aboutsummaryrefslogtreecommitdiff
diff options
context:
space:
mode:
authorKyle Evans <kevans@FreeBSD.org>2023-08-13 05:16:33 +0000
committerKyle Evans <kevans@FreeBSD.org>2023-08-13 05:16:33 +0000
commitacb089b983171667467adc66f56a723b609ed22e (patch)
tree114462653a60f601f6d215f0e18062bb0785e76f
parent022fad7dfc1d1e367dc331524ae91c74688a3098 (diff)
downloadsrc-vendor/NetBSD/libc-vis.tar.gz
src-vendor/NetBSD/libc-vis.zip
Vendor import of NetBSD's (un)vis(3) at 2023-08-13vendor/NetBSD/libc-vis/20230813vendor/NetBSD/libc-vis
-rw-r--r--unvis.324
-rw-r--r--unvis.c6
-rw-r--r--vis.34
-rw-r--r--vis.c116
-rw-r--r--vis.h4
5 files changed, 123 insertions, 31 deletions
diff --git a/unvis.3 b/unvis.3
index 0991f19cd3b8..6efe1982afbd 100644
--- a/unvis.3
+++ b/unvis.3
@@ -1,4 +1,4 @@
-.\" $NetBSD: unvis.3,v 1.29 2017/10/24 19:14:55 abhinav Exp $
+.\" $NetBSD: unvis.3,v 1.30 2019/05/08 15:37:41 bad Exp $
.\"
.\" Copyright (c) 1989, 1991, 1993
.\" The Regents of the University of California. All rights reserved.
@@ -29,7 +29,7 @@
.\"
.\" @(#)unvis.3 8.2 (Berkeley) 12/11/93
.\"
-.Dd March 12, 2011
+.Dd May 8, 2019
.Dt UNVIS 3
.Os
.Sh NAME
@@ -99,16 +99,18 @@ should be equal to the size of
.Pp
The
.Fn strunvisx
-function does the same as the
+and
+.Fn strnunvisx
+functions do the same as the
.Fn strunvis
-function,
-but it allows you to add a flag that specifies the style the string
+and
+.Fn strnunvis
+functions,
+but take a flag that specifies the style the string
.Ar src
is encoded with.
-Currently, the supported flags are:
-.Dv VIS_HTTPSTYLE
-and
-.Dv VIS_MIMESTYLE .
+The meaning of the flag is the same as explained below for
+.Fn unvis .
.Pp
The
.Fn unvis
@@ -157,6 +159,10 @@ The
.Fa flag
argument is also used to specify the encoding style of the source.
If set to
+.Dv VIS_NOESCAPE
+.Fn unvis
+will not decode backslash escapes.
+If set to
.Dv VIS_HTTPSTYLE
or
.Dv VIS_HTTP1808 ,
diff --git a/unvis.c b/unvis.c
index 30fad8154d75..039fea5e93f9 100644
--- a/unvis.c
+++ b/unvis.c
@@ -1,4 +1,4 @@
-/* $NetBSD: unvis.c,v 1.44 2014/09/26 15:43:36 roy Exp $ */
+/* $NetBSD: unvis.c,v 1.45 2022/04/19 20:32:15 rillig Exp $ */
/*-
* Copyright (c) 1989, 1993
@@ -34,7 +34,7 @@
#if 0
static char sccsid[] = "@(#)unvis.c 8.1 (Berkeley) 6/4/93";
#else
-__RCSID("$NetBSD: unvis.c,v 1.44 2014/09/26 15:43:36 roy Exp $");
+__RCSID("$NetBSD: unvis.c,v 1.45 2022/04/19 20:32:15 rillig Exp $");
#endif
#endif /* LIBC_SCCS and not lint */
@@ -505,7 +505,7 @@ strnunvisx(char *dst, size_t dlen, const char *src, int flag)
errno = ENOSPC; \
return -1; \
} \
- } while (/*CONSTCOND*/0)
+ } while (0)
while ((c = *src++) != '\0') {
again:
diff --git a/vis.3 b/vis.3
index 24cd2654d027..421f7d69827f 100644
--- a/vis.3
+++ b/vis.3
@@ -1,4 +1,4 @@
-.\" $NetBSD: vis.3,v 1.49 2017/08/05 20:22:29 wiz Exp $
+.\" $NetBSD: vis.3,v 1.50 2022/12/04 11:25:08 uwe Exp $
.\"
.\" Copyright (c) 1989, 1991, 1993
.\" The Regents of the University of California. All rights reserved.
@@ -179,7 +179,7 @@ and
functions return \-1 and set
.Va errno
to
-.Dv ENOSPC .
