Skip to content

Commit e1de3e0

Browse files
tatsuo-ishiitatsuo-ishii
committed
Implement following item in TODO:
* Reject character sequences those are not valid in their charset
1 parent d7f3cbc commit e1de3e0

File tree

3 files changed

+125
-56
lines changed

3 files changed

+125
-56
lines changed

‎src/backend/utils/mb/conv.c‎

Lines changed: 36 additions & 36 deletions
Original file line numberDiff line numberDiff line change
@@ -6,7 +6,7 @@
66
* WIN1250 client encoding support contributed by Pavel Behal
77
* SJIS UDC (NEC selection IBM kanji) support contributed by Eiji Tokuya
88
*
9-
* $Id: conv.c,v 1.27 2001/09/06 04:57:29 ishii Exp $
9+
* $Id: conv.c,v 1.28 2001/09/11 04:50:36 ishii Exp $
1010
*
1111
*
1212
*/
@@ -1633,48 +1633,48 @@ big5_to_utf(unsigned char *euc, unsigned char *utf, int len)
16331633
*/
16341634
pg_enconv pg_enconv_tbl[] =
16351635
{
1636-
{ PG_SQL_ASCII, ascii2mic, mic2ascii, ascii2utf, utf2ascii },
1637-
{ PG_EUC_JP, euc_jp2mic, mic2euc_jp, euc_jp_to_utf, utf_to_euc_jp },
1638-
{ PG_EUC_CN, euc_cn2mic, mic2euc_cn, euc_cn_to_utf, utf_to_euc_cn },
1639-
{ PG_EUC_KR, euc_kr2mic, mic2euc_kr, euc_kr_to_utf, utf_to_euc_kr },
1640-
{ PG_EUC_TW, euc_tw2mic, mic2euc_tw, euc_tw_to_utf, utf_to_euc_tw },
1641-
{ PG_UTF8, 0, 0, 0, 0 },
1642-
{ PG_MULE_INTERNAL, 0, 0, 0, 0 },
1643-
{ PG_LATIN1, latin12mic, mic2latin1, latin1_to_utf, utf_to_latin1 },
1644-
{ PG_LATIN2, latin22mic, mic2latin2, latin2_to_utf, utf_to_latin2 },
1645-
{ PG_LATIN3, latin32mic, mic2latin3, latin3_to_utf, utf_to_latin3 },
1646-
{ PG_LATIN4, latin42mic, mic2latin4, latin4_to_utf, utf_to_latin4 },
1647-
{ PG_LATIN5, iso2mic, mic2iso, latin5_to_utf, utf_to_latin5 },
1648-
{ PG_KOI8R, koi8r2mic, mic2koi8r, KOI8R_to_utf, utf_to_KOI8R },
1649-
{ PG_WIN1251, win12512mic, mic2win1251, WIN1251_to_utf, utf_to_WIN1251 },
1650-
{ PG_ALT, alt2mic, mic2alt, ALT_to_utf, utf_to_ALT },
1651-
{ PG_SJIS, sjis2mic, mic2sjis, sjis_to_utf, utf_to_sjis },
1636+
{ PG_SQL_ASCII, ascii2mic, mic2ascii, ascii2utf, utf2ascii},
1637+
{ PG_EUC_JP, euc_jp2mic, mic2euc_jp, euc_jp_to_utf, utf_to_euc_jp},
1638+
{ PG_EUC_CN, euc_cn2mic, mic2euc_cn, euc_cn_to_utf, utf_to_euc_cn},
1639+
{ PG_EUC_KR, euc_kr2mic, mic2euc_kr, euc_kr_to_utf, utf_to_euc_kr},
1640+
{ PG_EUC_TW, euc_tw2mic, mic2euc_tw, euc_tw_to_utf, utf_to_euc_tw},
1641+
{ PG_UTF8, 0, 0, 0, 0},
1642+
{ PG_MULE_INTERNAL, 0, 0, 0, 0},
1643+
{ PG_LATIN1, latin12mic, mic2latin1, latin1_to_utf, utf_to_latin1},
1644+
{ PG_LATIN2, latin22mic, mic2latin2, latin2_to_utf, utf_to_latin2},
1645+
{ PG_LATIN3, latin32mic, mic2latin3, latin3_to_utf, utf_to_latin3},
1646+
{ PG_LATIN4, latin42mic, mic2latin4, latin4_to_utf, utf_to_latin4},
1647+
{ PG_LATIN5, iso2mic, mic2iso, latin5_to_utf, utf_to_latin5},
1648+
{ PG_KOI8R, koi8r2mic, mic2koi8r, KOI8R_to_utf, utf_to_KOI8R},
1649+
{ PG_WIN1251, win12512mic, mic2win1251, WIN1251_to_utf, utf_to_WIN1251},
1650+
{ PG_ALT, alt2mic, mic2alt, ALT_to_utf, utf_to_ALT},
1651+
{ PG_SJIS, sjis2mic, mic2sjis, sjis_to_utf, utf_to_sjis},
16521652
{ PG_BIG5, big52mic, mic2big5, big5_to_utf, utf_to_big5},
1653-
{ PG_WIN1250, win12502mic, mic2win1250, 0, 0 },
1653+
{ PG_WIN1250, win12502mic, mic2win1250, 0, 0},
16541654
};
16551655

