Home | History | Annotate | Line # | Download | only in indent
lexi.c revision 1.32
      1 /*	$NetBSD: lexi.c,v 1.32 2021/03/11 21:47:36 rillig Exp $	*/
      2 
      3 /*-
      4  * SPDX-License-Identifier: BSD-4-Clause
      5  *
      6  * Copyright (c) 1985 Sun Microsystems, Inc.
      7  * Copyright (c) 1980, 1993
      8  *	The Regents of the University of California.  All rights reserved.
      9  * All rights reserved.
     10  *
     11  * Redistribution and use in source and binary forms, with or without
     12  * modification, are permitted provided that the following conditions
     13  * are met:
     14  * 1. Redistributions of source code must retain the above copyright
     15  *    notice, this list of conditions and the following disclaimer.
     16  * 2. Redistributions in binary form must reproduce the above copyright
     17  *    notice, this list of conditions and the following disclaimer in the
     18  *    documentation and/or other materials provided with the distribution.
     19  * 3. All advertising materials mentioning features or use of this software
     20  *    must display the following acknowledgement:
     21  *	This product includes software developed by the University of
     22  *	California, Berkeley and its contributors.
     23  * 4. Neither the name of the University nor the names of its contributors
     24  *    may be used to endorse or promote products derived from this software
     25  *    without specific prior written permission.
     26  *
     27  * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
     28  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
     29  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
     30  * ARE DISCLAIMED.  IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
     31  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
     32  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
     33  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
     34  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
     35  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
     36  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
     37  * SUCH DAMAGE.
     38  */
     39 
     40 #if 0
     41 #ifndef lint
     42 static char sccsid[] = "@(#)lexi.c	8.1 (Berkeley) 6/6/93";
     43 #endif /* not lint */
     44 #endif
     45 
     46 #include <sys/cdefs.h>
     47 #ifndef lint
     48 #if defined(__NetBSD__)
     49 __RCSID("$NetBSD: lexi.c,v 1.32 2021/03/11 21:47:36 rillig Exp $");
     50 #elif defined(__FreeBSD__)
     51 __FBSDID("$FreeBSD: head/usr.bin/indent/lexi.c 337862 2018-08-15 18:19:45Z pstef $");
     52 #endif
     53 #endif
     54 
     55 /*
     56  * Here we have the token scanner for indent.  It scans off one token and puts
     57  * it in the global variable "token".  It returns a code, indicating the type
     58  * of token scanned.
     59  */
     60 
     61 #include <assert.h>
     62 #include <err.h>
     63 #include <stdio.h>
     64 #include <ctype.h>
     65 #include <stdlib.h>
     66 #include <string.h>
     67 #include <sys/param.h>
     68 
     69 #include "indent.h"
     70 
     71 struct templ {
     72     const char *rwd;
     73     enum rwcode rwcode;
     74 };
     75 
     76 /*
     77  * This table has to be sorted alphabetically, because it'll be used in binary
     78  * search.
     79  */
     80 const struct templ specials[] =
     81 {
     82     {"_Bool", rw_type},
     83     {"_Complex", rw_type},
     84     {"_Imaginary", rw_type},
     85     {"auto", rw_storage_class},
     86     {"bool", rw_type},
     87     {"break", rw_jump},
     88     {"case", rw_case_or_default},
     89     {"char", rw_type},
     90     {"complex", rw_type},
     91     {"const", rw_type},
     92     {"continue", rw_jump},
     93     {"default", rw_case_or_default},
     94     {"do", rw_do_or_else},
     95     {"double", rw_type},
     96     {"else", rw_do_or_else},
     97     {"enum", rw_struct_or_union_or_enum},
     98     {"extern", rw_storage_class},
     99     {"float", rw_type},
    100     {"for", rw_for_or_if_or_while},
    101     {"global", rw_type},
    102     {"goto", rw_jump},
    103     {"if", rw_for_or_if_or_while},
    104     {"imaginary", rw_type},
    105     {"inline", rw_inline_or_restrict},
    106     {"int", rw_type},
    107     {"long", rw_type},
    108     {"offsetof", rw_offsetof},
    109     {"register", rw_storage_class},
    110     {"restrict", rw_inline_or_restrict},
    111     {"return", rw_jump},
    112     {"short", rw_type},
    113     {"signed", rw_type},
    114     {"sizeof", rw_sizeof},
    115     {"static", rw_storage_class},
    116     {"struct", rw_struct_or_union_or_enum},
    117     {"switch", rw_switch},
    118     {"typedef", rw_typedef},
    119     {"union", rw_struct_or_union_or_enum},
    120     {"unsigned", rw_type},
    121     {"void", rw_type},
    122     {"volatile", rw_type},
    123     {"while", rw_for_or_if_or_while}
    124 };
    125 
    126 const char **typenames;
    127 int         typename_count;
    128 int         typename_top = -1;
    129 
    130 /*
    131  * The transition table below was rewritten by hand from lx's output, given
    132  * the following definitions. lx is Katherine Flavel's lexer generator.
