regc_lex.c

从一个开源软件中摘取的正则表达式模块

				NOTE(REG_UBOUNDS);
				INTOCON(L_EBND);
				RET('{');
			}
			assert(NOTREACHED);
			break;
		case CHR('('):			/* parenthesis, or advanced extension */
			if ((v->cflags & REG_ADVF) && NEXT1('?'))
			{
				NOTE(REG_UNONPOSIX);
				v->now++;
				switch (*v->now++)
				{
					case CHR(':'):		/* non-capturing paren */
						RETV('(', 0);
						break;
					case CHR('#'):		/* comment */
						while (!ATEOS() && *v->now != CHR(')'))
							v->now++;
						if (!ATEOS())
							v->now++;
						assert(v->nexttype == v->lasttype);
						return next(v);
						break;
					case CHR('='):		/* positive lookahead */
						NOTE(REG_ULOOKAHEAD);
						RETV(LACON, 1);
						break;
					case CHR('!'):		/* negative lookahead */
						NOTE(REG_ULOOKAHEAD);
						RETV(LACON, 0);
						break;
					default:
						FAILW(REG_BADRPT);
						break;
				}
				assert(NOTREACHED);
			}
			if (v->cflags & REG_NOSUB)
				RETV('(', 0);	/* all parens non-capturing */
			else
				RETV('(', 1);
			break;
		case CHR(')'):
			if (LASTTYPE('('))
				NOTE(REG_UUNSPEC);
			RETV(')', c);
			break;
		case CHR('['):			/* easy except for [[:<:]] and [[:>:]] */
			if (HAVE(6) && *(v->now + 0) == CHR('[') &&
				*(v->now + 1) == CHR(':') &&
				(*(v->now + 2) == CHR('<') ||
				 *(v->now + 2) == CHR('>')) &&
				*(v->now + 3) == CHR(':') &&
				*(v->now + 4) == CHR(']') &&
				*(v->now + 5) == CHR(']'))
			{
				c = *(v->now + 2);
				v->now += 6;
				NOTE(REG_UNONPOSIX);
				RET((c == CHR('<')) ? '<' : '>');
			}
			INTOCON(L_BRACK);
			if (NEXT1('^'))
			{
				v->now++;
				RETV('[', 0);
			}
			RETV('[', 1);
			break;
		case CHR('.'):
			RET('.');
			break;
		case CHR('^'):
			RET('^');
			break;
		case CHR('$'):
			RET('$');
			break;
		case CHR('\\'): /* mostly punt backslashes to code below */
			if (ATEOS())
				FAILW(REG_EESCAPE);
			break;
		default:				/* ordinary character */
			RETV(PLAIN, c);
			break;
	}

	/* ERE/ARE backslash handling; backslash already eaten */
	assert(!ATEOS());
	if (!(v->cflags & REG_ADVF))
	{							/* only AREs have non-trivial escapes */
		if (iscalnum(*v->now))
		{
			NOTE(REG_UBSALNUM);
			NOTE(REG_UUNSPEC);
		}
		RETV(PLAIN, *v->now++);
	}
	(DISCARD) lexescape(v);
	if (ISERR())
		FAILW(REG_EESCAPE);
	if (v->nexttype == CCLASS)
	{							/* fudge at lexical level */
		switch (v->nextvalue)
		{
			case 'd':
				lexnest(v, backd, ENDOF(backd));
				break;
			case 'D':
				lexnest(v, backD, ENDOF(backD));
				break;
			case 's':
				lexnest(v, backs, ENDOF(backs));
				break;
			case 'S':
				lexnest(v, backS, ENDOF(backS));
				break;
			case 'w':
				lexnest(v, backw, ENDOF(backw));
				break;
			case 'W':
				lexnest(v, backW, ENDOF(backW));
				break;
			default:
				assert(NOTREACHED);
				FAILW(REG_ASSERT);
				break;
		}
		/* lexnest done, back up and try again */
		v->nexttype = v->lasttype;
		return next(v);
	}
	/* otherwise, lexescape has already done the work */
	return !ISERR();
}

