xref: /freebsd/usr.bin/indent/lexi.c (revision 06c3fb2749bda94cb5201f81ffdb8fa6c3161b2e)
1 /*-
2  * SPDX-License-Identifier: BSD-4-Clause
3  *
4  * Copyright (c) 1985 Sun Microsystems, Inc.
5  * Copyright (c) 1980, 1993
6  *	The Regents of the University of California.  All rights reserved.
7  * All rights reserved.
8  *
9  * Redistribution and use in source and binary forms, with or without
10  * modification, are permitted provided that the following conditions
11  * are met:
12  * 1. Redistributions of source code must retain the above copyright
13  *    notice, this list of conditions and the following disclaimer.
14  * 2. Redistributions in binary form must reproduce the above copyright
15  *    notice, this list of conditions and the following disclaimer in the
16  *    documentation and/or other materials provided with the distribution.
17  * 3. All advertising materials mentioning features or use of this software
18  *    must display the following acknowledgement:
19  *	This product includes software developed by the University of
20  *	California, Berkeley and its contributors.
21  * 4. Neither the name of the University nor the names of its contributors
22  *    may be used to endorse or promote products derived from this software
23  *    without specific prior written permission.
24  *
25  * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
26  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
27  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
28  * ARE DISCLAIMED.  IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
29  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
30  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
31  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
32  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
33  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
34  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
35  * SUCH DAMAGE.
36  */
37 
38 /*
39  * Here we have the token scanner for indent.  It scans off one token and puts
40  * it in the global variable "token".  It returns a code, indicating the type
41  * of token scanned.
42  */
43 
44 #include <err.h>
45 #include <stdio.h>
46 #include <ctype.h>
47 #include <stdlib.h>
48 #include <string.h>
49 #include <sys/param.h>
50 
51 #include "indent_globs.h"
52 #include "indent_codes.h"
53 #include "indent.h"
54 
55 struct templ {
56     const char *rwd;
57     int         rwcode;
58 };
59 
60 /*
61  * This table has to be sorted alphabetically, because it'll be used in binary
62  * search. For the same reason, string must be the first thing in struct templ.
63  */
64 struct templ specials[] =
65 {
66     {"_Bool", 4},
67     {"_Complex", 4},
68     {"_Imaginary", 4},
69     {"auto", 10},
70     {"bool", 4},
71     {"break", 9},
72     {"case", 8},
73     {"char", 4},
74     {"complex", 4},
75     {"const", 4},
76     {"continue", 12},
77     {"default", 8},
78     {"do", 6},
79     {"double", 4},
80     {"else", 6},
81     {"enum", 3},
82     {"extern", 10},
83     {"float", 4},
84     {"for", 5},
85     {"global", 4},
86     {"goto", 9},
87     {"if", 5},
88     {"imaginary", 4},
89     {"inline", 12},
90     {"int", 4},
91     {"long", 4},
92     {"offsetof", 1},
93     {"register", 10},
94     {"restrict", 12},
95     {"return", 9},
96     {"short", 4},
97     {"signed", 4},
98     {"sizeof", 2},
99     {"static", 10},
100     {"struct", 3},
101     {"switch", 7},
102     {"typedef", 11},
103     {"union", 3},
104     {"unsigned", 4},
105     {"void", 4},
106     {"volatile", 4},
107     {"while", 5}
108 };
109 
110 const char **typenames;
111 int         typename_count;
112 int         typename_top = -1;
113 
114 /*
115  * The transition table below was rewritten by hand from lx's output, given
116  * the following definitions. lx is Katherine Flavel's lexer generator.