    133  *
    134  * O  = /[0-7]/;        D  = /[0-9]/;          NZ = /[1-9]/;
    135  * H  = /[a-f0-9]/i;    B  = /[0-1]/;          HP = /0x/i;
    136  * BP = /0b/i;          E  = /e[+\-]?/i D+;    P  = /p[+\-]?/i D+;
    137  * FS = /[fl]/i;        IS = /u/i /(l|L|ll|LL)/? | /(l|L|ll|LL)/ /u/i?;
    138  *
    139  * D+           E  FS? -> $float;
    140  * D*    "." D+ E? FS? -> $float;
    141  * D+    "."    E? FS? -> $float;    HP H+           IS? -> $int;
    142  * HP H+        P  FS? -> $float;    NZ D*           IS? -> $int;
    143  * HP H* "." H+ P  FS? -> $float;    "0" O*          IS? -> $int;
    144  * HP H+ "."    P  FS  -> $float;    BP B+           IS? -> $int;
    145  */
    146 static char const *table[] = {
    147     /*                examples:
    148                                      00
    149              s                      0xx
    150              t                    00xaa
    151              a     11       101100xxa..
    152              r   11ee0001101lbuuxx.a.pp
    153              t.01.e+008bLuxll0Ll.aa.p+0
    154     states:  ABCDEFGHIJKLMNOPQRSTUVWXYZ */
    155     ['0'] = "CEIDEHHHIJQ  U  Q  VUVVZZZ",
    156     ['1'] = "DEIDEHHHIJQ  U  Q  VUVVZZZ",
    157     ['7'] = "DEIDEHHHIJ   U     VUVVZZZ",
    158     ['9'] = "DEJDEHHHJJ   U     VUVVZZZ",
    159     ['a'] = "             U     VUVV   ",
    160     ['b'] = "  K          U     VUVV   ",
    161     ['e'] = "  FFF   FF   U     VUVV   ",
    162     ['f'] = "    f  f     U     VUVV  f",
    163     ['u'] = "  MM    M  i  iiM   M     ",
    164     ['x'] = "  N                       ",
    165     ['p'] = "                    FFX   ",
    166     ['L'] = "  LLf  fL  PR   Li  L    f",
    167     ['l'] = "  OOf  fO   S P O i O    f",
    168     ['+'] = "     G                 Y  ",
    169     ['.'] = "B EE    EE   T      W     ",
    170     /*       ABCDEFGHIJKLMNOPQRSTUVWXYZ */
    171     [0]   = "uuiifuufiuuiiuiiiiiuiuuuuu",
    172 };
    173 
    174 static char
    175 inbuf_peek(void)
    176 {
    177     return *buf_ptr;
    178 }
    179 
    180 static void
    181 inbuf_skip(void)
    182 {
    183     buf_ptr++;
    184     if (buf_ptr >= buf_end)
    185 	fill_buffer();
    186 }
    187 
    188 static char
    189 inbuf_next(void)
    190 {
    191     char ch = inbuf_peek();
    192     inbuf_skip();
    193     return ch;
    194 }
    195 
    196 static void
    197 check_size_token(size_t desired_size)
    198 {
    199     if (e_token + (desired_size) >= l_token) {
    200 	int nsize = l_token - s_token + 400 + desired_size;
    201 	int token_len = e_token - s_token;
    202 	tokenbuf = (char *)realloc(tokenbuf, nsize);
    203 	if (tokenbuf == NULL)
    204 	    err(1, NULL);
    205 	e_token = tokenbuf + token_len + 1;
    206 	l_token = tokenbuf + nsize - 5;
    207 	s_token = tokenbuf + 1;
    208     }
    209 }
    210 
    211 static int
    212 compare_templ_array(const void *key, const void *elem)
    213 {
    214     return strcmp(key, ((const struct templ *)elem)->rwd);
    215 }
    216 
    217 static int
    218 compare_string_array(const void *key, const void *elem)
    219 {
    220     return strcmp(key, *((const char *const *)elem));
    221 }
    222 
    223 #ifdef debug
    224 const char *
    225 token_type_name(token_type tk)
    226 {
    227     static const char *const name[] = {
    228 	"end_of_file", "newline", "lparen", "rparen", "unary_op",