/*
 * lexescape - parse an ARE backslash escape (backslash already eaten)
 * Note slightly nonstandard use of the CCLASS type code.
 */
static int						/* not actually used, but convenient for RETV */
lexescape(struct vars * v)
{
	chr			c;
	static chr	alert[] = {
		CHR('a'), CHR('l'), CHR('e'), CHR('r'), CHR('t')
	};
	static chr	esc[] = {
		CHR('E'), CHR('S'), CHR('C')
	};
	chr		   *save;

	assert(v->cflags & REG_ADVF);

	assert(!ATEOS());
	c = *v->now++;
	if (!iscalnum(c))
		RETV(PLAIN, c);

	NOTE(REG_UNONPOSIX);
	switch (c)
	{
		case CHR('a'):
			RETV(PLAIN, chrnamed(v, alert, ENDOF(alert), CHR('\007')));
			break;
		case CHR('A'):
			RETV(SBEGIN, 0);
			break;
		case CHR('b'):
			RETV(PLAIN, CHR('\b'));
			break;
		case CHR('B'):
			RETV(PLAIN, CHR('\\'));
			break;
		case CHR('c'):
			NOTE(REG_UUNPORT);
			if (ATEOS())
				FAILW(REG_EESCAPE);
			RETV(PLAIN, (chr) (*v->now++ & 037));
			break;
		case CHR('d'):
			NOTE(REG_ULOCALE);
			RETV(CCLASS, 'd');
			break;
		case CHR('D'):
			NOTE(REG_ULOCALE);
			RETV(CCLASS, 'D');
			break;
		case CHR('e'):
			NOTE(REG_UUNPORT);
			RETV(PLAIN, chrnamed(v, esc, ENDOF(esc), CHR('\033')));
			break;
		case CHR('f'):
			RETV(PLAIN, CHR('\f'));
			break;
		case CHR('m'):
			RET('<');
			break;
		case CHR('M'):
			RET('>');
			break;
		case CHR('n'):
			RETV(PLAIN, CHR('\n'));
			break;
		case CHR('r'):
			RETV(PLAIN, CHR('\r'));
			break;
		case CHR('s'):
			NOTE(REG_ULOCALE);
			RETV(CCLASS, 's');
			break;
		case CHR('S'):
			NOTE(REG_ULOCALE);
			RETV(CCLASS, 'S');
			break;
		case CHR('t'):
			RETV(PLAIN, CHR('\t'));
			break;
		case CHR('u'):
			c = lexdigits(v, 16, 4, 4);
			if (ISERR())
				FAILW(REG_EESCAPE);
			RETV(PLAIN, c);
			break;
		case CHR('U'):
			c = lexdigits(v, 16, 8, 8);
			if (ISERR())
				FAILW(REG_EESCAPE);
			RETV(PLAIN, c);
			break;
		case CHR('v'):
			RETV(PLAIN, CHR('\v'));
			break;
		case CHR('w'):
			NOTE(REG_ULOCALE);
			RETV(CCLASS, 'w');
			break;
		case CHR('W'):
			NOTE(REG_ULOCALE);
			RETV(CCLASS, 'W');
			break;
		case CHR('x'):
			NOTE(REG_UUNPORT);
			c = lexdigits(v, 16, 1, 255);		/* REs >255 long outside spec */
			if (ISERR())
				FAILW(REG_EESCAPE);
			RETV(PLAIN, c);
			break;
		case CHR('y'):
			NOTE(REG_ULOCALE);
			RETV(WBDRY, 0);
			break;
		case CHR('Y'):
			NOTE(REG_ULOCALE);
			RETV(NWBDRY, 0);
			break;
		case CHR('Z'):
			RETV(SEND, 0);
			break;
		case CHR('1'):
		case CHR('2'):
		case CHR('3'):
		case CHR('4'):
		case CHR('5'):
		case CHR('6'):
		case CHR('7'):
		case CHR('8'):
		case CHR('9'):
			save = v->now;
			v->now--;			/* put first digit back */
			c = lexdigits(v, 10, 1, 255);		/* REs >255 long outside spec */
			if (ISERR())
				FAILW(REG_EESCAPE);
			/* ugly heuristic (first test is "exactly 1 digit?") */
			if (v->now - save == 0 || ((int) c > 0 && (int) c <= v->nsubexp))
			{
				NOTE(REG_UBACKREF);
				RETV(BACKREF, (chr) c);
			}
			/* oops, doesn't look like it's a backref after all... */
			v->now = save;
			/* and fall through into octal number */
		case CHR('0'):
			NOTE(REG_UUNPORT);
			v->now--;			/* put first digit back */
			c = lexdigits(v, 8, 1, 3);
			if (ISERR())
				FAILW(REG_EESCAPE);
			RETV(PLAIN, c);
			break;
		default:
			assert(iscalpha(c));
			FAILW(REG_EESCAPE); /* unknown alphabetic escape */
			break;
	}
	assert(NOTREACHED);
}

