multibyte.c

/*--------
 * Module :			multibyte.c
 *
 * Description:		New Mlutibyte related additional function.
 *
 *					Create 2001-03-03 Eiji Tokuya
 *					New Create 2001-09-16 Eiji Tokuya
 *--------
 */

#include "multibyte.h"
#include "misc.h"
#include "connection.h"
#include "pgapifunc.h"
#include <string.h>
#include <ctype.h>
#include <stdio.h>
#include <stdlib.h>
#ifndef	TRUE
#define	TRUE	1
#endif

static pg_CS CS_Table[] =
{
	{ "SQL_ASCII",	SQL_ASCII },
	{ "EUC_JP",	EUC_JP },
	{ "EUC_CN",	EUC_CN },
	{ "EUC_KR",	EUC_KR },
	{ "EUC_TW",	EUC_TW },
	{ "JOHAB",	JOHAB },	/* since 7.3 */
	{ "UTF8",	UTF8 },		/* since 7.2 */
	{ "MULE_INTERNAL",MULE_INTERNAL },
	{ "LATIN1",	LATIN1 },
	{ "LATIN2",	LATIN2 },
	{ "LATIN3",	LATIN3 },
	{ "LATIN4",	LATIN4 },
	{ "LATIN5",	LATIN5 },
	{ "LATIN6",	LATIN6 },
	{ "LATIN7",	LATIN7 },
	{ "LATIN8",	LATIN8 },
	{ "LATIN9",	LATIN9 },
	{ "LATIN10",	LATIN10 },
	{ "WIN1256",	WIN1256 },	/* Arabic since 7.3 */
	{ "WIN1258",	WIN1258 },	/* Vietnamese since 8.1 */
	{ "WIN866",	WIN866 },	/* since 8.1 */
	{ "WIN874",	WIN874 },	/* Thai since 7.3 */
	{ "KOI8",	KOI8R },
	{ "WIN1251",	WIN1251 },	/* Cyrillic */
	{ "WIN1252",	WIN1252 },	/* Western Europe since 8.1 */
	{ "ISO_8859_5", ISO_8859_5 },
	{ "ISO_8859_6", ISO_8859_6 },
	{ "ISO_8859_7", ISO_8859_7 },
	{ "ISO_8859_8", ISO_8859_8 },
	{ "WIN1250",	WIN1250 },	/* Central Europe */
	{ "WIN1253",	WIN1253 },	/* Greek since 8.2 */
	{ "WIN1254",	WIN1254 },	/* Turkish since 8.2 */
	{ "WIN1255",	WIN1255 },	/* Hebrew since 8.2 */
	{ "WIN1257",	WIN1257 },	/* Baltic(North Europe) since 8.2 */

	{ "EUC_JIS_2004", EUC_JIS_2004},	/* EUC for SHIFT-JIS-2004 Japanese, since 8.3 */
	{ "SJIS",	SJIS },
	{ "BIG5",	BIG5 },
	{ "GBK",	GBK },		/* since 7.3 */
	{ "UHC",	UHC },		/* since 7.3 */	
	{ "GB18030",	GB18030 },	/* since 7.3 */
	{ "SHIFT_JIS_2004", SHIFT_JIS_2004 },	/* SHIFT-JIS-2004 Japanese, standard JIS X 0213, since 8.3 */
	{ "OTHER",	OTHER }
};

static pg_CS CS_Alias[] =
{
	{ "UNICODE",	UTF8 },
	{ "TCVN",	WIN1258 },
	{ "ALT",	WIN866 },
	{ "WIN",	WIN1251 },
	{ "OTHER",	OTHER }
};

CSTR	OTHER_STRING = "OTHER";

int
pg_CS_code(const UCHAR *characterset_string)
{
	int i, c = -1;

	for(i = 0; CS_Table[i].code != OTHER; i++)
	{
		if (0 == stricmp(characterset_string, CS_Table[i].name))
		{
                       	c = CS_Table[i].code;
			break;
		}
	}
	if (c < 0)
	{
		for(i = 0; CS_Alias[i].code != OTHER; i++)
		{
			if (0 == stricmp(characterset_string, CS_Alias[i].name))
			{
                       		c = CS_Alias[i].code;
				break;
			}
		}
	}
	if (c < 0)
		c = OTHER;
	return (c);
}

