xref: /illumos-gate/usr/src/cmd/awk/awk.h (revision c3937c08b16d32677a7dc955a81325f2e7a61b5d)
1 /*
2  * Copyright (C) Lucent Technologies 1997
3  * All Rights Reserved
4  *
5  * Permission to use, copy, modify, and distribute this software and
6  * its documentation for any purpose and without fee is hereby
7  * granted, provided that the above copyright notice appear in all
8  * copies and that both that the copyright notice and this
9  * permission notice and warranty disclaimer appear in supporting
10  * documentation, and that the name Lucent Technologies or any of
11  * its entities not be used in advertising or publicity pertaining
12  * to distribution of the software without specific, written prior
13  * permission.
14  *
15  * LUCENT DISCLAIMS ALL WARRANTIES WITH REGARD TO THIS SOFTWARE,
16  * INCLUDING ALL IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS.
17  * IN NO EVENT SHALL LUCENT OR ANY OF ITS ENTITIES BE LIABLE FOR ANY
18  * SPECIAL, INDIRECT OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES
19  * WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER
20  * IN AN ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION,
21  * ARISING OUT OF OR IN CONNECTION WITH THE USE OR PERFORMANCE OF
22  * THIS SOFTWARE.
23  */
24 
25 /*
26  * CDDL HEADER START
27  *
28  * The contents of this file are subject to the terms of the
29  * Common Development and Distribution License (the "License").
30  * You may not use this file except in compliance with the License.
31  *
32  * You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE
33  * or http://www.opensolaris.org/os/licensing.
34  * See the License for the specific language governing permissions
35  * and limitations under the License.
36  *
37  * When distributing Covered Code, include this CDDL HEADER in each
38  * file and include the License file at usr/src/OPENSOLARIS.LICENSE.
39  * If applicable, add the following below this CDDL HEADER, with the
40  * fields enclosed by brackets "[]" replaced with your own identifying
41  * information: Portions Copyright [yyyy] [name of copyright owner]
42  *
43  * CDDL HEADER END
44  */
45 
46 /*
47  * Copyright (c) 1989, 2010, Oracle and/or its affiliates. All rights reserved.
48  */
49 
50 /*	Copyright (c) 1984, 1986, 1987, 1988, 1989 AT&T	*/
51 /*	  All Rights Reserved  	*/
52 
53 #ifndef AWK_H
54 #define	AWK_H
55 
56 #include <assert.h>
57 #include <sys/types.h>
58 #include <ctype.h>
59 #include <stdio.h>
60 #include <stdlib.h>
61 #include <string.h>
62 #include <libintl.h>
63 #include <limits.h>
64 
65 typedef double	Awkfloat;
66 
67 /* unsigned char is more trouble than it's worth */
68 
69 typedef	unsigned char uschar;
70 
71 #define	xfree(a)	{ if ((a) != NULL) { free((void *)(a)); (a) = NULL; } }
72 
73 /* guaranteed non-null for dprintf */
74 #define	NN(p)	((p) ? (p) : "(null)")
75 #define	DEBUG
76 #ifdef	DEBUG
77 			/* uses have to be doubly parenthesized */
78 #define	dprintf(x)	if (dbg) (void) printf x
79 #else
80 #define	dprintf(x)
81 #endif
82 
83 extern int	compile_time;	/* 1 if compiling, 0 if running */
84 extern int	safe;		/* 0 => unsafe, 1 => safe */
85 
86 #define	FLD_INCR	64
87 #define	LINE_INCR	256
88 #define	RECSIZE	(8 * 1024)	/* sets limit on records, fields, etc., etc. */
89 extern size_t	recsize;	/* size of current record, orig RECSIZE */
90 
91 /* ensure that there is extra 1 byte in the buffer */
92 #define	expand_buf(p, n, r)	\
93 	if (*(n) == 0 || (r) >= (*(n) - 1)) r_expand_buf(p, n, r)
94 
95 extern char	**FS;
96 extern char	**RS;
97 extern char	**ORS;
