||
self.to_re $1.to_sym
|
||||
when %r<(\d+)..(\d+)>
|
||||
RegExpr::Range.new $1.to_i, $2.to_i
|
||||
when %r<^(\d+,\d+|,\d+|\d+,?)>
|
||||
RegExpr::Repeat.new '', *$1.split( ',')
|
||||
else
|
||||
raise ArgumentError, 'Unknown form "%s"'% exp
|
||||
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 raise ArgumentError, 'Unknown form "%s"'% exp
|
||||
end
|
||||
end
|
||||
[ t, $' ]
|
||||
|
|
|
|||
|
|
@ -1,6 +0,0 @@
|
|||
class OptimizerTest < Test::Unit::TestCase
|
||||
context "Optimizer" do
|
||||
should "optimize many charblocks to one charblock" do
|
||||
/[a-dt-z]/.to_s == RegExpr[ main: '[a-d] | [t-z]'].to_r.to_s
|
||||
end
|
||||
end
|
||||
Loading…
Add table
Add a link
Reference in a new issue