Hi,
I can't input multibyte characters on Win32 console if 'encoding' is
different from current codepage. (e.g. :set enc=utf-8 tenc=cp932)
The encoding of typed characters is not properly converted from
'termencoding' to 'encoding'.
There are two problems in the current implementation.
1. mch_inchar() in os_win32.c uses convert_input() to convert typed characters,
but convert_input() discards the input if it is incomplete.
convert_input_safe() should be used instead of convert_input().
2. string_convert_ext() in mbyte.c does not return an error even if the input
is incomplete (e.g. only a leading byte of multibyte character is contained)
when CONV_CODEPAGE is used.
Attached patch fixes these problem.
For the first problem, mch_inchar() uses convert_input_safe() as
fill_input_buf() in ui.c.
For the second problem, MB_ERR_INVALID_CHARS flag is specified for
MultiByteToWideChar() to detect incomplete inputs if string_convert_ext()
is called from convert_input_safe().
Best regards,
Ken Takata
--
You received this message from the "vim_dev" maillist.
Do not top-post! Type your reply below the text you are replying to.
For more information, visit http://www.vim.org/maillist.php
# HG changeset patch
# Parent 04592728474a35794ae6397a0b90f2989864fb66
diff --git a/src/mbyte.c b/src/mbyte.c
--- a/src/mbyte.c
+++ b/src/mbyte.c
@@ -6205,8 +6205,23 @@
if (vcp->vc_cpfrom == 0)
tmp_len = utf8_to_utf16(ptr, len, NULL, NULL);
else
- tmp_len = MultiByteToWideChar(vcp->vc_cpfrom, 0,
- ptr, len, 0, 0);
+ {
+ tmp_len = MultiByteToWideChar(vcp->vc_cpfrom,
+ unconvlenp ? MB_ERR_INVALID_CHARS : 0,
+ ptr, len, 0, 0);
+ if (tmp_len == 0 &&
+ GetLastError() == ERROR_NO_UNICODE_TRANSLATION)
+ {
+ if (lenp != NULL)
+ *lenp = 0;
+ if (unconvlenp != NULL)
+ *unconvlenp = len;
+ retval = alloc(1);
+ if (retval)
+ retval[0] = NUL;
+ return retval;
+ }
+ }
tmp = (short_u *)alloc(sizeof(short_u) * tmp_len);
if (tmp == NULL)
break;
diff --git a/src/os_win32.c b/src/os_win32.c
--- a/src/os_win32.c
+++ b/src/os_win32.c
@@ -1431,6 +1431,11 @@
#define TYPEAHEADLEN 20
static char_u typeahead[TYPEAHEADLEN]; /* previously typed bytes. */
static int typeaheadlen = 0;
+#ifdef FEAT_MBYTE
+ static char_u *rest = NULL; /* unconverted rest of previous read */
+ static int restlen = 0;
+ int unconverted;
+#endif
/* First use any typeahead that was kept because "buf" was too small. */
if (typeaheadlen > 0)
@@ -1534,6 +1539,33 @@
c = tgetch(&modifiers, &ch2);
+# ifdef FEAT_MBYTE
+ /* stolen from fill_input_buf() in ui.c */
+ if (rest != NULL)
+ {
+ /* Use remainder of previous call, starts with an invalid
+ * character that may become valid when reading more. */
+ if (restlen > TYPEAHEADLEN - typeaheadlen)
+ unconverted = TYPEAHEADLEN - typeaheadlen;
+ else
+ unconverted = restlen;
+ mch_memmove(typeahead + typeaheadlen, rest, unconverted);
+ if (unconverted == restlen)
+ {
+ vim_free(rest);
+ rest = NULL;
+ }
+ else
+ {
+ restlen -= unconverted;
+ mch_memmove(rest, rest + unconverted, restlen);
+ }
+ typeaheadlen += unconverted;
+ }
+ else
+ unconverted = 0;
+#endif
+
if (typebuf_changed(tb_change_cnt))
{
/* "buf" may be invalid now if a client put something in the
@@ -1569,8 +1601,12 @@
* when 'tenc' is set. */
if (input_conv.vc_type != CONV_NONE
&& (ch2 == NUL || c != K_NUL))
- n = convert_input(typeahead + typeaheadlen, n,
- TYPEAHEADLEN - typeaheadlen);
+ {
+ typeaheadlen -= unconverted;
+ n = convert_input_safe(typeahead + typeaheadlen,
+ n + unconverted, TYPEAHEADLEN - typeaheadlen,
+ rest == NULL ? &rest : NULL, &restlen);
+ }
#endif
/* Use the ALT key to set the 8th bit of the character