UCHAR *check_client_encoding(const UCHAR *conn_settings)
{
	const UCHAR *cptr, *sptr = NULL;
	UCHAR	*rptr;
	BOOL	allowed_cmd = TRUE, in_quote = FALSE;
	int	step = 0;
	size_t	len = 0;

	for (cptr = conn_settings; *cptr; cptr++)
	{
		if (in_quote)
			if (LITERAL_QUOTE == *cptr)
			{
				in_quote = FALSE;
				continue;
			}
		if (';' == *cptr)
		{
			allowed_cmd = TRUE;
			step = 0;
			continue;
		}
		if (!allowed_cmd)
			continue;
		if (isspace(*cptr))
			continue;
		switch (step)
		{
			case 0:
				if (0 != strnicmp(cptr, "set", 3))
				{
					allowed_cmd = FALSE;
					continue;
				}
				step++;
				cptr += 3;
				break;
			case 1:
				if (0 != strnicmp(cptr, "client_encoding", 15))
				{
					allowed_cmd = FALSE;
					continue;
				}
				step++;
				cptr += 15;
				break;
			case 2:
				if (0 != strnicmp(cptr, "to", 2))
				{
					allowed_cmd = FALSE;
					continue;
				}
				step++;
				cptr += 2;
				break;
			case 3:
				if (LITERAL_QUOTE == *cptr)
				{
					cptr++;
					for (sptr = cptr; *cptr && *cptr != LITERAL_QUOTE; cptr++) ;
				}
				else
				{
					for (sptr = cptr; *cptr && !isspace(*cptr); cptr++) ;
				}
				len = cptr - sptr;
				step++;
				break;
		}
	}
	if (!sptr)
		return NULL;
	rptr = malloc(len + 1);
	memcpy(rptr, sptr, len);
	rptr[len] = '\0';
	mylog("extracted a client_encoding '%s' from conn_settings\n", rptr);
	return rptr;
}

const UCHAR *
pg_CS_name(int characterset_code)
{
	int i;
	for (i = 0; CS_Table[i].code != OTHER; i++)
	{
		if (CS_Table[i].code == characterset_code)
			return CS_Table[i].name;
	}
	return (OTHER_STRING);
}

static int
pg_mb_maxlen(characterset_code)
{
	switch (characterset_code)
	{
		case UTF8:
			return 6;
		case EUC_TW:
			return 4;
		case EUC_JIS_2004:
		case EUC_JP:
		case GB18030:
			return 3;
		case SHIFT_JIS_2004:
		case SJIS:
		case BIG5:
		case GBK:
		case UHC:
		case EUC_CN:
		case EUC_KR:
		case JOHAB:
			return 2;
		default:
			return 1;
	}
}

int
pg_CS_stat(int stat,unsigned int character,int characterset_code)
{
	if (character == 0)
		stat = 0;
	switch (characterset_code)
	{
		case UTF8:
			{
				if (stat < 2 &&
					character >= 0x80)
				{
					if (character >= 0xfc)
						stat = 6;
					else if (character >= 0xf8)
						stat = 5;
					else if (character >= 0xf0)
						stat = 4;
					else if (character >= 0xe0)
						stat = 3;
					else if (character >= 0xc0)
						stat = 2;
				}
				else if (stat >= 2 &&
					character > 0x7f)
					stat--;
				else
					stat=0;
			}
			break;
/* SHIFT_JIS_2004 Support. */
			case SHIFT_JIS_2004:
			{
				if (stat < 2 &&
					character >= 0x81 && character <= 0x9f)
					stat = 2;
				else if (stat < 2 &&
					character >= 0xe0 && character <= 0xef)
					stat = 2;
				else if (stat < 2 &&
					character >= 0xf0 && character <= 0xfc)
					stat = 2;
				else if (stat == 2)
					stat = 1;
				else
					stat = 0;
			}
			break;
/* Shift-JIS Support. */
			case SJIS:
			{
				if (stat < 2 &&
					character > 0x80 &&
					!(character > 0x9f &&
					character < 0xe0))
					stat = 2;
				else if (stat == 2)
					stat = 1;
				else
					stat = 0;
			}
			break;
/* Chinese Big5 Support. */
		case BIG5:
			{
				if (stat < 2 &&
					character > 0xA0)
					stat = 2;
				else if (stat == 2)
					stat = 1;
				else
					stat = 0;
			}
			break;
/* Chinese GBK Support. */
		case GBK:
			{
				if (stat < 2 &&
					character > 0x7F)
					stat = 2;
				else if (stat == 2)
					stat = 1;
				else
					stat = 0;
			}
			break;