/*
 * lexdigits - slurp up digits and return chr value
 */
static chr						/* chr value; errors signalled via ERR */
lexdigits(struct vars * v,
		  int base,
		  int minlen,
		  int maxlen)
{
	uchr		n;				/* unsigned to avoid overflow misbehavior */
	int			len;
	chr			c;
	int			d;
	const uchr	ub = (uchr) base;

	n = 0;
	for (len = 0; len < maxlen && !ATEOS(); len++)
	{
		c = *v->now++;
		switch (c)
		{
			case CHR('0'):
			case CHR('1'):
			case CHR('2'):
			case CHR('3'):
			case CHR('4'):
			case CHR('5'):
			case CHR('6'):
			case CHR('7'):
			case CHR('8'):
			case CHR('9'):
				d = DIGITVAL(c);
				break;
			case CHR('a'):
			case CHR('A'):
				d = 10;
				break;
			case CHR('b'):
			case CHR('B'):
				d = 11;
				break;
			case CHR('c'):
			case CHR('C'):
				d = 12;
				break;
			case CHR('d'):
			case CHR('D'):
				d = 13;
				break;
			case CHR('e'):
			case CHR('E'):
				d = 14;
				break;
			case CHR('f'):
			case CHR('F'):
				d = 15;
				break;
			default:
				v->now--;		/* oops, not a digit at all */
				d = -1;
				break;
		}

		if (d >= base)
		{						/* not a plausible digit */
			v->now--;
			d = -1;
		}
		if (d < 0)
			break;				/* NOTE BREAK OUT */
		n = n * ub + (uchr) d;
	}
	if (len < minlen)
		ERR(REG_EESCAPE);

	return (chr) n;
}

/*
 * brenext - get next BRE token
 *
 * This is much like EREs except for all the stupid backslashes and the
 * context-dependency of some things.
 */
static int						/* 1 normal, 0 failure */
brenext(struct vars * v,
		chr pc)
{
	chr			c = (chr) pc;

