Home | History | Annotate | Download | only in src
      1 /*	$OpenBSD: syn.c,v 1.28 2008/07/23 16:34:38 jaredy Exp $	*/
      2 
      3 /*-
      4  * Copyright (c) 2003, 2004, 2005, 2006, 2007, 2008, 2009
      5  *	Thorsten Glaser <tg (at) mirbsd.org>
      6  *
      7  * Provided that these terms and disclaimer and all copyright notices
      8  * are retained or reproduced in an accompanying document, permission
      9  * is granted to deal in this work without restriction, including un-
     10  * limited rights to use, publicly perform, distribute, sell, modify,
     11  * merge, give away, or sublicence.
     12  *
     13  * This work is provided "AS IS" and WITHOUT WARRANTY of any kind, to
     14  * the utmost extent permitted by applicable law, neither express nor
     15  * implied; without malicious intent or gross negligence. In no event
     16  * may a licensor, author or contributor be held liable for indirect,
     17  * direct, other damage, loss, or other issues arising in any way out
     18  * of dealing in the work, even if advised of the possibility of such
     19  * damage or existence of a defect, except proven that it results out
     20  * of said person's immediate fault when using the work as intended.
     21  */
     22 
     23 #include "sh.h"
     24 
     25 __RCSID("$MirOS: src/bin/mksh/syn.c,v 1.49 2010/07/17 22:09:39 tg Exp $");
     26 
     27 struct nesting_state {
     28 	int start_token;	/* token than began nesting (eg, FOR) */
     29 	int start_line;		/* line nesting began on */
     30 };
     31 
     32 static void yyparse(void);
     33 static struct op *pipeline(int);
     34 static struct op *andor(void);
     35 static struct op *c_list(int);
     36 static struct ioword *synio(int);
     37 static struct op *nested(int, int, int);
     38 static struct op *get_command(int);
     39 static struct op *dogroup(void);
     40 static struct op *thenpart(void);
     41 static struct op *elsepart(void);
     42 static struct op *caselist(void);
     43 static struct op *casepart(int);
     44 static struct op *function_body(char *, bool);
     45 static char **wordlist(void);
     46 static struct op *block(int, struct op *, struct op *, char **);
     47 static struct op *newtp(int);
     48 static void syntaxerr(const char *) MKSH_A_NORETURN;
     49 static void nesting_push(struct nesting_state *, int);
     50 static void nesting_pop(struct nesting_state *);
     51 static int assign_command(char *);
     52 static int inalias(struct source *);
     53 static Test_op dbtestp_isa(Test_env *, Test_meta);
     54 static const char *dbtestp_getopnd(Test_env *, Test_op, bool);
     55 static int dbtestp_eval(Test_env *, Test_op, const char *,
     56     const char *, bool);
     57 static void dbtestp_error(Test_env *, int, const char *) MKSH_A_NORETURN;
     58 
     59 static struct op *outtree;		/* yyparse output */
     60 static struct nesting_state nesting;	/* \n changed to ; */
     61 
     62 static int reject;		/* token(cf) gets symbol again */
     63 static int symbol;		/* yylex value */
     64 
     65 #define REJECT		(reject = 1)
     66 #define ACCEPT		(reject = 0)
     67 #define token(cf)	((reject) ? (ACCEPT, symbol) : (symbol = yylex(cf)))
     68 #define tpeek(cf)	((reject) ? (symbol) : (REJECT, symbol = yylex(cf)))
     69 #define musthave(c,cf)	do { if (token(cf) != (c)) syntaxerr(NULL); } while (0)
     70 
     71 static void
     72 yyparse(void)
     73 {
     74 	int c;
     75 
     76 	ACCEPT;
     77 
     78 	outtree = c_list(source->type == SSTRING);
     79 	c = tpeek(0);
     80 	if (c == 0 && !outtree)
     81 		outtree = newtp(TEOF);
     82 	else if (c != '\n' && c != 0)
     83 		syntaxerr(NULL);
     84 }
     85 
     86 static struct op *
     87 pipeline(int cf)
     88 {
     89 	struct op *t, *p, *tl = NULL;
     90 
     91 	t = get_command(cf);
     92 	if (t != NULL) {
