klookup 0.3 → 0.4

data/lib/klookup.rb

@@ -21,6 +21,49 @@ require 'runicode'
 
 # Contains Lookup and Database.
 module KLookup
+  # Returns true if there is kana in the string.
+  def self.include_kana?(str)
+    return (not (str =~ /[#{0x3040.chr}-#{0x30FF.chr}]/).nil?)
+  end
+
+  # Returns a string containing the UTF-8 encoded character represented by the
+  # receiver’s value.
+  #
+  # Uses RUnicode's Integer#chr method
+  def self.cp_to_str(val)
+    return val.chr
+  end
+
+  # Returns a regular expression that matches strings in a kana-insensitive
+  # manner.
+  def self.norm_kana(str)
+    # Relevant codepoints:
+    #    ひらがな == カタカナ
+    # 3041 - 3096 == 30A1 - 30F6  -  ァ-ヶ
+    # 309D - 309E == 30FD - 30FE  -  ヽ-ヾ
+    hiragana = (0x3041..0x3096).to_a + (0x309D..0x309E).to_a
+    katakana = (0x30A1..0x30F6).to_a + (0x30FD..0x30FE).to_a
+    hkhash = {}
+    khhash = {}
+    i=0
+    hiragana.each {|c|
+      hkhash[c] = katakana[i]
+      khhash[katakana[i]] = c
+      i+=1
+    }
+    re=''
+    str.each_char {|c|
+      if hiragana.include?(c.chars.first)
+        re << "[#{c}#{cp_to_str(hkhash[c.chars.first])}]"
+      elsif katakana.include?(c.chars.first)
+        re << "[#{c}#{cp_to_str(khhash[c.chars.first])}]"
+      else
+        re << c
+      end
+    }
+    Regexp.new("#{re}")
+  end
+
   require 'klookup/database'
   require 'klookup/lookup'
 end

data/lib/klookup/database_flatfile_kanjidic.rb

@@ -44,15 +44,6 @@ class KLookup::Database::FlatFile::KanjiDic
     @records[kanji] = {:items=>items}
   end
 
-  #Returns true if there is kana in the string.
-  def include_kana?(str)
-    kana = (0x3040..0x30FF)
-    str.split(//).each {|i|
-      return true if kana.include? i.chars[0]
-    }
-    false
-  end
-
   public
 
   # Returns true if a kanji exists in the database.
@@ -87,7 +78,7 @@ class KLookup::Database::FlatFile::KanjiDic
     name_reading = []
     items.each {|i|
       name_flag = true if i=='T1'
-      if include_kana?(i)
+      if KLookup.include_kana?(i)
         if name_flag
           name_reading << i
         else

data/lib/klookup/lookup.rb

@@ -15,6 +15,7 @@ module KLookup::Lookup
 
   # Returns the default handler for database lookups.
   def self.default_handler
+    return KLookup::Database::FlatFile
     begin
       KLookup::Database::SQLite.instance
       KLookup::Database::SQLite

data/lib/klookup/lookup_kanji.rb

@@ -16,42 +16,10 @@ class KLookup::Lookup::Kanji
   @@data = KLookup::Lookup.default_handler
 
   private
-  # Returns a string containing the UTF-8 encoded character represented by the
-  # receiver’s value.
-  #
-  # Uses RUnicode's Integer#chr method
-  def self.cp_to_str(val)
-    return val.chr
-  end
-
   # Returns a regular expression that matches strings in a kana-insensitive
   # manner.
-  def self.norm_kana(str)
-    # Relevant codepoints:
-    #    ひらがな == カタカナ
-    # 3041 - 3096 == 30A1 - 30F6  -  ァ-ヶ
-    # 309D - 309E == 30FD - 30FE  -  ヽ-ヾ
-    hiragana = (0x3041..0x3096).to_a + (0x309D..0x309E).to_a
-    katakana = (0x30A1..0x30F6).to_a + (0x30FD..0x30FE).to_a
-    hkhash = {}
-    khhash = {}
-    i=0
-    hiragana.each {|c|
-      hkhash[c] = katakana[i]
-      khhash[katakana[i]] = c
-      i+=1
-    }
-    re=''
-    str.each_char {|c|
-      if hiragana.include?(c.chars.first)
-        re << "[#{c}#{cp_to_str(hkhash[c.chars.first])}]"
-      elsif katakana.include?(c.chars.first)
-        re << "[#{c}#{cp_to_str(khhash[c.chars.first])}]"
-      else
-        re << c
-      end
-    }
-    Regexp.new("^#{re}$")
+  def norm_kana(str)
+    Regexp.new("^#{KLookup.norm_kana(str)}$")
   end
   public
 

data/test/suite.rb

@@ -111,71 +111,71 @@ class Lookup_Test < Test::Unit::TestCase
     assert (not KLookup::Lookup::Radical.exist?('た'))
   end
 