117  *
118  * O  = /[0-7]/;        D  = /[0-9]/;          NZ = /[1-9]/;
119  * H  = /[a-f0-9]/i;    B  = /[0-1]/;          HP = /0x/i;
120  * BP = /0b/i;          E  = /e[+\-]?/i D+;    P  = /p[+\-]?/i D+;
121  * FS = /[fl]/i;        IS = /u/i /(l|L|ll|LL)/? | /(l|L|ll|LL)/ /u/i?;
122  *
123  * D+           E  FS? -> $float;
124  * D*    "." D+ E? FS? -> $float;
125  * D+    "."    E? FS? -> $float;    HP H+           IS? -> $int;
126  * HP H+        P  FS? -> $float;    NZ D*           IS? -> $int;
127  * HP H* "." H+ P  FS? -> $float;    "0" O*          IS? -> $int;
128  * HP H+ "."    P  FS  -> $float;    BP B+           IS? -> $int;
129  */
130 static char const *table[] = {
131     /*                examples:
132                                      00
133              s                      0xx
134              t                    00xaa
135              a     11       101100xxa..
136              r   11ee0001101lbuuxx.a.pp
137              t.01.e+008bLuxll0Ll.aa.p+0
138     states:  ABCDEFGHIJKLMNOPQRSTUVWXYZ */
139     ['0'] = "CEIDEHHHIJQ  U  Q  VUVVZZZ",
140     ['1'] = "DEIDEHHHIJQ  U  Q  VUVVZZZ",
141     ['7'] = "DEIDEHHHIJ   U     VUVVZZZ",
142     ['9'] = "DEJDEHHHJJ   U     VUVVZZZ",
143     ['a'] = "             U     VUVV   ",
144     ['b'] = "  K          U     VUVV   ",
145     ['e'] = "  FFF   FF   U     VUVV   ",
146     ['f'] = "    f  f     U     VUVV  f",
147     ['u'] = "  MM    M  i  iiM   M     ",
148     ['x'] = "  N                       ",
149     ['p'] = "                    FFX   ",
150     ['L'] = "  LLf  fL  PR   Li  L    f",
151     ['l'] = "  OOf  fO   S P O i O    f",
152     ['+'] = "     G                 Y  ",
153     ['.'] = "B EE    EE   T      W     ",
154     /*       ABCDEFGHIJKLMNOPQRSTUVWXYZ */
155     [0]   = "uuiifuufiuuiiuiiiiiuiuuuuu",
156 };
157 
158 static int
159 strcmp_type(const void *e1, const void *e2)
160 {
161     return (strcmp(e1, *(const char * const *)e2));
162 }
163 
164 int
165 lexi(struct parser_state *state)
166 {
167     int         unary_delim;	/* this is set to 1 if the current token
168 				 * forces a following operator to be unary */
169     int         code;		/* internal code to be returned */
170     char        qchar;		/* the delimiter character for a string */
171 
172     e_token = s_token;		/* point to start of place to save token */
173     unary_delim = false;
174     state->col_1 = state->last_nl;	/* tell world that this token started
175 					 * in column 1 iff the last thing
176 					 * scanned was a newline */
177     state->last_nl = false;
178 
179     while (*buf_ptr == ' ' || *buf_ptr == '\t') {	/* get rid of blanks */
180 	state->col_1 = false;	/* leading blanks imply token is not in column
181 				 * 1 */
182 	if (++buf_ptr >= buf_end)
183 	    fill_buffer();
184     }
185 
186     /* Scan an alphanumeric token */
187     if (isalnum((unsigned char)*buf_ptr) ||
188 	*buf_ptr == '_' || *buf_ptr == '$' ||
189 	(buf_ptr[0] == '.' && isdigit((unsigned char)buf_ptr[1]))) {
190 	/*
191 	 * we have a character or number
192 	 */
193 	struct templ *p;
194 
195 	if (isdigit((unsigned char)*buf_ptr) ||
196 	    (buf_ptr[0] == '.' && isdigit((unsigned char)buf_ptr[1]))) {
197 	    char s;
198 	    unsigned char i;
199 