     93 		while (token(0) == '|') {
     94 			if ((p = get_command(CONTIN)) == NULL)
     95 				syntaxerr(NULL);
     96 			if (tl == NULL)
     97 				t = tl = block(TPIPE, t, p, NOWORDS);
     98 			else
     99 				tl = tl->right = block(TPIPE, tl->right, p, NOWORDS);
    100 		}
    101 		REJECT;
    102 	}
    103 	return (t);
    104 }
    105 
    106 static struct op *
    107 andor(void)
    108 {
    109 	struct op *t, *p;
    110 	int c;
    111 
    112 	t = pipeline(0);
    113 	if (t != NULL) {
    114 		while ((c = token(0)) == LOGAND || c == LOGOR) {
    115 			if ((p = pipeline(CONTIN)) == NULL)
    116 				syntaxerr(NULL);
    117 			t = block(c == LOGAND? TAND: TOR, t, p, NOWORDS);
    118 		}
    119 		REJECT;
    120 	}
    121 	return (t);
    122 }
    123 
    124 static struct op *
    125 c_list(int multi)
    126 {
    127 	struct op *t = NULL, *p, *tl = NULL;
    128 	int c, have_sep;
    129 
    130 	while (1) {
    131 		p = andor();
    132 		/* Token has always been read/rejected at this point, so
    133 		 * we don't worry about what flags to pass token()
    134 		 */
    135 		c = token(0);
    136 		have_sep = 1;
    137 		if (c == '\n' && (multi || inalias(source))) {
    138 			if (!p) /* ignore blank lines */
    139 				continue;
    140 		} else if (!p)
    141 			break;
    142 		else if (c == '&' || c == COPROC)
    143 			p = block(c == '&' ? TASYNC : TCOPROC,
    144 			    p, NOBLOCK, NOWORDS);
    145 		else if (c != ';')
    146 			have_sep = 0;
    147 		if (!t)
    148 			t = p;
    149 		else if (!tl)
    150 			t = tl = block(TLIST, t, p, NOWORDS);
    151 		else
    152 			tl = tl->right = block(TLIST, tl->right, p, NOWORDS);
    153 		if (!have_sep)
    154 			break;
    155 	}
    156 	REJECT;
    157 	return (t);
    158 }
    159 
    160 static struct ioword *
    161 synio(int cf)
    162 {
    163 	struct ioword *iop;
    164 	static struct ioword *nextiop = NULL;
    165 	bool ishere;
    166 
    167 	if (nextiop != NULL) {
    168 		iop = nextiop;
    169 		nextiop = NULL;
    170 		return (iop);
    171 	}
    172 
    173 	if (tpeek(cf) != REDIR)
    174 		return (NULL);
    175 	ACCEPT;
    176 	iop = yylval.iop;
    177 	ishere = (iop->flag&IOTYPE) == IOHERE;
    178 	musthave(LWORD, ishere ? HEREDELIM : 0);
    179 	if (ishere) {
    180 		iop->delim = yylval.cp;
    181 		if (*ident != 0) /* unquoted */
    182 			iop->flag |= IOEVAL;
    183 		if (herep > &heres[HERES - 1])
    184 			yyerror("too many <<s\n");
    185 		*herep++ = iop;
    186 	} else
    187 		iop->name = yylval.cp;
    188 
    189 	if (iop->flag & IOBASH) {
    190 		char *cp;
    191 
    192 		nextiop = alloc(sizeof(*iop), ATEMP);
    193 		nextiop->name = cp = alloc(5, ATEMP);
    194 
    195 		if (iop->unit > 9) {
    196 			*cp++ = CHAR;
    197 			*cp++ = '0' + (iop->unit / 10);
    198 		}
    199 		*cp++ = CHAR;
    200 		*cp++ = '0' + (iop->unit % 10);
    201 		*cp = EOS;
    202 
    203 		iop->flag &= ~IOBASH;
    204 		nextiop->unit = 2;
    205 		nextiop->flag = IODUP;
    206 		nextiop->delim = NULL;
    207 		nextiop->heredoc = NULL;
    208 	}
    209 	return (iop);
    210 }
    211 
    212 static struct op *
    213 nested(int type, int smark, int emark)
    214 {
    215 	struct op *t;
    216 	struct nesting_state old_nesting;
    217 
    218 	nesting_push(&old_nesting, smark);
    219 	t = c_list(true);
    220 	musthave(emark, KEYWORD|ALIAS);
    221 	nesting_pop(&old_nesting);
    222 	return (block(type, t, NOBLOCK, NOWORDS));
    223 }
    224 
    225 static struct op *
    226 get_command(int cf)
    227 {
    228 	struct op *t;
    229 	int c, iopn = 0, syniocf;
    230 	struct ioword *iop, **iops;
    231 	XPtrV args, vars;
    232 	struct nesting_state old_nesting;
    233 
    234 	iops = alloc((NUFILE + 1) * sizeof(struct ioword *), ATEMP);
    235 	XPinit(args, 16);
    236 	XPinit(vars, 16);
    237 
