Changeset 704

Timestamp:

10/25/07 16:12:27 (6 years ago)

Author:

gyuque

Message:

lang/javascript/ruby/syobodic/proc1.rb

Files:

• lang/ruby/syobodic/proc1.rb (modified) (4 diffs)

lang/ruby/syobodic/proc1.rb

@@ -1 +1 @@
 require 'net/http'
 require 'kconv'
-$KCODE = "UTF-8"
+require 'jcode'
+$KCODE = "UTF8"
 
 $SYOBO_SERVER = "cal.syoboi.jp"
 
+class Splitter
+        def initialize(t,k)
+                @text = t
+                @kana = k
+        end
+end
+
 class SyoboTitle
+        attr_reader :title
+        attr_reader :kana
+
         def read_from_line(ln)
+                ln.chomp!
+                ln.gsub!(/^[ *],/,"")
+
+                ln.gsub!(/^([0-9]+),/, "")
+                raise "cannot read tid" if !($&)
+                @tid = $1.to_i
+
+                ln.gsub!(/^\"([^\"]+)\",/, "")
+                raise "cannot read title" if !($&)
+                @title = $1
+
+                ln.gsub!(/^\"([^\"]+)\",/, "")
+                raise "cannot read kana" if !($&)
+                @kana = $1
+
+                ln.gsub!(/^\"([^\"]+)\"/, "")
+                @abbr = $1
+
+                @abbr = nil if @abbr == ""
+
         end
 
@@ -18 +49 @@
                         raise "cannot found h1" if !($&)
                         @title = $1
+                        @title.gsub!("&","&")
 
                         rbody =~ /<th>よみ<\/th><td>([^<]+)<\/td>/
@@ -33 +65 @@
         def print
                 mark = @abbr ? "*" : " "
-                puts "#{mark},#{@tid},\"#{@title}\",\"#{@kana}\",\"#{@abbr}\"".kconv(Kconv::SJIS, Kconv::UTF8)
+                puts "#{mark},#{@tid},\"#{@title}\",\"#{@kana}\",\"#{@abbr}\""
         end
 end
@@ -51 +83 @@
 end
 
-makeRawList(1200, 1250)
+def splitWords(rawlist_fn)
+        list = []
+
+        File.open(rawlist_fn) {|f|
+                f.each {|ln|
+                        t = SyoboTitle.new
+                        t.read_from_line(ln)
+
+                        s = Splitter.new(t.title, t.kana)
+                }
+        }
+end
+
+# -----------------------------
+
+if ARGV.length == 1
+        splitWords(ARGV[0])
+else
+        makeRawList(1150, 1250)
+end