hisui · November 23, 2011 15:59
diff --git a/rjquery.rb b/rjquery.rb
 # coding: utf-8

 require "strscan"
 require "set"


 # HTMLを解析する
 class RjStAX
 	attr_reader :scan, :node, :name, :attrs, :text

 	def initialize(html)
 		@scan = StringScanner.new html
 	end

 	def next_node       # 1                     2    3      4
 		if @scan.scan %r{^((?:.*?(?:<!.*?>)?)*?)(<\s*(/)?\s*((?:\w+:)?\w+))}m
 			unless @scan[1].empty?
 				@text = @scan[1]
 				@node = :text
 				@scan.pos -= @scan[2].size
 				return true
 			end
 			if @scan[4]
 				@name = @scan[4].downcase
 				if @scan[3].nil?
 					@attrs = {}        #    1                    2      3
 					while @scan.scan %r{^\s*([^\s/=>]+)(?:\s*=\s*(['"]?)((?=['"])(?=\2)|(?!\2).*?|[^\s>]*)\2)?}m
 						@attrs[@scan[1]] = @scan[3] || @scan[1]
 					end
 					if @scan.scan %r{^\s*(/)?\s*>}m
 						@node = @scan[1] ? :empty : :open
 						return true
 					end
 				elsif scan.scan /^\s*>/m
 					@node = :close
 					return true
 				end
 			end
 		end
 		unless scan.eos?
 			@text = scan.rest
 			@node = :text
 			scan.pos += @text.size
 			return true
 		end
 		nil
 	end
 end


 # 要素の集合を表す
 class RjNodeSet
 	include Enumerable
 	
 	class ArraySet < RjNodeSet
 		def initialize(list)
 			@list = list
 		end
 		
 		def each(&block)
 			@list.each &block
 		end
 		
 		def to_a
 			@list.dup
 		end
 	end
 	
 	def self.from_a(a)
 		a.size == 1 ? a[0]: ArraySet.new(a)
 	end
 	
 	def method_missing(key, *args)
 		first or raise NoMethodError.new
 		first.__send__ key, args
 	end
 	
 	def find(query)
 		filter " #{query}"
 	end
 	
 	def filter(query)
 		scan = StringScanner.new query
 		def scan.*(pattern)
 			scan pattern
 		end
 		RjNodeSet.from_a eval_query(scan, to_a, /^$/)
 	end
 	