    238 	syniocf = KEYWORD|ALIAS;
    239 	switch (c = token(cf|KEYWORD|ALIAS|VARASN)) {
    240 	default:
    241 		REJECT;
    242 		afree(iops, ATEMP);
    243 		XPfree(args);
    244 		XPfree(vars);
    245 		return (NULL); /* empty line */
    246 
    247 	case LWORD:
    248 	case REDIR:
    249 		REJECT;
    250 		syniocf &= ~(KEYWORD|ALIAS);
    251 		t = newtp(TCOM);
    252 		t->lineno = source->line;
    253 		while (1) {
    254 			cf = (t->u.evalflags ? ARRAYVAR : 0) |
    255 			    (XPsize(args) == 0 ? ALIAS|VARASN : CMDWORD);
    256 			switch (tpeek(cf)) {
    257 			case REDIR:
    258 				while ((iop = synio(cf)) != NULL) {
    259 					if (iopn >= NUFILE)
    260 						yyerror("too many redirections\n");
    261 					iops[iopn++] = iop;
    262 				}
    263 				break;
    264 
    265 			case LWORD:
    266 				ACCEPT;
    267 				/* the iopn == 0 and XPsize(vars) == 0 are
    268 				 * dubious but AT&T ksh acts this way
    269 				 */
    270 				if (iopn == 0 && XPsize(vars) == 0 &&
    271 				    XPsize(args) == 0 &&
    272 				    assign_command(ident))
    273 					t->u.evalflags = DOVACHECK;
    274 				if ((XPsize(args) == 0 || Flag(FKEYWORD)) &&
    275 				    is_wdvarassign(yylval.cp))
    276 					XPput(vars, yylval.cp);
    277 				else
    278 					XPput(args, yylval.cp);
    279 				break;
    280 
    281 			case '(':
    282 				/* Check for "> foo (echo hi)" which AT&T ksh
    283 				 * allows (not POSIX, but not disallowed)
    284 				 */
    285 				afree(t, ATEMP);
    286 				if (XPsize(args) == 0 && XPsize(vars) == 0) {
    287 					ACCEPT;
    288 					goto Subshell;
    289 				}
    290 #ifndef MKSH_SMALL
    291 				if ((XPsize(args) == 0 || Flag(FKEYWORD)) &&
    292 				    XPsize(vars) == 1 && is_wdvarassign(yylval.cp))
    293 					goto is_wdarrassign;
    294 #endif
    295 				/* Must be a function */
    296 				if (iopn != 0 || XPsize(args) != 1 ||
    297 				    XPsize(vars) != 0)
    298 					syntaxerr(NULL);
    299 				ACCEPT;
    300 				/*(*/
    301 				musthave(')', 0);
    302 				t = function_body(XPptrv(args)[0], false);
    303 				goto Leave;
    304 #ifndef MKSH_SMALL
    305  is_wdarrassign:
    306 			{
    307 				static const char set_cmd0[] = {
    308 					CHAR, 'e', CHAR, 'v',
    309 					CHAR, 'a', CHAR, 'l', EOS
    310 				};
    311 				static const char set_cmd1[] = {
    312 					CHAR, 's', CHAR, 'e',
    313 					CHAR, 't', CHAR, ' ',
    314 					CHAR, '-', CHAR, 'A', EOS
    315 				};
    316 				static const char set_cmd2[] = {
    317 					CHAR, '-', CHAR, '-', EOS
    318 				};
    319 				char *tcp;
    320 				XPfree(vars);
    321 				XPinit(vars, 16);
    322 				/*
    323 				 * we know (or rather hope) that yylval.cp
    324 				 * contains a string "varname="
    325 				 */
    326 				tcp = wdcopy(yylval.cp, ATEMP);
    327 				tcp[wdscan(tcp, EOS) - tcp - 3] = EOS;
    328 				/* now make an array assignment command */
    329 				t = newtp(TCOM);
    330 				t->lineno = source->line;
    331 				ACCEPT;
    332 				XPput(args, wdcopy(set_cmd0, ATEMP));
    333 				XPput(args, wdcopy(set_cmd1, ATEMP));
    334 				XPput(args, tcp);
    335 				XPput(args, wdcopy(set_cmd2, ATEMP));
    336 				musthave(LWORD,LETARRAY);
    337 				XPput(args, yylval.cp);
    338 				break;
    339 			}
    340 #endif
    341 
    342 			default:
    343 				goto Leave;
    344 			}
    345 		}
    346  Leave:
    347 		break;
    348 
    349 	case '(':
    350  Subshell:
    351 		t = nested(TPAREN, '(', ')');
    352 		break;
    353 
    354 	case '{': /*}*/
    355 		t = nested(TBRACE, '{', '}');
    356 		break;
    357 
    358 	case MDPAREN: {
    359 		int lno;
    360 		static const char let_cmd[] = {
    361 			CHAR, 'l', CHAR, 'e',
    362 			CHAR, 't', EOS
    363 		};
    364 
    365 		/* Leave KEYWORD in syniocf (allow if (( 1 )) then ...) */
    366 		lno = source->line;
