xref: /src/usr.bin/indent/lexi.c

/*	$NetBSD: lexi.c,v 1.172 2022/02/13 12:43:26 rillig Exp $	*/

/*-
 * SPDX-License-Identifier: BSD-4-Clause
 *
 * Copyright (c) 1985 Sun Microsystems, Inc.
 * Copyright (c) 1980, 1993
 *	The Regents of the University of California.  All rights reserved.
 * All rights reserved.
 *
 * Redistribution and use in source and binary forms, with or without
 * modification, are permitted provided that the following conditions
 * are met:
 * 1. Redistributions of source code must retain the above copyright
 *    notice, this list of conditions and the following disclaimer.
 * 2. Redistributions in binary form must reproduce the above copyright
 *    notice, this list of conditions and the following disclaimer in the
 *    documentation and/or other materials provided with the distribution.
 * 3. All advertising materials mentioning features or use of this software
 *    must display the following acknowledgement:
 *	This product includes software developed by the University of
 *	California, Berkeley and its contributors.
 * 4. Neither the name of the University nor the names of its contributors
 *    may be used to endorse or promote products derived from this software
 *    without specific prior written permission.
 *
 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
 * ARE DISCLAIMED.  IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
 * SUCH DAMAGE.
 */

#if 0
static char sccsid[] = "@(#)lexi.c	8.1 (Berkeley) 6/6/93";
#endif

#include <sys/cdefs.h>
#if defined(__NetBSD__)
__RCSID("$NetBSD: lexi.c,v 1.172 2022/02/13 12:43:26 rillig Exp $");
#elif defined(__FreeBSD__)
__FBSDID("$FreeBSD: head/usr.bin/indent/lexi.c 337862 2018-08-15 18:19:45Z pstef $");
#endif

#include <stdlib.h>
#include <string.h>

#include "indent.h"

/*
 * While inside lexi_alnum, this constant just marks a type, independently of
 * the parentheses level.
 */
#define lsym_type lsym_type_outside_parentheses

/* must be sorted alphabetically, is used in binary search */
static const struct keyword {
    const char *name;
    lexer_symbol lsym;
} keywords[] = {
    {"_Bool", lsym_type},
    {"_Complex", lsym_type},
    {"_Imaginary", lsym_type},
    {"auto", lsym_storage_class},
    {"bool", lsym_type},
    {"break", lsym_word},
    {"case", lsym_case_label},
    {"char", lsym_type},
    {"complex", lsym_type},
    {"const", lsym_type},
    {"continue", lsym_word},
    {"default", lsym_case_label},
    {"do", lsym_do},
    {"double", lsym_type},
    {"else", lsym_else},
    {"enum", lsym_tag},
    {"extern", lsym_storage_class},
    {"float", lsym_type},
    {"for", lsym_for},
    {"goto", lsym_word},
    {"if", lsym_if},
    {"imaginary", lsym_type},
    {"inline", lsym_word},
    {"int", lsym_type},
    {"long", lsym_type},
    {"offsetof", lsym_offsetof},
    {"register", lsym_storage_class},
    {"restrict", lsym_word},
    {"return", lsym_return},
    {"short", lsym_type},
    {"signed", lsym_type},
    {"sizeof", lsym_sizeof},
    {"static", lsym_storage_class},
    {"struct", lsym_tag},
    {"switch", lsym_switch},
    {"typedef", lsym_typedef},
    {"union", lsym_tag},
    {"unsigned", lsym_type},
    {"void", lsym_type},
    {"volatile", lsym_type},
    {"while", lsym_while}
};

static struct {
    const char **items;
    unsigned int len;
    unsigned int cap;
} typenames;