 	# jQuery風のクエリを実行する
 	# 参考: http://semooh.jp/jquery/api/selectors/
 	def eval_query(scan, list, terminator)
 		#p ["QUERY:", scan.rest, terminator]
 		until scan.scan terminator
 			# 検索範囲を限定
 			case
 			when scan * /^\s+/; # descendings
 				done = Set.new list
 				list.map! {|node| node.children }.flatten!
 				list.each {|node|
 					next if done.member? node
 					done << node
 					list.concat node.children
 				}
 			when scan * /^\+/; list.map! {|node| node.succ or[]}.flatten!
 			when scan * /^\-/; list.map! {|node| node.prev or[]}.flatten!
 			when scan * /^\>/; list.map! {|node| node.children }.flatten!
 			when scan * /^\~/; list.map! {|node| node.siblings }.flatten!
 				list.uniq! # {|node| node.object_id }
 			end
 			# フィルタリング
 			loop {
 				case
 				when scan *      /^\*/; nil
 				when scan * /^\.(\w+)/; list.select! {|node| node.attrs["class"] =~/\b#{scan[1]}\b/ }
 				when scan * /^\#(\w+)/; list.select! {|node| node.attrs["id"] == scan[1] }
 				when scan *   /^(\w+)/; list.select! {|node| node.name == scan[1] }
 				when scan * /^:parent/; list.select! {|node| node.children.empty? }
 				when scan *  /^:empty/; list.reject! {|node| node.children.empty? }
 				# リスト操作タイプ
 				when scan *        /^:first/; list = [list[ 0]] unless list.empty?
 				when scan *         /^:last/; list = [list[-1]] unless list.empty?
 				when scan *  /^:eq\((.*?)\)/; list = list[scan[1].to_i, 1] || []
 				when scan *  /^:gt\((.*?)\)/; list.slice! scan[1].to_i..-1
 				when scan *  /^:lt\((.*?)\)/; list.slice! 0 ..scan[1].to_i
 				# 子要素フィルタ(要素インデックスと違って1-origin)
 				when scan * /^:nth-child\((\d+)(n(?:\+(\d+))?)?\)/
 					i = scan[1].to_i
 					j = scan[3].to_i
 					list.select! &(scan[2] ?
 						lambda {|node| (node.index+1) % i == j }:
 						lambda {|node|  node.index+1      == i })
 				# 属性フィルタ
 				when scan * /^\[(\w+)(?:([!^$*]?)=(.*?))?\]/;
 					key = scan[1]
 					val = scan[3]
 					list.select! {|node| node.attrs[key] }
 					case scan[2]
 					when  ""; list.select! {|node| node.attrs[key] == val }
 					when "!"; list.select! {|node| node.attrs[key] != val }
 					when "^"; list.select! {|node| node.attrs[key] =~/^#{val}/  }
 					when "$"; list.select! {|node| node.attrs[key] =~ /#{val}$/ }
 					when "*"; list.select! {|node| node.attrs[key].include? val }
 					end
 				# 与えられた文字列を持つ要素を(σ・∀・)σｹﾞｯﾂ!!
 				when scan * /^:contains\((.*?)\)/;
 					list.select! {|node| node.inner_text.include? scan[1] }
 				# 再帰するやつ(適当なのでなまら重い)
 				when scan * /^:not\(/; list -= eval_query(scan, list.dup, /^\)/)
 				when scan * /^:has\(/;
 					pos = scan.pos
 					list.reject! {|node|
 						scan.pos = pos
 						eval_query(scan, node.children.dup, /^\)/).empty?
 					}
 					eval_query(scan, [], /^\)/) if pos == scan.pos
 				
 				else break
 				end
 			}
 		end
 		list
 	end
 end


 # HTMLの要素(タグ)を表現
 class RjNode < RjNodeSet
 	attr_accessor :name, :attrs, :all, :parent

 	def initialize(name, attrs)
 		def (@attrs = attrs).to_s
 			map {|key, val| "#{key}=\"#{val}\"" }.join " "
 		end
 		@name = name
 		@all  = [] # TODO: linked list
 	end
 	
 	def method_missing(key, *args)
 		@attrs[key.to_s] or raise NoMethodError.new
 	end
 	
 	def each
 		yield self
 	end
 	
 	def children
 		@tags_cache ||= @all.reject {|node| node.instance_of? RjTextNode }
 	end
 	
 	def inner_text
 		@text_cache ||= @all.map {|node| node.inner_text }.join
 	end
 	
 	def index
 		parent and parent.children.index(self)
 	end
 	
 	def succ; parent.children[index+1] rescue nil end
 	def prev; parent.children[index-1] rescue nil end
 	
 	def siblings
 		parent ? parent.children[(index+1)..-1]: []
 	end
 	
 	def <<(child)
 		child.parent.remove child if child.parent
 		child.parent = self
 		@all << child
 		@tags_cache = nil
 		@text_cache = nil
 	end
 	
 	def remove(child)
 		child.parent = nil
 		@all.delete child
 		@tags_cache = nil
 		@text_cache = nil
 	end
 	
 	def to_s
 		html = "<#{name} #{attrs}>\n"
 		lifo = [[0, self]]
 		until lifo.empty?
 			i, node = lifo.last
 			if i >= node.all.size
 				lifo.pop
 				html += "    " * lifo.size + "</#{node.name}>\n"
 				next
 			end
 			node = node.all[i]
 			lifo.last[0] += 1
 			html += "    " * lifo.size
 			if node.all.empty?
 				html += node.instance_of?(RjTextNode) ? node.text: "<#{node.name} #{node.attrs}/>\n"
 				next
 			end
 			html += "<#{node.name} #{node.attrs}>\n"
 			lifo << [0, node]
 		end
 		html
 	end
 	