    367 		ACCEPT;
    368 		switch (token(LETEXPR)) {
    369 		case LWORD:
    370 			break;
    371 		case '(':	/* ) */
    372 			goto Subshell;
    373 		default:
    374 			syntaxerr(NULL);
    375 		}
    376 		t = newtp(TCOM);
    377 		t->lineno = lno;
    378 		XPput(args, wdcopy(let_cmd, ATEMP));
    379 		XPput(args, yylval.cp);
    380 		break;
    381 	}
    382 
    383 	case DBRACKET: /* [[ .. ]] */
    384 		/* Leave KEYWORD in syniocf (allow if [[ -n 1 ]] then ...) */
    385 		t = newtp(TDBRACKET);
    386 		ACCEPT;
    387 		{
    388 			Test_env te;
    389 
    390 			te.flags = TEF_DBRACKET;
    391 			te.pos.av = &args;
    392 			te.isa = dbtestp_isa;
    393 			te.getopnd = dbtestp_getopnd;
    394 			te.eval = dbtestp_eval;
    395 			te.error = dbtestp_error;
    396 
    397 			test_parse(&te);
    398 		}
    399 		break;
    400 
    401 	case FOR:
    402 	case SELECT:
    403 		t = newtp((c == FOR) ? TFOR : TSELECT);
    404 		musthave(LWORD, ARRAYVAR);
    405 		if (!is_wdvarname(yylval.cp, true))
    406 			yyerror("%s: bad identifier\n",
    407 			    c == FOR ? "for" : "select");
    408 		strdupx(t->str, ident, ATEMP);
    409 		nesting_push(&old_nesting, c);
    410 		t->vars = wordlist();
    411 		t->left = dogroup();
    412 		nesting_pop(&old_nesting);
    413 		break;
    414 
    415 	case WHILE:
    416 	case UNTIL:
    417 		nesting_push(&old_nesting, c);
    418 		t = newtp((c == WHILE) ? TWHILE : TUNTIL);
    419 		t->left = c_list(true);
    420 		t->right = dogroup();
    421 		nesting_pop(&old_nesting);
    422 		break;
    423 
    424 	case CASE:
    425 		t = newtp(TCASE);
    426 		musthave(LWORD, 0);
    427 		t->str = yylval.cp;
    428 		nesting_push(&old_nesting, c);
    429 		t->left = caselist();
    430 		nesting_pop(&old_nesting);
    431 		break;
    432 
    433 	case IF:
    434 		nesting_push(&old_nesting, c);
    435 		t = newtp(TIF);
    436 		t->left = c_list(true);
    437 		t->right = thenpart();
    438 		musthave(FI, KEYWORD|ALIAS);
    439 		nesting_pop(&old_nesting);
    440 		break;
    441 
    442 	case BANG:
    443 		syniocf &= ~(KEYWORD|ALIAS);
    444 		t = pipeline(0);
    445 		if (t == NULL)
    446 			syntaxerr(NULL);
    447 		t = block(TBANG, NOBLOCK, t, NOWORDS);
    448 		break;
    449 
    450 	case TIME:
    451 		syniocf &= ~(KEYWORD|ALIAS);
    452 		t = pipeline(0);
    453 		if (t) {
    454 			t->str = alloc(2, ATEMP);
    455 			t->str[0] = '\0';	/* TF_* flags */
    456 			t->str[1] = '\0';
    457 		}
    458 		t = block(TTIME, t, NOBLOCK, NOWORDS);
    459 		break;
    460 
    461 	case FUNCTION:
    462 		musthave(LWORD, 0);
    463 		t = function_body(yylval.cp, true);
    464 		break;
    465 	}
    466 
    467 	while ((iop = synio(syniocf)) != NULL) {
    468 		if (iopn >= NUFILE)
    469 			yyerror("too many redirections\n");
    470 		iops[iopn++] = iop;
    471 	}
    472 
    473 	if (iopn == 0) {
    474 		afree(iops, ATEMP);
    475 		t->ioact = NULL;
    476 	} else {
    477 		iops[iopn++] = NULL;
    478 		iops = aresize(iops, iopn * sizeof(struct ioword *), ATEMP);
    479 		t->ioact = iops;
    480 	}
    481 
    482 	if (t->type == TCOM || t->type == TDBRACKET) {
    483 		XPput(args, NULL);
    484 		t->args = (const char **)XPclose(args);
    485 		XPput(vars, NULL);
    486 		t->vars = (char **) XPclose(vars);
    487 	} else {
    488 		XPfree(args);
    489 		XPfree(vars);
    490 	}
    491 
    492 	return (t);
    493 }
    494 
    495 static struct op *
    496 dogroup(void)
    497 {
    498 	int c;
    499 	struct op *list;
    500 
    501 	c = token(CONTIN|KEYWORD|ALIAS);
    502 	/* A {...} can be used instead of do...done for for/select loops
    503 	 * but not for while/until loops - we don't need to check if it
    504 	 * is a while loop because it would have been parsed as part of
    505 	 * the conditional command list...