    229 	"binary_op", "postfix_op", "question", "case_label", "colon",
    230 	"semicolon", "lbrace", "rbrace", "ident", "comma",
    231 	"comment", "switch_expr", "preprocessing", "form_feed", "decl",
    232 	"keyword_for_if_while", "keyword_do_else",
    233 	"if_expr", "while_expr", "for_exprs",
    234 	"stmt", "stmt_list", "keyword_else", "keyword_do", "do_stmt",
    235 	"if_expr_stmt", "if_expr_stmt_else", "period", "string_prefix",
    236 	"storage_class", "funcname", "type_def", "keyword_struct_union_enum"
    237     };
    238 
    239     assert(0 <= tk && tk < sizeof name / sizeof name[0]);
    240 
    241     return name[tk];
    242 }
    243 
    244 static void
    245 print_buf(const char *name, const char *s, const char *e)
    246 {
    247     if (s == e)
    248 	return;
    249 
    250     printf(" %s \"", name);
    251     for (const char *p = s; p < e; p++) {
    252 	if (isprint((unsigned char)*p) && *p != '\\' && *p != '"')
    253 	    printf("%c", *p);
    254 	else if (*p == '\n')
    255 	    printf("\\n");
    256 	else if (*p == '\t')
    257 	    printf("\\t");
    258 	else
    259 	    printf("\\x%02x", *p);
    260     }
    261     printf("\"");
    262 }
    263 
    264 static token_type
    265 lexi_end(token_type code)
    266 {
    267     printf("in line %d, lexi returns '%s'", line_no, token_type_name(code));
    268     print_buf("token", s_token, e_token);
    269     print_buf("label", s_lab, e_lab);
    270     print_buf("code", s_code, e_code);
    271     print_buf("comment", s_com, e_com);
    272     printf("\n");
    273 
    274     return code;
    275 }
    276 #else
    277 #  define lexi_end(tk) (tk)
    278 #endif
    279 
    280 token_type
    281 lexi(struct parser_state *state)
    282 {
    283     int         unary_delim;	/* this is set to 1 if the current token
    284 				 * forces a following operator to be unary */
    285     token_type  code;		/* internal code to be returned */
    286     char        qchar;		/* the delimiter character for a string */
    287 
    288     e_token = s_token;		/* point to start of place to save token */
    289     unary_delim = false;
    290     state->col_1 = state->last_nl;	/* tell world that this token started
    291 					 * in column 1 iff the last thing
    292 					 * scanned was a newline */
    293     state->last_nl = false;
    294 
    295     while (*buf_ptr == ' ' || *buf_ptr == '\t') {	/* get rid of blanks */
    296 	state->col_1 = false;	/* leading blanks imply token is not in column
    297 				 * 1 */
    298 	inbuf_skip();
    299     }
    300 
    301     /* Scan an alphanumeric token */
    302     if (isalnum((unsigned char)*buf_ptr) ||
    303 	*buf_ptr == '_' || *buf_ptr == '$' ||
    304 	(buf_ptr[0] == '.' && isdigit((unsigned char)buf_ptr[1]))) {
    305 	/*
    306 	 * we have a character or number
    307 	 */
    308 	struct templ *p;
    309 
    310 	if (isdigit((unsigned char)*buf_ptr) ||
    311 	    (buf_ptr[0] == '.' && isdigit((unsigned char)buf_ptr[1]))) {
    312 	    char s;
    313 	    unsigned char i;
    314 
    315 	    for (s = 'A'; s != 'f' && s != 'i' && s != 'u'; ) {
    316 		i = (unsigned char)*buf_ptr;
    317 		if (i >= nitems(table) || table[i] == NULL ||
    318 		    table[i][s - 'A'] == ' ') {
    319 		    s = table[0][s - 'A'];
    320 		    break;
    321 		}
    322 		s = table[i][s - 'A'];
    323 		check_size_token(1);
    324 		*e_token++ = inbuf_next();
