Home | History | Annotate | Line # | Download | only in indent
lexi.c revision 1.46
      1 /*	$NetBSD: lexi.c,v 1.46 2021/09/24 18:47:29 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 1.46 2021/09/24 18:47:29 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 /* Initialize constant transition table */
    175 void
    176 init_constant_tt(void)
    177 {
    178     table['-'] = table['+'];
    179     table['8'] = table['9'];
    180     table['2'] = table['3'] = table['4'] = table['5'] = table['6'] = table['7'];
    181     table['A'] = table['C'] = table['D'] = table['c'] = table['d'] = table['a'];
    182     table['B'] = table['b'];
    183     table['E'] = table['e'];
    184     table['U'] = table['u'];
    185     table['X'] = table['x'];
    186     table['P'] = table['p'];
    187     table['F'] = table['f'];
    188 }
    189 
    190 static char
    191 inbuf_peek(void)
    192 {
    193     return *buf_ptr;
    194 }
    195 
    196 static void
    197 inbuf_skip(void)
    198 {
    199     buf_ptr++;
    200     if (buf_ptr >= buf_end)
    201 	fill_buffer();
    202 }
    203 
    204 static char
    205 inbuf_next(void)
    206 {
    207     char ch = inbuf_peek();
    208     inbuf_skip();
    209     return ch;
    210 }
    211 
    212 static void
    213 check_size_token(size_t desired_size)
    214 {
    215     if (e_token + (desired_size) < l_token)
    216         return;
    217 
    218     size_t nsize = l_token - s_token + 400 + desired_size;
    219     size_t token_len = e_token - s_token;
    220     tokenbuf = realloc(tokenbuf, nsize);
    221     if (tokenbuf == NULL)
    222 	err(1, NULL);
    223     e_token = tokenbuf + token_len + 1;
    224     l_token = tokenbuf + nsize - 5;
    225     s_token = tokenbuf + 1;
    226 }
    227 
    228 static int
    229 compare_templ_array(const void *key, const void *elem)
    230 {
    231     return strcmp(key, ((const struct templ *)elem)->rwd);
    232 }
    233 
    234 static int
    235 compare_string_array(const void *key, const void *elem)
    236 {
    237     return strcmp(key, *((const char *const *)elem));
    238 }
    239 
    240 #ifdef debug
    241 const char *
    242 token_type_name(token_type tk)
    243 {
    244     static const char *const name[] = {
    245 	"end_of_file", "newline", "lparen", "rparen", "unary_op",
    246 	"binary_op", "postfix_op", "question", "case_label", "colon",
    247 	"semicolon", "lbrace", "rbrace", "ident", "comma",
    248 	"comment", "switch_expr", "preprocessing", "form_feed", "decl",
    249 	"keyword_for_if_while", "keyword_do_else",
    250 	"if_expr", "while_expr", "for_exprs",
    251 	"stmt", "stmt_list", "keyword_else", "keyword_do", "do_stmt",
    252 	"if_expr_stmt", "if_expr_stmt_else", "period", "string_prefix",
    253 	"storage_class", "funcname", "type_def", "keyword_struct_union_enum"
    254     };
    255 
    256     assert(0 <= tk && tk < sizeof name / sizeof name[0]);
    257 
    258     return name[tk];
    259 }
    260 
    261 static void
    262 print_buf(const char *name, const char *s, const char *e)
    263 {
    264     if (s < e) {
    265 	debug_printf(" %s ", name);
    266 	debug_vis_range("\"", s, e, "\"");
    267     }
    268 }
    269 
    270 static token_type
    271 lexi_end(token_type code)
    272 {
    273     debug_printf("in line %d, lexi returns '%s'",
    274 	line_no, token_type_name(code));
    275     print_buf("token", s_token, e_token);
    276     print_buf("label", lab.s, lab.e);
    277     print_buf("code", s_code, e_code);
    278     print_buf("comment", com.s, com.e);
    279     debug_printf("\n");
    280 
    281     return code;
    282 }
    283 #else
    284 #  define lexi_end(tk) (tk)
    285 #endif
    286 