    506 	 */
    507 	if (c == DO)
    508 		c = DONE;
    509 	else if (c == '{')
    510 		c = '}';
    511 	else
    512 		syntaxerr(NULL);
    513 	list = c_list(true);
    514 	musthave(c, KEYWORD|ALIAS);
    515 	return (list);
    516 }
    517 
    518 static struct op *
    519 thenpart(void)
    520 {
    521 	struct op *t;
    522 
    523 	musthave(THEN, KEYWORD|ALIAS);
    524 	t = newtp(0);
    525 	t->left = c_list(true);
    526 	if (t->left == NULL)
    527 		syntaxerr(NULL);
    528 	t->right = elsepart();
    529 	return (t);
    530 }
    531 
    532 static struct op *
    533 elsepart(void)
    534 {
    535 	struct op *t;
    536 
    537 	switch (token(KEYWORD|ALIAS|VARASN)) {
    538 	case ELSE:
    539 		if ((t = c_list(true)) == NULL)
    540 			syntaxerr(NULL);
    541 		return (t);
    542 
    543 	case ELIF:
    544 		t = newtp(TELIF);
    545 		t->left = c_list(true);
    546 		t->right = thenpart();
    547 		return (t);
    548 
    549 	default:
    550 		REJECT;
    551 	}
    552 	return (NULL);
    553 }
    554 
    555 static struct op *
    556 caselist(void)
    557 {
    558 	struct op *t, *tl;
    559 	int c;
    560 
    561 	c = token(CONTIN|KEYWORD|ALIAS);
    562 	/* A {...} can be used instead of in...esac for case statements */
    563 	if (c == IN)
    564 		c = ESAC;
    565 	else if (c == '{')
    566 		c = '}';
    567 	else
    568 		syntaxerr(NULL);
    569 	t = tl = NULL;
    570 	while ((tpeek(CONTIN|KEYWORD|ESACONLY)) != c) { /* no ALIAS here */
    571 		struct op *tc = casepart(c);
    572 		if (tl == NULL)
    573 			t = tl = tc, tl->right = NULL;
    574 		else
    575 			tl->right = tc, tl = tc;
    576 	}
    577 	musthave(c, KEYWORD|ALIAS);
    578 	return (t);
    579 }
    580 
    581 static struct op *
    582 casepart(int endtok)
    583 {
    584 	struct op *t;
    585 	XPtrV ptns;
    586 
    587 	XPinit(ptns, 16);
    588 	t = newtp(TPAT);
    589 	/* no ALIAS here */
    590 	if (token(CONTIN | KEYWORD) != '(')
    591 		REJECT;
    592 	do {
    593 		musthave(LWORD, 0);
    594 		XPput(ptns, yylval.cp);
    595 	} while (token(0) == '|');
    596 	REJECT;
    597 	XPput(ptns, NULL);
    598 	t->vars = (char **) XPclose(ptns);
    599 	musthave(')', 0);
    600 
    601 	t->left = c_list(true);
    602 	/* Note: POSIX requires the ;; */
    603 	if ((tpeek(CONTIN|KEYWORD|ALIAS)) != endtok)
    604 		musthave(BREAK, CONTIN|KEYWORD|ALIAS);
    605 	return (t);
    606 }
    607 
    608 static struct op *
    609 function_body(char *name,
    610     bool ksh_func)		/* function foo { ... } vs foo() { .. } */
    611 {
    612 	char *sname, *p;
    613 	struct op *t;
    614 	bool old_func_parse;
    615 
    616 	sname = wdstrip(name, false, false);
    617 	/* Check for valid characters in name. POSIX and AT&T ksh93 say only
    618 	 * allow [a-zA-Z_0-9] but this allows more as old pdkshs have
    619 	 * allowed more (the following were never allowed:
    620 	 *	NUL TAB NL SP " $ & ' ( ) ; < = > \ ` |
    621 	 * C_QUOTE covers all but adds # * ? [ ]
    622 	 */
    623 	for (p = sname; *p; p++)
    624 		if (ctype(*p, C_QUOTE))
    625 			yyerror("%s: invalid function name\n", sname);
    626 
    627 	/* Note that POSIX allows only compound statements after foo(), sh and
    628 	 * AT&T ksh allow any command, go with the later since it shouldn't
    629 	 * break anything. However, for function foo, AT&T ksh only accepts
    630 	 * an open-brace.