/* Korian UHC Support. */
		case UHC:
			{
				if (stat < 2 &&
					character > 0x7F)
					stat = 2;
				else if (stat == 2)
					stat = 1;
				else
					stat = 0;
			}
			break;

		case EUC_JIS_2004:
			/* 0x8f is JIS X 0212 + JIS X 0213(2) 3 byte */
			/* 0x8e is JIS X 0201 2 byte */
			/* 0xa0-0xff is JIS X 0213(1) 2 byte */
		case EUC_JP:
			/* 0x8f is JIS X 0212 3 byte */
			/* 0x8e is JIS X 0201 2 byte */
			/* 0xa0-0xff is JIS X 0208 2 byte */
			{
				if (stat < 3 && 
					character == 0x8f)	/* JIS X 0212 */
					stat = 3;
				else
				if (stat != 2 && 
					(character == 0x8e ||
					character > 0xa0))	/* Half Katakana HighByte & Kanji HighByte */
					stat = 2;
				else if (stat == 2)
					stat = 1;
				else
					stat = 0;
			}
			break;

/* EUC_CN, EUC_KR, JOHAB Support */
		case EUC_CN:
		case EUC_KR:
		case JOHAB:
			{
				if (stat < 2 &&
					character > 0xa0)
					stat = 2;
				else if (stat == 2)
					stat = 1;
				else
					stat = 0;
			}
			break;
		case EUC_TW:
			{
				if (stat < 4 &&
					character == 0x8e)
					stat = 4;
				else if (stat == 4 &&
					character > 0xa0)
					stat = 3;
				else if ((stat == 3 ||
					stat < 2) &&
					character > 0xa0)
					stat = 2;
				else if (stat == 2)
					stat = 1;
				else
					stat = 0;
			}
			break;
			/*Chinese GB18030 support.Added by Bill Huang <bhuang@redhat.com> <bill_huanghb@ybb.ne.jp>*/
		case GB18030:
			{
				if (stat < 2 && character > 0x80)
					stat = 2;
				else if (stat == 2)
				{
					if (character >= 0x30 && character <= 0x39)
						stat = 3;
					else
						stat = 1;
				}
				else if (stat == 3)
				{
					if (character >= 0x30 && character <= 0x39)
						stat = 1;
					else
						stat = 3;
				}
				else
					stat = 0;
			}
			break;
		default:
			{
				stat = 0;
			}
			break;
	}
	return stat;
}


UCHAR *
pg_mbschr(int csc, const UCHAR *string, unsigned int character)
{
	int			mb_st = 0;
	const UCHAR *s, *rs = NULL;

	for(s = string; *s ; s++) 
	{
		mb_st = pg_CS_stat(mb_st, (UCHAR) *s, csc);
		if (mb_st == 0 && (*s == character))
		{
			rs = s;
			break;
		}
	}
	return ((UCHAR *) rs);
}

size_t
pg_mbslen(int csc, const UCHAR *string)
{
	UCHAR *s;
	size_t	len;
	int	cs_stat;
	for (len = 0, cs_stat = 0, s = (UCHAR *) string; *s != 0; s++)
	{
		cs_stat = pg_CS_stat(cs_stat,(unsigned int) *s, csc);
		if (cs_stat < 2)
			len++;
	}
	return len;
}

