UCW: Added some ff-unicode test cases.

diff --git a/lib/Makefile b/lib/Makefile
index f579a34399e65ab0ef9470a466268b3b26995d6f..432a14ec97456998a3dd6062c77e2245b47842bb 100644 (file)
--- a/lib/Makefile
+++ b/lib/Makefile
@@ -100,7 +100,7 @@ $(o)/lib/kmp-test: $(o)/lib/kmp-test.o $(LIBUCW) $(LIBCHARSET)
 $(o)/lib/ipaccess-test: $(o)/lib/ipaccess-test.o $(LIBUCW)
 
 TESTS+=$(addprefix $(o)/lib/,regex.test unicode.test hash-test.test mempool.test stkstring.test \
-    slists.test kmp-test.test bbuf.test getopt.test fastbuf.test eltpool.test)
+    slists.test kmp-test.test bbuf.test getopt.test fastbuf.test ff-unicode.test eltpool.test)
 
 $(o)/lib/regex.test: $(o)/lib/regex-t
 $(o)/lib/unicode.test: $(o)/lib/unicode-t
@@ -113,6 +113,7 @@ $(o)/lib/kmp-test.test: $(o)/lib/kmp-test
 $(o)/lib/bbuf.test: $(o)/lib/bbuf-t
 $(o)/lib/getopt.test: $(o)/lib/getopt-t
 $(o)/lib/fastbuf.test: $(o)/lib/fb-file-t $(o)/lib/fb-grow-t $(o)/lib/fb-pool-t
+$(o)/lib/ff-unicode.test: $(o)/lib/ff-unicode-t
 $(o)/lib/eltpool.test: $(o)/lib/eltpool-t
 
 ifdef CONFIG_UCW_THREADS

diff --git a/lib/ff-unicode.c b/lib/ff-unicode.c
index 280b3b75d319111b51408c9c6d022899b14264b4..6057e2408dd0c3b9328079d2089dc3077fe97dcd 100644 (file)
--- a/lib/ff-unicode.c
+++ b/lib/ff-unicode.c
@@ -238,3 +238,109 @@ bput_utf16_le_slow(struct fastbuf *b, uns u)
   else
     ASSERT(0);
 }
+
+#ifdef TEST
+
+#include <stdlib.h>
+#include <stdio.h>
+
+int main(int argc, char **argv)
+{
+#define FUNCS \
+  F(BGET_UTF8) F(BGET_UTF8_32) F(BGET_UTF16_BE) F(BGET_UTF16_LE) \
+  F(BPUT_UTF8) F(BPUT_UTF8_32) F(BPUT_UTF16_BE) F(BPUT_UTF16_LE)
+
+  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;
+    }
+
+  struct fastbuf *b = fbgrow_create(8);
+  if (func < FUNC_BPUT_UTF8)
+    {
+      uns u;
+      while (scanf("%x", &u) == 1)
+       bputc(b, u);
+      fbgrow_rewind(b);
+      while (bpeekc(b) >= 0)
+        {
+         if (btell(b))
+           putchar(' ');
+         switch (func)
+           {
+             case FUNC_BGET_UTF8:
+               u = bget_utf8_slow(b, UNI_REPLACEMENT);
+               break;
+             case FUNC_BGET_UTF8_32:
+               u = bget_utf8_32_slow(b, UNI_REPLACEMENT);
+               break;
+             case FUNC_BGET_UTF16_BE:
+               u = bget_utf16_be_slow(b, UNI_REPLACEMENT);
+               break;
+             case FUNC_BGET_UTF16_LE:
+               u = bget_utf16_le_slow(b, UNI_REPLACEMENT);
+               break;
+             default:
+               ASSERT(0);
+           }
+         printf("%04x", u);
+       }
+      putchar('\n');
+    }
+  else
+    {
+      uns u, i = 0;
+      while (scanf("%x", &u) == 1)
+        {
+         switch (func)
+           {
+             case FUNC_BPUT_UTF8:
+               bput_utf8_slow(b, u);
+               break;
+             case FUNC_BPUT_UTF8_32:
+               bput_utf8_32_slow(b, u);
+               break;
+             case FUNC_BPUT_UTF16_BE:
+                bput_utf16_be_slow(b, u);
+               break;
+             case FUNC_BPUT_UTF16_LE:
+               bput_utf16_le_slow(b, u);
+               break;
+             default:
+               ASSERT(0);
+           }
+         fbgrow_rewind(b);
+         u = 0;
+         while (bpeekc(b) >= 0)
+           {
+             if (i++)
+               putchar(' ');
+             printf("%02x", bgetc(b));
+           }
+         fbgrow_reset(b);
+       }
+      putchar('\n');
+    }
+  bclose(b);
+
+  return 0;
+}
+
+#endif

diff --git a/lib/ff-unicode.t b/lib/ff-unicode.t
new file mode 100644 (file)
index 0000000..2a8b3df
--- /dev/null
+++ b/lib/ff-unicode.t
@@ -0,0 +1,41 @@
+# Tests for the Unicode module
+
+Name:  bput_utf8
+Run:   ../obj/lib/ff-unicode-t bput_utf8
+In:    0041 0048 004f 004a
+Out:   41 48 4f 4a
+
+Name:   bget_utf8_32
+Run:    ../obj/lib/ff-unicode-t bget_utf8_32
+In:     fe 83 81
+Out:    fffc
+
+Name:   bput_utf16_be
+Run:    ../obj/lib/ff-unicode-t bput_utf16_be
+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:   bput_utf16_le
+Run:    ../obj/lib/ff-unicode-t bput_utf16_le
+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:   bget_utf16_be (1)
+Run:    ../obj/lib/ff-unicode-t bget_utf16_be
+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:   bget_utf16_be (2)
+Run:    ../obj/lib/ff-unicode-t bget_utf16_be
+In:     dc 1a 2a 5f d8 01 d8 01 2a 5f d8 01
+Out:    fffc 2a5f fffc 2a5f fffc
+
+Name:   bget_utf16_le (1)
+Run:    ../obj/lib/ff-unicode-t bget_utf16_le
+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:   bget_utf16_le (2)
+Run:    ../obj/lib/ff-unicode-t bget_utf16_le
+In:     1a dc 5f 2a 01 d8 01 d8 5f 2a 01 d8
+Out:    fffc 2a5f fffc 2a5f fffc