200 	    for (s = 'A'; s != 'f' && s != 'i' && s != 'u'; ) {
201 		i = (unsigned char)*buf_ptr;
202 		if (i >= nitems(table) || table[i] == NULL ||
203 		    table[i][s - 'A'] == ' ') {
204 		    s = table[0][s - 'A'];
205 		    break;
206 		}
207 		s = table[i][s - 'A'];
208 		CHECK_SIZE_TOKEN(1);
209 		*e_token++ = *buf_ptr++;
210 		if (buf_ptr >= buf_end)
211 		    fill_buffer();
212 	    }
213 	    /* s now indicates the type: f(loating), i(integer), u(nknown) */
214 	}
215 	else
216 	    while (isalnum((unsigned char)*buf_ptr) ||
217 	        *buf_ptr == BACKSLASH ||
218 		*buf_ptr == '_' || *buf_ptr == '$') {
219 		/* fill_buffer() terminates buffer with newline */
220 		if (*buf_ptr == BACKSLASH) {
221 		    if (*(buf_ptr + 1) == '\n') {
222 			buf_ptr += 2;
223 			if (buf_ptr >= buf_end)
224 			    fill_buffer();
225 			} else
226 			    break;
227 		}
228 		CHECK_SIZE_TOKEN(1);
229 		/* copy it over */
230 		*e_token++ = *buf_ptr++;
231 		if (buf_ptr >= buf_end)
232 		    fill_buffer();
233 	    }
234 	*e_token = '\0';
235 
236 	if (s_token[0] == 'L' && s_token[1] == '\0' &&
237 	      (*buf_ptr == '"' || *buf_ptr == '\''))
238 	    return (strpfx);
239 
240 	while (*buf_ptr == ' ' || *buf_ptr == '\t') {	/* get rid of blanks */
241 	    if (++buf_ptr >= buf_end)
242 		fill_buffer();
243 	}
244 	state->keyword = 0;
245 	if (state->last_token == structure && !state->p_l_follow) {
246 				/* if last token was 'struct' and we're not
247 				 * in parentheses, then this token
248 				 * should be treated as a declaration */
249 	    state->last_u_d = true;
250 	    return (decl);
251 	}
252 	/*
253 	 * Operator after identifier is binary unless last token was 'struct'
254 	 */
255 	state->last_u_d = (state->last_token == structure);
256 
257 	p = bsearch(s_token,
258 	    specials,
259 	    sizeof(specials) / sizeof(specials[0]),
260 	    sizeof(specials[0]),
261 	    strcmp_type);
262 	if (p == NULL) {	/* not a special keyword... */
263 	    char *u;
264 
265 	    /* ... so maybe a type_t or a typedef */
266 	    if ((opt.auto_typedefs && ((u = strrchr(s_token, '_')) != NULL) &&
267 	        strcmp(u, "_t") == 0) || (typename_top >= 0 &&
268 		  bsearch(s_token, typenames, typename_top + 1,
269 		    sizeof(typenames[0]), strcmp_type))) {
270 		state->keyword = 4;	/* a type name */
271 		state->last_u_d = true;
272 	        goto found_typename;
273 	    }
274 	} else {			/* we have a keyword */
275 	    state->keyword = p->rwcode;
276 	    state->last_u_d = true;
277 	    switch (p->rwcode) {
278 	    case 7:		/* it is a switch */
279 		return (swstmt);
280 	    case 8:		/* a case or default */
281 		return (casestmt);
282 
283 	    case 3:		/* a "struct" */
284 		/* FALLTHROUGH */
285 	    case 4:		/* one of the declaration keywords */
286 	    found_typename:
287 		if (state->p_l_follow) {
288 		    /* inside parens: cast, param list, offsetof or sizeof */
289 		    state->cast_mask |= (1 << state->p_l_follow) & ~state->not_cast_mask;
290 		}
291 		if (state->last_token == period || state->last_token == unary_op) {
292 		    state->keyword = 0;
293 		    break;
294 		}
295 		if (p != NULL && p->rwcode == 3)
296 		    return (structure);
297 		if (state->p_l_follow)
298 		    break;
