From 7ec29364432b993336caceb155c2605890e91bd7 Mon Sep 17 00:00:00 2001
From: Yohei YASUKAWA <yasulab@gmail.com>
Date: Sun, 21 Oct 2012 15:42:36 +0900
Subject: [PATCH] Parse lookup table

---
 yohei_tangstad/a_to_map.rb     | 14 +++++++++++
 yohei_tangstad/data.txt        | 11 +++++++++
 yohei_tangstad/dict-creator.rb | 45 ++++++++++++++++++++++++++++++++++
 yohei_tangstad/reader.rb       | 21 ++++++++++++++++
 4 files changed, 91 insertions(+)
 create mode 100644 yohei_tangstad/a_to_map.rb
 create mode 100644 yohei_tangstad/data.txt
 create mode 100644 yohei_tangstad/dict-creator.rb
 create mode 100644 yohei_tangstad/reader.rb

diff --git a/yohei_tangstad/a_to_map.rb b/yohei_tangstad/a_to_map.rb
new file mode 100644
index 0000000..780a0dd
--- /dev/null
+++ b/yohei_tangstad/a_to_map.rb
@@ -0,0 +1,14 @@
+require 'rspec'
+
+def a_to_map(a)
+	target = a[0]
+	m = Hash.new
+	a.slice(1..-1).each {|value| m[value] = target}
+	m
+end
+
+describe "a_to_map" do
+	it "should convert array with number to map to number" do
+		a_to_map(["0", "a", "b"]).should == {"a" => "0", "b" => "0"}
+	end
+end
diff --git a/yohei_tangstad/data.txt b/yohei_tangstad/data.txt
new file mode 100644
index 0000000..dd15fcb
--- /dev/null
+++ b/yohei_tangstad/data.txt
@@ -0,0 +1,11 @@
+0      | まる、ま                  | れい、れ                 | おう、ぜろ、ぜ 
+1      | ひとつ、ひと、ひ          | いち、い                 | わん 
+2      | ふたつ、ふた、ふ          | に                       | つ 
+3      | みつ、み                  | さん、さ                 | すりー 
+4      | よん、よ、よつ            | し                       | ふぉー 
+5      | いつつ、いつ              | ご、こ                   | ふぁいぶ、ふぁいぶ
+6      | むつ、む                  | ろく、ろ                 | しっくす
+7      | ななつ、なな、な          | しち                     | せぶん、せゔん
+8      | やつ、や                  | はち、は、ば             | えーと
+9      | ここのつ、こ              | きゅう、く               | ないん
+10     | とお                      | じゅう、じ               | てん
diff --git a/yohei_tangstad/dict-creator.rb b/yohei_tangstad/dict-creator.rb
new file mode 100644
index 0000000..eb66564
--- /dev/null
+++ b/yohei_tangstad/dict-creator.rb
@@ -0,0 +1,45 @@
+# -*- coding: utf-8 -*-
+
+require 'rspec'
+
+class String
+  def split_characters(chars)
+    array = [strip]
+    chars.each_char{ |c|
+      temp_array = []
+      array.each{|s|
+        temp_array.concat s.split(c)
+	temp_array.delete_if{|a| a==""}
+      }
+      array = temp_array
+    }
+    # result = split('|')
+    return array.map{|l| l.strip}
+  end
+end
+
+#"".split_characters
+given_data = open('./data.txt').read
+
+given_data.each_line {|d|
+  puts d.split_characters('|、').to_s
+}
+
+
+
+describe "split_characters" do
+	it "should split on character" do
+		"one two".split_characters(" ").should == ["one", "two"]
+	end
+	it "should split on multiple characters" do
+		"one two,three, four".split_characters(" ,").should == ["one", "two", "three", "four"]
+	end
+	it "should split our input properly" do
+		"0      | まる、ま                  | れい、れ                 | おう 、ぜろ、ぜ ".split_characters("| 、").should == ["0", "まる", "ま", "れい", "れ", "おう", "ぜろ", "ぜ"]
+	end
+end
+
+
+
+
+
diff --git a/yohei_tangstad/reader.rb b/yohei_tangstad/reader.rb
new file mode 100644
index 0000000..25372fb
--- /dev/null
+++ b/yohei_tangstad/reader.rb
@@ -0,0 +1,21 @@
+# encoding: UTF-8
+
+require 'rspec'
+
+class TableReader
+	def parseline(line)
+		a = Array.new
+		a.push(line)
+		a
+	end
+end
+
+describe TableReader do
+	describe "#parseline" do
+		it "should return all hiragana words in a line" do
+			tr = TableReader.new
+			tr.parseline("ひらがな").should == ["ひらがな"]
+		end
+	end
+end
+