+.Er ENOSPC .
The
.Fn strenvisx
function takes an additional argument,
diff --git a/vis.c b/vis.c
index 3c965d5e77d9..6d77016272c6 100644
--- a/vis.c
+++ b/vis.c
@@ -1,4 +1,4 @@
-/* $NetBSD: vis.c,v 1.74 2017/11/27 16:37:21 christos Exp $ */
+/* $NetBSD: vis.c,v 1.83 2023/08/12 12:48:52 riastradh Exp $ */
/*-
* Copyright (c) 1989, 1993
@@ -57,7 +57,7 @@
#include <sys/cdefs.h>
#if defined(LIBC_SCCS) && !defined(lint)
-__RCSID("$NetBSD: vis.c,v 1.74 2017/11/27 16:37:21 christos Exp $");
+__RCSID("$NetBSD: vis.c,v 1.83 2023/08/12 12:48:52 riastradh Exp $");
#endif /* LIBC_SCCS and not lint */
#ifdef __FBSDID
__FBSDID("$FreeBSD$");
@@ -353,12 +353,15 @@ makeextralist(int flags, const char *src)
wchar_t *dst, *d;
size_t len;
const wchar_t *s;
+ mbstate_t mbstate;
len = strlen(src);
if ((dst = calloc(len + MAXEXTRAS, sizeof(*dst))) == NULL)
return NULL;
- if ((flags & VIS_NOLOCALE) || mbstowcs(dst, src, len) == (size_t)-1) {
+ memset(&mbstate, 0, sizeof(mbstate));
+ if ((flags & VIS_NOLOCALE)
+ || mbsrtowcs(dst, &src, len, &mbstate) == (size_t)-1) {
size_t i;
for (i = 0; i < len; i++)
dst[i] = (wchar_t)(u_char)src[i];
@@ -393,20 +396,23 @@ static int
istrsenvisx(char **mbdstp, size_t *dlen, const char *mbsrc, size_t mblength,
int flags, const char *mbextra, int *cerr_ptr)
{
+ char mbbuf[MB_LEN_MAX];
wchar_t *dst, *src, *pdst, *psrc, *start, *extra;
size_t len, olen;
uint64_t bmsk, wmsk;
wint_t c;
visfun_t f;
int clen = 0, cerr, error = -1, i, shft;
- char *mbdst, *mdst;
- ssize_t mbslength, maxolen;
+ char *mbdst, *mbwrite, *mdst;
+ size_t mbslength;
+ size_t maxolen;
+ mbstate_t mbstate;
_DIAGASSERT(mbdstp != NULL);
_DIAGASSERT(mbsrc != NULL || mblength == 0);
_DIAGASSERT(mbextra != NULL);
- mbslength = (ssize_t)mblength;
+ mbslength = mblength;
/*
* When inputing a single character, must also read in the
* next character for nextc, the look-ahead character.
@@ -427,6 +433,14 @@ istrsenvisx(char **mbdstp, size_t *dlen, const char *mbsrc, size_t mblength,
* return to the caller.
*/
+ /*
+ * Guarantee the arithmetic on input to calloc won't overflow.
+ */
+ if (mbslength > (SIZE_MAX - 1)/16) {
+ errno = ENOMEM;
+ return -1;
+ }
+
/* Allocate space for the wide char strings */
psrc = pdst = extra = NULL;
mdst = NULL;
@@ -458,10 +472,18 @@ istrsenvisx(char **mbdstp, size_t *dlen, const char *mbsrc, size_t mblength,
* stop at NULs because we may be processing a block of data
* that includes NULs.
*/
+ memset(&mbstate, 0, sizeof(mbstate));
while (mbslength > 0) {
/* Convert one multibyte character to wchar_t. */
- if (!cerr)
- clen = mbtowc(src, mbsrc, MB_LEN_MAX);
+ if (!cerr) {
+ clen = mbrtowc(src, mbsrc,
+ (mbslength < MB_LEN_MAX
+ ? mbslength
+ : MB_LEN_MAX),
+ &mbstate);
+ assert(clen < 0 || (size_t)clen <= mbslength);
+ assert(clen <= MB_LEN_MAX);
+ }
if (cerr || clen < 0) {
/* Conversion error, process as a byte instead. */
*src = (wint_t)(u_char)*mbsrc;
@@ -475,6 +497,20 @@ istrsenvisx(char **mbdstp, size_t *dlen, const char *mbsrc, size_t mblength,
*/
clen = 1;
}
+ /*
+ * Let n := MIN(mbslength, MB_LEN_MAX). We have:
+ *
+ * mbslength >= 1
+ * mbrtowc(..., n, &mbstate) <= n,
+ * by the contract of mbrtowc
+ *
+ * clen is either
+ * (a) mbrtowc(..., n, &mbstate), in which case
+ * clen <= n <= mbslength; or
+ * (b) 1, in which case clen = 1 <= mbslength.