 	def inspect
 		"<#{name} #{attrs}> ... </#{name}>"
 	end
 end


 class RjTextNode
 	attr_accessor :text, :parent

 	def initialize(text)
 		@text = text
 	end
 	
 	def inner_text
 		@text
 	end
 	
 	def to_s
 		@text
 	end
 	
 	def inspect
 		"text:#{@text.inspect}"
 	end
 end


 # HTMLテキストからDOMを生成
 def RjQuery(html)
 	parser = RjStAX.new html
 	lifo = []
 	root = []
 	while parser.next_node
 		case parser.node
 		when :text then
 			if lifo.last and (text = parser.text.strip) != ""
 				lifo.last << RjTextNode.new(text)
 			end
 		when :open, :empty then
 			node = RjNode.new parser.name, parser.attrs || {}
 			(lifo.last || root) << node
 			lifo << node if parser.node == :open
 		when :close then
 			lifo.slice!(lifo.rindex {|node| parser.name == node.name }..-1) rescue nil
 		end
 	end
 	#puts root[0]
 	RjNodeSet.from_a root
 end


 ##
 ## ## ここからテスト用 ##
 ##

 if $0 == __FILE__

 	page = RjQuery(<<-"__TEXT__")
 	<html lang="ja">
 		<head>
 			<meta http-equiv="Content-Type" content="text/html;charset=UTF-8" />
 			<meta http-equiv="Content-Script-Type" content="text/javascript" />
 			<title>WELCOME to Hisui's HOME PAGE!</title>
 			<script src="jquery-1.7.1.js"></script>
 		</head>
 		<body>
 			<h1>Welcome to Underground</h1>
 			<p>
 				中学時代のパソコンの授業でインターネットを使った時 
 				みんなが自分の好きな漫画や野球のページを見てる時に 
 				自分だけこれみよがしに2ｃｈにつないでAAとかを周りに見せてたこと 
 				しかも「このページって何？」って聞かれた時に
 				「ヤバイ奴らの集会所みたいなもん」とか答えたこと
 				さらに友達に2ｃｈへの行きかたを教えるためにヤフーで2ｃｈって検索させて 
 				でてきたリンクをクリックして2ｃｈのトップページが表示された瞬間に 
 				そいつの耳元で「Welcome to　Underground」ってささやいたこと 
 			</p>
 			<div>
 				<h3>環境変数一覧</h3>
 				<table id="env">
 					<tr> <td>キー</td><td>値</td> </tr>
 					<tr> <td>HTTP_ACCEPT_LANGUAGE</td><td>ja,en-US;q=0.8,en;q=0.6</td> </tr>
 					<tr> <td>HTTP_ACCEPT_ENCODING</td><td>gzip,deflate,sdch</td> </tr>
 					<tr> <td>HTTP_USER_AGENT</td><td>Mozilla/5.0 (Windows NT 6.1; WOW64) AppleWebKit/535.2 (KHTML, like Gecko) Chrome/15.0.874.121 Safari/535.2</td> </tr>
 				</table>
 				<h3>最近の書き込み</h3>
 				<table id="bbs">
 					<tbody>
 						<tr>
 							<td>名前</td><td>内容</td>
 						</tr>
 					</tbody>
 					<tbody>
 						<tr colspan="2"><td>荒らしの書き込み</td></tr>
 						<tr><td>山ちゃん</td><td>オラオラオラ！</td></tr>
 						<tr><td>山ちゃん</td><td>オラオラオラ！</td></tr>
 						<tr><td>山ちゃん</td><td>オラオラオラ！</td></tr>
 						<tr colspan="2"><td>ユーザーの書き込み</td></tr>
 						<tr><td id="name_1"><a href="mailto:[email protected]">名無しさん</a></a></td><td>ヾ(*´∀｀*)ﾉｷｬｯｷｬ</td></tr>
 						<tr><td id="name_2"><a href="mailto:[email protected]">イケメン</a></td><td>よろしく～</td></tr>
 						<tr><td id="name_3"><a href="mailto:[email protected]">たぬき</a></td><td>おなかへった</td></tr>
 						<tr><td id="name_4"><a href="mailto:[email protected]">ボブ</a></td><td>こん！</td></tr>
 						<tr><td id="name_5"><a href="mailto:[email protected]">たぬき</a></td><td>こん～</td></tr>
 						<tr><td id="name_6"><a href="mailto:[email protected]">山ちゃん</a></td><td>オラオラオラ！</td></tr>
 					</tbody>
 				</table>
 			</div>
 		</body>
 	</html>
 	__TEXT__

 	nodes = page.find("table#bbs tr:has(td:contains(ユーザーの書き込み))~tr:nth-child(2n+1) td[id^=name]>a")
 	nodes.to_a.each{|node|
 		puts "#{node.inner_text} : #{node.attrs["href"]}"
 	}

 end
	# coding: utf-8

	require "strscan"
	require "set"