16561656
#else
16571657

16581658
pg_enconv pg_enconv_tbl[] =
16591659
{
1660-
{ PG_SQL_ASCII, ascii2mic, mic2ascii, 0, 0 },
1661-
{ PG_EUC_JP, euc_jp2mic, mic2euc_jp, 0, 0 },
1662-
{ PG_EUC_CN, euc_cn2mic, mic2euc_cn, 0, 0 },
1663-
{ PG_EUC_KR, euc_kr2mic, mic2euc_kr, 0, 0 },
1664-
{ PG_EUC_TW, euc_tw2mic, mic2euc_tw, 0, 0 },
1665-
{ PG_UTF8, 0, 0, 0, 0 },
1666-
{ PG_MULE_INTERNAL, 0, 0, 0, 0 },
1667-
{ PG_LATIN1, latin12mic, mic2latin1, 0, 0 },
1668-
{ PG_LATIN2, latin22mic, mic2latin2, 0, 0 },
1669-
{ PG_LATIN3, latin32mic, mic2latin3, 0, 0 },
1670-
{ PG_LATIN4, latin42mic, mic2latin4, 0, 0 },
1671-
{ PG_LATIN5, iso2mic, mic2iso, 0, 0 },
1672-
{ PG_KOI8R, koi8r2mic, mic2koi8r, 0, 0 },
1673-
{ PG_WIN1251, win12512mic, mic2win1251, 0, 0 },
1674-
{ PG_ALT, alt2mic, mic2alt, 0, 0 },
1675-
{ PG_SJIS, sjis2mic, mic2sjis, 0, 0 },
1676-
{ PG_BIG5, big52mic, mic2big5, 0, 0 },
1677-
{ PG_WIN1250, win12502mic, mic2win1250, 0, 0 },
1660+
{ PG_SQL_ASCII, ascii2mic, mic2ascii, 0, 0},
1661+
{ PG_EUC_JP, euc_jp2mic, mic2euc_jp, 0, 0},
1662+
{ PG_EUC_CN, euc_cn2mic, mic2euc_cn, 0, 0},
1663+
{ PG_EUC_KR, euc_kr2mic, mic2euc_kr, 0, 0},
1664+
{ PG_EUC_TW, euc_tw2mic, mic2euc_tw, 0, 0},
1665+
{ PG_UTF8, 0, 0, 0, 0},
1666+
{ PG_MULE_INTERNAL, 0, 0, 0, 0},
1667+
{ PG_LATIN1, latin12mic, mic2latin1, 0, 0},
1668+
{ PG_LATIN2, latin22mic, mic2latin2, 0, 0},
1669+
{ PG_LATIN3, latin32mic, mic2latin3, 0, 0},
1670+
{ PG_LATIN4, latin42mic, mic2latin4, 0, 0},
1671+
{ PG_LATIN5, iso2mic, mic2iso, 0, 0},
1672+
{ PG_KOI8R, koi8r2mic, mic2koi8r, 0, 0},
1673+
{ PG_WIN1251, win12512mic, mic2win1251, 0, 0},
1674+
{ PG_ALT, alt2mic, mic2alt, 0, 0},
1675+
{ PG_SJIS, sjis2mic, mic2sjis, 0, 0},
1676+
{ PG_BIG5, big52mic, mic2big5, 0, 0},
1677+
{ PG_WIN1250, win12502mic, mic2win1250, 0, 0},
16781678
};
16791679