UCHAR *
pg_mbsinc(int csc, const UCHAR *current )
{
	int mb_stat = 0;
	if (*current != 0)
	{
		mb_stat = (int) pg_CS_stat(mb_stat, *current, csc);
		if (mb_stat == 0)
			mb_stat = 1;
		return ((UCHAR *) current + mb_stat);
	}
	else
		return NULL;
}

static char *
CC_lookup_cs_new(ConnectionClass *self)
{
	char		*encstr = NULL;
	QResultClass	*res;

	res = CC_send_query(self, "select pg_client_encoding()", NULL, IGNORE_ABORT_ON_CONN | ROLLBACK_ON_ERROR, NULL);
	if (QR_command_maybe_successful(res))
	{
		const char 	*enc = QR_get_value_backend_text(res, 0, 0);

		if (enc)
			encstr = strdup(enc);
	}
	QR_Destructor(res);
	return encstr;
}
static char *
CC_lookup_cs_old(ConnectionClass *self)
{
	char		*encstr = NULL;
	HSTMT		hstmt;
	RETCODE		result;

	result = PGAPI_AllocStmt(self, &hstmt, 0);
	if (!SQL_SUCCEEDED(result))
		return encstr;

	result = PGAPI_ExecDirect(hstmt, "Show Client_Encoding", SQL_NTS, 0);
	if (result == SQL_SUCCESS_WITH_INFO)
	{
		char sqlState[8], errormsg[128], enc[32];

		if (PGAPI_Error(NULL, NULL, hstmt, sqlState, NULL, errormsg,
			sizeof(errormsg), NULL) == SQL_SUCCESS &&
		    sscanf(errormsg, "%*s %*s %*s %*s %*s %s", enc) > 0)
			encstr = strdup(enc);
	}
	PGAPI_FreeStmt(hstmt, SQL_DROP);
	return encstr;
}

/*
 *	This function works under Windows or Unicode case only.
 *	Simply returns NULL under other OSs.
 */
const char * get_environment_encoding(const ConnectionClass *conn, const char *setenc, const char *currenc, BOOL bStartup)
{
	const char *wenc = NULL;
#ifdef	WIN32
	int	acp;
#endif /* WIN32 */

#ifdef	UNICODE_SUPPORT
	if (CC_is_in_unicode_driver(conn))
		return "UTF8";
#endif /* UNICODE_SUPPORT */
	if (setenc && stricmp(setenc, OTHER_STRING))
		return setenc;
	if (wenc = getenv("PGCLIENTENCODING"), NULL != wenc)
		return wenc;
#ifdef	WIN32
	acp = GetACP();
	if (acp >= 1251 && acp <= 1258)
	{
		if (bStartup ||
		    stricmp(currenc, "SQL_ASCII") == 0)
			return wenc;
	}
	switch (acp)
	{
		case 932:
			wenc = "SJIS";
			break;
		case 936:
			if (!bStartup && PG_VERSION_GT(conn, 7.2))
				wenc = "GBK";
			break;
		case 949:
			if (!bStartup && PG_VERSION_GT(conn, 7.2))  
				wenc = "UHC";
			break;
		case 950:
			wenc = "BIG5";
			break;
		case 1250:
			wenc = "WIN1250";
			break;
		case 1251:
			wenc = "WIN1251";
			break;
		case 1256:
			if (PG_VERSION_GE(conn, 7.3))
				wenc = "WIN1256";
			break;
		case 1252:
			if (strnicmp(currenc, "LATIN", 5) == 0)
				break;
			if (PG_VERSION_GE(conn, 8.1))
				wenc = "WIN1252";
			else
				wenc = "LATIN1";
			break;
		case 1258:
			if (PG_VERSION_GE(conn, 8.1))
				wenc = "WIN1258";
			break;
		case 1253:
			if (PG_VERSION_GE(conn, 8.2))
				wenc = "WIN1253";
			break;
		case 1254:
			if (PG_VERSION_GE(conn, 8.2))
				wenc = "WIN1254";
			break;
		case 1255:
			if (PG_VERSION_GE(conn, 8.2))
				wenc = "WIN1255";
			break;
		case 1257:
			if (PG_VERSION_GE(conn, 8.2))
				wenc = "WIN1257";
			break;
	}
#endif /* WIN32 */
	return wenc;
}