	switch (c)
	{
		case CHR('*'):
			if (LASTTYPE(EMPTY) || LASTTYPE('(') || LASTTYPE('^'))
				RETV(PLAIN, c);
			RET('*');
			break;
		case CHR('['):
			if (HAVE(6) && *(v->now + 0) == CHR('[') &&
				*(v->now + 1) == CHR(':') &&
				(*(v->now + 2) == CHR('<') ||
				 *(v->now + 2) == CHR('>')) &&
				*(v->now + 3) == CHR(':') &&
				*(v->now + 4) == CHR(']') &&
				*(v->now + 5) == CHR(']'))
			{
				c = *(v->now + 2);
				v->now += 6;
				NOTE(REG_UNONPOSIX);
				RET((c == CHR('<')) ? '<' : '>');
			}
			INTOCON(L_BRACK);
			if (NEXT1('^'))
			{
				v->now++;
				RETV('[', 0);
			}
			RETV('[', 1);
			break;
		case CHR('.'):
			RET('.');
			break;
		case CHR('^'):
			if (LASTTYPE(EMPTY))
				RET('^');
			if (LASTTYPE('('))
			{
				NOTE(REG_UUNSPEC);
				RET('^');
			}
			RETV(PLAIN, c);
			break;
		case CHR('$'):
			if (v->cflags & REG_EXPANDED)
				skip(v);
			if (ATEOS())
				RET('$');
			if (NEXT2('\\', ')'))
			{
				NOTE(REG_UUNSPEC);
				RET('$');
			}
			RETV(PLAIN, c);
			break;
		case CHR('\\'):
			break;				/* see below */
		default:
			RETV(PLAIN, c);
			break;
	}

	assert(c == CHR('\\'));

	if (ATEOS())
		FAILW(REG_EESCAPE);

	c = *v->now++;
	switch (c)
	{
		case CHR('{'):
			INTOCON(L_BBND);
			NOTE(REG_UBOUNDS);
			RET('{');
			break;
		case CHR('('):
			RETV('(', 1);
			break;
		case CHR(')'):
			RETV(')', c);
			break;
		case CHR('<'):
			NOTE(REG_UNONPOSIX);
			RET('<');
			break;
		case CHR('>'):
			NOTE(REG_UNONPOSIX);
			RET('>');
			break;
		case CHR('1'):
		case CHR('2'):
		case CHR('3'):
		case CHR('4'):
		case CHR('5'):
		case CHR('6'):
		case CHR('7'):
		case CHR('8'):
		case CHR('9'):
			NOTE(REG_UBACKREF);
			RETV(BACKREF, (chr) DIGITVAL(c));
			break;
		default:
			if (iscalnum(c))
			{
				NOTE(REG_UBSALNUM);
				NOTE(REG_UUNSPEC);
			}
			RETV(PLAIN, c);
			break;
	}

	assert(NOTREACHED);
	return 0;
}

/*
 * skip - skip white space and comments in expanded form
 */
static void
skip(struct vars * v)
{
	chr		   *start = v->now;

	assert(v->cflags & REG_EXPANDED);

	for (;;)
	{
		while (!ATEOS() && iscspace(*v->now))
			v->now++;
		if (ATEOS() || *v->now != CHR('#'))
			break;				/* NOTE BREAK OUT */
		assert(NEXT1('#'));
		while (!ATEOS() && *v->now != CHR('\n'))
			v->now++;
		/* leave the newline to be picked up by the iscspace loop */
	}

	if (v->now != start)
		NOTE(REG_UNONPOSIX);
}

/*
 * newline - return the chr for a newline
 *
 * This helps confine use of CHR to this source file.
 */
static chr
newline(void)
{
	return CHR('\n');
}

/*
 * chrnamed - return the chr known by a given (chr string) name
 *
 * The code is a bit clumsy, but this routine gets only such specialized
 * use that it hardly matters.
 */
static chr
chrnamed(struct vars * v,
		 chr *startp,			/* start of name */
		 chr *endp,				/* just past end of name */
		 chr lastresort)		/* what to return if name lookup fails */
{
	celt		c;
	int			errsave;
	int			e;
	struct cvec *cv;

	errsave = v->err;
	v->err = 0;
	c = element(v, startp, endp);
	e = v->err;
	v->err = errsave;

	if (e != 0)
		return (chr) lastresort;

	cv = range(v, c, c, 0);
	if (cv->nchrs == 0)
		return (chr) lastresort;
	return cv->chrs[0];
}