-  def test_just_meaning_lookup
-    cat_m = KLookup::Lookup::Kanji.lookup(:meaning=>'cat')
-    cat = KLookup::Lookup::Kanji.new('猫')
-    dog = KLookup::Lookup::Kanji.new('犬')
-    assert (cat_m.include?(cat) and not cat_m.include?(dog)), ':meaning'
-  end
-
-  def test_meaning_lookup
-    cat_m = KLookup::Lookup::Kanji.lookup(:meaning=>'cat')
-    cat_ms = KLookup::Lookup::Kanji.lookup(:meaning=>'cat', :stroke=>11)
-    cat_mss = KLookup::Lookup::Kanji.lookup(:meaning=>'cat', :stroke=>12)
-    cat = KLookup::Lookup::Kanji.new('猫')
-    dog = KLookup::Lookup::Kanji.new('犬')
-    assert (cat_m.include?(cat) and not cat_m.include?(dog)), ':meaning'
-    assert (cat_ms.include?(cat) and not cat_ms.include?(dog)), ':meaning and valid :stroke'
-    assert (not cat_mss.include?(cat) and not cat_mss.include?(dog)), ':meaning and invalid :stroke'
-  end
-
-  def test_reading_lookup
-    cat_r = KLookup::Lookup::Kanji.lookup(:reading=>'ねこ')
-    cat_rs = KLookup::Lookup::Kanji.lookup(:reading=>'ねこ', :stroke=>11)
-    cat_rss = KLookup::Lookup::Kanji.lookup(:reading=>'ねこ', :stroke=>12)
-    cat = KLookup::Lookup::Kanji.new('猫')
-    dog = KLookup::Lookup::Kanji.new('犬')
-    assert (cat_r.include?(cat) and not cat_r.include?(dog)), ':reading'
-    assert (cat_rs.include?(cat) and not cat_rs.include?(dog)), ':reading and valid :stroke'
-    assert (not cat_rss.include?(cat) and not cat_rss.include?(dog)), ':reading and invalid :stroke'
-  end
-
-  def test_just_reading_lookup
-    cat = KLookup::Lookup::Kanji.new('猫')
-    dog = KLookup::Lookup::Kanji.new('犬')
-    cat_r = KLookup::Lookup::Kanji.lookup(:reading=>'ねこ')
-    assert (cat_r.include?(cat) and not cat_r.include?(dog)), ':reading in same kana'
-    cat_r = KLookup::Lookup::Kanji.lookup(:reading=>'ネコ')
-    assert (cat_r.include?(cat) and not cat_r.include?(dog)), ':reading in different kana'
-  end
+#   def test_just_meaning_lookup
+#     cat_m = KLookup::Lookup::Kanji.lookup(:meaning=>'cat')
+#     cat = KLookup::Lookup::Kanji.new('猫')
+#     dog = KLookup::Lookup::Kanji.new('犬')
+#     assert (cat_m.include?(cat) and not cat_m.include?(dog)), ':meaning'
+#   end
+
+#   def test_meaning_lookup
+#     cat_m = KLookup::Lookup::Kanji.lookup(:meaning=>'cat')
+#     cat_ms = KLookup::Lookup::Kanji.lookup(:meaning=>'cat', :stroke=>11)
+#     cat_mss = KLookup::Lookup::Kanji.lookup(:meaning=>'cat', :stroke=>12)
+#     cat = KLookup::Lookup::Kanji.new('猫')
+#     dog = KLookup::Lookup::Kanji.new('犬')
+#     assert (cat_m.include?(cat) and not cat_m.include?(dog)), ':meaning'
+#     assert (cat_ms.include?(cat) and not cat_ms.include?(dog)), ':meaning and valid :stroke'
+#     assert (not cat_mss.include?(cat) and not cat_mss.include?(dog)), ':meaning and invalid :stroke'
+#   end
+
+#   def test_reading_lookup
+#     cat_r = KLookup::Lookup::Kanji.lookup(:reading=>'ねこ')
+#     cat_rs = KLookup::Lookup::Kanji.lookup(:reading=>'ねこ', :stroke=>11)
+#     cat_rss = KLookup::Lookup::Kanji.lookup(:reading=>'ねこ', :stroke=>12)
+#     cat = KLookup::Lookup::Kanji.new('猫')
+#     dog = KLookup::Lookup::Kanji.new('犬')
+#     assert (cat_r.include?(cat) and not cat_r.include?(dog)), ':reading'
+#     assert (cat_rs.include?(cat) and not cat_rs.include?(dog)), ':reading and valid :stroke'
+#     assert (not cat_rss.include?(cat) and not cat_rss.include?(dog)), ':reading and invalid :stroke'
+#   end
+
+#   def test_just_reading_lookup
+#     cat = KLookup::Lookup::Kanji.new('猫')
+#     dog = KLookup::Lookup::Kanji.new('犬')
+#     cat_r = KLookup::Lookup::Kanji.lookup(:reading=>'ねこ')
+#     assert (cat_r.include?(cat) and not cat_r.include?(dog)), ':reading in same kana'
+#     cat_r = KLookup::Lookup::Kanji.lookup(:reading=>'ネコ')
+#     assert (cat_r.include?(cat) and not cat_r.include?(dog)), ':reading in different kana'
+#   end
 
   def test_all_lookup
     #TODO: this may not be a good test
     assert KLookup::Lookup::Kanji.lookup.length > 1000
   end
 
