summaryrefslogtreecommitdiff
path: root/test/ruby/test_gb18030.rb
diff options
context:
space:
mode:
authormame <mame@b2dd03c8-39d4-4d8f-98ff-823fe69b080e>2008-06-02 13:30:38 +0000
committermame <mame@b2dd03c8-39d4-4d8f-98ff-823fe69b080e>2008-06-02 13:30:38 +0000
commit65670f9400fc16c0cb900b9b0f6851c3af93df2b (patch)
treea99dd7d4373ca760ad01269967b713afda2c6971 /test/ruby/test_gb18030.rb
parente58dc6a96818c17dd32b3929fc0d9ad2b3a2bbd9 (diff)
* enc/iso_8859_5.c: Large omicron should lowercase to small omicron.
* test/ruby/test_big5.rb, test/ruby/test_cp949.rb, test/ruby/test_euc_jp.rb, test/ruby/test_euc_kr.rb, test/ruby/test_euc_tw.rb, test/ruby/test_gb18030.rb, test/ruby/test_gbk.rb, test/ruby/test_iso_8859.rb, test/ruby/test_koi8.rb, test/ruby/test_shift_jis.rb, test/ruby/test_windows_1251.rb: new tests for encoding. * test/ruby/test_utf16.rb, test/ruby/test_utf32.rb, test/ruby/test_regexp.rb: add tests. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@16759 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
Diffstat (limited to 'test/ruby/test_gb18030.rb')
-rw-r--r--test/ruby/test_gb18030.rb125
1 files changed, 125 insertions, 0 deletions
diff --git a/test/ruby/test_gb18030.rb b/test/ruby/test_gb18030.rb
new file mode 100644
index 0000000000..a33a9eb28e
--- /dev/null
+++ b/test/ruby/test_gb18030.rb
@@ -0,0 +1,125 @@
+require "test/unit"
+
+class TestGB18030 < Test::Unit::TestCase
+ def s(s)
+ s.force_encoding("gb18030")
+ end
+
+ def test_mbc_enc_len
+ assert_equal(1, s("\x81\x40").size)
+ assert_equal(1, s("\x81\x30\x81\x30").size)
+ end
+
+ def test_mbc_to_code
+ assert_equal(0x8140, s("\x81\x40").ord)
+ end
+
+ def test_code_to_mbc
+ assert_equal(s("\x81\x40"), 0x8140.chr("gb18030"))
+ end
+
+ def test_mbc_case_fold
+ r = Regexp.new(s("(\x81\x40)\\1"), "i")
+ assert_match(r, s("\x81\x40\x81\x40"))
+ end
+
+ def scheck(c, i)
+ assert_equal(s(c.reverse.take(c.size - i).join), s(c.reverse.join).chop)
+ end
+
+ def fcheck(c)
+ assert_raise(ArgumentError) { s(c.reverse.join).chop }
+ end
+
+ def test_left_adjust_char_head
+ # C1: 00-2f, 3a-3f, 7f, ff
+ # C2: 40-7e, 80
+ # C4: 30-39
+ # CM: 81-fe
+ c1 = "\x2f"
+ c2 = "\x40"
+ c4 = "\x30"
+ cm = "\x81"
+