    631 	 */
    632 	if (ksh_func) {
    633 		if (tpeek(CONTIN|KEYWORD|ALIAS) == '(' /* ) */) {
    634 			struct tbl *tp;
    635 
    636 			/* function foo () { */
    637 			ACCEPT;
    638 			musthave(')', 0);
    639 			/* degrade to POSIX function */
    640 			ksh_func = false;
    641 			if ((tp = ktsearch(&aliases, sname, hash(sname))))
    642 				ktdelete(tp);
    643 		}
    644 		musthave('{', CONTIN|KEYWORD|ALIAS); /* } */
    645 		REJECT;
    646 	}
    647 
    648 	t = newtp(TFUNCT);
    649 	t->str = sname;
    650 	t->u.ksh_func = ksh_func;
    651 	t->lineno = source->line;
    652 
    653 	old_func_parse = e->flags & EF_FUNC_PARSE;
    654 	e->flags |= EF_FUNC_PARSE;
    655 	if ((t->left = get_command(CONTIN)) == NULL) {
    656 		char *tv;
    657 		/*
    658 		 * Probably something like foo() followed by eof or ;.
    659 		 * This is accepted by sh and ksh88.
    660 		 * To make "typeset -f foo" work reliably (so its output can
    661 		 * be used as input), we pretend there is a colon here.
    662 		 */
    663 		t->left = newtp(TCOM);
    664 		t->left->args = alloc(2 * sizeof(char *), ATEMP);
    665 		t->left->args[0] = tv = alloc(3, ATEMP);
    666 		tv[0] = CHAR;
    667 		tv[1] = ':';
    668 		tv[2] = EOS;
    669 		t->left->args[1] = NULL;
    670 		t->left->vars = alloc(sizeof(char *), ATEMP);
    671 		t->left->vars[0] = NULL;
    672 		t->left->lineno = 1;
    673 	}
    674 	if (!old_func_parse)
    675 		e->flags &= ~EF_FUNC_PARSE;
    676 
    677 	return (t);
    678 }
    679 
    680 static char **
    681 wordlist(void)
    682 {
    683 	int c;
    684 	XPtrV args;
    685 
    686 	XPinit(args, 16);
    687 	/* POSIX does not do alias expansion here... */
    688 	if ((c = token(CONTIN|KEYWORD|ALIAS)) != IN) {
    689 		if (c != ';') /* non-POSIX, but AT&T ksh accepts a ; here */
    690 			REJECT;
    691 		return (NULL);
    692 	}
    693 	while ((c = token(0)) == LWORD)
    694 		XPput(args, yylval.cp);
    695 	if (c != '\n' && c != ';')
    696 		syntaxerr(NULL);
    697 	if (XPsize(args) == 0) {
    698 		XPfree(args);
    699 		return (NULL);
    700 	} else {
    701 		XPput(args, NULL);
    702 		return ((char **)XPclose(args));
    703 	}
    704 }
    705 
    706 /*
    707  * supporting functions
    708  */
    709 
    710 static struct op *
    711 block(int type, struct op *t1, struct op *t2, char **wp)
    712 {
    713 	struct op *t;
    714 
    715 	t = newtp(type);
    716 	t->left = t1;
    717 	t->right = t2;
    718 	t->vars = wp;
    719 	return (t);
    720 }
    721 
    722 const struct tokeninfo {
    723 	const char *name;
    724 	short val;
    725 	short reserved;
    726 } tokentab[] = {
    727 	/* Reserved words */
    728 	{ "if",		IF,	true },
    729 	{ "then",	THEN,	true },
    730 	{ "else",	ELSE,	true },
    731 	{ "elif",	ELIF,	true },
    732 	{ "fi",		FI,	true },
    733 	{ "case",	CASE,	true },
    734 	{ "esac",	ESAC,	true },
    735 	{ "for",	FOR,	true },
    736 	{ "select",	SELECT,	true },
    737 	{ "while",	WHILE,	true },
    738 	{ "until",	UNTIL,	true },
    739 	{ "do",		DO,	true },
    740 	{ "done",	DONE,	true },
    741 	{ "in",		IN,	true },
    742 	{ "function",	FUNCTION, true },
    743 	{ "time",	TIME,	true },
    744 	{ "{",		'{',	true },
    745 	{ "}",		'}',	true },
    746 	{ "!",		BANG,	true },
    747 	{ "[[",		DBRACKET, true },
    748 	/* Lexical tokens (0[EOF], LWORD and REDIR handled specially) */
    749 	{ "&&",		LOGAND,	false },
    750 	{ "||",		LOGOR,	false },
    751 	{ ";;",		BREAK,	false },
    752 	{ "((",		MDPAREN, false },
    753 	{ "|&",		COPROC,	false },
    754 	/* and some special cases... */
    755 	{ "newline",	'\n',	false },
    756 	{ NULL,		0,	false }
    757 };
    758 
    759 void
    760 initkeywords(void)
    761 {
    762 	struct tokeninfo const *tt;
    763 	struct tbl *p;
    764 
    765 	ktinit(&keywords, APERM,