    325 	    }
    326 	    /* s now indicates the type: f(loating), i(integer), u(nknown) */
    327 	}
    328 	else
    329 	    while (isalnum((unsigned char)*buf_ptr) ||
    330 	        *buf_ptr == '\\' ||
    331 		*buf_ptr == '_' || *buf_ptr == '$') {
    332 		/* fill_buffer() terminates buffer with newline */
    333 		if (*buf_ptr == '\\') {
    334 		    if (*(buf_ptr + 1) == '\n') {
    335 			buf_ptr += 2;
    336 			if (buf_ptr >= buf_end)
    337 			    fill_buffer();
    338 			} else
    339 			    break;
    340 		}
    341 		check_size_token(1);
    342 		*e_token++ = inbuf_next();
    343 	    }
    344 	*e_token = '\0';
    345 
    346 	if (s_token[0] == 'L' && s_token[1] == '\0' &&
    347 	      (*buf_ptr == '"' || *buf_ptr == '\''))
    348 	    return lexi_end(string_prefix);
    349 
    350 	while (*buf_ptr == ' ' || *buf_ptr == '\t')	/* get rid of blanks */
    351 	    inbuf_next();
    352 	state->keyword = rw_0;
    353 	if (state->last_token == keyword_struct_union_enum &&
    354 	    !state->p_l_follow) {
    355 	    /* if last token was 'struct' and we're not in parentheses, then
    356 	     * this token should be treated as a declaration */
    357 	    state->last_u_d = true;
    358 	    return lexi_end(decl);
    359 	}
    360 	/*
    361 	 * Operator after identifier is binary unless last token was 'struct'
    362 	 */
    363 	state->last_u_d = (state->last_token == keyword_struct_union_enum);
    364 
    365 	p = bsearch(s_token, specials, sizeof specials / sizeof specials[0],
    366 	    sizeof specials[0], compare_templ_array);
    367 	if (p == NULL) {	/* not a special keyword... */
    368 	    char *u;
    369 
    370 	    /* ... so maybe a type_t or a typedef */
    371 	    if ((opt.auto_typedefs && ((u = strrchr(s_token, '_')) != NULL) &&
    372 	        strcmp(u, "_t") == 0) || (typename_top >= 0 &&
    373 		  bsearch(s_token, typenames, typename_top + 1,
    374 		    sizeof typenames[0], compare_string_array))) {
    375 		state->keyword = rw_type;
    376 		state->last_u_d = true;
    377 	        goto found_typename;
    378 	    }
    379 	} else {			/* we have a keyword */
    380 	    state->keyword = p->rwcode;
    381 	    state->last_u_d = true;
    382 	    switch (p->rwcode) {
    383 	    case rw_switch:
    384 		return lexi_end(switch_expr);
    385 	    case rw_case_or_default:
    386 		return lexi_end(case_label);
    387 	    case rw_struct_or_union_or_enum:
    388 	    case rw_type:
    389 	    found_typename:
    390 		if (state->p_l_follow) {
    391 		    /* inside parens: cast, param list, offsetof or sizeof */
    392 		    state->cast_mask |= (1 << state->p_l_follow) & ~state->not_cast_mask;
    393 		}
    394 		if (state->last_token == period || state->last_token == unary_op) {
    395 		    state->keyword = rw_0;
    396 		    break;
    397 		}
    398 		if (p != NULL && p->rwcode == rw_struct_or_union_or_enum)
    399 		    return lexi_end(keyword_struct_union_enum);
    400 		if (state->p_l_follow)
    401 		    break;
    402 		return lexi_end(decl);
    403 
    404 	    case rw_for_or_if_or_while:
    405 		return lexi_end(keyword_for_if_while);
    406 
    407 	    case rw_do_or_else:
    408 		return lexi_end(keyword_do_else);
    409 
    410 	    case rw_storage_class:
    411 		return lexi_end(storage_class);
    412 
    413 	    case rw_typedef:
    414 		return lexi_end(type_def);
    415 