/*
 * The transition table below was rewritten by hand from lx's output, given
 * the following definitions. lx is Katherine Flavel's lexer generator.
 *
 * O  = /[0-7]/;        D  = /[0-9]/;          NZ = /[1-9]/;
 * H  = /[a-f0-9]/i;    B  = /[0-1]/;          HP = /0x/i;
 * BP = /0b/i;          E  = /e[+\-]?/i D+;    P  = /p[+\-]?/i D+;
 * FS = /[fl]/i;        IS = /u/i /(l|L|ll|LL)/? | /(l|L|ll|LL)/ /u/i?;
 *
 * D+           E  FS? -> $float;
 * D*    "." D+ E? FS? -> $float;
 * D+    "."    E? FS? -> $float;    HP H+           IS? -> $int;
 * HP H+        P  FS? -> $float;    NZ D*           IS? -> $int;
 * HP H* "." H+ P  FS? -> $float;    "0" O*          IS? -> $int;
 * HP H+ "."    P  FS  -> $float;    BP B+           IS? -> $int;
 */
/* INDENT OFF */
static const unsigned char lex_number_state[][26] = {
    /*                examples:
                                     00
             s                      0xx
             t                    00xaa
             a     11       101100xxa..
             r   11ee0001101lbuuxx.a.pp
             t.01.e+008bLuxll0Ll.aa.p+0
    states:  ABCDEFGHIJKLMNOPQRSTUVWXYZ */
    [0] =   "uuiifuufiuuiiuiiiiiuiuuuuu",	/* (other) */
    [1] =   "CEIDEHHHIJQ  U  Q  VUVVZZZ",	/* 0 */
    [2] =   "DEIDEHHHIJQ  U  Q  VUVVZZZ",	/* 1 */
    [3] =   "DEIDEHHHIJ   U     VUVVZZZ",	/* 2 3 4 5 6 7 */
    [4] =   "DEJDEHHHJJ   U     VUVVZZZ",	/* 8 9 */
    [5] =   "             U     VUVV   ",	/* A a C c D d */
    [6] =   "  K          U     VUVV   ",	/* B b */
    [7] =   "  FFF   FF   U     VUVV   ",	/* E e */
    [8] =   "    f  f     U     VUVV  f",	/* F f */
    [9] =   "  LLf  fL  PR   Li  L    f",	/* L */
    [10] =  "  OOf  fO   S P O i O    f",	/* l */
    [11] =  "                    FFX   ",	/* P p */
    [12] =  "  MM    M  i  iiM   M     ",	/* U u */
    [13] =  "  N                       ",	/* X x */
    [14] =  "     G                 Y  ",	/* + - */
    [15] =  "B EE    EE   T      W     ",	/* . */
    /*       ABCDEFGHIJKLMNOPQRSTUVWXYZ */
};
/* INDENT ON */

static const unsigned char lex_number_row[] = {
    ['0'] = 1,
    ['1'] = 2,
    ['2'] = 3, ['3'] = 3, ['4'] = 3, ['5'] = 3, ['6'] = 3, ['7'] = 3,
    ['8'] = 4, ['9'] = 4,
    ['A'] = 5, ['a'] = 5, ['C'] = 5, ['c'] = 5, ['D'] = 5, ['d'] = 5,
    ['B'] = 6, ['b'] = 6,
    ['E'] = 7, ['e'] = 7,
    ['F'] = 8, ['f'] = 8,
    ['L'] = 9,
    ['l'] = 10,
    ['P'] = 11, ['p'] = 11,
    ['U'] = 12, ['u'] = 12,
    ['X'] = 13, ['x'] = 13,
    ['+'] = 14, ['-'] = 14,
    ['.'] = 15,
};

static void
check_size_token(size_t desired_size)
{
    if (token.e + desired_size >= token.l)
	buf_expand(&token, desired_size);
}

static void
token_add_char(char ch)
{
    check_size_token(1);
    *token.e++ = ch;
}

#ifdef debug
static const char *
lsym_name(lexer_symbol sym)
{
    static const char *const name[] = {
	"eof",
	"preprocessing",
	"newline",
	"form_feed",
	"comment",
	"lparen_or_lbracket",
	"rparen_or_rbracket",
	"lbrace",
	"rbrace",
	"period",
	"unary_op",
	"binary_op",
	"postfix_op",
	"question",
	"colon",
	"comma",
	"semicolon",
	"typedef",
	"storage_class",
	"type_outside_parentheses",
	"type_in_parentheses",
	"tag",
	"case_label",
	"sizeof",
	"offsetof",
	"word",
	"funcname",
	"do",
	"else",
	"for",
	"if",
	"switch",
	"while",
	"return",
    };

    return name[sym];
}

static void
debug_print_buf(const char *name, const struct buffer *buf)
{
    if (buf->s < buf->e) {
	debug_printf("%s ", name);
	debug_vis_range("\"", buf->s, buf->e, "\"\n");
    }
}

