ICU-942 fully handle <subchar1> and |2 in converters

X-SVN-Rev: 13522

icu4c/source/test/cintltst/ncnvfbts.c

@@ -709,7 +709,7 @@ static void TestConvertFallBackWithBufferSizes(int32_t outsize, int32_t insize )
                 0x01, 0x02, 0x03, 0x0a, 0x01, 0x02, 0x03, 0x0e, 0x01, 0x02, 0x03, 0x0d, 0x03, 0x01, 0x02, 0x03, 0x0c,};
         const UChar expectedUnicode[] = 
                 { 0x20ac, 0x0005, 0x0006, 0x000b, 0xdbc4, 0xde34, 0xdbba, 0xdfcd,
-                  0xd84d, 0xdc56, 0x000e, 0xd891, 0xdd67, 0xfffd, 0xfffd}; 
+                  0xd84d, 0xdc56, 0x000e, 0xd891, 0xdd67, 0x1a, 0xfffd}; 
         int32_t fromtest4Offs[]       = 
                 { 0, 1, 2, 3, 7, 7, 8, 8, 9, 9, 13, 17, 17, 21, 22,};
 

icu4c/source/test/testdata/conversion.txt

@@ -136,6 +136,42 @@ conversion {
     fromUnicode {
       Headers { "charset", "unicode", "bytes", "offsets", "flush", "fallbacks", "errorCode", "callback", "invalidUChars" }
       Cases {
+        // <subchar1> from |2 mappings
+        {
+          "ibm-1390",
+          "\x0e\x0f\u0901\U00050000\uffe8\uffee",
+          :bin{ 3f3f0efefefefe0f3f3f },
+          :intvector{ 0, 1, 2, 2, 2, 3, 3, 5, 5, 6 },
+          :int{1}, :int{1}, "", "?", ""
+        }
+
+        // <subchar1> from |2 mappings, and also contains a fallback to 00
+        {
+          "*test4",
+          "\u20ac\u20ad\U00050005\U00023456\U0010ffff\x30",
+          :bin{ 0000e10102030affff },
+          :intvector{ 0, 1, 2, 4, 4, 4, 4, 6, 8 },
+          :int{1}, :int{1}, "", "?", ""
+        }
+
+        // setting a <subchar> resets the <subchar1>
+        {
+          "*test4",
+          "\u20ac\u20ad\U00050005\U00023456\U0010ffff\x30",
+          :bin{ 00000102030f0102030a0102030f0102030f },
+          :intvector{ 0, 1, 2, 2, 2, 2, 4, 4, 4, 4, 6, 6, 6, 6, 8, 8, 8, 8 },
+          :int{1}, :int{1}, "", "?\x00\x01\x02\x03\x0f", ""
+        }
+
+        // fallback to 00 with old single-byte data structure
+        {
+          "*test1",
+          "\u20ac\u20ad\U00101234\U00050000",
+          :bin{ 000007ff },
+          :intvector{ 0, 1, 2, 4 },
+          :int{1}, :int{1}, "", "?", ""
+        }
+
         // extensions
         {
           "ibm-1390",

icu4c/source/test/testdata/test1.ucm

@@ -6,7 +6,6 @@
 # test1.ucm
 #
 # Test file for MBCS conversion with single-byte codepage data.
-# Also contains extension mappings (m:n).
 
 <code_set_name>     "test1"
 <mb_cur_max>        1
@@ -19,6 +18,7 @@ CHARMAP
 
 # fromUnicode result is zero byte from other than U+0000
 <U20ac>     \x00 |0
+<U20ad>     \x00 |1
 
 # nothing special
 <U0005>     \x05 |0

icu4c/source/test/testdata/test4.ucm

@@ -11,17 +11,25 @@
 <mb_cur_max>        4
 <mb_cur_min>        1
 <uconv_class>       "MBCS"
+
+# both subchars are single-byters, which does not make sense
+# but works - adding subchar1 for tests but don't want to
+# change old tests for a new subchar -- markus 20031028
 <subchar>           \xff
-<icu:state>         0, 1:1, 5-9, ff
+<subchar1>          \xe1
+<icu:state>         0, 1:1, 5-9, e1, ff
 <icu:state>         2:2
 <icu:state>         3:3
-<icu:state>         a-f.p
+<icu:state>         a-f.p, ff
 
 CHARMAP
 
 # fromUnicode result is zero byte from other than U+0000
 <U20ac>     \x00 |0
 
+# fallback from non-zero to zero possible with extension table
+<U20ad>     \x00 |1
+
 # nothing special
 <U0005>     \x05 |0
 
@@ -42,4 +50,7 @@ CHARMAP
 <U000e>     \x01\x02\x03\x0e |3
 #unassigned \x01\x02\x03\x0f
 
+# <subchar1> non-mapping
+<U50005>    \xe1 |2
+
 END CHARMAP