    416 	    default:		/* all others are treated like any other
    417 				 * identifier */
    418 		return lexi_end(ident);
    419 	    }			/* end of switch */
    420 	}			/* end of if (found_it) */
    421 	if (*buf_ptr == '(' && state->tos <= 1 && state->ind_level == 0 &&
    422 	    state->in_parameter_declaration == 0 && state->block_init == 0) {
    423 	    char *tp = buf_ptr;
    424 	    while (tp < buf_end)
    425 		if (*tp++ == ')' && (*tp == ';' || *tp == ','))
    426 		    goto not_proc;
    427 	    strncpy(state->procname, token, sizeof state->procname - 1);
    428 	    if (state->in_decl)
    429 		state->in_parameter_declaration = 1;
    430 	    return lexi_end(funcname);
    431     not_proc:;
    432 	}
    433 	/*
    434 	 * The following hack attempts to guess whether or not the current
    435 	 * token is in fact a declaration keyword -- one that has been
    436 	 * typedefd
    437 	 */
    438 	else if (!state->p_l_follow && !state->block_init &&
    439 	    !state->in_stmt &&
    440 	    ((*buf_ptr == '*' && buf_ptr[1] != '=') ||
    441 		isalpha((unsigned char)*buf_ptr)) &&
    442 	    (state->last_token == semicolon || state->last_token == lbrace ||
    443 		state->last_token == rbrace)) {
    444 	    state->keyword = rw_type;
    445 	    state->last_u_d = true;
    446 	    return lexi_end(decl);
    447 	}
    448 	if (state->last_token == decl)	/* if this is a declared variable,
    449 					 * then following sign is unary */
    450 	    state->last_u_d = true;	/* will make "int a -1" work */
    451 	return lexi_end(ident);		/* the ident is not in the list */
    452     }				/* end of procesing for alpanum character */
    453 
    454     /* Scan a non-alphanumeric token */
    455 
    456     check_size_token(3);	/* things like "<<=" */
    457     *e_token++ = inbuf_next();	/* if it is only a one-character token, it is
    458 				 * moved here */
    459     *e_token = '\0';
    460 
    461     switch (*token) {
    462     case '\n':
    463 	unary_delim = state->last_u_d;
    464 	state->last_nl = true;	/* remember that we just had a newline */
    465 	code = (had_eof ? end_of_file : newline);
    466 
    467 	/*
    468 	 * if data has been exhausted, the newline is a dummy, and we should
    469 	 * return code to stop
    470 	 */
    471 	break;
    472 
    473     case '\'':			/* start of quoted character */
    474     case '"':			/* start of string */
    475 	qchar = *token;
    476 	do {			/* copy the string */
    477 	    while (1) {		/* move one character or [/<char>]<char> */
    478 		if (*buf_ptr == '\n') {
    479 		    diag(1, "Unterminated literal");
    480 		    goto stop_lit;
    481 		}
    482 		check_size_token(2);
    483 		*e_token = inbuf_next();
    484 		if (*e_token == '\\') {		/* if escape, copy extra char */
    485 		    if (*buf_ptr == '\n')	/* check for escaped newline */
    486 			++line_no;
    487 		    *++e_token = inbuf_next();
    488 		    ++e_token;	/* we must increment this again because we
    489 				 * copied two chars */
    490 		}
    491 		else
    492 		    break;	/* we copied one character */
    493 	    }			/* end of while (1) */
    494 	} while (*e_token++ != qchar);
    495 stop_lit:
    496 	code = ident;
    497 	break;
    498 
    499     case ('('):
    500     case ('['):
    501 	unary_delim = true;
    502 	code = lparen;
    503 	break;
    504 
    505     case (')'):
    506     case (']'):