static bool
debug_full_parser_state(void)
{
    return true;
}

#define debug_ps_bool(name) \
        if (ps.name != prev_ps.name) \
	    debug_println("[%c] -> [%c] ps." #name, \
		prev_ps.name ? 'x' : ' ', ps.name ? 'x' : ' '); \
	else if (debug_full_parser_state()) \
	    debug_println("       [%c] ps." #name, ps.name ? 'x' : ' ')
#define debug_ps_int(name) \
	if (ps.name != prev_ps.name) \
	    debug_println("%3d -> %3d ps." #name, prev_ps.name, ps.name); \
	else if (debug_full_parser_state()) \
	    debug_println("       %3d ps." #name, ps.name)

static bool
ps_paren_has_changed(const struct parser_state *prev_ps)
{
    const paren_level_props *prev = prev_ps->paren, *curr = ps.paren;

    if (prev_ps->nparen != ps.nparen)
	return true;

    for (int i = 0; i < ps.nparen; i++) {
	if (curr[i].indent != prev[i].indent ||
	    curr[i].maybe_cast != prev[i].maybe_cast ||
	    curr[i].no_cast != prev[i].no_cast)
	    return true;
    }
    return false;
}

static void
debug_ps_paren(const struct parser_state *prev_ps)
{
    if (!debug_full_parser_state() && !ps_paren_has_changed(prev_ps))
	return;

    debug_printf("           ps.paren:");
    for (int i = 0; i < ps.nparen; i++) {
	const paren_level_props *props = ps.paren + i;
	const char *cast = props->no_cast ? "(no cast)"
	    : props->maybe_cast ? "(cast)"
	    : "";
	debug_printf(" %s%d", cast, props->indent);
    }
    if (ps.nparen == 0)
	debug_printf(" none");
    debug_println("");
}

static void
debug_lexi(lexer_symbol lsym)
{
    /*
     * Watch out for 'rolled back parser state' in the debug output; the
     * differences around these are unreliable.
     */
    static struct parser_state prev_ps;

    debug_println("");
    debug_printf("line %d: %s", line_no, lsym_name(lsym));
    debug_vis_range(" \"", token.s, token.e, "\"\n");

    debug_print_buf("label", &lab);
    debug_print_buf("code", &code);
    debug_print_buf("comment", &com);

    debug_println("           ps.prev_token = %s", lsym_name(ps.prev_token));
    debug_ps_bool(next_col_1);
    debug_ps_bool(curr_col_1);
    debug_ps_bool(next_unary);
    debug_ps_bool(is_function_definition);
    debug_ps_bool(want_blank);
    debug_ps_int(line_start_nparen);
    debug_ps_int(nparen);
    debug_ps_paren(&prev_ps);

    debug_ps_int(comment_delta);
    debug_ps_int(n_comment_delta);
    debug_ps_int(com_ind);

    debug_ps_bool(block_init);
    debug_ps_int(block_init_level);
    debug_ps_bool(init_or_struct);

    debug_ps_int(ind_level);
    debug_ps_int(ind_level_follow);

    debug_ps_int(decl_level);
    debug_ps_bool(decl_on_line);
    debug_ps_bool(in_decl);
    debug_ps_int(just_saw_decl);
    debug_ps_bool(in_func_def_params);
    debug_ps_bool(decl_indent_done);

    debug_ps_bool(in_stmt_or_decl);
    debug_ps_bool(in_stmt_cont);
    debug_ps_bool(is_case_label);

    debug_ps_bool(search_stmt);

    prev_ps = ps;
}
#endif

static lexer_symbol
lexi_end(lexer_symbol lsym)
{
#ifdef debug
    debug_lexi(lsym);
#endif
    return lsym;
}

static void
lex_number(void)
{
    for (unsigned char s = 'A'; s != 'f' && s != 'i' && s != 'u';) {
	unsigned char ch = (unsigned char)inp_peek();
	if (ch >= array_length(lex_number_row) || lex_number_row[ch] == 0)
	    break;