	# HTMLを解析する
	class RjStAX
	attr_reader :scan, :node, :name, :attrs, :text

	def initialize(html)
	@scan = StringScanner.new html
	end

	def next_node # 1 2 3 4
	if @scan.scan %r{^((?:.?(?:<!.?>)?)?)(<\s(/)?\s*((?:\w+:)?\w+))}m
	unless @scan[1].empty?
	@text = @scan[1]
	@node = :text
	@scan.pos -= @scan[2].size
	return true
	end
	if @scan[4]
	@name = @scan[4].downcase
	if @scan[3].nil?
	@attrs = {} # 1 2 3
	while @scan.scan %r{^\s([^\s/=>]+)(?:\s=\s(['"]?)((?=['"])(?=\2)\|(?!\2).?\|[^\s>]*)\2)?}m
	@attrs[@scan[1]] = @scan[3] \|\| @scan[1]
	end
	if @scan.scan %r{^\s(/)?\s>}m
	@node = @scan[1] ? :empty : :open
	return true
	end
	elsif scan.scan /^\s*>/m
	@node = :close
	return true
	end
	end
	end
	unless scan.eos?
	@text = scan.rest
	@node = :text
	scan.pos += @text.size
	return true
	end
	nil
	end
	end


	# 要素の集合を表す
	class RjNodeSet
	include Enumerable

	class ArraySet < RjNodeSet
	def initialize(list)
	@list = list
	end

	def each(&block)
	@list.each &block
	end

	def to_a
	@list.dup
	end
	end

	def self.from_a(a)
	a.size == 1 ? a[0]: ArraySet.new(a)
	end

	def method_missing(key, *args)
	first or raise NoMethodError.new
	first.__send__ key, args
	end

	def find(query)
	filter " #{query}"
	end

	def filter(query)
	scan = StringScanner.new query
	def scan.*(pattern)
	scan pattern
	end
	RjNodeSet.from_a eval_query(scan, to_a, /^$/)
	end

