On 05.04.2012 14:32, Bruno Haible wrote: > Hi Vladimir, > >> I'm not sure if previous time I sent with or without \0 bugfix. Resending > > > - The function __argp_get_display_len looks very similar to mbsnwidth(), > from module 'mbswidth'. Could you use that function? One of the gnulib > principles is to reuse code that is already in gnulib, where it makes sense. Remaining is the issue due to escape sequences. > - You have shown a test case as a Cyrillic string. But what is the C code > to make the behaviour explicit? Could you add this code to > tests/test-argp.c, > or create a new test file tests/test-argp-3.c? Done but the test is valid only for UTF-8 locales. Should I force some specific locale? It's impossible to make a test working in all locales since in case of e.g. ASCII we don't have such characters at all.
-- Regards Vladimir 'φ-coder/phcoder' Serbinenko
diff --git a/lib/argp-fmtstream.c b/lib/argp-fmtstream.c
index 70e3eb8..d47e75f 100644
--- a/lib/argp-fmtstream.c
+++ b/lib/argp-fmtstream.c
@@ -29,6 +29,7 @@
#include <errno.h>
#include <stdarg.h>
#include <ctype.h>
+#include <wchar.h>
#include "argp-fmtstream.h"
#include "argp-namefrob.h"
@@ -116,6 +117,70 @@ weak_alias (__argp_fmtstream_free, argp_fmtstream_free)
#endif
#endif
+
+/* Return the space needed to display the range
+ [beg, min (end, beg + strlen (beg))). */
+size_t
+__argp_get_display_width (const char *beg, const char *end)
+{
+ const char *ptr;
+ size_t r = 0;
+ mbstate_t ps;
+
+ memset (&ps, 0, sizeof (ps));
+
+ for (ptr = beg; *ptr && ptr < end; )
+ {
+ wchar_t wc;
+ size_t s;
+
+ s = mbrtowc (&wc, ptr, end - ptr, &ps);
+ if (s == (size_t) -1)
+ break;
+ if (s == (size_t) -2)
+ {
+ ptr++;
+ r++;
+ continue;
+ }
+ r += wcwidth (wc);
+ ptr += s;
+ }
+ return r;
+}
+
+/* Return the pointer to the first character that doesn't fit in l columns. */
+static inline const ptrdiff_t
+add_width (const char *ptr, const char *end, size_t l)
+{
+ mbstate_t ps;
+ const char *ptr0 = ptr;
+
+ memset (&ps, 0, sizeof (ps));
+
+ while (ptr < end)
+ {
+ wchar_t wc;
+ size_t s, k;
+
+ s = mbrtowc (&wc, ptr, end - ptr, &ps);
+ if (s == (size_t) -1)
+ break;
+ if (s == (size_t) -2)
+ {
+ s = 1;
+ k = 1;
+ }
+ else
+ k = wcwidth (wc);
+ if (k >= l)
+ break;
+ l -= k;
+ ptr += s;
+ }
+ return ptr - ptr0;
+}
+
/* Process FS's buffer so that line wrapping is done from POINT_OFFS to the
end of its buffer. This code is mostly from glibc stdio/linewrap.c. */
void
@@ -168,14 +233,15 @@ __argp_fmtstream_update (argp_fmtstream_t fs)
if (!nl)
{
+ size_t display_width = __argp_get_display_width (buf, fs->p);
/* The buffer ends in a partial line. */
- if (fs->point_col + len < fs->rmargin)
+ if (fs->point_col + display_width < fs->rmargin)
{
/* The remaining buffer text is a partial line and fits
within the maximum line width. Advance point for the
characters to be written and stop scanning. */
- fs->point_col += len;
+ fs->point_col += display_width;
break;
}
else
@@ -183,14 +249,18 @@ __argp_fmtstream_update (argp_fmtstream_t fs)
the end of the buffer. */
nl = fs->p;
}
- else if (fs->point_col + (nl - buf) < (ssize_t) fs->rmargin)
- {
- /* The buffer contains a full line that fits within the maximum
- line width. Reset point and scan the next line. */
- fs->point_col = 0;
- buf = nl + 1;
- continue;
- }
+ else
+ {
+ size_t display_width = __argp_get_display_width (buf, nl);
+ if (display_width < (ssize_t) fs->rmargin)
+ {
+ /* The buffer contains a full line that fits within the maximum
+ line width. Reset point and scan the next line. */
+ fs->point_col = 0;
+ buf = nl + 1;
+ continue;
+ }
+ }
/* This line is too long. */
r = fs->rmargin - 1;
@@ -226,7 +296,7 @@ __argp_fmtstream_update (argp_fmtstream_t fs)
char *p, *nextline;
int i;
- p = buf + (r + 1 - fs->point_col);
+ p = buf + add_width (buf, fs->p, (r + 1 - fs->point_col));
while (p >= buf && !isblank ((unsigned char) *p))
--p;
nextline = p + 1; /* This will begin the next line. */
@@ -244,7 +314,7 @@ __argp_fmtstream_update (argp_fmtstream_t fs)
{
/* A single word that is greater than the maximum line width.