299 		return (decl);
300 
301 	    case 5:		/* if, while, for */
302 		return (sp_paren);
303 
304 	    case 6:		/* do, else */
305 		return (sp_nparen);
306 
307 	    case 10:		/* storage class specifier */
308 		return (storage);
309 
310 	    case 11:		/* typedef */
311 		return (type_def);
312 
313 	    default:		/* all others are treated like any other
314 				 * identifier */
315 		return (ident);
316 	    }			/* end of switch */
317 	}			/* end of if (found_it) */
318 	if (*buf_ptr == '(' && state->tos <= 1 && state->ind_level == 0 &&
319 	    state->in_parameter_declaration == 0 && state->block_init == 0) {
320 	    char *tp = buf_ptr;
321 	    while (tp < buf_end)
322 		if (*tp++ == ')' && (*tp == ';' || *tp == ','))
323 		    goto not_proc;
324 	    strncpy(state->procname, token, sizeof state->procname - 1);
325 	    if (state->in_decl)
326 		state->in_parameter_declaration = 1;
327 	    return (funcname);
328     not_proc:;
329 	}
330 	/*
331 	 * The following hack attempts to guess whether or not the current
332 	 * token is in fact a declaration keyword -- one that has been
333 	 * typedefd
334 	 */
335 	else if (!state->p_l_follow && !state->block_init &&
336 	    !state->in_stmt &&
337 	    ((*buf_ptr == '*' && buf_ptr[1] != '=') ||
338 		isalpha((unsigned char)*buf_ptr)) &&
339 	    (state->last_token == semicolon || state->last_token == lbrace ||
340 		state->last_token == rbrace)) {
341 	    state->keyword = 4;	/* a type name */
342 	    state->last_u_d = true;
343 	    return decl;
344 	}
345 	if (state->last_token == decl)	/* if this is a declared variable,
346 					 * then following sign is unary */
347 	    state->last_u_d = true;	/* will make "int a -1" work */
348 	return (ident);		/* the ident is not in the list */
349     }				/* end of processing for alpanum character */
350 
351     /* Scan a non-alphanumeric token */
352 
353     CHECK_SIZE_TOKEN(3);		/* things like "<<=" */
354     *e_token++ = *buf_ptr;		/* if it is only a one-character token, it is
355 				 * moved here */
356     *e_token = '\0';
357     if (++buf_ptr >= buf_end)
358 	fill_buffer();
359 
360     switch (*token) {
361     case '\n':
362 	unary_delim = state->last_u_d;
363 	state->last_nl = true;	/* remember that we just had a newline */
364 	code = (had_eof ? 0 : newline);
365 
366 	/*
367 	 * if data has been exhausted, the newline is a dummy, and we should
368 	 * return code to stop
369 	 */
370 	break;
371 
372     case '\'':			/* start of quoted character */
373     case '"':			/* start of string */
374 	qchar = *token;
375 	do {			/* copy the string */
376 	    while (1) {		/* move one character or [/<char>]<char> */
377 		if (*buf_ptr == '\n') {
378 		    diag2(1, "Unterminated literal");
379 		    goto stop_lit;
380 		}
381 		CHECK_SIZE_TOKEN(2);
382 		*e_token = *buf_ptr++;
383 		if (buf_ptr >= buf_end)
384 		    fill_buffer();
385 		if (*e_token == BACKSLASH) {	/* if escape, copy extra char */
386 		    if (*buf_ptr == '\n')	/* check for escaped newline */
387 			++line_no;
388 		    *++e_token = *buf_ptr++;
389 		    ++e_token;	/* we must increment this again because we
390 				 * copied two chars */
391 		    if (buf_ptr >= buf_end)
392 			fill_buffer();
393 		}
394 		else
395 		    break;	/* we copied one character */