    287 static void
    288 lex_number(void)
    289 {
    290     char s;
    291     unsigned char i;
    292 
    293     for (s = 'A'; s != 'f' && s != 'i' && s != 'u'; ) {
    294 	i = (unsigned char)*buf_ptr;
    295 	if (i >= nitems(table) || table[i] == NULL ||
    296 	    table[i][s - 'A'] == ' ') {
    297 	    s = table[0][s - 'A'];
    298 	    break;
    299 	}
    300 	s = table[i][s - 'A'];
    301 	check_size_token(1);
    302 	*e_token++ = inbuf_next();
    303     }
    304     /* s now indicates the type: f(loating), i(integer), u(nknown) */
    305 }
    306 
    307 static void
    308 lex_word(void)
    309 {
    310     while (isalnum((unsigned char)*buf_ptr) ||
    311 	   *buf_ptr == '\\' ||
    312 	   *buf_ptr == '_' || *buf_ptr == '$') {
    313 	/* fill_buffer() terminates buffer with newline */
    314 	if (*buf_ptr == '\\') {
    315 	    if (buf_ptr[1] == '\n') {
    316 		buf_ptr += 2;
    317 		if (buf_ptr >= buf_end)
    318 		    fill_buffer();
    319 	    } else
    320 		break;
    321 	}
    322 	check_size_token(1);
    323 	*e_token++ = inbuf_next();
    324     }
    325 }
    326 
    327 static void
    328 lex_char_or_string(void)
    329 {
    330     char delim;
    331 
    332     delim = *token;
    333     do {			/* copy the string */
    334 	for (;;) {		/* move one character or [/<char>]<char> */
    335 	    if (*buf_ptr == '\n') {
    336 		diag(1, "Unterminated literal");
    337 		return;
    338 	    }
    339 	    check_size_token(2);
    340 	    *e_token = inbuf_next();
    341 	    if (*e_token == '\\') {	/* if escape, copy extra char */
    342 		if (*buf_ptr == '\n')	/* check for escaped newline */
    343 		    ++line_no;
    344 		*++e_token = inbuf_next();
    345 		++e_token;	/* we must increment this again because we
    346 				 * copied two chars */
    347 	    } else
    348 		break;		/* we copied one character */
    349 	}			/* end of while (1) */
    350     } while (*e_token++ != delim);
    351 }
    352 
    353 token_type
    354 lexi(struct parser_state *state)
    355 {
    356     int         unary_delim;	/* this is set to 1 if the current token
    357 				 * forces a following operator to be unary */
    358     token_type  code;		/* internal code to be returned */
    359 
    360     e_token = s_token;		/* point to start of place to save token */
    361     unary_delim = false;
    362     state->col_1 = state->last_nl;	/* tell world that this token started
    363 					 * in column 1 iff the last thing
    364 					 * scanned was a newline */
    365     state->last_nl = false;
    366 
    367     while (*buf_ptr == ' ' || *buf_ptr == '\t') {	/* get rid of blanks */
    368 	state->col_1 = false;	/* leading blanks imply token is not in column
    369 				 * 1 */
    370 	inbuf_skip();
    371     }
    372 
    373     /* Scan an alphanumeric token */
    374     if (isalnum((unsigned char)*buf_ptr) ||
    375 	*buf_ptr == '_' || *buf_ptr == '$' ||
    376 	(buf_ptr[0] == '.' && isdigit((unsigned char)buf_ptr[1]))) {
    377 	/*
    378 	 * we have a letter or number
    379 	 */
    380 	struct templ *p;
    381 
    382 	if (isdigit((unsigned char)*buf_ptr) ||
    383 	    (buf_ptr[0] == '.' && isdigit((unsigned char)buf_ptr[1]))) {
    384 	    lex_number();
    385 	} else {
    386 	    lex_word();
    387 	}
    388 	*e_token = '\0';
    389 
    390 	if (s_token[0] == 'L' && s_token[1] == '\0' &&
    391 	      (*buf_ptr == '"' || *buf_ptr == '\''))
    392 	    return lexi_end(string_prefix);
    393 
    394 	while (*buf_ptr == ' ' || *buf_ptr == '\t')	/* get rid of blanks */
    395 	    inbuf_next();
    396 	state->keyword = rw_0;
    397 	if (state->last_token == keyword_struct_union_enum &&
    398 	    !state->p_l_follow) {
