fis-gtm/sr_port/advancewindow.c

/****************************************************************
 *								*
 *	Copyright 2001, 2011 Fidelity Information Services, Inc	*
 *								*
 *	This source code contains the intellectual property	*
 *	of its copyright holder(s), and is made available	*
 *	under a license.  If you do not know the terms of	*
 *	the license, please stop and do not read further.	*
 *								*
 ****************************************************************/

#include "mdef.h"

#include "gtm_string.h"

#include "compiler.h"
#include "toktyp.h"
#include "stringpool.h"
#include "gtm_caseconv.h"
#include "advancewindow.h"
#include "show_source_line.h"

#ifdef UNICODE_SUPPORTED
#include "gtm_utf8.h"
#include "gtm_icu_api.h"	/* U_ISPRINT() needs this header */
#endif

GBLREF	unsigned char	*source_buffer;
GBLREF	short int	source_column;
GBLREF	char		window_token;
GBLREF	mval		window_mval;
GBLREF	char		director_token;
GBLREF	mval		director_mval;
GBLREF	char		*lexical_ptr;
GBLREF	spdesc		stringpool;
GBLREF	boolean_t	gtm_utf8_mode;
GBLREF	boolean_t	run_time;
GBLREF	mident		window_ident;	/* the current identifier */
GBLREF	mident		director_ident;	/* the look-ahead identifier */

LITREF	char		ctypetab[NUM_CHARS];

error_def(ERR_LITNONGRAPH);

static readonly unsigned char apos_ok[] =
{
	0,TK_NEXCLAIMATION,0,0,0,0,TK_NAMPERSAND,0
	,0,0,0,0,0,0,0,0
	,0,0,0,0,0,0,0,0
	,0,0,0,0,TK_NLESS,TK_NEQUAL,TK_NGREATER,TK_NQUESTION
	,0,0,0,0,0,0,0,0
	,0,0,0,0,0,0,0,0
	,0,0,0,0,0,0,0,0
	,0,0,0,TK_NLBRACKET,0,TK_NRBRACKET,0,0
};

void advancewindow(void)
{
	error_def(ERR_NUMOFLOW);
	unsigned char	*cp1, *cp2, *cp3, x;
	char		*tmp, source_line_buff[MAX_SRCLINE + SIZEOF(ARROW)];
	int		y, charlen;
#	ifdef UNICODE_SUPPORTED
	uint4		ch;
	unsigned char	*cptr;
#	endif
	DCL_THREADGBL_ACCESS;

	SETUP_THREADGBL_ACCESS;
	TREF(last_source_column) = source_column;
	source_column = (unsigned char *)lexical_ptr - source_buffer + 1;
	window_token = director_token;
	window_mval = director_mval;
	director_mval.mvtype = 0;	/* keeps mval from being GC'd since it is not useful until re-used */

	/* It is more efficient to swich buffers between window_ident and director_ident
	 * instead of copying the text from director_ident to window_ident. This can be
	 * done by exchanging the pointers window_ident.addr and director_ident.addr */
	tmp = window_ident.addr;
	window_ident = director_ident;
	director_ident.addr = tmp;

	x = *lexical_ptr;
	switch (y = ctypetab[x])
	{
		case TK_QUOTE:
			ENSURE_STP_FREE_SPACE(MAX_SRCLINE);
			cp1 = (unsigned char *)lexical_ptr + 1;
			cp2 = cp3 = stringpool.free;
			for (;;)
			{