396 	    }			/* end of while (1) */
397 	} while (*e_token++ != qchar);
398 stop_lit:
399 	code = ident;
400 	break;
401 
402     case ('('):
403     case ('['):
404 	unary_delim = true;
405 	code = lparen;
406 	break;
407 
408     case (')'):
409     case (']'):
410 	code = rparen;
411 	break;
412 
413     case '#':
414 	unary_delim = state->last_u_d;
415 	code = preesc;
416 	break;
417 
418     case '?':
419 	unary_delim = true;
420 	code = question;
421 	break;
422 
423     case (':'):
424 	code = colon;
425 	unary_delim = true;
426 	break;
427 
428     case (';'):
429 	unary_delim = true;
430 	code = semicolon;
431 	break;
432 
433     case ('{'):
434 	unary_delim = true;
435 
436 	/*
437 	 * if (state->in_or_st) state->block_init = 1;
438 	 */
439 	/* ?	code = state->block_init ? lparen : lbrace; */
440 	code = lbrace;
441 	break;
442 
443     case ('}'):
444 	unary_delim = true;
445 	/* ?	code = state->block_init ? rparen : rbrace; */
446 	code = rbrace;
447 	break;
448 
449     case 014:			/* a form feed */
450 	unary_delim = state->last_u_d;
451 	state->last_nl = true;	/* remember this so we can set 'state->col_1'
452 				 * right */
453 	code = form_feed;
454 	break;
455 
456     case (','):
457 	unary_delim = true;
458 	code = comma;
459 	break;
460 
461     case '.':
462 	unary_delim = false;
463 	code = period;
464 	break;
465 
466     case '-':
467     case '+':			/* check for -, +, --, ++ */
468 	code = (state->last_u_d ? unary_op : binary_op);
469 	unary_delim = true;
470 
471 	if (*buf_ptr == token[0]) {
472 	    /* check for doubled character */
473 	    *e_token++ = *buf_ptr++;
474 	    /* buffer overflow will be checked at end of loop */
475 	    if (state->last_token == ident || state->last_token == rparen) {
476 		code = (state->last_u_d ? unary_op : postop);
477 		/* check for following ++ or -- */
478 		unary_delim = false;
479 	    }
480 	}
481 	else if (*buf_ptr == '=')
482 	    /* check for operator += */
483 	    *e_token++ = *buf_ptr++;
484 	else if (*buf_ptr == '>') {
485 	    /* check for operator -> */
486 	    *e_token++ = *buf_ptr++;
487 	    if (!opt.pointer_as_binop) {
488 		unary_delim = false;
489 		code = unary_op;
490 		state->want_blank = false;
491 	    }
492 	}
493 	break;			/* buffer overflow will be checked at end of
494 				 * switch */
495 
496     case '=':
497 	if (state->in_or_st)
498 	    state->block_init = 1;
499 	if (*buf_ptr == '=') {/* == */
500 	    *e_token++ = '=';	/* Flip =+ to += */
501 	    buf_ptr++;
502 	    *e_token = 0;
503 	}
504 	code = binary_op;
505 	unary_delim = true;
506 	break;
507 	/* can drop thru!!! */
508 
509     case '>':
510     case '<':
511     case '!':			/* ops like <, <<, <=, !=, etc */
512 	if (*buf_ptr == '>' || *buf_ptr == '<' || *buf_ptr == '=') {
513 	    *e_token++ = *buf_ptr;
514 	    if (++buf_ptr >= buf_end)
515 		fill_buffer();
516 	}
517 	if (*buf_ptr == '=')
518 	    *e_token++ = *buf_ptr++;
519 	code = (state->last_u_d ? unary_op : binary_op);
520 	unary_delim = true;
521 	break;
522 
523     case '*':
524 	unary_delim = true;
525 	if (!state->last_u_d) {
526 	    if (*buf_ptr == '=')
527 		*e_token++ = *buf_ptr++;
528 	    code = binary_op;
529 	    break;
530 	}