98 extern char	**OFS;
99 extern char	**OFMT;
100 extern Awkfloat *NR;
101 extern Awkfloat *FNR;
102 extern Awkfloat *NF;
103 extern char	**FILENAME;
104 extern char	**SUBSEP;
105 extern Awkfloat *RSTART;
106 extern Awkfloat *RLENGTH;
107 
108 extern char	*record;	/* points to $0 */
109 extern size_t	recsize;
110 extern int	errorflag;	/* 1 if error has occurred */
111 extern int	donefld;	/* 1 if record broken into fields */
112 extern int	donerec;	/* 1 if record is valid (no fld has changed */
113 
114 extern	char	*patbeg;	/* beginning of pattern matched */
115 extern	int	patlen;		/* length of pattern matched. set in b.c */
116 
117 /* Cell:  all information about a variable or constant */
118 
119 typedef struct Cell {
120 	uschar	ctype;		/* OCELL, OBOOL, OJUMP, etc. */
121 	uschar	csub;		/* CCON, CTEMP, CFLD, etc. */
122 	char	*nval;		/* name, for variables only */
123 	char	*sval;		/* string value */
124 	Awkfloat fval;		/* value as number */
125 	int	 tval;
126 		/* type info: STR|NUM|ARR|FCN|FLD|CON|DONTFREE|CONVC|CONVO */
127 	char	*fmt;
128 		/* CONVFMT/OFMT value used to convert from number */
129 	struct Cell *cnext;	/* ptr to next if chained */
130 } Cell;
131 
132 typedef struct Array {		/* symbol table array */
133 	int	nelem;		/* elements in table right now */
134 	int	size;		/* size of tab */
135 	Cell	**tab;		/* hash table pointers */
136 } Array;
137 
138 #define	NSYMTAB	50	/* initial size of a symbol table */
139 extern Array	*symtab, *makesymtab(int);
140 extern Cell	*setsymtab(const char *, const char *, Awkfloat,
141 		    unsigned int, Array *);
142 extern Cell	*lookup(const char *, Array *);
143 
144 extern Cell	*recloc;	/* location of input record */
145 extern Cell	*nrloc;		/* NR */
146 extern Cell	*fnrloc;	/* FNR */
147 extern Cell	*fsloc;		/* FS */
148 extern Cell	*nfloc;		/* NF */
149 extern Cell	*ofsloc;	/* OFS */
150 extern Cell	*orsloc;	/* ORS */
151 extern Cell	*rsloc;		/* RS */
152 extern Cell	*rtloc;		/* RT */
153 extern Cell	*rstartloc;	/* RSTART */
154 extern Cell	*rlengthloc;	/* RLENGTH */
155 extern Cell	*subseploc;	/* SUBSEP */
156 extern Cell	*symtabloc;	/* SYMTAB */
157 
158 /* Cell.tval values: */
159 #define	NUM	01	/* number value is valid */
160 #define	STR	02	/* string value is valid */
161 #define	DONTFREE 04	/* string space is not freeable */
162 #define	CON	010	/* this is a constant */
163 #define	ARR	020	/* this is an array */
164 #define	FCN	040	/* this is a function name */
165 #define	FLD	0100	/* this is a field $1, $2, ... */
166 #define	REC	0200	/* this is $0 */
167 #define	CONVC	0400	/* string was converted from number via CONVFMT */
168 #define	CONVO	01000	/* string was converted from number via OFMT */
169 
170 
171 extern Awkfloat	setfval(Cell *, Awkfloat);
172 extern Awkfloat	getfval(Cell *);
173 extern char	*setsval(Cell *, const char *);
174 extern char	*getsval(Cell *);
175 extern char	*getpssval(Cell *);	/* for print */
176 extern char	*tostring(const char *);
177 extern char	*tokname(int);
178 extern char	*qstring(const char *, int);
179 
180 /* function types */
181 #define	FLENGTH	1
182 #define	FSQRT	2
183 #define	FEXP	3
184 #define	FLOG	4
185 #define	FINT	5
186 #define	FSYSTEM	6
187 #define	FRAND	7
188 #define	FSRAND	8
189 #define	FSIN	9
190 #define	FCOS	10
191 #define	FATAN	11
192 #define	FTOUPPER 12
193 #define	FTOLOWER 13
194 #define	FFLUSH	14
195 
196 /* Node:  parse tree is made of nodes, with Cell's at bottom */
197 
198 typedef struct Node {
199 	int	ntype;
200 	struct	Node *nnext;
201 	off_t	lineno;
202 	int	nobj;
203 	struct	Node *narg[1];
