xref: /openbsd-src/usr.bin/less/decode.c (revision a4afd6dad3fba28f80e70208181c06c482259988)
1 /*
2  * Copyright (c) 1984,1985,1989,1994,1995  Mark Nudelman
3  * All rights reserved.
4  *
5  * Redistribution and use in source and binary forms, with or without
6  * modification, are permitted provided that the following conditions
7  * are met:
8  * 1. Redistributions of source code must retain the above copyright
9  *    notice, this list of conditions and the following disclaimer.
10  * 2. Redistributions in binary form must reproduce the above copyright
11  *    notice in the documentation and/or other materials provided with
12  *    the distribution.
13  *
14  * THIS SOFTWARE IS PROVIDED BY THE AUTHOR ``AS IS'' AND ANY
15  * EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
16  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR
17  * PURPOSE ARE DISCLAIMED.  IN NO EVENT SHALL THE AUTHOR BE LIABLE
18  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
19  * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT
20  * OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR
21  * BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY,
22  * WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
23  * OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN
24  * IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
25  */
26 
27 
28 /*
29  * Routines to decode user commands.
30  *
31  * This is all table driven.
32  * A command table is a sequence of command descriptors.
33  * Each command descriptor is a sequence of bytes with the following format:
34  *	<c1><c2>...<cN><0><action>
35  * The characters c1,c2,...,cN are the command string; that is,
36  * the characters which the user must type.
37  * It is terminated by a null <0> byte.
38  * The byte after the null byte is the action code associated
39  * with the command string.
40  * If an action byte is OR-ed with A_EXTRA, this indicates
41  * that the option byte is followed by an extra string.
42  *
43  * There may be many command tables.
44  * The first (default) table is built-in.
45  * Other tables are read in from "lesskey" files.
46  * All the tables are linked together and are searched in order.
47  */
48 
49 #include "less.h"
50 #include "cmd.h"
51 #include "lesskey.h"
52 
53 extern int erase_char, kill_char;
54 
55 /*
56  * Command table is ordered roughly according to expected
57  * frequency of use, so the common commands are near the beginning.
58  */
59 static unsigned char cmdtable[] =
60 {
61 	'\r',0,				A_F_LINE,
62 	'\n',0,				A_F_LINE,
63 	'e',0,				A_F_LINE,
64 	'j',0,				A_F_LINE,
65 	CONTROL('E'),0,			A_F_LINE,
66 	CONTROL('N'),0,			A_F_LINE,
67 	'k',0,				A_B_LINE,
68 	'y',0,				A_B_LINE,
69 	CONTROL('Y'),0,			A_B_LINE,
70 	CONTROL('K'),0,			A_B_LINE,
71 	CONTROL('P'),0,			A_B_LINE,
72 	'J',0,				A_FF_LINE,
73 	'K',0,				A_BF_LINE,
74 	'Y',0,				A_BF_LINE,
75 	'd',0,				A_F_SCROLL,
76 	CONTROL('D'),0,			A_F_SCROLL,
77 	'u',0,				A_B_SCROLL,
78 	CONTROL('U'),0,			A_B_SCROLL,
79 	' ',0,				A_F_SCREEN,
80 	'f',0,				A_F_SCREEN,
81 	CONTROL('F'),0,			A_F_SCREEN,
82 	CONTROL('V'),0,			A_F_SCREEN,