531 	while (*buf_ptr == '*' || isspace((unsigned char)*buf_ptr)) {
532 	    if (*buf_ptr == '*') {
533 		CHECK_SIZE_TOKEN(1);
534 		*e_token++ = *buf_ptr;
535 	    }
536 	    if (++buf_ptr >= buf_end)
537 		fill_buffer();
538 	}
539 	if (ps.in_decl) {
540 	    char *tp = buf_ptr;
541 
542 	    while (isalpha((unsigned char)*tp) ||
543 		   isspace((unsigned char)*tp)) {
544 		if (++tp >= buf_end)
545 		    fill_buffer();
546 	    }
547 	    if (*tp == '(')
548 		ps.procname[0] = ' ';
549 	}
550 	code = unary_op;
551 	break;
552 
553     default:
554 	if (token[0] == '/' && *buf_ptr == '*') {
555 	    /* it is start of comment */
556 	    *e_token++ = '*';
557 
558 	    if (++buf_ptr >= buf_end)
559 		fill_buffer();
560 
561 	    code = comment;
562 	    unary_delim = state->last_u_d;
563 	    break;
564 	}
565 	while (*(e_token - 1) == *buf_ptr || *buf_ptr == '=') {
566 	    /*
567 	     * handle ||, &&, etc, and also things as in int *****i
568 	     */
569 	    CHECK_SIZE_TOKEN(1);
570 	    *e_token++ = *buf_ptr;
571 	    if (++buf_ptr >= buf_end)
572 		fill_buffer();
573 	}
574 	code = (state->last_u_d ? unary_op : binary_op);
575 	unary_delim = true;
576 
577 
578     }				/* end of switch */
579     if (buf_ptr >= buf_end)	/* check for input buffer empty */
580 	fill_buffer();
581     state->last_u_d = unary_delim;
582     CHECK_SIZE_TOKEN(1);
583     *e_token = '\0';		/* null terminate the token */
584     return (code);
585 }
586 
587 /* Initialize constant transition table */
588 void
589 init_constant_tt(void)
590 {
591     table['-'] = table['+'];
592     table['8'] = table['9'];
593     table['2'] = table['3'] = table['4'] = table['5'] = table['6'] = table['7'];
594     table['A'] = table['C'] = table['D'] = table['c'] = table['d'] = table['a'];
595     table['B'] = table['b'];
596     table['E'] = table['e'];
597     table['U'] = table['u'];
598     table['X'] = table['x'];
599     table['P'] = table['p'];
600     table['F'] = table['f'];
601 }
602 
603 void
604 alloc_typenames(void)
605 {
606 
607     typenames = (const char **)malloc(sizeof(typenames[0]) *
608         (typename_count = 16));
609     if (typenames == NULL)
610 	err(1, NULL);
611 }
612 
613 void
614 add_typename(const char *key)
615 {
616     int comparison;
617     const char *copy;
618 
619     if (typename_top + 1 >= typename_count) {
620 	typenames = realloc((void *)typenames,
621 	    sizeof(typenames[0]) * (typename_count *= 2));
622 	if (typenames == NULL)
623 	    err(1, NULL);
624     }
625     if (typename_top == -1)
626 	typenames[++typename_top] = copy = strdup(key);
627     else if ((comparison = strcmp(key, typenames[typename_top])) >= 0) {
628 	/* take advantage of sorted input */
629 	if (comparison == 0)	/* remove duplicates */
630 	    return;
631 	typenames[++typename_top] = copy = strdup(key);
632     }
633     else {
634 	int p;
635 
636 	for (p = 0; (comparison = strcmp(key, typenames[p])) > 0; p++)
637 	    /* find place for the new key */;
638 	if (comparison == 0)	/* remove duplicates */
639 	    return;
640 	memmove(&typenames[p + 1], &typenames[p],
641 	    sizeof(typenames[0]) * (++typename_top - p));
642 	typenames[p] = copy = strdup(key);
643     }
644 
645     if (copy == NULL)
646 	err(1, NULL);
647 }
648