#ifdef UNICODE_SUPPORTED
				if (gtm_utf8_mode)
					cptr = (unsigned char *)UTF8_MBTOWC((sm_uc_ptr_t)cp1, source_buffer + MAX_SRCLINE, ch);
#endif
				x = *cp1++;
				if ((SP > x) UNICODE_ONLY(|| (gtm_utf8_mode && !(U_ISPRINT(ch)))))
				{
					TREF(last_source_column) = cp1 - source_buffer;
					if ('\0' == x)
					{
						director_token = window_token = TK_ERROR;
						return;
					}
					if (!run_time)
					{
						show_source_line(source_line_buff, SIZEOF(source_line_buff), TRUE);
						dec_err(VARLSTCNT(1) ERR_LITNONGRAPH);
					}
				}
				if ('\"' == x)
				{
					UNICODE_ONLY(assert(!gtm_utf8_mode || (cp1 == cptr)));
					if ('\"' == *cp1)
						cp1++;
					else
						break;
				}
				*cp2++ = x;
#ifdef UNICODE_SUPPORTED
				if (gtm_utf8_mode && (cptr > cp1))
				{
					assert(4 > (cptr - cp1));
					for (; cptr > cp1;)
						*cp2++ = *cp1++;
				}
#endif
				assert(cp2 <= stringpool.top);
			}
			lexical_ptr = (char *)cp1;
			director_token = TK_STRLIT;
			director_mval.mvtype = MV_STR;
			director_mval.str.addr = (char *)cp3;
			director_mval.str.len = INTCAST(cp2 - cp3);
			stringpool.free = cp2;
			s2n(&director_mval);
#ifdef UNICODE_SUPPORTED
			if (gtm_utf8_mode && !run_time)
			{	/* UTF8 mode and not compiling an indirect gets an optimization to set the
				   (true) length of the string into the mval
				*/
				charlen = utf8_len_stx(&director_mval.str);
				if (0 > charlen)	/* got a BADCHAR error */
					director_token = TK_ERROR;
				else
				{
					assert(charlen == director_mval.str.char_len);
					director_mval.mvtype |= MV_UTF_LEN;
				}
			}
#endif
			return;
		case TK_LOWER:
		case TK_PERCENT:
		case TK_UPPER:
			cp2 = (unsigned char *)director_ident.addr;
			cp3 = cp2 + MAX_MIDENT_LEN;
			for (;;)
			{
				if (cp2 < cp3)
					*cp2++ = x;
				y = ctypetab[x = *++lexical_ptr];
				if ((TK_UPPER != y) && (TK_DIGIT != y) && (TK_LOWER != y))
					break;
			}
			director_ident.len = INTCAST(cp2 - (unsigned char*)director_ident.addr);
			director_token = TK_IDENT;
			return;
		case TK_PERIOD:
			if (ctypetab[x = *(lexical_ptr + 1)] != TK_DIGIT)
				break;
		case TK_DIGIT:
			director_mval.str.addr = lexical_ptr;
			director_mval.str.len = MAX_SRCLINE;
			director_mval.mvtype = MV_STR;
			lexical_ptr = (char *)s2n(&director_mval);
			if (!(director_mval.mvtype &= MV_NUM_MASK))
			{
				stx_error(ERR_NUMOFLOW);
				director_token = TK_ERROR;
				return;
			}
			if (TREF(s2n_intlit))
			{
				director_token = TK_NUMLIT ;
				n2s(&director_mval);
			} else
			{
				director_token = TK_INTLIT ;
				director_mval.str.len = INTCAST(lexical_ptr - director_mval.str.addr);
				ENSURE_STP_FREE_SPACE(director_mval.str.len);
				memcpy(stringpool.free, director_mval.str.addr, director_mval.str.len);
				assert (stringpool.free <= stringpool.top) ;
			}
			return;
		case TK_APOSTROPHE:
			if (( x = *++lexical_ptr) >= 32)
			{
				x -= 32;
				if (x < SIZEOF(apos_ok) / SIZEOF(unsigned char))
				{
					if (y = apos_ok[x])
					{
						if (DEL < (x = *++lexical_ptr))
						{
							director_token = TK_ERROR;
							return;
						}
						if (TK_RBRACKET == ctypetab[x])
						{
							lexical_ptr++;
							y = TK_NSORTS_AFTER;
						}
						director_token = y;
						return;
					}
				}
			}
			director_token = TK_APOSTROPHE;
			return;
		case TK_SEMICOLON:
			while (*++lexical_ptr) ;
			y = TK_EOL;
			break;
		case TK_ASTERISK:
			if (DEL < (x = *(lexical_ptr + 1)))
			{
				director_token = TK_ERROR;
				return;
			}
			if (TK_ASTERISK == ctypetab[x])
			{
				lexical_ptr++;
				y = TK_EXPONENT;
			}
			break;
		case TK_RBRACKET:
			if ((x = *(lexical_ptr + 1)) > DEL)
			{
				director_token = TK_ERROR;
				return;
			}
			if (TK_RBRACKET == ctypetab[x])
			{
				lexical_ptr++;
				y = TK_SORTS_AFTER;
			}
			break;
		default:
			;
	}
	lexical_ptr++;
	director_token = y;
	return;
}

#ifdef GTM_TRIGGER
/* The M standard does not allow the '#' character to appear inside mnames but in specific places, we want to allow this
 * so that triggers, which have the imbedded '#' character in their routine names, can be debugged and printed. The places
 * where this is allowed follow.
 *
 *   1. $TEXT()
 *   2. ZBREAK
 *   3. ZPRINT
 *
 * All other uses still prohibit '#' from being in an MNAME. Routines that need to allow # in a name can call this routine to
 * recombine the existing token and the look-ahead (director) token such that '#' is considered part of an mident.
 */
void advwindw_hash_in_mname_allowed(void)
{
	unsigned char	*cp2, *cp3, x;
	unsigned char	ident_buffer[SIZEOF(mident_fixed)];
	int		ident_len, ch;

	assert(TK_IDENT == window_token);
	assert(TK_HASH == director_token);
	/* First copy the existing token we want to expand into our safe-haven */
	memcpy(ident_buffer, window_ident.addr, window_ident.len);
	/* Now parse further until we run out of [m]ident */
	cp2 = ident_buffer + window_ident.len;
	cp3 = ident_buffer + MAX_MIDENT_LEN;
	*cp2++ = '#';	/* We are only called if director token is '#' so put that char in buffer now */
	/* Start processing with the token following the '#' */
	for (x = *lexical_ptr, ch = ctypetab[x];
	     ((TK_UPPER == ch) || (TK_DIGIT == ch) || (TK_LOWER == ch) || (TK_HASH == ch));
	     x = *++lexical_ptr, ch = ctypetab[x])
	{
		if (cp2 < cp3)
			*cp2++ = x;
	}
	director_ident.len = INTCAST(cp2 - ident_buffer);
	director_token = TK_IDENT;
	memcpy(director_ident.addr, ident_buffer, director_ident.len);
	advancewindow();	/* Makes the homogenized token the current token (again) and prereads next token */
}
#endif