83 	'b',0,				A_B_SCREEN,
84 	CONTROL('B'),0,			A_B_SCREEN,
85 	ESC,'v',0,			A_B_SCREEN,
86 	'z',0,				A_F_WINDOW,
87 	'w',0,				A_B_WINDOW,
88 	'F',0,				A_F_FOREVER,
89 	'R',0,				A_FREPAINT,
90 	'r',0,				A_REPAINT,
91 	CONTROL('R'),0,			A_REPAINT,
92 	CONTROL('L'),0,			A_REPAINT,
93 	ESC,'u',0,			A_UNDO_SEARCH,
94 	'g',0,				A_GOLINE,
95 	'<',0,				A_GOLINE,
96 	ESC,'<',0,			A_GOLINE,
97 	'p',0,				A_PERCENT,
98 	'%',0,				A_PERCENT,
99 	'{',0,				A_F_BRACKET|A_EXTRA,	'{','}',0,
100 	'}',0,				A_B_BRACKET|A_EXTRA,	'{','}',0,
101 	'(',0,				A_F_BRACKET|A_EXTRA,	'(',')',0,
102 	')',0,				A_B_BRACKET|A_EXTRA,	'(',')',0,
103 	'[',0,				A_F_BRACKET|A_EXTRA,	'[',']',0,
104 	']',0,				A_B_BRACKET|A_EXTRA,	'[',']',0,
105 	ESC,CONTROL('F'),0,		A_F_BRACKET,
106 	ESC,CONTROL('B'),0,		A_B_BRACKET,
107 	'G',0,				A_GOEND,
108 	ESC,'>',0,			A_GOEND,
109 	'>',0,				A_GOEND,
110 	'P',0,				A_GOPOS,
111 
112 	'0',0,				A_DIGIT,
113 	'1',0,				A_DIGIT,
114 	'2',0,				A_DIGIT,
115 	'3',0,				A_DIGIT,
116 	'4',0,				A_DIGIT,
117 	'5',0,				A_DIGIT,
118 	'6',0,				A_DIGIT,
119 	'7',0,				A_DIGIT,
120 	'8',0,				A_DIGIT,
121 	'9',0,				A_DIGIT,
122 
123 	'=',0,				A_STAT,
124 	CONTROL('G'),0,			A_STAT,
125 	':','f',0,			A_STAT,
126 	'/',0,				A_F_SEARCH,
127 	'?',0,				A_B_SEARCH,
128 	ESC,'/',0,			A_F_SEARCH|A_EXTRA,	'*',0,
129 	ESC,'?',0,			A_B_SEARCH|A_EXTRA,	'*',0,
130 	'n',0,				A_AGAIN_SEARCH,
131 	ESC,'n',0,			A_T_AGAIN_SEARCH,
132 	'N',0,				A_REVERSE_SEARCH,
133 	ESC,'N',0,			A_T_REVERSE_SEARCH,
134 	'm',0,				A_SETMARK,
135 	'\'',0,				A_GOMARK,
136 	CONTROL('X'),CONTROL('X'),0,	A_GOMARK,
137 	'E',0,				A_EXAMINE,
138 	':','e',0,			A_EXAMINE,
139 	CONTROL('X'),CONTROL('V'),0,	A_EXAMINE,
140 	':','n',0,			A_NEXT_FILE,
141 	':','p',0,			A_PREV_FILE,
142 	':','x',0,			A_INDEX_FILE,
143 	'-',0,				A_OPT_TOGGLE,
144 	':','t',0,			A_OPT_TOGGLE|A_EXTRA,	't',0,
145 	's',0,				A_OPT_TOGGLE|A_EXTRA,	'o',0,
146 	'_',0,				A_DISP_OPTION,
147 	'|',0,				A_PIPE,
148 	'v',0,				A_VISUAL,
149 	'!',0,				A_SHELL,
150 	'+',0,				A_FIRSTCMD,
151 
152 	'H',0,				A_HELP,
153 	'h',0,				A_HELP,
154 	'V',0,				A_VERSION,
155 	'q',0,				A_QUIT,
156 	':','q',0,			A_QUIT,
157 	':','Q',0,			A_QUIT,
158 	'Z','Z',0,			A_QUIT
159 };
160 
161 static unsigned char edittable[] =
162 {
163 	'\t',0,	    		EC_F_COMPLETE,	/* TAB */
164 	'\17',0,		EC_B_COMPLETE,	/* BACKTAB */
165 	'\14',0,		EC_EXPAND,	/* CTRL-L */
166 	CONTROL('V'),0,		EC_LITERAL,	/* BACKSLASH */
167 	CONTROL('A'),0,		EC_LITERAL,	/* BACKSLASH */
168    	ESC,'l',0,		EC_RIGHT,	/* ESC l */
169 	ESC,'h',0,		EC_LEFT,	/* ESC h */
170 	ESC,'b',0,		EC_W_LEFT,	/* ESC b */
171 	ESC,'w',0,		EC_W_RIGHT,	/* ESC w */
172 	ESC,'i',0,		EC_INSERT,	/* ESC i */
173 	ESC,'x',0,		EC_DELETE,	/* ESC x */
174 	ESC,'X',0,		EC_W_DELETE,	/* ESC X */
175 	ESC,'\b',0,		EC_W_BACKSPACE,	/* ESC BACKSPACE */
176 	ESC,'0',0,		EC_HOME,	/* ESC 0 */
177 	ESC,'$',0,		EC_END,		/* ESC $ */
178 	ESC,'k',0,		EC_UP,		/* ESC k */
179 	ESC,'j',0,		EC_DOWN,	/* ESC j */
180 	ESC,'\t',0,		EC_B_COMPLETE	/* ESC TAB */
181 };
182 
183 /*
184  * Structure to support a list of command tables.