    507 	code = rparen;
    508 	break;
    509 
    510     case '#':
    511 	unary_delim = state->last_u_d;
    512 	code = preprocessing;
    513 	break;
    514 
    515     case '?':
    516 	unary_delim = true;
    517 	code = question;
    518 	break;
    519 
    520     case (':'):
    521 	code = colon;
    522 	unary_delim = true;
    523 	break;
    524 
    525     case (';'):
    526 	unary_delim = true;
    527 	code = semicolon;
    528 	break;
    529 
    530     case ('{'):
    531 	unary_delim = true;
    532 
    533 	/*
    534 	 * if (state->in_or_st) state->block_init = 1;
    535 	 */
    536 	/* ?	code = state->block_init ? lparen : lbrace; */
    537 	code = lbrace;
    538 	break;
    539 
    540     case ('}'):
    541 	unary_delim = true;
    542 	/* ?	code = state->block_init ? rparen : rbrace; */
    543 	code = rbrace;
    544 	break;
    545 
    546     case 014:			/* a form feed */
    547 	unary_delim = state->last_u_d;
    548 	state->last_nl = true;	/* remember this so we can set 'state->col_1'
    549 				 * right */
    550 	code = form_feed;
    551 	break;
    552 
    553     case (','):
    554 	unary_delim = true;
    555 	code = comma;
    556 	break;
    557 
    558     case '.':
    559 	unary_delim = false;
    560 	code = period;
    561 	break;
    562 
    563     case '-':
    564     case '+':			/* check for -, +, --, ++ */
    565 	code = (state->last_u_d ? unary_op : binary_op);
    566 	unary_delim = true;
    567 
    568 	if (*buf_ptr == token[0]) {
    569 	    /* check for doubled character */
    570 	    *e_token++ = *buf_ptr++;
    571 	    /* buffer overflow will be checked at end of loop */
    572 	    if (state->last_token == ident || state->last_token == rparen) {
    573 		code = (state->last_u_d ? unary_op : postfix_op);
    574 		/* check for following ++ or -- */
    575 		unary_delim = false;
    576 	    }
    577 	}
    578 	else if (*buf_ptr == '=')
    579 	    /* check for operator += */
    580 	    *e_token++ = *buf_ptr++;
    581 	else if (*buf_ptr == '>') {
    582 	    /* check for operator -> */
    583 	    *e_token++ = *buf_ptr++;
    584 	    unary_delim = false;
    585 	    code = unary_op;
    586 	    state->want_blank = false;
    587 	}
    588 	break;			/* buffer overflow will be checked at end of
    589 				 * switch */
    590 
    591     case '=':
    592 	if (state->in_or_st)
    593 	    state->block_init = 1;
    594 	if (*buf_ptr == '=') {	/* == */
    595 	    *e_token++ = '=';	/* Flip =+ to += */
    596 	    buf_ptr++;
    597 	    *e_token = 0;
    598 	}
    599 	code = binary_op;
    600 	unary_delim = true;
    601 	break;
    602 	/* can drop thru!!! */
    603 
    604     case '>':
    605     case '<':
    606     case '!':			/* ops like <, <<, <=, !=, etc */
    607 	if (*buf_ptr == '>' || *buf_ptr == '<' || *buf_ptr == '=')
    608 	    *e_token++ = inbuf_next();
    609 	if (*buf_ptr == '=')
    610 	    *e_token++ = *buf_ptr++;
    611 	code = (state->last_u_d ? unary_op : binary_op);
    612 	unary_delim = true;
    613 	break;
    614 
    615     case '*':
    616 	unary_delim = true;
    617 	if (!state->last_u_d) {
    618 	    if (*buf_ptr == '=')
    619 		*e_token++ = *buf_ptr++;
    620 	    code = binary_op;
    621 	    break;
    622 	}
    623 	while (*buf_ptr == '*' || isspace((unsigned char)*buf_ptr)) {
    624 	    if (*buf_ptr == '*') {
    625 		check_size_token(1);