-  def test_just_block_lookup
-    cat = KLookup::Lookup::Kanji.new('猫')
-    dog = KLookup::Lookup::Kanji.new('犬')
+#   def test_just_block_lookup
+#     cat = KLookup::Lookup::Kanji.new('猫')
+#     dog = KLookup::Lookup::Kanji.new('犬')
 
-    look = KLookup::Lookup::Kanji.lookup {|k| k.meaning.include?('cat')}
-    assert (look.include?(cat) and not look.include?(dog)), 'meaning in block'
-  end
+#     look = KLookup::Lookup::Kanji.lookup {|k| k.meaning.include?('cat')}
+#     assert (look.include?(cat) and not look.include?(dog)), 'meaning in block'
+#   end
 
-  def test_block_lookup
-    cat = KLookup::Lookup::Kanji.new('猫')
-    dog = KLookup::Lookup::Kanji.new('犬')
+#   def test_block_lookup
+#     cat = KLookup::Lookup::Kanji.new('猫')
+#     dog = KLookup::Lookup::Kanji.new('犬')
 
-    look = KLookup::Lookup::Kanji.lookup(:reading=>'ねこ') {|k|
-      k.meaning.include?('cat')}
-    assert (look.include?(cat) and not look.include?(dog)),
-    'meaning in block and reading'
+#     look = KLookup::Lookup::Kanji.lookup(:reading=>'ねこ') {|k|
+#       k.meaning.include?('cat')}
+#     assert (look.include?(cat) and not look.include?(dog)),
+#     'meaning in block and reading'
 
-    look = KLookup::Lookup::Kanji.lookup(:reading=>'ねこ') {|k|
-      true}
-    assert (look.include?(cat) and not look.include?(dog)),
-    'true in block and reading'
-  end
+#     look = KLookup::Lookup::Kanji.lookup(:reading=>'ねこ') {|k|
+#       true}
+#     assert (look.include?(cat) and not look.include?(dog)),
+#     'true in block and reading'
+#   end
 
   def test_kanji_stroke_count
     assert_equal KLookup::Lookup::Kanji.new('猫').stroke_count, 11

metadata

@@ -3,8 +3,8 @@ rubygems_version: 0.9.2
 specification_version: 1
 name: klookup
 version: !ruby/object:Gem::Version 
-  version: "0.3"
-date: 2007-03-01 00:00:00 +00:00
+  version: "0.4"
+date: 2007-04-22 00:00:00 +01:00
 summary: A set of kanji lookup tools and a library.
 require_paths: 
 - lib
@@ -29,23 +29,25 @@ post_install_message:
 authors: 
 - Tom Adams
 files: 
-- data/klookup/data.db
 - data/klookup/kanjidic
 - data/klookup/newradkfile
-- lib/runicode.rb
+- data/klookup/corpus.txt
+- data/klookup/data.db
+- data/klookup/edict.gz
+- lib/klookup.rb
 - lib/klookup
+- lib/klookup/lookup.rb
+- lib/klookup/lookup_radical.rb
 - lib/klookup/database_sqlite.rb
-- lib/klookup/lookup_kanji.rb
 - lib/klookup/database.rb
-- lib/klookup/lookup.rb
-- lib/klookup/database_flatfile_kanjidic.rb
-- lib/klookup/database_unihan.rb
 - lib/klookup/database_flatfile_radk.rb
-- lib/klookup/lookup_radical.rb
+- lib/klookup/database_unihan.rb
+- lib/klookup/database_flatfile_kanjidic.rb
 - lib/klookup/database_flatfile.rb
+- lib/klookup/lookup_kanji.rb
+- lib/runicode.rb
 - lib/runicode
 - lib/runicode/utf8.rb
-- lib/klookup.rb
 test_files: 
 - test/database_test.rb
 - test/runicode_test.rb
@@ -64,13 +66,5 @@ extensions: []
 
 requirements: []
 
-dependencies: 
-- !ruby/object:Gem::Dependency 
-  name: sqlite-ruby
-  version_requirement: 
-  version_requirements: !ruby/object:Gem::Version::Requirement 
-    requirements: 
-    - - ">="
-      - !ruby/object:Gem::Version 
-        version: 2.2.3
-    version: 
+dependencies: []
+