    399 	    /* if last token was 'struct' and we're not in parentheses, then
    400 	     * this token should be treated as a declaration */
    401 	    state->last_u_d = true;
    402 	    return lexi_end(decl);
    403 	}
    404 	/*
    405 	 * Operator after identifier is binary unless last token was 'struct'
    406 	 */
    407 	state->last_u_d = (state->last_token == keyword_struct_union_enum);
    408 
    409 	p = bsearch(s_token, specials, sizeof specials / sizeof specials[0],
    410 	    sizeof specials[0], compare_templ_array);
    411 	if (p == NULL) {	/* not a special keyword... */
    412 	    char *u;
    413 
    414 	    /* ... so maybe a type_t or a typedef */
    415 	    if ((opt.auto_typedefs && ((u = strrchr(s_token, '_')) != NULL) &&
    416 	        strcmp(u, "_t") == 0) || (typename_top >= 0 &&
    417 		  bsearch(s_token, typenames, (size_t)typename_top + 1,
    418 		    sizeof typenames[0], compare_string_array))) {
    419 		state->keyword = rw_type;
    420 		state->last_u_d = true;
    421 	        goto found_typename;
    422 	    }
    423 	} else {			/* we have a keyword */
    424 	    state->keyword = p->rwcode;
    425 	    state->last_u_d = true;
    426 	    switch (p->rwcode) {
    427 	    case rw_switch:
    428 		return lexi_end(switch_expr);
    429 	    case rw_case_or_default:
    430 		return lexi_end(case_label);
    431 	    case rw_struct_or_union_or_enum:
    432 	    case rw_type:
    433 	    found_typename:
    434 		if (state->p_l_follow) {
    435 		    /* inside parens: cast, param list, offsetof or sizeof */
    436 		    state->cast_mask |= (1 << state->p_l_follow) & ~state->not_cast_mask;
    437 		}
    438 		if (state->last_token == period || state->last_token == unary_op) {
    439 		    state->keyword = rw_0;
    440 		    break;
    441 		}
    442 		if (p != NULL && p->rwcode == rw_struct_or_union_or_enum)
    443 		    return lexi_end(keyword_struct_union_enum);
    444 		if (state->p_l_follow)
    445 		    break;
    446 		return lexi_end(decl);
    447 
    448 	    case rw_for_or_if_or_while:
    449 		return lexi_end(keyword_for_if_while);
    450 
    451 	    case rw_do_or_else:
    452 		return lexi_end(keyword_do_else);
    453 
    454 	    case rw_storage_class:
    455 		return lexi_end(storage_class);
    456 
    457 	    case rw_typedef:
    458 		return lexi_end(type_def);
    459 
    460 	    default:		/* all others are treated like any other
    461 				 * identifier */
    462 		return lexi_end(ident);
    463 	    }			/* end of switch */
    464 	}			/* end of if (found_it) */
    465 	if (*buf_ptr == '(' && state->tos <= 1 && state->ind_level == 0 &&
    466 	    state->in_parameter_declaration == 0 && state->block_init == 0) {
    467 	    char *tp = buf_ptr;
    468 	    while (tp < buf_end)
    469 		if (*tp++ == ')' && (*tp == ';' || *tp == ','))
    470 		    goto not_proc;
    471 	    strncpy(state->procname, token, sizeof state->procname - 1);
    472 	    if (state->in_decl)
    473 		state->in_parameter_declaration = 1;
    474 	    return lexi_end(funcname);
    475     not_proc:;
    476 	}
    477 	/*
    478 	 * The following hack attempts to guess whether or not the current
    479 	 * token is in fact a declaration keyword -- one that has been
    480 	 * typedefd
    481 	 */
    482 	else if (!state->p_l_follow && !state->block_init &&
    483 	    !state->in_stmt &&
    484 	    ((*buf_ptr == '*' && buf_ptr[1] != '=') ||
    485 		isalpha((unsigned char)*buf_ptr)) &&
    486 	    (state->last_token == semicolon || state->last_token == lbrace ||
    487 		state->last_token == rbrace)) {
    488 	    state->keyword = rw_type;
    489 	    state->last_u_d = true;
    490 	    return lexi_end(decl);