void
CC_lookup_characterset(ConnectionClass *self)
{
	char	*encspec = NULL, *currenc = NULL, *tencstr;
	CSTR func = "CC_lookup_characterset";

	mylog("%s: entering...\n", func);
	if (self->original_client_encoding)
		encspec = strdup(self->original_client_encoding);
	if (self->current_client_encoding)
		currenc = strdup(self->current_client_encoding);
	else if (PG_VERSION_LT(self, 7.2))
		currenc = CC_lookup_cs_old(self);
	else
		currenc = CC_lookup_cs_new(self);
	tencstr = encspec ? encspec : currenc;
	if (self->original_client_encoding)
	{
		if (stricmp(self->original_client_encoding, tencstr))
		{
			char msg[256];

			snprintf(msg, sizeof(msg), "The client_encoding '%s' was changed to '%s'", self->original_client_encoding, tencstr);
			CC_set_error(self, CONN_OPTION_VALUE_CHANGED, msg, func);
		}
		free(self->original_client_encoding);
	}
#ifndef	UNICODE_SUPPORT
	else
	{
		const char *wenc = get_environment_encoding(self, encspec, currenc, FALSE);
		if (wenc && (!tencstr || stricmp(tencstr, wenc)))
		{
			QResultClass	*res;
			char		query[64];
			int		errnum = CC_get_errornumber(self);
			BOOL		cmd_success;

			sprintf(query, "set client_encoding to '%s'", wenc);
			res = CC_send_query(self, query, NULL, IGNORE_ABORT_ON_CONN | ROLLBACK_ON_ERROR, NULL);
			cmd_success = QR_command_maybe_successful(res);
			QR_Destructor(res);
			CC_set_errornumber(self, errnum);
			if (cmd_success)
			{
				self->original_client_encoding = strdup(wenc);
				self->ccsc = pg_CS_code(self->original_client_encoding);
				if (encspec)
					free(encspec);
				if (currenc)
					free(currenc);
				return;
			}
		}
	}
#endif /* UNICODE_SUPPORT */
	if (tencstr)
	{
		self->original_client_encoding = tencstr;
		if (encspec && currenc)
			free(currenc);
		self->ccsc = pg_CS_code(tencstr);
		qlog("    [ Client encoding = '%s' (code = %d) ]\n", self->original_client_encoding, self->ccsc);
		if (self->ccsc < 0)
		{
			char msg[256];

			snprintf(msg, sizeof(msg), "would handle the encoding '%s' like ASCII", tencstr);
			CC_set_error(self, CONN_OPTION_VALUE_CHANGED, msg, func); 
		}
	}
	else
	{
		self->ccsc = SQL_ASCII;
		self->original_client_encoding = NULL;
	}
	self->mb_maxbyte_per_char = pg_mb_maxlen(self->ccsc);
}

void encoded_str_constr(encoded_str *encstr, int ccsc, const char *str)
{
	encstr->ccsc = ccsc;
	encstr->encstr = str;
	encstr->pos = -1;
	encstr->ccst = 0;
}
int encoded_nextchar(encoded_str *encstr)
{
	int	chr;

	chr = encstr->encstr[++encstr->pos]; 
	encstr->ccst = pg_CS_stat(encstr->ccst, (unsigned int) chr, encstr->ccsc);
	return chr; 
}
ssize_t encoded_position_shift(encoded_str *encstr, size_t shift)
{
	encstr->pos += shift; 
	return encstr->pos; 
}
int encoded_byte_check(encoded_str *encstr, size_t abspos)
{
	int	chr;

	chr = encstr->encstr[encstr->pos = abspos]; 
	encstr->ccst = pg_CS_stat(encstr->ccst, (unsigned int) chr, encstr->ccsc);
	return chr; 
}