    766 	    /* must be 80% of 2^n (currently 20 keywords) */ 32);
    767 	for (tt = tokentab; tt->name; tt++) {
    768 		if (tt->reserved) {
    769 			p = ktenter(&keywords, tt->name, hash(tt->name));
    770 			p->flag |= DEFINED|ISSET;
    771 			p->type = CKEYWD;
    772 			p->val.i = tt->val;
    773 		}
    774 	}
    775 }
    776 
    777 static void
    778 syntaxerr(const char *what)
    779 {
    780 	char redir[6];	/* 2<<- is the longest redirection, I think */
    781 	const char *s;
    782 	struct tokeninfo const *tt;
    783 	int c;
    784 
    785 	if (!what)
    786 		what = "unexpected";
    787 	REJECT;
    788 	c = token(0);
    789  Again:
    790 	switch (c) {
    791 	case 0:
    792 		if (nesting.start_token) {
    793 			c = nesting.start_token;
    794 			source->errline = nesting.start_line;
    795 			what = "unmatched";
    796 			goto Again;
    797 		}
    798 		/* don't quote the EOF */
    799 		yyerror("%s: unexpected EOF\n", T_synerr);
    800 		/* NOTREACHED */
    801 
    802 	case LWORD:
    803 		s = snptreef(NULL, 32, "%S", yylval.cp);
    804 		break;
    805 
    806 	case REDIR:
    807 		s = snptreef(redir, sizeof(redir), "%R", yylval.iop);
    808 		break;
    809 
    810 	default:
    811 		for (tt = tokentab; tt->name; tt++)
    812 			if (tt->val == c)
    813 			    break;
    814 		if (tt->name)
    815 			s = tt->name;
    816 		else {
    817 			if (c > 0 && c < 256) {
    818 				redir[0] = c;
    819 				redir[1] = '\0';
    820 			} else
    821 				shf_snprintf(redir, sizeof(redir),
    822 					"?%d", c);
    823 			s = redir;
    824 		}
    825 	}
    826 	yyerror("%s: '%s' %s\n", T_synerr, s, what);
    827 }
    828 
    829 static void
    830 nesting_push(struct nesting_state *save, int tok)
    831 {
    832 	*save = nesting;
    833 	nesting.start_token = tok;
    834 	nesting.start_line = source->line;
    835 }
    836 
    837 static void
    838 nesting_pop(struct nesting_state *saved)
    839 {
    840 	nesting = *saved;
    841 }
    842 
    843 static struct op *
    844 newtp(int type)
    845 {
    846 	struct op *t;
    847 
    848 	t = alloc(sizeof(struct op), ATEMP);
    849 	t->type = type;
    850 	t->u.evalflags = 0;
    851 	t->args = NULL;
    852 	t->vars = NULL;
    853 	t->ioact = NULL;
    854 	t->left = t->right = NULL;
    855 	t->str = NULL;
    856 	return (t);
    857 }
    858 
    859 struct op *
    860 compile(Source *s)
    861 {
    862 	nesting.start_token = 0;
    863 	nesting.start_line = 0;
    864 	herep = heres;
    865 	source = s;
    866 	yyparse();
    867 	return (outtree);
    868 }
    869 
    870 /* This kludge exists to take care of sh/AT&T ksh oddity in which
    871  * the arguments of alias/export/readonly/typeset have no field
    872  * splitting, file globbing, or (normal) tilde expansion done.
    873  * AT&T ksh seems to do something similar to this since
    874  *	$ touch a=a; typeset a=[ab]; echo "$a"
    875  *	a=[ab]
    876  *	$ x=typeset; $x a=[ab]; echo "$a"
    877  *	a=a
    878  *	$
    879  */
    880 static int
    881 assign_command(char *s)
    882 {
    883 	if (!*s)
    884 		return (0);
    885 	return ((strcmp(s, "alias") == 0) ||
    886 	    (strcmp(s, "export") == 0) ||
    887 	    (strcmp(s, "readonly") == 0) ||
    888 	    (strcmp(s, T_typeset) == 0));
    889 }
    890 
    891 /* Check if we are in the middle of reading an alias */
    892 static int
    893 inalias(struct source *s)
    894 {
    895 	for (; s && s->type == SALIAS; s = s->next)
    896 		if (!(s->flags & SF_ALIASEND))
    897 			return (1);
    898 	return (0);
    899 }
    900 
    901 
    902 /* Order important - indexed by Test_meta values
    903  * Note that ||, &&, ( and ) can't appear in as unquoted strings
    904  * in normal shell input, so these can be interpreted unambiguously
    905  * in the evaluation pass.