    491 	}
    492 	if (state->last_token == decl)	/* if this is a declared variable,
    493 					 * then following sign is unary */
    494 	    state->last_u_d = true;	/* will make "int a -1" work */
    495 	return lexi_end(ident);		/* the ident is not in the list */
    496     }				/* end of procesing for alpanum character */
    497 
    498     /* Scan a non-alphanumeric token */
    499 
    500     check_size_token(3);	/* things like "<<=" */
    501     *e_token++ = inbuf_next();	/* if it is only a one-character token, it is
    502 				 * moved here */
    503     *e_token = '\0';
    504 
    505     switch (*token) {
    506     case '\n':
    507 	unary_delim = state->last_u_d;
    508 	state->last_nl = true;	/* remember that we just had a newline */
    509 	code = (had_eof ? end_of_file : newline);
    510 
    511 	/*
    512 	 * if data has been exhausted, the newline is a dummy, and we should
    513 	 * return code to stop
    514 	 */
    515 	break;
    516 
    517     case '\'':
    518     case '"':
    519 	lex_char_or_string();
    520 	code = ident;
    521 	break;
    522 
    523     case '(':
    524     case '[':
    525 	unary_delim = true;
    526 	code = lparen;
    527 	break;
    528 
    529     case ')':
    530     case ']':
    531 	code = rparen;
    532 	break;
    533 
    534     case '#':
    535 	unary_delim = state->last_u_d;
    536 	code = preprocessing;
    537 	break;
    538 
    539     case '?':
    540 	unary_delim = true;
    541 	code = question;
    542 	break;
    543 
    544     case ':':
    545 	code = colon;
    546 	unary_delim = true;
    547 	break;
    548 
    549     case ';':
    550 	unary_delim = true;
    551 	code = semicolon;
    552 	break;
    553 
    554     case '{':
    555 	unary_delim = true;
    556 
    557 	/*
    558 	 * if (state->in_or_st) state->block_init = 1;
    559 	 */
    560 	/* ?	code = state->block_init ? lparen : lbrace; */
    561 	code = lbrace;
    562 	break;
    563 
    564     case '}':
    565 	unary_delim = true;
    566 	/* ?	code = state->block_init ? rparen : rbrace; */
    567 	code = rbrace;
    568 	break;
    569 
    570     case 014:			/* a form feed */
    571 	unary_delim = state->last_u_d;
    572 	state->last_nl = true;	/* remember this so we can set 'state->col_1'
    573 				 * right */
    574 	code = form_feed;
    575 	break;
    576 
    577     case ',':
    578 	unary_delim = true;
    579 	code = comma;
    580 	break;
    581 
    582     case '.':
    583 	unary_delim = false;
    584 	code = period;
    585 	break;
    586 
    587     case '-':
    588     case '+':			/* check for -, +, --, ++ */
    589 	code = (state->last_u_d ? unary_op : binary_op);
    590 	unary_delim = true;
    591 
    592 	if (*buf_ptr == token[0]) {
    593 	    /* check for doubled character */
    594 	    *e_token++ = *buf_ptr++;
    595 	    /* buffer overflow will be checked at end of loop */
    596 	    if (state->last_token == ident || state->last_token == rparen) {
    597 		code = (state->last_u_d ? unary_op : postfix_op);
    598 		/* check for following ++ or -- */
    599 		unary_delim = false;
    600 	    }
    601 	} else if (*buf_ptr == '=')
    602 	    /* check for operator += */
    603 	    *e_token++ = *buf_ptr++;
    604 	else if (*buf_ptr == '>') {
    605 	    /* check for operator -> */
    606 	    *e_token++ = *buf_ptr++;
    607 	    unary_delim = false;
    608 	    code = unary_op;
    609 	    state->want_blank = false;
    610 	}
    611 	break;			/* buffer overflow will be checked at end of
    612 				 * switch */
    613 
    614     case '=':
    615 	if (state->in_or_st)
    616 	    state->block_init = 1;
    617 	if (*buf_ptr == '=') {	/* == */
