UCW: Added UTF-16 test cases.

diff --git a/lib/unicode.c b/lib/unicode.c
index e4a33af8495f216bcbade228c55ce271a8f0f23c..c5ffb8267bc89ad736dcc724abfc2554e875a723 100644 (file)
--- a/lib/unicode.c
+++ b/lib/unicode.c
@@ -37,55 +37,107 @@ utf8_strnlen(const byte *str, uns n)
 }
 
 #ifdef TEST
+
 #include <string.h>
 #include <stdio.h>
+
 int main(int argc, char **argv)
 {
   byte buf[256];
-  if (argc > 1 && !strncmp(argv[1], "get", 3))
+
+#define FUNCS \
+  F(UTF8_GET) F(UTF8_32_GET) F(UTF16_BE_GET) F(UTF16_LE_GET) \
+  F(UTF8_PUT) F(UTF8_32_PUT) F(UTF16_BE_PUT) F(UTF16_LE_PUT)
+
+  enum {
+#define F(x) FUNC_##x,
+    FUNCS
+#undef F
+  };
+  char *names[] = {
+#define F(x) [FUNC_##x] = #x,
+    FUNCS
+#undef F
+  };
+
+  uns func = ~0U;
+  if (argc > 1)
+    for (uns i = 0; i < ARRAY_SIZE(names); i++)
+      if (!strcasecmp(names[i], argv[1]))
+       func = i;
+  if (!~func)
+    {
+      fprintf(stderr, "Invalid usage!\n");
+      return 1;
+    }
+
+  if (func < FUNC_UTF8_PUT)
     {
-      int f32 = !strcmp(argv[1], "get32");
-      byte *p = buf;
+      byte *p = buf, *q = buf, *last;
       uns u;
+      bzero(buf, sizeof(buf));
       while (scanf("%x", &u) == 1)
-       *p++ = u;
-      *p = 0;
-      p = buf;
-      while (*p)
+       *q++ = u;
+      while (p < q)
        {
+         last = p;
          if (p != buf)
            putchar(' ');
-         if (f32)
-           GET_UTF8_32(p, u);
-         else
-           GET_UTF8(p, u);
+         switch (func)
+           {
+             case FUNC_UTF8_GET:
+               p = utf8_get(p, &u);
+               break;
+             case FUNC_UTF8_32_GET:
+               p = utf8_32_get(p, &u);
+               break;
+             case FUNC_UTF16_BE_GET:
+               p = utf16_be_get(p, &u);
+               break;
+             case FUNC_UTF16_LE_GET:
+               p = utf16_le_get(p, &u);
+               break;
+             default:
+               ASSERT(0);
+           }
          printf("%04x", u);
+         ASSERT(last < p && p <= q);
        }
       putchar('\n');
     }
-  else if (argc > 1 && !strncmp(argv[1], "put", 3))
+  else
     {
       uns u, i=0;
-      int f32 = !strcmp(argv[1], "put32");
       while (scanf("%x", &u) == 1)
        {
-         byte *p = buf;
-         if (f32)
-           PUT_UTF8_32(p, u);
-         else
-           PUT_UTF8(p, u);
-         *p = 0;
-         for (p=buf; *p; p++)
+         byte *p = buf, *q = buf;
+         switch (func)
+           {
+             case FUNC_UTF8_PUT:
+               p = utf8_put(p, u);
+               break;
+             case FUNC_UTF8_32_PUT:
+               p = utf8_32_put(p, u);
+               break;
+             case FUNC_UTF16_BE_PUT:
+               p = utf16_be_put(p, u);
+               break;
+             case FUNC_UTF16_LE_PUT:
+               p = utf16_le_put(p, u);
+               break;
+             default:
+               ASSERT(0);
+           }
+         while (q < p)
            {
              if (i++)
                putchar(' ');
-             printf("%02x", *p);
+             printf("%02x", *q++);
            }
        }
       putchar('\n');
     }
-  else
-    puts("?");
   return 0;
 }
+
 #endif

diff --git a/lib/unicode.t b/lib/unicode.t
index 132bbbd2fb31e6bc49d2beea25a19759d2096389..1b5549ef738261ff019fb366d1a189ea3182a4d6 100644 (file)
--- a/lib/unicode.t
+++ b/lib/unicode.t
@@ -1,41 +1,71 @@
-# Tests for the Unicode UTF-8 module
+# Tests for the Unicode module
 
-Name:  put1
-Run:   ../obj/lib/unicode-utf8-t put
+Name:  utf8_put (1)
+Run:   ../obj/lib/unicode-t utf8_put
 In:    0041 0048 004f 004a
 Out:   41 48 4f 4a
 
-Name:  put2
-Run:   ../obj/lib/unicode-utf8-t put
+Name:  utf8_put (2)
+Run:   ../obj/lib/unicode-t utf8_put
 In:    00aa 01aa 02a5 05a5 0a5a 15a5 2a5a 5a5a a5a5
 Out:   c2 aa c6 aa ca a5 d6 a5 e0 a9 9a e1 96 a5 e2 a9 9a e5 a9 9a ea 96 a5
 