204 		/* variable: actual size set by calling malloc */
205 } Node;
206 
207 #define	NIL	((Node *)0)
208 
209 extern Node	*winner;
210 extern Node	*nullstat;
211 extern Node	*nullnode;
212 
213 /* ctypes */
214 #define	OCELL	1
215 #define	OBOOL	2
216 #define	OJUMP	3
217 
218 /* Cell subtypes: csub */
219 #define	CFREE	7
220 #define	CCOPY	6
221 #define	CCON	5
222 #define	CTEMP	4
223 #define	CNAME	3
224 #define	CVAR	2
225 #define	CFLD	1
226 #define	CUNK	0
227 
228 /* bool subtypes */
229 #define	BTRUE	11
230 #define	BFALSE	12
231 
232 /* jump subtypes */
233 #define	JEXIT	21
234 #define	JNEXT	22
235 #define	JBREAK	23
236 #define	JCONT	24
237 #define	JRET	25
238 #define	JNEXTFILE	26
239 
240 /* node types */
241 #define	NVALUE	1
242 #define	NSTAT	2
243 #define	NEXPR	3
244 #define	NFIELD	4
245 
246 extern	Cell	*(*proctab[])(Node **, int);
247 extern	Cell	*nullproc(Node **, int);
248 extern	int	*pairstack, paircnt;
249 
250 extern	Node	*stat1(int, Node *), *stat2(int, Node *, Node *);
251 extern	Node	*stat3(int, Node *, Node *, Node *);
252 extern	Node	*stat4(int, Node *, Node *, Node *, Node *);
253 extern	Node	*pa2stat(Node *, Node *, Node *);
254 extern	Node	*op1(int, Node *), *op2(int, Node *, Node *);
255 extern	Node	*op3(int, Node *, Node *, Node *);
256 extern	Node	*op4(int, Node *, Node *, Node *, Node *);
257 extern	Node	*linkum(Node *, Node *), *celltonode(Cell *, int);
258 extern	Node	*rectonode(void), *exptostat(Node *);
259 extern	Node	*makearr(Node *);
260 
261 #define	notlegal(n)	\
262 	(n <= FIRSTTOKEN || n >= LASTTOKEN || proctab[n-FIRSTTOKEN] == nullproc)
263 #define	isvalue(n)	((n)->ntype == NVALUE)
264 #define	isexpr(n)	((n)->ntype == NEXPR)
265 #define	isjump(n)	((n)->ctype == OJUMP)
266 #define	isexit(n)	((n)->csub == JEXIT)
267 #define	isbreak(n)	((n)->csub == JBREAK)
268 #define	iscont(n)	((n)->csub == JCONT)
269 #define	isnext(n)	((n)->csub == JNEXT || (n)->csub == JNEXTFILE)
270 #define	isret(n)	((n)->csub == JRET)
271 #define	isrec(n)	((n)->tval & REC)
272 #define	isfld(n)	((n)->tval & FLD)
273 #define	isstr(n)	((n)->tval & STR)
274 #define	isnum(n)	((n)->tval & NUM)
275 #define	isarr(n)	((n)->tval & ARR)
276 #define	isfcn(n)	((n)->tval & FCN)
277 #define	istrue(n)	((n)->csub == BTRUE)
278 #define	istemp(n)	((n)->csub == CTEMP)
279 #define	freeable(p)	(((p)->tval & (STR|DONTFREE)) == STR)
280 
281 /* structures used by regular expression matching machinery, mostly b.c: */
282 
283 /* 256 handles 8-bit chars; 128 does 7-bit */
284 /* watch out in match(), etc. */
285 #define	NCHARS	(256+3)
286 #define	NSTATES	32
287 
288 typedef struct rrow {
289 	long	ltype;	/* long avoids pointer warnings on 64-bit */
290 	union {
291 		int i;
292 		Node *np;
293 		uschar *up;
294 	} lval;		/* because Al stores a pointer in it! */
295 	int	*lfollow;
296 } rrow;
297 
298 typedef struct fa {
299 	uschar	gototab[NSTATES][NCHARS];
300 	uschar	out[NSTATES];
301 	uschar	*restr;
302 	int	*posns[NSTATES];
303 	int	anchor;
304 	int	use;
305 	int	initstat;
306 	int	curstat;
307 	int	accept;
308 	int	reset;
309 	/* re is variable: actual size set by calling malloc */
310 	struct	rrow re[1];
311 } fa;
312 
313 /* lex.c */
314 extern	int	yylex(void);
315 extern	void	startreg(void);
316 extern	int	input(void);
317 extern	void	unput(int);
318 extern	void	unputstr(const char *);
319 extern	int	yylook(void);
320 extern	int	yyback(int *, int);
321 extern	int	yyinput(void);
322 
323 /* parse.c */
324 extern	void	defn(Cell *, Node *, Node *);
325 extern	int	ptoi(void *);
326 extern	Node	*itonp(int);
327 extern	int	isarg(const char *);