185  */
186 struct tablelist
187 {
188 	struct tablelist *t_next;
189 	char *t_start;
190 	char *t_end;
191 };
192 
193 /*
194  * List of command tables and list of line-edit tables.
195  */
196 static struct tablelist *list_fcmd_tables = NULL;
197 static struct tablelist *list_ecmd_tables = NULL;
198 
199 
200 /*
201  * Initialize the command lists.
202  */
203 	public void
204 init_cmds()
205 {
206 	/*
207 	 * Add the default command tables.
208 	 */
209 	add_fcmd_table((char*)cmdtable, sizeof(cmdtable));
210 	add_ecmd_table((char*)edittable, sizeof(edittable));
211 	get_editkeys();
212 #if USERFILE
213 	/*
214 	 * Try to add the tables in the standard lesskey file "$HOME/.less".
215 	 */
216 	add_hometable();
217 #endif
218 }
219 
220 /*
221  *
222  */
223 	static int
224 add_cmd_table(tlist, buf, len)
225 	struct tablelist **tlist;
226 	char *buf;
227 	int len;
228 {
229 	register struct tablelist *t;
230 
231 	if (len == 0)
232 		return (0);
233 	/*
234 	 * Allocate a tablelist structure, initialize it,
235 	 * and link it into the list of tables.
236 	 */
237 	if ((t = (struct tablelist *)
238 			calloc(1, sizeof(struct tablelist))) == NULL)
239 	{
240 		return (-1);
241 	}
242 	t->t_start = buf;
243 	t->t_end = buf + len;
244 	t->t_next = *tlist;
245 	*tlist = t;
246 	return (0);
247 }
248 
249 /*
250  * Add a command table.
251  */
252 	public void
253 add_fcmd_table(buf, len)
254 	char *buf;
255 	int len;
256 {
257 	if (add_cmd_table(&list_fcmd_tables, buf, len) < 0)
258 		error("Warning: some commands disabled", NULL_PARG);
259 }
260 
261 /*
262  * Add an editing command table.
263  */
264 	public void
265 add_ecmd_table(buf, len)
266 	char *buf;
267 	int len;
268 {
269 	if (add_cmd_table(&list_ecmd_tables, buf, len) < 0)
270 		error("Warning: some edit commands disabled", NULL_PARG);
271 }
272 
273 /*
274  * Search a single command table for the command string in cmd.
275  */
276 	public int
277 cmd_search(cmd, table, endtable, sp)
278 	char *cmd;
279 	char *table;
280 	char *endtable;
281 	char **sp;
282 {
283 	register char *p;
284 	register char *q;
285 	register int a;
286 
287 	for (p = table, q = cmd;  p < endtable;  p++, q++)
288 	{
289 		if (*p == *q)
290 		{
291 			/*
292 			 * Current characters match.
293 			 * If we're at the end of the string, we've found it.
294 			 * Return the action code, which is the character
295 			 * after the null at the end of the string
296 			 * in the command table.
297 			 */
298 			if (*p == '\0')
299 			{
300 				a = *++p & 0377;
301 				if (a == A_END_LIST)
302 				{
303 					/*
304 					 * We get here only if the original
305 					 * cmd string passed in was empty ("").
306 					 * I don't think that can happen,
307 					 * but just in case ...
308 					 */
309 					return (A_UINVALID);
310 				}
311 				/*
312 				 * Check for an "extra" string.
313 				 */
314 				if (a & A_EXTRA)
315 				{
316 					*sp = ++p;
317 					a &= ~A_EXTRA;
318 				} else
319 					*sp = NULL;
320 				return (a);
321 			}
322 		} else if (*q == '\0')
323 		{
324 			/*
325 			 * Hit the end of the user's command,
326 			 * but not the end of the string in the command table.
327 			 * The user's command is incomplete.
328 			 */
329 			return (A_PREFIX);
330 		} else
331 		{
332 			/*
333 			 * Not a match.
334 			 * Skip ahead to the next command in the
335 			 * command table, and reset the pointer
336 			 * to the beginning of the user's command.