+ # S_START-c1
+ # S_START-c2-S_one_C2-0
+ # S_START-c2-S_one_C2-c1
+ # S_START-c2-S_one_C2-cm-S_odd_CM_one_CX-c1
+ # S_START-c2-S_one_C2-cm-S_odd_CM_one_CX-cm-S_even_CM_one_CX-c1
+ # S_START-c2-S_one_C2-cm-S_odd_CM_one_CX-cm-S_even_CM_one_CX-cm-S_odd_CM_one_CX(rec)
+ # S_START-c4-S_one_C4-c1
+ # S_START-c4-S_one_C4-cm-S_one_CMC4-c1
+ # S_START-c4-S_one_C4-cm-S_one_CMC4-c4-S_one_C4_odd_CMC4-c1
+ # S_START-c4-S_one_C4-cm-S_one_CMC4-c4-S_one_C4_odd_CMC4-cm-S_even_CMC4-c1
+ # S_START-c4-S_one_C4-cm-S_one_CMC4-c4-S_one_C4_odd_CMC4-cm-S_even_CMC4-c4-S_one_C4_even_CMC4-c1
+ # S_START-c4-S_one_C4-cm-S_one_CMC4-c4-S_one_C4_odd_CMC4-cm-S_even_CMC4-c4-S_one_C4_even_CMC4-cm-S_odd_CMC4-c1
+ # S_START-c4-S_one_C4-cm-S_one_CMC4-c4-S_one_C4_odd_CMC4-cm-S_even_CMC4-c4-S_one_C4_even_CMC4-cm-S_odd_CMC4-c4-S_one_C4_odd_CMC4(rec)
+ # S_START-c4-S_one_C4-cm-S_one_CMC4-c4-S_one_C4_odd_CMC4-cm-S_even_CMC4-c4-S_one_C4_even_CMC4-cm-S_odd_CMC4-cm-S_odd_CM_odd_CMC4-c1
+ # S_START-c4-S_one_C4-cm-S_one_CMC4-c4-S_one_C4_odd_CMC4-cm-S_even_CMC4-c4-S_one_C4_even_CMC4-cm-S_odd_CMC4-cm-S_odd_CM_odd_CMC4-cm-S_even_CM_odd_CMC4-c1
+ # S_START-c4-S_one_C4-cm-S_one_CMC4-c4-S_one_C4_odd_CMC4-cm-S_even_CMC4-c4-S_one_C4_even_CMC4-cm-S_odd_CMC4-cm-S_odd_CM_odd_CMC4-cm-S_even_CM_odd_CMC4-cm-S_odd_CM_odd_CMC4(rec)
+ # S_START-c4-S_one_C4-cm-S_one_CMC4-c4-S_one_C4_odd_CMC4-cm-S_even_CMC4-cm-S_odd_CM_even_CMC4-c1
+ # S_START-c4-S_one_C4-cm-S_one_CMC4-c4-S_one_C4_odd_CMC4-cm-S_even_CMC4-cm-S_odd_CM_even_CMC4-cm-S_even_CM_even_CMC4-c1
+ # S_START-c4-S_one_C4-cm-S_one_CMC4-c4-S_one_C4_odd_CMC4-cm-S_even_CMC4-cm-S_odd_CM_even_CMC4-cm-S_even_CM_even_CMC4-cm-S_odd_CM_even_CMC4(rec)
+ # S_START-c4-S_one_C4-cm-S_one_CMC4-cm-S_even_CM_one_CX(rec)
+ # S_START-cm-S_one_CM-c1
+ # S_START-cm-S_one_CM-c4-S_odd_C4CM-c1
+ # S_START-cm-S_one_CM-c4-S_odd_C4CM-cm-S_one_CM_odd_C4CM-c1
+ # S_START-cm-S_one_CM-c4-S_odd_C4CM-cm-S_one_CM_odd_C4CM-c4-S_even_C4CM-c1
+ # S_START-cm-S_one_CM-c4-S_odd_C4CM-cm-S_one_CM_odd_C4CM-c4-S_even_C4CM-cm-S_one_CM_even_C4CM-c1
+ # S_START-cm-S_one_CM-c4-S_odd_C4CM-cm-S_one_CM_odd_C4CM-c4-S_even_C4CM-cm-S_one_CM_even_C4CM-c4-S_odd_C4CM(rec)
+ # S_START-cm-S_one_CM-c4-S_odd_C4CM-cm-S_one_CM_odd_C4CM-c4-S_even_C4CM-cm-S_one_CM_even_C4CM-cm-S_even_CM_even_C4CM-c1