    906  */
    907 static const char dbtest_or[] = { CHAR, '|', CHAR, '|', EOS };
    908 static const char dbtest_and[] = { CHAR, '&', CHAR, '&', EOS };
    909 static const char dbtest_not[] = { CHAR, '!', EOS };
    910 static const char dbtest_oparen[] = { CHAR, '(', EOS };
    911 static const char dbtest_cparen[] = { CHAR, ')', EOS };
    912 const char *const dbtest_tokens[] = {
    913 	dbtest_or, dbtest_and, dbtest_not,
    914 	dbtest_oparen, dbtest_cparen
    915 };
    916 const char db_close[] = { CHAR, ']', CHAR, ']', EOS };
    917 const char db_lthan[] = { CHAR, '<', EOS };
    918 const char db_gthan[] = { CHAR, '>', EOS };
    919 
    920 /*
    921  * Test if the current token is a whatever. Accepts the current token if
    922  * it is. Returns 0 if it is not, non-zero if it is (in the case of
    923  * TM_UNOP and TM_BINOP, the returned value is a Test_op).
    924  */
    925 static Test_op
    926 dbtestp_isa(Test_env *te, Test_meta meta)
    927 {
    928 	int c = tpeek(ARRAYVAR | (meta == TM_BINOP ? 0 : CONTIN));
    929 	int uqword;
    930 	char *save = NULL;
    931 	Test_op ret = TO_NONOP;
    932 
    933 	/* unquoted word? */
    934 	uqword = c == LWORD && *ident;
    935 
    936 	if (meta == TM_OR)
    937 		ret = c == LOGOR ? TO_NONNULL : TO_NONOP;
    938 	else if (meta == TM_AND)
    939 		ret = c == LOGAND ? TO_NONNULL : TO_NONOP;
    940 	else if (meta == TM_NOT)
    941 		ret = (uqword && !strcmp(yylval.cp,
    942 		    dbtest_tokens[(int)TM_NOT])) ? TO_NONNULL : TO_NONOP;
    943 	else if (meta == TM_OPAREN)
    944 		ret = c == '(' /*)*/ ? TO_NONNULL : TO_NONOP;
    945 	else if (meta == TM_CPAREN)
    946 		ret = c == /*(*/ ')' ? TO_NONNULL : TO_NONOP;
    947 	else if (meta == TM_UNOP || meta == TM_BINOP) {
    948 		if (meta == TM_BINOP && c == REDIR &&
    949 		    (yylval.iop->flag == IOREAD || yylval.iop->flag == IOWRITE)) {
    950 			ret = TO_NONNULL;
    951 			save = wdcopy(yylval.iop->flag == IOREAD ?
    952 			    db_lthan : db_gthan, ATEMP);
    953 		} else if (uqword && (ret = test_isop(meta, ident)))
    954 			save = yylval.cp;
    955 	} else /* meta == TM_END */
    956 		ret = (uqword && !strcmp(yylval.cp,
    957 		    db_close)) ? TO_NONNULL : TO_NONOP;
    958 	if (ret != TO_NONOP) {
    959 		ACCEPT;
    960 		if (meta < NELEM(dbtest_tokens))
    961 			save = wdcopy(dbtest_tokens[(int)meta], ATEMP);
    962 		if (save)
    963 			XPput(*te->pos.av, save);
    964 	}
    965 	return (ret);
    966 }
    967 
    968 static const char *
    969 dbtestp_getopnd(Test_env *te, Test_op op MKSH_A_UNUSED,
    970     bool do_eval MKSH_A_UNUSED)
    971 {
    972 	int c = tpeek(ARRAYVAR);
    973 
    974 	if (c != LWORD)
    975 		return (NULL);
    976 
    977 	ACCEPT;
    978 	XPput(*te->pos.av, yylval.cp);
    979 
    980 	return (null);
    981 }
    982 
    983 static int
    984 dbtestp_eval(Test_env *te MKSH_A_UNUSED, Test_op op MKSH_A_UNUSED,
    985     const char *opnd1 MKSH_A_UNUSED, const char *opnd2 MKSH_A_UNUSED,
    986     bool do_eval MKSH_A_UNUSED)
    987 {
    988 	return (1);
    989 }
    990 
    991 static void
    992 dbtestp_error(Test_env *te, int offset, const char *msg)
    993 {
    994 	te->flags |= TEF_ERROR;
    995 
    996 	if (offset < 0) {
    997 		REJECT;
    998 		/* Kludgy to say the least... */
    999 		symbol = LWORD;
   1000 		yylval.cp = *(XPptrv(*te->pos.av) + XPsize(*te->pos.av) +
   1001 		    offset);
   1002 	}
   1003 	syntaxerr(msg);
   1004 }
   1005