	# jQuery風のクエリを実行する
	# 参考: http://semooh.jp/jquery/api/selectors/
	def eval_query(scan, list, terminator)
	#p ["QUERY:", scan.rest, terminator]
	until scan.scan terminator
	# 検索範囲を限定
	case
	when scan * /^\s+/; # descendings
	done = Set.new list
	list.map! {\|node\| node.children }.flatten!
	list.each {\|node\|
	next if done.member? node
	done << node
	list.concat node.children
	}
	when scan * /^\+/; list.map! {\|node\| node.succ or[]}.flatten!
	when scan * /^\-/; list.map! {\|node\| node.prev or[]}.flatten!
	when scan * /^\>/; list.map! {\|node\| node.children }.flatten!
	when scan * /^\~/; list.map! {\|node\| node.siblings }.flatten!
	list.uniq! # {\|node\| node.object_id }
	end
	# フィルタリング
	loop {
	case
	when scan * /^\*/; nil
	when scan * /^\.(\w+)/; list.select! {\|node\| node.attrs["class"] =~/\b#{scan[1]}\b/ }
	when scan * /^\#(\w+)/; list.select! {\|node\| node.attrs["id"] == scan[1] }
	when scan * /^(\w+)/; list.select! {\|node\| node.name == scan[1] }
	when scan * /^:parent/; list.select! {\|node\| node.children.empty? }
	when scan * /^:empty/; list.reject! {\|node\| node.children.empty? }
	# リスト操作タイプ
	when scan * /^:first/; list = [list[ 0]] unless list.empty?
	when scan * /^:last/; list = [list[-1]] unless list.empty?
	when scan * /^:eq\((.*?)\)/; list = list[scan[1].to_i, 1] \|\| []
	when scan * /^:gt\((.*?)\)/; list.slice! scan[1].to_i..-1
	when scan * /^:lt\((.*?)\)/; list.slice! 0 ..scan[1].to_i
	# 子要素フィルタ(要素インデックスと違って1-origin)
	when scan * /^:nth-child\((\d+)(n(?:\+(\d+))?)?\)/
	i = scan[1].to_i
	j = scan[3].to_i
	list.select! &(scan[2] ?
	lambda {\|node\| (node.index+1) % i == j }:
	lambda {\|node\| node.index+1 == i })
	# 属性フィルタ
	when scan * /^\[(\w+)(?:([!^$]?)=(.?))?\]/;
	key = scan[1]
	val = scan[3]
	list.select! {\|node\| node.attrs[key] }
	case scan[2]
	when ""; list.select! {\|node\| node.attrs[key] == val }
	when "!"; list.select! {\|node\| node.attrs[key] != val }
	when "^"; list.select! {\|node\| node.attrs[key] =~/^#{val}/ }
	when "$"; list.select! {\|node\| node.attrs[key] =~ /#{val}$/ }
	when "*"; list.select! {\|node\| node.attrs[key].include? val }
	end
	# 与えられた文字列を持つ要素を(σ・∀・)σｹﾞｯﾂ!!
	when scan * /^:contains\((.*?)\)/;
	list.select! {\|node\| node.inner_text.include? scan[1] }
	# 再帰するやつ(適当なのでなまら重い)
	when scan * /^:not\(/; list -= eval_query(scan, list.dup, /^\)/)
	when scan * /^:has\(/;
	pos = scan.pos
	list.reject! {\|node\|
	scan.pos = pos
	eval_query(scan, node.children.dup, /^\)/).empty?
	}
	eval_query(scan, [], /^\)/) if pos == scan.pos

	else break
	end
	}
	end
	list
	end
	end


	# HTMLの要素(タグ)を表現
	class RjNode < RjNodeSet
	attr_accessor :name, :attrs, :all, :parent

	def initialize(name, attrs)
	def (@attrs = attrs).to_s
	map {\|key, val\| "#{key}=\"#{val}\"" }.join " "
	end
	@name = name
	@all = [] # TODO: linked list
	end

	def method_missing(key, *args)
	@attrs[key.to_s] or raise NoMethodError.new
	end

	def each
	yield self
	end

	def children
	@tags_cache \|\|= @all.reject {\|node\| node.instance_of? RjTextNode }
	end

	def inner_text
	@text_cache \|\|= @all.map {\|node\| node.inner_text }.join
	end

	def index
	parent and parent.children.index(self)
	end

	def succ; parent.children[index+1] rescue nil end
	def prev; parent.children[index-1] rescue nil end

	def siblings
	parent ? parent.children[(index+1)..-1]: []
	end

	def <<(child)
	child.parent.remove child if child.parent
	child.parent = self
	@all << child
	@tags_cache = nil
	@text_cache = nil
	end

	def remove(child)
	child.parent = nil
	@all.delete child
	@tags_cache = nil
	@text_cache = nil
	end

	def to_s
	html = "<#{name} #{attrs}>\n"
	lifo = [[0, self]]
	until lifo.empty?
	i, node = lifo.last
	if i >= node.all.size
	lifo.pop
	html += " " * lifo.size + "</#{node.name}>\n"
	next
	end
	node = node.all[i]
	lifo.last[0] += 1
	html += " " * lifo.size
	if node.all.empty?
	html += node.instance_of?(RjTextNode) ? node.text: "<#{node.name} #{node.attrs}/>\n"
	next
	end
	html += "<#{node.name} #{node.attrs}>\n"
	lifo << [0, node]
	end
	html
	end

	def inspect
	"<#{name} #{attrs}> ... </#{name}>"
	end
	end


	class RjTextNode
	attr_accessor :text, :parent

	def initialize(text)
	@text = text
	end

	def inner_text
	@text
	end

	def to_s
	@text
	end

	def inspect
	"text:#{@text.inspect}"
	end
	end


	# HTMLテキストからDOMを生成
	def RjQuery(html)
	parser = RjStAX.new html
	lifo = []
	root = []
	while parser.next_node
	case parser.node
	when :text then
	if lifo.last and (text = parser.text.strip) != ""
	lifo.last << RjTextNode.new(text)
	end
	when :open, :empty then
	node = RjNode.new parser.name, parser.attrs \|\| {}
	(lifo.last \|\| root) << node
	lifo << node if parser.node == :open
	when :close then
	lifo.slice!(lifo.rindex {\|node\| parser.name == node.name }..-1) rescue nil
	end
	end
	#puts root[0]
	RjNodeSet.from_a root
	end