337 			 */
338 			if (*p == '\0' && p[1] == A_END_LIST)
339 			{
340 				/*
341 				 * A_END_LIST is a special marker that tells
342 				 * us to abort the cmd search.
343 				 */
344 				return (A_UINVALID);
345 			}
346 			while (*p++ != '\0') ;
347 			if (*p & A_EXTRA)
348 				while (*++p != '\0') ;
349 			q = cmd-1;
350 		}
351 	}
352 	/*
353 	 * No match found in the entire command table.
354 	 */
355 	return (A_INVALID);
356 }
357 
358 /*
359  * Decode a command character and return the associated action.
360  * The "extra" string, if any, is returned in sp.
361  */
362 	static int
363 cmd_decode(tlist, cmd, sp)
364 	struct tablelist *tlist;
365 	char *cmd;
366 	char **sp;
367 {
368 	register struct tablelist *t;
369 	register int action = A_INVALID;
370 
371 	/*
372 	 * Search thru all the command tables.
373 	 * Stop when we find an action which is not A_INVALID.
374 	 */
375 	for (t = tlist;  t != NULL;  t = t->t_next)
376 	{
377 		action = cmd_search(cmd, t->t_start, t->t_end, sp);
378 		if (action != A_INVALID)
379 			break;
380 	}
381 	return (action);
382 }
383 
384 /*
385  * Decode a command from the cmdtables list.
386  */
387 	public int
388 fcmd_decode(cmd, sp)
389 	char *cmd;
390 	char **sp;
391 {
392 	return (cmd_decode(list_fcmd_tables, cmd, sp));
393 }
394 
395 /*
396  * Decode a command from the edittables list.
397  */
398 	public int
399 ecmd_decode(cmd, sp)
400 	char *cmd;
401 	char **sp;
402 {
403 	return (cmd_decode(list_ecmd_tables, cmd, sp));
404 }
405 
406 #if USERFILE
407 	static int
408 gint(sp)
409 	char **sp;
410 {
411 	int n;
412 
413 	n = *(*sp)++;
414 	n += *(*sp)++ * KRADIX;
415 	return (n);
416 }
417 
418 	static int
419 old_lesskey(buf, len)
420 	char *buf;
421 	int len;
422 {
423 	/*
424 	 * Old-style lesskey file.
425 	 * The file must end with either
426 	 *     ...,cmd,0,action
427 	 * or  ...,cmd,0,action|A_EXTRA,string,0
428 	 * So the last byte or the second to last byte must be zero.
429 	 */
430 	if (buf[len-1] != '\0' && buf[len-2] != '\0')
431 		return (-1);
432 	add_fcmd_table(buf, len);
433 	return (0);
434 }
435 
436 	static int
437 new_lesskey(buf, len)
438 	char *buf;
439 	int len;
440 {
441 	char *p;
442 	register int c;
443 	register int done;
444 	register int n;
445 
446 	/*
447 	 * New-style lesskey file.
448 	 * Extract the pieces.
449 	 */
450 	if (buf[len-3] != C0_END_LESSKEY_MAGIC ||
451 	    buf[len-2] != C1_END_LESSKEY_MAGIC ||
452 	    buf[len-1] != C2_END_LESSKEY_MAGIC)
453 		return (-1);
454 	p = buf + 4;
455 	done = 0;
456 	while (!done)
457 	{
458 		c = *p++;
459 		switch (c)
460 		{
461 		case CMD_SECTION:
462 			n = gint(&p);
463 			add_fcmd_table(p, n);
464 			p += n;
465 			break;
466 		case EDIT_SECTION:
467 			n = gint(&p);
468 			add_ecmd_table(p, n);
469 			p += n;
470 			break;
471 		case END_SECTION:
472 			done = 1;
473 			break;
474 		default:
475 			free(buf);
476 			return (-1);
477 		}
478 	}
479 	return (0);
480 }
481 
482 /*
483  * Set up a user command table, based on a "lesskey" file.
484  */
485 	public int
486 lesskey(filename)
487 	char *filename;
488 {
489 	register char *buf;
490 	register POSITION len;
491 	register long n;
492 	register int f;
493 
494 	/*
495 	 * Try to open the lesskey file.
496 	 */
497 	f = open(filename, OPEN_READ);
498 	if (f < 0)
499 		return (1);
500 
501 	/*
502 	 * Read the file into a buffer.
503 	 * We first figure out the size of the file and allocate space for it.
504 	 * {{ Minimal error checking is done here.