	unsigned char row = lex_number_row[ch];
	if (lex_number_state[row][s - 'A'] == ' ') {
	    /*-
	     * lex_number_state[0][s - 'A'] now indicates the type:
	     * f = floating, i = integer, u = unknown
	     */
	    return;
	}

	s = lex_number_state[row][s - 'A'];
	token_add_char(inp_next());
    }
}

static bool
is_identifier_start(char ch)
{
    return ch_isalpha(ch) || ch == '_' || ch == '$';
}

static bool
is_identifier_part(char ch)
{
    return ch_isalnum(ch) || ch == '_' || ch == '$';
}

static void
lex_word(void)
{
    for (;;) {
	if (is_identifier_part(inp_peek()))
	    token_add_char(inp_next());
	else if (inp_peek() == '\\' && inp_lookahead(1) == '\n') {
	    inp_skip();
	    inp_skip();
	} else
	    return;
    }
}

static void
lex_char_or_string(void)
{
    for (char delim = token.e[-1];;) {
	if (inp_peek() == '\n') {
	    diag(1, "Unterminated literal");
	    return;
	}

	token_add_char(inp_next());
	if (token.e[-1] == delim)
	    return;

	if (token.e[-1] == '\\') {
	    if (inp_peek() == '\n')
		++line_no;
	    token_add_char(inp_next());
	}
    }
}

/* Guess whether the current token is a declared type. */
static bool
probably_typename(void)
{
    if (ps.prev_token == lsym_storage_class)
	return true;
    if (ps.block_init)
	return false;
    if (ps.in_stmt_or_decl)	/* XXX: this condition looks incorrect */
	return false;
    if (inp_peek() == '*' && inp_lookahead(1) != '=')
	goto maybe;
    /* XXX: is_identifier_start */
    if (ch_isalpha(inp_peek()))
	goto maybe;
    return false;
maybe:
    return ps.prev_token == lsym_semicolon ||
	ps.prev_token == lsym_lbrace ||
	ps.prev_token == lsym_rbrace;
}

static int
bsearch_typenames(const char *key)
{
    const char **arr = typenames.items;
    int lo = 0;
    int hi = (int)typenames.len - 1;

    while (lo <= hi) {
	int mid = (int)((unsigned)(lo + hi) >> 1);
	int cmp = strcmp(arr[mid], key);
	if (cmp < 0)
	    lo = mid + 1;
	else if (cmp > 0)
	    hi = mid - 1;
	else
	    return mid;
    }
    return -(lo + 1);
}

static bool
is_typename(void)
{
    if (opt.auto_typedefs &&
	token.e - token.s >= 2 && memcmp(token.e - 2, "_t", 2) == 0)
	return true;

    return bsearch_typenames(token.s) >= 0;
}

static int
cmp_keyword_by_name(const void *key, const void *elem)
{
    return strcmp(key, ((const struct keyword *)elem)->name);
}

/*
 * Looking at something like 'function_name(...)' in a line, guess whether
 * this starts a function definition or a declaration.
 */
static bool
probably_looking_at_definition(void)
{
    int paren_level = 0;
    for (const char *p = inp_p(), *e = inp_line_end(); p < e; p++) {
	if (*p == '(')
	    paren_level++;
	if (*p == ')' && --paren_level == 0) {
	    p++;

	    while (p < e && (ch_isspace(*p) || is_identifier_part(*p)))
		p++;		/* '__dead' or '__unused' */

	    if (p == e)		/* func(...) */
		break;
	    if (*p == ';')	/* func(...); */
		return false;
	    if (*p == ',')	/* double abs(), pi; */
		return false;
	    if (*p == '(')	/* func(...) __attribute__((...)) */
		paren_level++;	/* func(...) __printflike(...) */
	    else
		break;		/* func(...) { ... */
	}
    }

    /*
     * To further reduce the cases where indent wrongly treats an incomplete
     * function declaration as a function definition, thus adding a newline
     * before the function name, it may be worth looking for parameter names,
     * as these are often omitted in function declarations and only included
     * in function definitions. Or just increase the lookahead to more than
     * just the current line of input, until the next '{'.
     */
    return true;
}