	##
	## ## ここからテスト用 ##
	##

	if $0 == __FILE__

	page = RjQuery(<<-"__TEXT__")
	<html lang="ja">
	<head>
	<meta http-equiv="Content-Type" content="text/html;charset=UTF-8" />
	<meta http-equiv="Content-Script-Type" content="text/javascript" />
	<title>WELCOME to Hisui's HOME PAGE!</title>
	<script src="jquery-1.7.1.js"></script>
	</head>
	<body>
	<h1>Welcome to Underground</h1>
	<p>
	中学時代のパソコンの授業でインターネットを使った時
	みんなが自分の好きな漫画や野球のページを見てる時に
	自分だけこれみよがしに2ｃｈにつないでAAとかを周りに見せてたこと
	しかも「このページって何？」って聞かれた時に
	「ヤバイ奴らの集会所みたいなもん」とか答えたこと
	さらに友達に2ｃｈへの行きかたを教えるためにヤフーで2ｃｈって検索させて
	でてきたリンクをクリックして2ｃｈのトップページが表示された瞬間に
	そいつの耳元で「Welcome to　Underground」ってささやいたこと
	</p>
	<div>
	<h3>環境変数一覧</h3>
	<table id="env">
	<tr> <td>キー</td><td>値</td> </tr>
	<tr> <td>HTTP_ACCEPT_LANGUAGE</td><td>ja,en-US;q=0.8,en;q=0.6</td> </tr>
	<tr> <td>HTTP_ACCEPT_ENCODING</td><td>gzip,deflate,sdch</td> </tr>
	<tr> <td>HTTP_USER_AGENT</td><td>Mozilla/5.0 (Windows NT 6.1; WOW64) AppleWebKit/535.2 (KHTML, like Gecko) Chrome/15.0.874.121 Safari/535.2</td> </tr>
	</table>
	<h3>最近の書き込み</h3>
	<table id="bbs">
	<tbody>
	<tr>
	<td>名前</td><td>内容</td>
	</tr>
	</tbody>
	<tbody>
	<tr colspan="2"><td>荒らしの書き込み</td></tr>
	<tr><td>山ちゃん</td><td>オラオラオラ！</td></tr>
	<tr><td>山ちゃん</td><td>オラオラオラ！</td></tr>
	<tr><td>山ちゃん</td><td>オラオラオラ！</td></tr>
	<tr colspan="2"><td>ユーザーの書き込み</td></tr>
	<tr><td id="name_1"><a href="mailto:[email protected]">名無しさん</a></a></td><td>ヾ(´∀｀)ﾉｷｬｯｷｬ</td></tr>
	<tr><td id="name_2"><a href="mailto:[email protected]">イケメン</a></td><td>よろしく～</td></tr>
	<tr><td id="name_3"><a href="mailto:[email protected]">たぬき</a></td><td>おなかへった</td></tr>
	<tr><td id="name_4"><a href="mailto:[email protected]">ボブ</a></td><td>こん！</td></tr>
	<tr><td id="name_5"><a href="mailto:[email protected]">たぬき</a></td><td>こん～</td></tr>
	<tr><td id="name_6"><a href="mailto:[email protected]">山ちゃん</a></td><td>オラオラオラ！</td></tr>
	</tbody>
	</table>
	</div>
	</body>
	</html>
	__TEXT__

	nodes = page.find("table#bbs tr:has(td:contains(ユーザーの書き込み))~tr:nth-child(2n+1) td[id^=name]>a")
	nodes.to_a.each{\|node\|
	puts "#{node.inner_text} : #{node.attrs["href"]}"
	}

	end