16801680
#endif /* UNICODE_CONVERSION */

‎src/backend/utils/mb/wchar.c‎

Lines changed: 84 additions & 19 deletions
Original file line numberDiff line numberDiff line change
@@ -1,7 +1,7 @@
11
/*
22
* conversion functions between pg_wchar and multi-byte streams.
33
* Tatsuo Ishii
4-
* $Id: wchar.c,v 1.19 2001/09/06 04:57:29 ishii Exp $
4+
* $Id: wchar.c,v 1.20 2001/09/11 04:50:36 ishii Exp $
55
*
66
* WIN1250 client encoding updated by Pavel Behal
77
*
@@ -458,24 +458,24 @@ pg_big5_mblen(const unsigned char *s)
458458
}
459459

460460
pg_wchar_tbl pg_wchar_table[] = {
461-
{pg_ascii2wchar_with_len, pg_ascii_mblen}, /* 0; PG_SQL_ASCII */
462-
{pg_eucjp2wchar_with_len, pg_eucjp_mblen}, /* 1; PG_EUC_JP */
463-
{pg_euccn2wchar_with_len, pg_euccn_mblen}, /* 2; PG_EUC_CN */
464-
{pg_euckr2wchar_with_len, pg_euckr_mblen}, /* 3; PG_EUC_KR */
465-
{pg_euctw2wchar_with_len, pg_euctw_mblen}, /* 4; PG_EUC_TW */
466-
{pg_utf2wchar_with_len, pg_utf_mblen}, /* 5; PG_UNICODE */
467-
{pg_mule2wchar_with_len, pg_mule_mblen}, /* 6; PG_MULE_INTERNAL */
468-
{pg_latin12wchar_with_len, pg_latin1_mblen}, /* 7; PG_LATIN1 */
469-
{pg_latin12wchar_with_len, pg_latin1_mblen}, /* 8; PG_LATIN2 */
470-
{pg_latin12wchar_with_len, pg_latin1_mblen}, /* 9; PG_LATIN3 */
471-
{pg_latin12wchar_with_len, pg_latin1_mblen}, /* 10; PG_LATIN4 */
472-
{pg_latin12wchar_with_len, pg_latin1_mblen}, /* 11; PG_LATIN5 */
473-
{pg_latin12wchar_with_len, pg_latin1_mblen}, /* 12; PG_KOI8 */
474-
{pg_latin12wchar_with_len, pg_latin1_mblen}, /* 13; PG_WIN1251 */
475-
{pg_latin12wchar_with_len, pg_latin1_mblen}, /* 14; PG_ALT */
476-
{0, pg_sjis_mblen}, /* 15; PG_SJIS */
477-
{0, pg_big5_mblen}, /* 17; PG_BIG5 */
478-
{pg_latin12wchar_with_len, pg_latin1_mblen} /* 18; PG_WIN1250 */
461+
{pg_ascii2wchar_with_len, pg_ascii_mblen, 1}, /* 0; PG_SQL_ASCII */
462+
{pg_eucjp2wchar_with_len, pg_eucjp_mblen, 3}, /* 1; PG_EUC_JP */
463+
{pg_euccn2wchar_with_len, pg_euccn_mblen, 3}, /* 2; PG_EUC_CN */
464+
{pg_euckr2wchar_with_len, pg_euckr_mblen, 3}, /* 3; PG_EUC_KR */
465+
{pg_euctw2wchar_with_len, pg_euctw_mblen, 3}, /* 4; PG_EUC_TW */
466+
{pg_utf2wchar_with_len, pg_utf_mblen, 3}, /* 5; PG_UNICODE */
467+
{pg_mule2wchar_with_len, pg_mule_mblen, 3}, /* 6; PG_MULE_INTERNAL */
468+
{pg_latin12wchar_with_len, pg_latin1_mblen, 1}, /* 7; PG_LATIN1 */
469+
{pg_latin12wchar_with_len, pg_latin1_mblen, 1}, /* 8; PG_LATIN2 */
470+
{pg_latin12wchar_with_len, pg_latin1_mblen, 1}, /* 9; PG_LATIN3 */
471+
{pg_latin12wchar_with_len, pg_latin1_mblen, 1}, /* 10; PG_LATIN4 */
472+
{pg_latin12wchar_with_len, pg_latin1_mblen, 1}, /* 11; PG_LATIN5 */
473+
{pg_latin12wchar_with_len, pg_latin1_mblen, 1}, /* 12; PG_KOI8 */
474+
{pg_latin12wchar_with_len, pg_latin1_mblen, 1}, /* 13; PG_WIN1251 */
475+
{pg_latin12wchar_with_len, pg_latin1_mblen, 1}, /* 14; PG_ALT */
476+
{0, pg_sjis_mblen, 2}, /* 15; PG_SJIS */
477+
{0, pg_big5_mblen, 2}, /* 17; PG_BIG5 */
478+
{pg_latin12wchar_with_len, pg_latin1_mblen, 1} /* 18; PG_WIN1250 */
479479
};
480480