    618 	    *e_token++ = '=';	/* Flip =+ to += */
    619 	    buf_ptr++;
    620 	    *e_token = 0;
    621 	}
    622 	code = binary_op;
    623 	unary_delim = true;
    624 	break;
    625 	/* can drop thru!!! */
    626 
    627     case '>':
    628     case '<':
    629     case '!':			/* ops like <, <<, <=, !=, etc */
    630 	if (*buf_ptr == '>' || *buf_ptr == '<' || *buf_ptr == '=')
    631 	    *e_token++ = inbuf_next();
    632 	if (*buf_ptr == '=')
    633 	    *e_token++ = *buf_ptr++;
    634 	code = (state->last_u_d ? unary_op : binary_op);
    635 	unary_delim = true;
    636 	break;
    637 
    638     case '*':
    639 	unary_delim = true;
    640 	if (!state->last_u_d) {
    641 	    if (*buf_ptr == '=')
    642 		*e_token++ = *buf_ptr++;
    643 	    code = binary_op;
    644 	    break;
    645 	}
    646 	while (*buf_ptr == '*' || isspace((unsigned char)*buf_ptr)) {
    647 	    if (*buf_ptr == '*') {
    648 		check_size_token(1);
    649 		*e_token++ = *buf_ptr;
    650 	    }
    651 	    inbuf_skip();
    652 	}
    653 	if (ps.in_decl) {
    654 	    char *tp = buf_ptr;
    655 
    656 	    while (isalpha((unsigned char)*tp) ||
    657 		   isspace((unsigned char)*tp)) {
    658 		if (++tp >= buf_end)
    659 		    fill_buffer();
    660 	    }
    661 	    if (*tp == '(')
    662 		ps.procname[0] = ' ';
    663 	}
    664 	code = unary_op;
    665 	break;
    666 
    667     default:
    668 	if (token[0] == '/' && (*buf_ptr == '*' || *buf_ptr == '/')) {
    669 	    /* it is start of comment */
    670 	    *e_token++ = inbuf_next();
    671 
    672 	    code = comment;
    673 	    unary_delim = state->last_u_d;
    674 	    break;
    675 	}
    676 	while (e_token[-1] == *buf_ptr || *buf_ptr == '=') {
    677 	    /*
    678 	     * handle ||, &&, etc, and also things as in int *****i
    679 	     */
    680 	    check_size_token(1);
    681 	    *e_token++ = inbuf_next();
    682 	}
    683 	code = (state->last_u_d ? unary_op : binary_op);
    684 	unary_delim = true;
    685 
    686 
    687     }				/* end of switch */
    688     if (buf_ptr >= buf_end)	/* check for input buffer empty */
    689 	fill_buffer();
    690     state->last_u_d = unary_delim;
    691     check_size_token(1);
    692     *e_token = '\0';		/* null terminate the token */
    693     return lexi_end(code);
    694 }
    695 
    696 void
    697 alloc_typenames(void)
    698 {
    699 
    700     typenames = malloc(sizeof(typenames[0]) * (typename_count = 16));
    701     if (typenames == NULL)
    702 	err(1, NULL);
    703 }
    704 
    705 void
    706 add_typename(const char *key)
    707 {
    708     int comparison;
    709     const char *copy;
    710 
    711     if (typename_top + 1 >= typename_count) {
    712 	typenames = realloc((void *)typenames,
    713 	    sizeof(typenames[0]) * (typename_count *= 2));
    714 	if (typenames == NULL)
    715 	    err(1, NULL);
    716     }
    717     if (typename_top == -1)
    718 	typenames[++typename_top] = copy = strdup(key);
    719     else if ((comparison = strcmp(key, typenames[typename_top])) >= 0) {
    720 	/* take advantage of sorted input */
    721 	if (comparison == 0)	/* remove duplicates */
    722 	    return;
    723 	typenames[++typename_top] = copy = strdup(key);
    724     } else {
    725 	int p;
    726 
    727 	for (p = 0; (comparison = strcmp(key, typenames[p])) > 0; p++)
    728 	    /* find place for the new key */;
    729 	if (comparison == 0)	/* remove duplicates */
    730 	    return;
    731 	memmove(&typenames[p + 1], &typenames[p],
    732 	    sizeof(typenames[0]) * (++typename_top - p));
    733 	typenames[p] = copy = strdup(key);
    734     }
    735 
    736     if (copy == NULL)
    737 	err(1, NULL);
    738 }
    739