328 
329 /* b.c */
330 extern	fa	*makedfa(const char *, int);
331 extern	int	nematch(fa *, const char *);
332 extern	int	match(fa *, const char *);
333 extern	int	pmatch(fa *, const char *);
334 
335 /* lib.c */
336 
337 extern	void	SYNTAX(const char *, ...);
338 extern	void	FATAL(const char *, ...) __attribute__((__noreturn__));
339 extern	void	WARNING(const char *, ...);
340 extern	void	error(void);
341 extern	void	nextfile(void);
342 extern	void	savefs(void);
343 
344 extern	int	isclvar(const char *);
345 extern	int	is_number(const char *);
346 extern	void	setclvar(char *);
347 extern	int	readrec(char **, size_t *, FILE *);
348 extern	void	bracecheck(void);
349 extern	void	recinit(unsigned int n);
350 extern	void	syminit(void);
351 extern	void	yyerror(const char *);
352 extern	void	fldbld(void);
353 extern	void	recbld(void);
354 extern	int	getrec(char **, size_t *, int);
355 extern	Cell	*fieldadr(int);
356 extern	void	newfld(int);
357 extern	int	fldidx(Cell *);
358 extern	double	errcheck(double, const char *);
359 extern	void	fpecatch(int);
360 extern	void	r_expand_buf(char **, size_t *, size_t);
361 extern	void	makefields(int, int);
362 extern	void	growfldtab(int n);
363 extern	void	setlastfld(int n);
364 
365 /* main.c */
366 extern	int	dbg;
367 extern	char	*lexprog;
368 extern	int	compile_time;
369 extern	char	*cursource(void);
370 extern	int	pgetc(void);
371 
372 /* tran.c */
373 extern	void	syminit(void);
374 extern	void	arginit(int, char **);
375 extern	void	envinit(char **);
376 extern	void	freesymtab(Cell *);
377 extern	void	freeelem(Cell *, const char *);
378 extern	void	funnyvar(Cell *, const char *);
379 extern	int	hash(const char *, int);
380 extern	Awkfloat *ARGC;
381 
382 /* run.c */
383 extern	void		run(Node *);
384 extern	const char	*filename(FILE *);
385 extern	int		adjbuf(char **pb, size_t *sz, size_t min, size_t q,
386 			    char **pbp, const char *what);
387 
388 extern	int	paircnt;
389 extern	Node	*winner;
390 
391 #ifndef input
392 extern	int	input(void);
393 #endif
394 extern	int	yyparse(void);
395 extern	FILE	*yyin;
396 extern	off_t	lineno;
397 
398 /* proc */
399 extern Cell *nullproc(Node **, int);
400 extern Cell *program(Node **, int);
401 extern Cell *boolop(Node **, int);
402 extern Cell *relop(Node **, int);
403 extern Cell *array(Node **, int);
404 extern Cell *indirect(Node **, int);
405 extern Cell *substr(Node **, int);
406 extern Cell *sub(Node **, int);
407 extern Cell *gsub(Node **, int);
408 extern Cell *sindex(Node **, int);
409 extern Cell *awksprintf(Node **, int);
410 extern Cell *arith(Node **, int);
411 extern Cell *incrdecr(Node **, int);
412 extern Cell *cat(Node **, int);
413 extern Cell *pastat(Node **, int);
414 extern Cell *dopa2(Node **, int);
415 extern Cell *matchop(Node **, int);
416 extern Cell *intest(Node **, int);
417 extern Cell *awkprintf(Node **, int);
418 extern Cell *printstat(Node **, int);
419 extern Cell *closefile(Node **, int);
420 extern Cell *awkdelete(Node **, int);
421 extern Cell *split(Node **, int);
422 extern Cell *assign(Node **, int);
423 extern Cell *condexpr(Node **, int);
424 extern Cell *ifstat(Node **, int);
425 extern Cell *whilestat(Node **, int);
426 extern Cell *forstat(Node **, int);
427 extern Cell *dostat(Node **, int);
428 extern Cell *instat(Node **, int);
429 extern Cell *jump(Node **, int);
430 extern Cell *bltin(Node **, int);
431 extern Cell *call(Node **, int);
432 extern Cell *arg(Node **, int);
433 extern Cell *getnf(Node **, int);
434 extern Cell *awkgetline(Node **, int);
435 
436 #endif /* AWK_H */
437