Check behaviour when converting from UTF-7.

diff --git a/ChangeLog b/ChangeLog
index cc9e8a54bb673629cbdff7aec75eb676c0008916..3c7252160e2a1c3939cf9fba2043f33b838e5eb2 100644 (file)
--- a/ChangeLog
+++ b/ChangeLog
@@ -1,5 +1,8 @@
 2008-09-07  Bruno Haible  <bruno@clisp.org>
 
+       * tests/test-striconveh.c (main): Check behaviour when converting from
+       UTF-7.
+
        Make striconveh work better with stateful encodings.
        * lib/striconveh.c (iconv_carefully, iconv_carefully_1): Don't assume
        that iconv does not increment the inptr when returning -1/EINVAL.

diff --git a/tests/test-striconveh.c b/tests/test-striconveh.c
index 2557baa48ea2200186bf69da69c3e568c9a2a7c3..0f9e22a52a14c91fa6e06440baf6daf6c3b45e81 100644 (file)
--- a/tests/test-striconveh.c
+++ b/tests/test-striconveh.c
@@ -71,6 +71,7 @@ main ()
   iconv_t cd_utf8_to_88591 = iconv_open ("ISO-8859-1", "UTF-8");
   iconv_t cd_88592_to_utf8 = iconv_open ("UTF-8", "ISO-8859-2");
   iconv_t cd_utf8_to_88592 = iconv_open ("ISO-8859-2", "UTF-8");
+  iconv_t cd_utf7_to_utf8 = iconv_open ("UTF-8", "UTF-7");
 
   ASSERT (cd_88591_to_utf8 != (iconv_t)(-1));
   ASSERT (cd_utf8_to_88591 != (iconv_t)(-1));
@@ -335,6 +336,79 @@ main ()
        }
     }
 
+  if (cd_utf7_to_utf8 != (iconv_t)(-1))
+    {
+      /* Disabled on Solaris, because Solaris 9 iconv() is buggy: it returns
+        -1 / EILSEQ when converting the 7th byte of the input "+VDLYP9hA".  */
+# if !(defined __sun && !defined _LIBICONV_VERSION)
+      /* Test conversion from UTF-7 to UTF-8 with EINVAL.  */
+      for (h = 0; h < SIZEOF (handlers); h++)
+       {
+         enum iconv_ilseq_handler handler = handlers[h];
+         /* This is base64 encoded 0x54 0x32 0xD8 0x3F 0xD8 0x40.  It would
+            convert to U+5432 U+D83F U+D840 but these are Unicode surrogates.  */
+         static const char input[] = "+VDLYP9hA";
+         static const char expected1[] = "\345\220\262"; /* 吲 glibc */
+         static const char expected2[] = ""; /* libiconv */
+         char *result = NULL;
+         size_t length = 0;
+         int retval = mem_cd_iconveh (input, 7,
+                                      cd_utf7_to_utf8,
+                                      cd_utf7_to_utf8, (iconv_t)(-1),
+                                      handler,
+                                      NULL,
+                                      &result, &length);
+         ASSERT (retval == 0);
+         ASSERT (length == strlen (expected1) || length == strlen (expected2));
+         ASSERT (result != NULL);
+         if (length == strlen (expected1))
+           ASSERT (memcmp (result, expected1, strlen (expected1)) == 0);
+         else
+           ASSERT (memcmp (result, expected2, strlen (expected2)) == 0);
+         free (result);
+       }
+
+      /* Test conversion from UTF-7 to UTF-8 with EILSEQ.  */
+      for (h = 0; h < SIZEOF (handlers); h++)
+       {
+         enum iconv_ilseq_handler handler = handlers[h];
+         /* This is base64 encoded 0xD8 0x3F 0xD8 0x40 0xD8 0x41.  It would
+            convert to U+D83F U+D840 U+D841 but these are Unicode surrogates.  */
+         static const char input[] = "+2D/YQNhB";
+         char *result = NULL;
+         size_t length = 0;
+         int retval = mem_cd_iconveh (input, strlen (input),
+                                      cd_utf7_to_utf8,
+                                      cd_utf7_to_utf8, (iconv_t)(-1),
+                                      handler,
+                                      NULL,
+                                      &result, &length);
+         switch (handler)
+           {
+           case iconveh_error:
+             ASSERT (retval == -1 && errno == EILSEQ);
+             ASSERT (result == NULL);
+             break;
+           case iconveh_question_mark:
+           case iconveh_escape_sequence:
+             {
+               static const char expected1[] = "?????"; /* glibc */
+               static const char expected2[] = "?2D/YQNhB"; /* libiconv */
+               ASSERT (retval == 0);
+               ASSERT (length == strlen (expected1) || length == strlen (expected2));
+               ASSERT (result != NULL);
+               if (length == strlen (expected1))
+                 ASSERT (memcmp (result, expected1, strlen (expected1)) == 0);
+               else
+                 ASSERT (memcmp (result, expected2, strlen (expected2)) == 0);
+               free (result);
+             }
+             break;
+           }
+       }
+# endif
+    }
+
   /* ------------------------ Test str_cd_iconveh() ------------------------ */
 
   /* Test conversion from ISO-8859-2 to ISO-8859-1 with no errors.  */