From 0f10d2ed312115998d5ce1dc88a8d9207c9e4959 Mon Sep 17 00:00:00 2001 From: Jeremy Allison Date: Thu, 21 Jun 2007 17:25:13 +0000 Subject: r23572: Ensure we obey Unicode consortium restrictions. Code based on patch from MORIYAMA Masayuki . Jeremy. --- source/lib/iconv.c | 37 ++++++++++++++++++++++--------------- 1 file changed, 22 insertions(+), 15 deletions(-) (limited to 'source/lib/iconv.c') diff --git a/source/lib/iconv.c b/source/lib/iconv.c index 6e040b77f17..90e2faab6fb 100644 --- a/source/lib/iconv.c +++ b/source/lib/iconv.c @@ -544,6 +544,8 @@ static size_t utf8_pull(void *cd, const char **inbuf, size_t *inbytesleft, uint8 *uc = (uint8 *)*outbuf; while (in_left >= 1 && out_left >= 2) { + unsigned int codepoint; + if ((c[0] & 0x80) == 0) { uc[0] = c[0]; uc[1] = 0; @@ -560,8 +562,14 @@ static size_t utf8_pull(void *cd, const char **inbuf, size_t *inbytesleft, errno = EILSEQ; goto error; } - uc[1] = (c[0]>>2) & 0x7; - uc[0] = (c[0]<<6) | (c[1]&0x3f); + codepoint = (c[1]&0x3f) | ((c[0]&0x1f)<<6); + if (codepoint < 0x80) { + /* don't accept UTF-8 characters that are not minimally packed */ + errno = EILSEQ; + goto error; + } + uc[1] = codepoint >> 8; + uc[0] = codepoint & 0xff; c += 2; in_left -= 2; out_left -= 2; @@ -576,8 +584,14 @@ static size_t utf8_pull(void *cd, const char **inbuf, size_t *inbytesleft, errno = EILSEQ; goto error; } - uc[1] = ((c[0]&0xF)<<4) | ((c[1]>>2)&0xF); - uc[0] = (c[1]<<6) | (c[2]&0x3f); + codepoint = (c[2]&0x3f) | ((c[1]&0x3f)<<6) | ((c[0]&0xf)<<12); + if (codepoint < 0x800) { + /* don't accept UTF-8 characters that are not minimally packed */ + errno = EILSEQ; + goto error; + } + uc[1] = codepoint >> 8; + uc[0] = codepoint & 0xff; c += 3; in_left -= 3; out_left -= 2; @@ -586,7 +600,6 @@ static size_t utf8_pull(void *cd, const char **inbuf, size_t *inbytesleft, } if ((c[0] & 0xf8) == 0xf0) { - unsigned int codepoint; if (in_left < 4 || (c[1] & 0xc0) != 0x80 || (c[2] & 0xc0) != 0x80 || @@ -599,16 +612,10 @@ static size_t utf8_pull(void *cd, const char **inbuf, size_t *inbytesleft, ((c[2]&0x3f)<<6) | ((c[1]&0x3f)<<12) | ((c[0]&0x7)<<18); - if (codepoint < 0x10000) { - /* accept UTF-8 characters that are not - minimally packed, but pack the result */ - uc[0] = (codepoint & 0xFF); - uc[1] = (codepoint >> 8); - c += 4; - in_left -= 4; - out_left -= 2; - uc += 2; - continue; + if (codepoint < 0x10000 || codepoint > 0x10ffff) { + /* don't accept UTF-8 characters that are not minimally packed */ + errno = EILSEQ; + goto error; } codepoint -= 0x10000; -- cgit