505 	 *    A garbage .less file will produce strange results.
506 	 *    To avoid a large amount of error checking code here, we
507 	 *    rely on the lesskey program to generate a good .less file. }}
508 	 */
509 	len = filesize(f);
510 	if (len == NULL_POSITION || len < 3)
511 	{
512 		/*
513 		 * Bad file (valid file must have at least 3 chars).
514 		 */
515 		close(f);
516 		return (-1);
517 	}
518 	if ((buf = (char *) calloc((int)len, sizeof(char))) == NULL)
519 	{
520 		close(f);
521 		return (-1);
522 	}
523 	if (lseek(f, (off_t)0, 0) == BAD_LSEEK)
524 	{
525 		free(buf);
526 		close(f);
527 		return (-1);
528 	}
529 	n = read(f, buf, (unsigned int) len);
530 	close(f);
531 	if (n != len)
532 	{
533 		free(buf);
534 		return (-1);
535 	}
536 
537 	/*
538 	 * Figure out if this is an old-style (before version 241)
539 	 * or new-style lesskey file format.
540 	 */
541 	if (buf[0] != C0_LESSKEY_MAGIC || buf[1] != C1_LESSKEY_MAGIC ||
542 	    buf[2] != C2_LESSKEY_MAGIC || buf[3] != C3_LESSKEY_MAGIC)
543 		return (old_lesskey(buf, (int)len));
544 	return (new_lesskey(buf, (int)len));
545 }
546 
547 /*
548  * Add the standard lesskey file "$HOME/.less"
549  */
550 	public void
551 add_hometable()
552 {
553 	char *filename;
554 	PARG parg;
555 
556 	filename = homefile(LESSKEYFILE);
557 	if (filename == NULL)
558 		return;
559 	if (lesskey(filename) < 0)
560 	{
561 		parg.p_string = filename;
562 		error("Cannot use lesskey file \"%s\"", &parg);
563 	}
564 	free(filename);
565 }
566 #endif
567 
568 /*
569  * See if a char is a special line-editing command.
570  */
571 	public int
572 editchar(c, flags)
573 	int c;
574 	int flags;
575 {
576 	int action;
577 	int nch;
578 	char *s;
579 	char usercmd[MAX_CMDLEN+1];
580 
581 	/*
582 	 * An editing character could actually be a sequence of characters;
583 	 * for example, an escape sequence sent by pressing the uparrow key.
584 	 * To match the editing string, we use the command decoder
585 	 * but give it the edit-commands command table
586 	 * This table is constructed to match the user's keyboard.
587 	 */
588 	if (c == erase_char)
589 		return (EC_BACKSPACE);
590 	if (c == kill_char)
591 		return (EC_LINEKILL);
592 
593 	/*
594 	 * Collect characters in a buffer.
595 	 * Start with the one we have, and get more if we need them.
596 	 */
597 	nch = 0;
598 	do {
599 	  	if (nch > 0)
600 			c = getcc();
601 		usercmd[nch] = c;
602 		usercmd[nch+1] = '\0';
603 		nch++;
604 		action = ecmd_decode(usercmd, &s);
605 	} while (action == A_PREFIX);
606 
607 	if (flags & EC_NOHISTORY)
608 	{
609 		/*
610 		 * The caller says there is no history list.
611 		 * Reject any history-manipulation action.
612 		 */
613 		switch (action)
614 		{
615 		case EC_UP:
616 		case EC_DOWN:
617 			action = A_INVALID;
618 			break;
619 		}
620 	}
621 	if (flags & EC_NOCOMPLETE)
622 	{
623 		/*
624 		 * The caller says we don't want any filename completion cmds.
625 		 * Reject them.
626 		 */
627 		switch (action)
628 		{
629 		case EC_F_COMPLETE:
630 		case EC_B_COMPLETE:
631 		case EC_EXPAND:
632 			action = A_INVALID;
633 			break;
634 		}
635 	}
636 	if ((flags & EC_PEEK) || action == A_INVALID)
637 	{
638 		/*
639 		 * We're just peeking, or we didn't understand the command.
640 		 * Unget all the characters we read in the loop above.
641 		 * This does NOT include the original character that was
642 		 * passed in as a parameter.
643 		 */
644 		while (nch > 1) {
645 			ungetcc(usercmd[--nch]);
646 		}
647 	} else
648 	{
649 		if (s != NULL)
650 			ungetsc(s);
651 	}
652 	return action;
653 }
654 
655