    626 		*e_token++ = *buf_ptr;
    627 	    }
    628 	    inbuf_skip();
    629 	}
    630 	if (ps.in_decl) {
    631 	    char *tp = buf_ptr;
    632 
    633 	    while (isalpha((unsigned char)*tp) ||
    634 		   isspace((unsigned char)*tp)) {
    635 		if (++tp >= buf_end)
    636 		    fill_buffer();
    637 	    }
    638 	    if (*tp == '(')
    639 		ps.procname[0] = ' ';
    640 	}
    641 	code = unary_op;
    642 	break;
    643 
    644     default:
    645 	if (token[0] == '/' && (*buf_ptr == '*' || *buf_ptr == '/')) {
    646 	    /* it is start of comment */
    647 	    *e_token++ = inbuf_next();
    648 
    649 	    code = comment;
    650 	    unary_delim = state->last_u_d;
    651 	    break;
    652 	}
    653 	while (*(e_token - 1) == *buf_ptr || *buf_ptr == '=') {
    654 	    /*
    655 	     * handle ||, &&, etc, and also things as in int *****i
    656 	     */
    657 	    check_size_token(1);
    658 	    *e_token++ = inbuf_next();
    659 	}
    660 	code = (state->last_u_d ? unary_op : binary_op);
    661 	unary_delim = true;
    662 
    663 
    664     }				/* end of switch */
    665     if (buf_ptr >= buf_end)	/* check for input buffer empty */
    666 	fill_buffer();
    667     state->last_u_d = unary_delim;
    668     check_size_token(1);
    669     *e_token = '\0';		/* null terminate the token */
    670     return lexi_end(code);
    671 }
    672 
    673 /* Initialize constant transition table */
    674 void
    675 init_constant_tt(void)
    676 {
    677     table['-'] = table['+'];
    678     table['8'] = table['9'];
    679     table['2'] = table['3'] = table['4'] = table['5'] = table['6'] = table['7'];
    680     table['A'] = table['C'] = table['D'] = table['c'] = table['d'] = table['a'];
    681     table['B'] = table['b'];
    682     table['E'] = table['e'];
    683     table['U'] = table['u'];
    684     table['X'] = table['x'];
    685     table['P'] = table['p'];
    686     table['F'] = table['f'];
    687 }
    688 
    689 void
    690 alloc_typenames(void)
    691 {
    692 
    693     typenames = (const char **)malloc(sizeof(typenames[0]) *
    694         (typename_count = 16));
    695     if (typenames == NULL)
    696 	err(1, NULL);
    697 }
    698 
    699 void
    700 add_typename(const char *key)
    701 {
    702     int comparison;
    703     const char *copy;
    704 
    705     if (typename_top + 1 >= typename_count) {
    706 	typenames = realloc((void *)typenames,
    707 	    sizeof(typenames[0]) * (typename_count *= 2));
    708 	if (typenames == NULL)
    709 	    err(1, NULL);
    710     }
    711     if (typename_top == -1)
    712 	typenames[++typename_top] = copy = strdup(key);
    713     else if ((comparison = strcmp(key, typenames[typename_top])) >= 0) {
    714 	/* take advantage of sorted input */
    715 	if (comparison == 0)	/* remove duplicates */
    716 	    return;
    717 	typenames[++typename_top] = copy = strdup(key);
    718     }
    719     else {
    720 	int p;
    721 
    722 	for (p = 0; (comparison = strcmp(key, typenames[p])) > 0; p++)
    723 	    /* find place for the new key */;
    724 	if (comparison == 0)	/* remove duplicates */
    725 	    return;
    726 	memmove(&typenames[p + 1], &typenames[p],
    727 	    sizeof(typenames[0]) * (++typename_top - p));
    728 	typenames[p] = copy = strdup(key);
    729     }
    730 
    731     if (copy == NULL)
    732 	err(1, NULL);
    733 }
    734