yahoo_jp_transit 0.1.0 → 0.1.1

data/README.txt

@@ -4,14 +4,21 @@ Copyright (c) 2007 SUGAWARA Genki <sgwr_dts@yahoo.co.jp>
 
 == Example
 
+    $KCODE = 's'
     require 'yahoo_jp_transit'
 
     searcher = YahooJpTransit::Searcher.new
-    result = searcher.query('大宮(埼玉県)', '東京')
+
+    selected = searcher.select('大宮', '東京')
+    puts selected
+
+    from = selected.from[0]
+    to = selected.to[0]
+
+    result = searcher.query(from, to)
 
     exit if result.empty?
 
     begin
       puts result
     end while result = result.next_pagerequire
-

data/lib/yahoo_jp_transit.rb

@@ -5,10 +5,18 @@
 #
 # == Example
 #
+#     $KCODE = 's'
 #     require 'yahoo_jp_transit'
 #
 #     searcher = YahooJpTransit::Searcher.new
-#     result = searcher.query('大宮(埼玉県)', '東京')
+#
+#     selected = searcher.select('大宮', '東京')
+#     puts selected
+#
+#     from = selected.from[0]
+#     to = selected.to[0]
+#
+#     result = searcher.query(from, to)
 #
 #     exit if result.empty?
 #
@@ -47,9 +55,26 @@ module YahooJpTransit
     end
 
     def query(from, to, options = {})
+      options = prepare_options(:result, from, to, options)
+
+      open("#{@@url}?#{query_string(options)}", { :proxy => @proxy }) do |f|
+        ResultParser.new(f, self, from, to, options).parse
+      end
+    end
+
+    def select(from, to, options = {})
+      options = prepare_options(:select, from, to, options)
+
+      open("#{@@url}?#{query_string(options)}", { :proxy => @proxy }) do |f|
+        SelectParser.new(f, self).parse
+      end
+    end
+
+    private
+    def prepare_options(htmb, from, to, options)
       options = options.dup
       defaults = {
-        :val_htmb          => :result,
+        :val_htmb          => htmb,
         :from              => from,
         :p                 => to,
         :sort              => 0,
@@ -76,12 +101,9 @@ module YahooJpTransit
         defaults.delete(:val_dsmask_air)
       end
 
-      open("#{@@url}?#{query_string(defaults.merge(options))}", { :proxy => @proxy }) do |f|
-        Parser.new(f, self, from, to, options).parse
-      end
-    end # def query
+      return defaults.merge(options)
+    end
 
-    private
     def query_string(params)
       params.map {|k, v|
         k = CGI.escape("#{k}")
@@ -103,7 +125,7 @@ module YahooJpTransit
     end
   end # class Searcher
 
-  class Parser
+  class ResultParser
     def initialize(input_stream, searcher, from, to, search_options)
       @input_stream = input_stream
       @searcher = searcher
@@ -174,19 +196,19 @@ module YahooJpTransit
     private
     def parse_result_header(header_class)
       kwargs = {}
-      kwargs[:start] = search('"start') {|l| l.match(/>([^<]+)</e)[1] }
-      kwargs[:reach] = search('"reach') {|l| l.match(/>([^<]+)</e)[1] }
-      kwargs[:lapse] = search('"result_lapse') do |l|
+      kwargs[:start] = search('"start', '"result_head_right_cont"') {|l| l.match(/>([^<]+)</e)[1] }
+      kwargs[:reach] = search('"reach', '"result_head_right_cont"') {|l| l.match(/>([^<]+)</e)[1] }
+      kwargs[:lapse] = search('"result_lapse', '<table') do |l|
         l.match(%r|>(.+)</div>|e)[1].gsub(%r|</?span[^>]*>|e, '')
       end
-      kwargs[:distance] = search('"result_distance') {|l| l.match(/>([^<]+)</e)[1] }
-      kwargs[:passage] = search('"result_passage') do |l|
+      kwargs[:distance] = search('"result_distance', '<table') {|l| l.match(/>([^<]+)</e)[1] }
+      kwargs[:passage] = search('"result_passage', '<table') do |l|
         l.match(%r|>(.+)</div>|e)[1].gsub(%r|</?span[^>]*>|e, '')
       end
-      kwargs[:connection] = search('"result_connection') do |l|
+      kwargs[:connection] = search('"result_connection', '<table') do |l|
         l.match(%r|>(.+)</div>|e)[1].gsub(%r|</?span[^>]*>|e, '')
       end
-      kwargs[:pass] = search('"result_pass') do |l|
+      kwargs[:pass] = search('"result_pass', '<table', nil) do |l|
         l.match(%r|>(.+)</div>|e)[1].gsub(%r|</?em>|e, '')
       end
 