+ # S_START-cm-S_one_CM-c4-S_odd_C4CM-cm-S_one_CM_odd_C4CM-c4-S_even_C4CM-cm-S_one_CM_even_C4CM-cm-S_even_CM_even_C4CM-cm-S_odd_CM_even_C4CM-c1
+ # S_START-cm-S_one_CM-c4-S_odd_C4CM-cm-S_one_CM_odd_C4CM-c4-S_even_C4CM-cm-S_one_CM_even_C4CM-cm-S_even_CM_even_C4CM-cm-S_odd_CM_even_C4CM-cm-S_even_CM_even_C4CM(rec)
+ # S_START-cm-S_one_CM-c4-S_odd_C4CM-cm-S_one_CM_odd_C4CM-cm-S_even_CM_odd_C4CM-c1
+ # S_START-cm-S_one_CM-c4-S_odd_C4CM-cm-S_one_CM_odd_C4CM-cm-S_even_CM_odd_C4CM-cm-S_odd_CM_odd_C4CM-c1
+ # S_START-cm-S_one_CM-c4-S_odd_C4CM-cm-S_one_CM_odd_C4CM-cm-S_even_CM_odd_C4CM-cm-S_odd_CM_odd_C4CM-cm-S_even_CM_odd_C4CM(rec)
+ # S_START-cm-S_one_CM-cm-S_odd_CM_one_CX(rec)
+
+ scheck([c1], 1)
+ scheck([c2], 1)
+ scheck([c2, c1], 1)
+ scheck([c2, cm, c1], 2)
+ scheck([c2, cm, cm, c1], 1)
+ scheck([c2, cm, cm, cm], 2)
+ scheck([c4], 1)
+ scheck([c4, c1], 1)
+ scheck([c4, cm], 2)
+ fcheck([c4, cm, c1])
+ fcheck([c4, cm, c4, c1])
+ scheck([c4, cm, c4, cm], 4)
+ scheck([c4, cm, c4, cm, c1], 4)
+ scheck([c4, cm, c4, cm, c4], 4)
+ scheck([c4, cm, c4, cm, c4, c1], 4)
+ fcheck([c4, cm, c4, cm, c4, cm])
+ fcheck([c4, cm, c4, cm, c4, cm, c1])
+ fcheck([c4, cm, c4, cm, c4, cm, c4])
+ scheck([c4, cm, c4, cm, c4, cm, cm, c1], 4)
+ fcheck([c4, cm, c4, cm, c4, cm, cm, cm])
+ fcheck([c4, cm, c4, cm, c4, cm, cm, cm, c1])
+ scheck([c4, cm, c4, cm, c4, cm, cm, cm, cm], 4)
+ fcheck([c4, cm, c4, cm, cm, c1])
+ scheck([c4, cm, c4, cm, cm, cm], 4)
+ scheck([c4, cm, c4, cm, cm, cm, c1], 4)
+ fcheck([c4, cm, c4, cm, cm, cm, cm])
+ scheck([c4, cm, cm], 1)
+ scheck([cm], 1)
+ fcheck([cm, c1])
+ fcheck([cm, c4, c1])
+ scheck([cm, c4, cm], 3)
+ fcheck([cm, c4, cm, c1])
+ fcheck([cm, c4, cm, c4])
+ fcheck([cm, c4, cm, c4, c1])
+ fcheck([cm, c4, cm, c4, cm])
+ fcheck([cm, c4, cm, c4, cm, c1])
+ fcheck([cm, c4, cm, c4, cm, c4])
+ fcheck([cm, c4, cm, c4, cm, cm, c1])
+ fcheck([cm, c4, cm, c4, cm, cm, cm])
+ fcheck([cm, c4, cm, c4, cm, cm, cm, c1])
+ fcheck([cm, c4, cm, c4, cm, cm, cm, cm])
+ fcheck([cm, c4, cm, cm, c1])
+ fcheck([cm, c4, cm, cm, cm])
+ fcheck([cm, c4, cm, cm, cm, c1])
+ fcheck([cm, c4, cm, cm, cm, cm])
+ scheck([cm, cm], 2)
+ end
+end