481481
/* returns the byte length of a word for mule internal code */
@@ -498,3 +498,68 @@ pg_encoding_mblen(int encoding, const unsigned char *mbstr)
498498
((*pg_wchar_table[encoding].mblen) (mbstr)) :
499499
((*pg_wchar_table[PG_SQL_ASCII].mblen) (mbstr)));
500500
}
501+
502+
#ifndef FRONTEND
503+
/*
504+
* Verify mbstr to make sure that it has a valid character sequence.
505+
* mbstr is not necessarily NULL terminated. length of mbstr is
506+
* specified by len. If an error was found, returns an error message.
507+
* Note that the message is kept in a static buffer, the next invocation
508+
* might break the message.
509+
* If no error was found, this function returns NULL.
510+
*/
511+
char *
512+
pg_verifymbstr(const unsigned char *mbstr, int len)
513+
{
514+
int l;
515+
int i, j;
516+
static char buf[256];
517+
int slen = 0;
518+
519+
/* we do not check single byte encodings */
520+
if (pg_wchar_table[GetDatabaseEncoding()].maxmblen <= 1)
521+
return NULL;
522+
523+
while (len > 0 && *mbstr)
524+
{
525+
l = pg_mblen(mbstr);
526+
527+
/* multi-byte letter? */
528+
if (l > 1)
529+
{
530+
for (i=1;i<l;i++)
531+
{
532+
if (i > len || *(mbstr+i) == '\0' ||
533+
/* we assume that every muti-byte letter
534+
* consists of bytes being the 8th bit set
535+
*/
536+
((*(mbstr+i) & 0x80) == 0))
537+
{
538+
int remains = sizeof(buf);
539+
char *p = buf;
540+
541+
slen = snprintf(p, remains, "Invalid %s character sequence found (0x",
542+
GetDatabaseEncodingName());
543+
p += slen;
544+
remains -= slen;
545+
546+
i = ((*(mbstr+i) & 0x80) == 0)?l:i;
547+
548+
for (j=0;j<i;j++)
549+
{
550+
slen = snprintf(p, remains, "%02x",
551+
*(mbstr+j));
552+
p += slen;
553+
remains -= slen;
554+
}
555+
snprintf(p, remains, ")");
556+
return(buf);
557+
}
558+
}
559+
}
560+
len -= l;
561+
mbstr += l;
562+
}
563+
return NULL;
564+
}
565+
#endif

‎src/include/mb/pg_wchar.h‎

Lines changed: 5 additions & 1 deletion
Original file line numberDiff line numberDiff line change
@@ -1,4 +1,4 @@
1-
/* $Id: pg_wchar.h,v 1.29 2001/09/06 04:57:29 ishii Exp $ */
1+
/* $Id: pg_wchar.h,v 1.30 2001/09/11 04:50:36 ishii Exp $ */
22

33
#ifndef PG_WCHAR_H
44
#define PG_WCHAR_H
@@ -182,6 +182,8 @@ typedef struct
182182
int (*mb2wchar_with_len) (); /* convert a multi-byte
183183
* string to a wchar */
184184
int (*mblen) (); /* returns the length of a multi-byte word */
185+
int maxmblen; /* max bytes for a letter in this charset */
186+
185187
} pg_wchar_tbl;
186188

187189
extern pg_wchar_tbl pg_wchar_table[];
@@ -240,6 +242,8 @@ extern unsigned char *pg_server_to_client(unsigned char *, int);
240242
extern unsigned short BIG5toCNS(unsigned short, unsigned char *);
241243
extern unsigned short CNStoBIG5(unsigned short, unsigned char);
242244

245+
char *pg_verifymbstr(const unsigned char *, int);
246+
243247
#endif /* MULTIBYTE */
244248

245249
#endif /* PG_WCHAR_H */

0 commit comments

Comments
 (0)