@@ -201,15 +223,15 @@ module YahooJpTransit
 
     def parse_result_path(path_class, out)
       kwargs = {}
-      kwargs[:station] = search(['[', "\241\316"]) {|l| l.match(%r|<em>(.+)</em>|e)[1] }
+      kwargs[:station] = search(['[', "\241\316"], "<!--\265\376-->") {|l| l.match(%r|<em>(.+)</em>|e)[1] }
 
       line = ''
-      line = @input_stream.gets until line["\265\376"]
+      line = @input_stream.gets until line["<!--\265\376-->"]
       line = @input_stream.gets until line =~ /\A<tr>/e || line =~ %r|\A</div>|e
 
       if has_next = (line =~ /\A<tr>/e)
-        kwargs[:time] = search('<small>') {|l| l.match(%r|<small>([^<]+)</small>|e)[1] }
-        kwargs[:course], kwargs[:fare] = search('<small>') do |l|
+        kwargs[:time] = search('<small>', "<!--\265\376-->") {|l| l.match(%r|<small>([^<]+)</small>|e)[1] }
+        kwargs[:course], kwargs[:fare] = search('<small>', "<!--\265\376-->") do |l|
           desc = l.match(%r|>([^<]+)</font>|e)[1]
           fare = (m = l.match(%r|>([^<]+)</td>|e)) ? m[1] : nil
           [desc, fare]
@@ -246,22 +268,86 @@ module YahooJpTransit
       return errmsg
     end
 
-    def search(strs, limit = 64)
+    def search(strs, limit, default = false)
       if strs.kind_of?(String)
         strs = [strs]
       end
 
-      limit.times do
-        line = @input_stream.gets
-
+      until (line = @input_stream.gets)[limit]
         if strs.any? {|str| line[str] }
           return yield(line)
         end
       end
 
-      raise "cannot find strings: #{strs.join(', ')}"
+      if default == false
+        raise "cannot find strings: #{strs.map{|i| i.inspect }.join(', ')}"
+      else
+        return default
+      end
+    end
+  end # class ResultParser
+
+  class SelectParser
+    def initialize(input_stream, searcher)
+      @input_stream = input_stream
+      @searcher = searcher
+    end
+
+    def parse
+      from = []
+      to = []
+
+      @input_stream.each do |line|
+        raise(SearchError, parse_error) if line['"error_container"']
+        break if line["<h3>2. \222T\215\365\223\372\216\236</h3>"]
+
+        if line["<em>\275\320\310\257\261\330</em>"]
+          parse_options(from)
+        end
+
+        if line["<em>\305\376\303\345\261\330</em>"]
+          parse_options(to)
+        end
+      end
+
+      selected = Selected.new(from, to)
+
+      if @searcher.kcode
+        selected.kconv(@searcher.kcode)
+      end
+
+      return selected
+    end
+
+    private
+    def parse_options(list)
+      until (line = @input_stream.gets)['</select>']
+        if (m = %r|^<option[^>]*>([^<]+)</option>|.match(line))
+          list << m[1]
+        end
+      end
+    end
+
+    def parse_error
+      errs = []
+
+      @input_stream.each do |line|
+        break if line['</div>']
+
+        if line["\241\246"]
+          errs << line.gsub("\241\246", '').gsub(%r|</?[^>]+>|e, '').strip
+        end
+      end
+
+      if (errmsg = errs.join(', ')).empty?
+        errmsg = 'search error'
+      elsif @searcher.kcode
+        errmsg = Kconv.kconv(errmsg, @searcher.kcode, Kconv::EUC)
+      end
+
+      return errmsg
     end
-  end # class Parser
+  end # class SelectParser
 
   class Result
     class Header
@@ -347,4 +433,34 @@ module YahooJpTransit
       to_yaml
     end
   end # class Result
+
+  class Selected
+    attr_reader :from, :to
+
+    def initialize(from, to)
+      @from = from
+      @to = to
+    end
+
+    def kconv(kcode)
+      @from = @from.map {|i| Kconv.kconv(i, kcode, Kconv::EUC) }
+      @to = @to.map {|i| Kconv.kconv(i, kcode, Kconv::EUC) }
+    end
+
+    def to_yaml
+      yaml = ['---']
+
+      yaml << 'from:'
+      @from.each {|i| yaml << "  - #{i}" }
+
+      yaml << 'to:'
+      @to.each {|i| yaml << "  - #{i}" }
+
+      yaml.join($/)
+    end
+
+    def to_s
+      to_yaml
+    end
+  end # class Selected
 end # module YahooJpTransit

metadata

@@ -1,10 +1,10 @@
 --- !ruby/object:Gem::Specification 
-rubygems_version: 0.9.2
+rubygems_version: 0.9.4
 specification_version: 1
 name: yahoo_jp_transit
 version: !ruby/object:Gem::Version 
-  version: 0.1.0
-date: 2007-10-22 00:00:00 +09:00
+  version: 0.1.1
+date: 2007-10-24 00:00:00 +09:00
 summary: Scraping library for Yahoo! route information.(http://transit.yahoo.co.jp/)
 require_paths: 
 - lib