+ */
+ assert(clen > 0);
+ assert((size_t)clen <= mbslength);
/* Advance buffer character pointer. */
src++;
/* Advance input pointer by number of bytes read. */
@@ -532,11 +568,49 @@ istrsenvisx(char **mbdstp, size_t *dlen, const char *mbsrc, size_t mblength,
* output byte-by-byte here. Else use wctomb().
*/
len = wcslen(start);
- maxolen = dlen ? *dlen : (wcslen(start) * MB_LEN_MAX + 1);
+ if (dlen) {
+ maxolen = *dlen;
+ if (maxolen == 0) {
+ errno = ENOSPC;
+ goto out;
+ }
+ } else {
+ if (len > (SIZE_MAX - 1)/MB_LEN_MAX) {
+ errno = ENOSPC;
+ goto out;
+ }
+ maxolen = len*MB_LEN_MAX + 1;
+ }
olen = 0;
+ memset(&mbstate, 0, sizeof(mbstate));
for (dst = start; len > 0; len--) {
- if (!cerr)
- clen = wctomb(mbdst, *dst);
+ if (!cerr) {
+ /*
+ * If we have at least MB_CUR_MAX bytes in the buffer,
+ * we'll just do the conversion in-place into mbdst. We
+ * need to be a little more conservative when we get to
+ * the end of the buffer, as we may not have MB_CUR_MAX
+ * bytes but we may not need it.
+ */
+ if (maxolen - olen > MB_CUR_MAX)
+ mbwrite = mbdst;
+ else
+ mbwrite = mbbuf;
+ clen = wcrtomb(mbwrite, *dst, &mbstate);
+ if (clen > 0 && mbwrite != mbdst) {
+ /*
+ * Don't break past our output limit, noting
+ * that maxolen includes the nul terminator so
+ * we can't write past maxolen - 1 here.
+ */
+ if (olen + clen >= maxolen) {
+ errno = ENOSPC;
+ goto out;
+ }
+
+ memcpy(mbdst, mbwrite, clen);
+ }
+ }
if (cerr || clen < 0) {
/*
* Conversion error, process as a byte(s) instead.
@@ -551,16 +625,27 @@ istrsenvisx(char **mbdstp, size_t *dlen, const char *mbsrc, size_t mblength,
shft = i * NBBY;
bmsk = (uint64_t)0xffLL << shft;
wmsk |= bmsk;
- if ((*dst & wmsk) || i == 0)
+ if ((*dst & wmsk) || i == 0) {
+ if (olen + clen + 1 >= maxolen) {
+ errno = ENOSPC;
+ goto out;
+ }
+
mbdst[clen++] = (char)(
(uint64_t)(*dst & bmsk) >>
shft);
+ }
}
cerr = 1;
}
- /* If this character would exceed our output limit, stop. */
- if (olen + clen > (size_t)maxolen)
- break;
+
+ /*
+ * We'll be dereferencing mbdst[clen] after this to write the
+ * nul terminator; the above paths should have checked for a
+ * possible overflow already.
+ */
+ assert(olen + clen < maxolen);
+
/* Advance output pointer by number of bytes written. */
mbdst += clen;
/* Advance buffer character pointer. */
@@ -570,6 +655,7 @@ istrsenvisx(char **mbdstp, size_t *dlen, const char *mbsrc, size_t mblength,
}
/* Terminate the output string. */
+ assert(olen < maxolen);
*mbdst = '\0';
if (flags & VIS_NOLOCALE) {
diff --git a/vis.h b/vis.h
index 7fc880cbfad5..7a5322e58f76 100644
--- a/vis.h
+++ b/vis.h
@@ -1,4 +1,4 @@
-/* $NetBSD: vis.h,v 1.25 2017/04/23 01:57:36 christos Exp $ */
+/* $NetBSD: vis.h,v 1.26 2022/05/20 21:31:24 andvar Exp $ */
/*-
* Copyright (c) 1990, 1993
@@ -40,7 +40,7 @@
* to select alternate encoding format
*/
#define VIS_OCTAL 0x0001 /* use octal \ddd format */
-#define VIS_CSTYLE 0x0002 /* use \[nrft0..] where appropiate */
+#define VIS_CSTYLE 0x0002 /* use \[nrft0..] where appropriate */
/*
* to alter set of characters encoded (default is to encode all