-Name:  get1
-Run:   ../obj/lib/unicode-utf8-t get
+Name:  utf8_get (1)
+Run:   ../obj/lib/unicode-t utf8_get
 In:    41 48 4f 4a
 Out:   0041 0048 004f 004a
 
-Name:  get2
-Run:   ../obj/lib/unicode-utf8-t get
+Name:  utf8_get (2)
+Run:   ../obj/lib/unicode-t utf8_get
 In:    c2 aa c6 aa ca a5 d6 a5 e0 a9 9a e1 96 a5 e2 a9 9a e5 a9 9a ea 96 a5
 Out:   00aa 01aa 02a5 05a5 0a5a 15a5 2a5a 5a5a a5a5
 
-Name:  get3
-Run:   ../obj/lib/unicode-utf8-t get
+Name:  utf8_get (3)
+Run:   ../obj/lib/unicode-t utf8_get
 In:    84 ff f9 f8 c2 aa 41
 Out:   fffc fffc fffc fffc 00aa 0041
 
-Name:  put32
-Run:   ../obj/lib/unicode-utf8-t put32
+Name:  utf8_32_put
+Run:   ../obj/lib/unicode-t utf8_32_put
 In:    15a5a 2a5a5 5a5a5 a5a5a 15a5a5 2a5a5a 5a5a5a a5a5a5 15a5a5a 2a5a5a5 5a5a5a5 a5a5a5a 15a5a5a5 2a5a5a5a 5a5a5a5a
 Out:   f0 95 a9 9a f0 aa 96 a5 f1 9a 96 a5 f2 a5 a9 9a f5 9a 96 a5 f8 8a a5 a9 9a f8 96 a5 a9 9a f8 a9 9a 96 a5 f9 96 a5 a9 9a fa a9 9a 96 a5 fc 85 a9 9a 96 a5 fc 8a 96 a5 a9 9a fc 95 a9 9a 96 a5 fc aa 96 a5 a9 9a fd 9a 96 a5 a9 9a
 
-Name:  get32a
-Run:   ../obj/lib/unicode-utf8-t get32
+Name:  utf8_32_get (1)
+Run:   ../obj/lib/unicode-t utf8_32_get
 In:    f0 95 a9 9a f0 aa 96 a5 f1 9a 96 a5 f2 a5 a9 9a f5 9a 96 a5 f8 8a a5 a9 9a f8 96 a5 a9 9a f8 a9 9a 96 a5 f9 96 a5 a9 9a fa a9 9a 96 a5 fc 85 a9 9a 96 a5 fc 8a 96 a5 a9 9a fc 95 a9 9a 96 a5 fc aa 96 a5 a9 9a fd 9a 96 a5 a9 9a
 Out:   15a5a 2a5a5 5a5a5 a5a5a 15a5a5 2a5a5a 5a5a5a a5a5a5 15a5a5a 2a5a5a5 5a5a5a5 a5a5a5a 15a5a5a5 2a5a5a5a 5a5a5a5a
 
-Name:  get32b
-Run:   ../obj/lib/unicode-utf8-t get32
+Name:  utf8_32_get (2)
+Run:   ../obj/lib/unicode-t utf8_32_get
 In:    fe 83 81
 Out:   fffc fffc fffc
+
+Name:  utf16_be_put
+Run:   ../obj/lib/unicode-t utf16_be_put
+In:    0041 004a 2a5f feff 0000 10ffff ffff 10000
+Out:   00 41 00 4a 2a 5f fe ff 00 00 db ff df ff ff ff d8 00 dc 00
+
+Name:  utf16_le_put
+Run:   ../obj/lib/unicode-t utf16_le_put
+In:    0041 004a 2a5f feff 0000 10ffff ffff 10000
+Out:   41 00 4a 00 5f 2a ff fe 00 00 ff db ff df ff ff 00 d8 00 dc
+
+Name:  utf16_be_get (1)
+Run:   ../obj/lib/unicode-t utf16_be_get
+In:    00 41 00 4a 2a 5f fe ff 00 00 db ff df ff ff ff d8 00 dc 00
+Out:   0041 004a 2a5f feff 0000 10ffff ffff 10000
+
+Name:  utf16_be_get (2)
+Run:   ../obj/lib/unicode-t utf16_be_get
+In:    dc 1a 2a 5f d8 01 d8 01 2a 5f d8 01
+Out:   fffc 2a5f fffc fffc 2a5f fffc
+
+Name:  utf16_le_get (1)
+Run:   ../obj/lib/unicode-t utf16_le_get
+In:    41 00 4a 00 5f 2a ff fe 00 00 ff db ff df ff ff 00 d8 00 dc
+Out:   0041 004a 2a5f feff 0000 10ffff ffff 10000
+
+Name:  utf16_le_get (2)
+Run:   ../obj/lib/unicode-t utf16_le_get
+In:    1a dc 5f 2a 01 d8 01 d8 5f 2a 01 d8
+Out:   fffc 2a5f fffc fffc 2a5f fffc