Oh well. Put it on an overlong line by itself. */
- p = buf + (r + 1 - fs->point_col);
+ p = buf + add_width (buf, fs->p, (r + 1 - fs->point_col));
/* Find the end of the long word. */
if (p < nl)
do
@@ -278,7 +348,7 @@ __argp_fmtstream_update (argp_fmtstream_t fs)
&& fs->p > nextline)
{
/* The margin needs more blanks than we removed. */
- if (fs->end - fs->p > fs->wmargin + 1)
+ if (__argp_get_display_width (fs->p, fs->end) > fs->wmargin + 1)
/* Make some space for them. */
{
size_t mv = fs->p - nextline;
diff --git a/lib/argp-fmtstream.h b/lib/argp-fmtstream.h
index a891c3d..618e1f1 100644
--- a/lib/argp-fmtstream.h
+++ b/lib/argp-fmtstream.h
@@ -349,6 +349,9 @@ __argp_fmtstream_point (argp_fmtstream_t __fs)
#endif /* __OPTIMIZE__ */
+size_t
+__argp_get_display_width (const char *beg, const char *end);
+
#endif /* ARGP_FMTSTREAM_USE_LINEWRAP */
#endif /* argp-fmtstream.h */
diff --git a/lib/argp-help.c b/lib/argp-help.c
index a126acb..fe3db39 100644
--- a/lib/argp-help.c
+++ b/lib/argp-help.c
@@ -1451,7 +1451,7 @@ argp_args_usage (const struct argp *argp, const struct argp_state *state,
/* Manually do line wrapping so that it (probably) won't get wrapped at
any embedded spaces. */
- space (stream, 1 + nl - cp);
+ space (stream, 1 + __argp_get_display_width (cp, nl));
__argp_fmtstream_write (stream, cp, nl - cp);
}
diff --git a/tests/test-argp-2.sh b/tests/test-argp-2.sh
index 49eff9c..935cce5 100755
--- a/tests/test-argp-2.sh
+++ b/tests/test-argp-2.sh
@@ -33,10 +33,11 @@ func_compare() {
####
# Test --usage output
cat > $TMP <<EOT
-Usage: test-argp [-tvCSOlp?V] [-f FILE] [-r FILE] [-o[ARG]] [--test]
- [--file=FILE] [--input=FILE] [--read=FILE] [--verbose] [--cantiga]
- [--sonet] [--option] [--optional[=ARG]] [--limerick] [--poem]
- [--help] [--usage] [--version] ARGS...
+Usage: test-argp [-tvCSOlp?V] [-c ÐиÑиллиÑа] [-f FILE] [-r FILE] [-o[ARG]]
+ [--ÑÑÑÑкий=ÐиÑиллиÑа] [--test] [--file=FILE] [--input=FILE]
+ [--read=FILE] [--verbose] [--cantiga] [--sonet] [--option]
+ [--optional[=ARG]] [--limerick] [--poem] [--help] [--usage]
+ [--version] ARGS...
EOT
./test-argp$EXEEXT --usage | func_compare || ERR=1
@@ -45,10 +46,10 @@ EOT
# Test working usage-indent format
cat > $TMP <<EOT
-Usage: test-argp [-tvCSOlp?V] [-f FILE] [-r FILE] [-o[ARG]] [--test]
-[--file=FILE] [--input=FILE] [--read=FILE] [--verbose] [--cantiga] [--sonet]
-[--option] [--optional[=ARG]] [--limerick] [--poem] [--help] [--usage]
-[--version] ARGS...
+Usage: test-argp [-tvCSOlp?V] [-c ÐиÑиллиÑа] [-f FILE] [-r FILE] [-o[ARG]]
+[--ÑÑÑÑкий=ÐиÑиллиÑа] [--test] [--file=FILE] [--input=FILE] [--read=FILE]
+[--verbose] [--cantiga] [--sonet] [--option] [--optional[=ARG]] [--limerick]
+[--poem] [--help] [--usage] [--version] ARGS...
EOT
ARGP_HELP_FMT='usage-indent=0' ./test-argp$EXEEXT --usage | func_compare || ERR=1
@@ -60,6 +61,8 @@ Usage: test-argp [OPTION...] ARGS...
documentation string
Main options
+ -c, --ÑÑÑÑкий=ÐиÑиллиÑа ÐлиннÑй ÑекÑÑ, доÑÑаÑоÑно длиннÑй, ÑÑÐ¾Ð±Ñ Ð¿ÑиÑлоÑÑ
+ пеÑенеÑÑи его на вÑоÑÑÑ ÑÑÑокÑ
-t, --test
Option Group 1
diff --git a/tests/test-argp.c b/tests/test-argp.c
index 6ba55e9..1a3f7dc 100644
--- a/tests/test-argp.c
+++ b/tests/test-argp.c
@@ -22,6 +22,7 @@
#include <stdio.h>
#include <stdlib.h>
#include <string.h>
+#include <locale.h>
#if HAVE_STRINGS_H
# include <strings.h>
#endif
@@ -213,6 +214,8 @@ struct argp_child group2_1_child = {
static struct argp_option main_options[] = {
{ NULL, 0, NULL, 0, "Main options", 0 },
{ "test", 't', NULL, 0, NULL, 1 },
+ { "ÑÑÑÑкий", 'c', "ÐиÑиллиÑа", 0, "ÐлиннÑй ÑекÑÑ, доÑÑаÑоÑно длиннÑй, "
+ "ÑÑÐ¾Ð±Ñ Ð¿ÑиÑлоÑÑ Ð¿ÐµÑенеÑÑи его на вÑоÑÑÑ ÑÑÑокÑ", 1},
{ NULL, 0, NULL, 0, NULL, 0 }
};
@@ -459,6 +462,7 @@ main (int argc, char **argv)
struct argp_child argp_children[3], group1_children[2], group2_children[2];
test_fp *fun;
+ setlocale (LC_ALL, "");
set_program_name (argv[0]);
group1_children[0] = group1_1_child;
signature.asc
Description: OpenPGP digital signature