/* Read an alphanumeric token into 'token', or return lsym_eof. */
static lexer_symbol
lexi_alnum(void)
{
    if (ch_isdigit(inp_peek()) ||
	    (inp_peek() == '.' && ch_isdigit(inp_lookahead(1)))) {
	lex_number();
    } else if (is_identifier_start(inp_peek())) {
	lex_word();

	if (token.s[0] == 'L' && token.e - token.s == 1 &&
		(inp_peek() == '"' || inp_peek() == '\'')) {
	    token_add_char(inp_next());
	    lex_char_or_string();
	    ps.next_unary = false;

	    check_size_token(1);
	    *token.e = '\0';

	    return lsym_word;
	}
    } else
	return lsym_eof;	/* just as a placeholder */

    *token.e = '\0';

    while (ch_isblank(inp_peek()))
	inp_skip();

    ps.next_unary = ps.prev_token == lsym_tag;	/* for 'struct s *' */

    if (ps.prev_token == lsym_tag && ps.nparen == 0)
	return lsym_type_outside_parentheses;

    const struct keyword *kw = bsearch(token.s, keywords,
	array_length(keywords), sizeof(keywords[0]), cmp_keyword_by_name);
    bool is_type = false;
    if (kw == NULL) {
	if (is_typename()) {
	    is_type = true;
	    ps.next_unary = true;
	    if (ps.in_enum == in_enum_enum)
		ps.in_enum = in_enum_type;
	    goto found_typename;
	}

    } else {			/* we have a keyword */
	is_type = kw->lsym == lsym_type;
	ps.next_unary = true;
	if (kw->lsym != lsym_tag && kw->lsym != lsym_type)
	    return kw->lsym;

found_typename:
	if (ps.nparen > 0) {
	    /* inside parentheses: cast, param list, offsetof or sizeof */
	    if (!ps.paren[ps.nparen - 1].no_cast)
		ps.paren[ps.nparen - 1].maybe_cast = true;
	}
	if (ps.prev_token != lsym_period && ps.prev_token != lsym_unary_op) {
	    if (kw != NULL && kw->lsym == lsym_tag) {
		if (token.s[0] == 'e' /* enum */)
		    ps.in_enum = in_enum_enum;
		return lsym_tag;
	    }
	    if (ps.nparen == 0)
		return lsym_type_outside_parentheses;
	}
    }

    if (inp_peek() == '(' && ps.tos <= 1 && ps.ind_level == 0 &&
	!ps.in_func_def_params && !ps.block_init) {

	if (ps.nparen == 0 && probably_looking_at_definition()) {
	    ps.is_function_definition = true;
	    if (ps.in_decl)
		ps.in_func_def_params = true;
	    return lsym_funcname;
	}

    } else if (ps.nparen == 0 && probably_typename()) {
	ps.next_unary = true;
	return lsym_type_outside_parentheses;
    }

    return is_type ? lsym_type_in_parentheses : lsym_word;
}

static bool
is_asterisk_unary(void)
{
    if (ps.next_unary || ps.in_func_def_params)
	return true;
    if (ps.prev_token == lsym_word ||
	    ps.prev_token == lsym_rparen_or_rbracket)
	return false;
    return ps.in_decl && ps.nparen > 0;
}

static void
lex_asterisk_unary(void)
{
    while (inp_peek() == '*' || ch_isspace(inp_peek())) {
	if (inp_peek() == '*')
	    token_add_char('*');
	inp_skip();
    }

    if (ps.in_decl) {
	const char *tp = inp_p(), *e = inp_line_end();

	while (tp < e) {
	    if (ch_isspace(*tp))
		tp++;
	    else if (is_identifier_start(*tp)) {
		tp++;
		while (tp < e && is_identifier_part(*tp))
		    tp++;
	    } else
		break;
	}

	if (tp < e && *tp == '(')
	    ps.is_function_definition = true;
    }
}

/* Reads the next token, placing it in the global variable "token". */
lexer_symbol
lexi(void)
{
    token.e = token.s;
    ps.curr_col_1 = ps.next_col_1;
    ps.next_col_1 = false;

    while (ch_isblank(inp_peek())) {
	ps.curr_col_1 = false;
	inp_skip();
    }

    lexer_symbol alnum_lsym = lexi_alnum();
    if (alnum_lsym != lsym_eof)
	return lexi_end(alnum_lsym);

    /* Scan a non-alphanumeric token */

    check_size_token(3);	/* for things like "<<=" */
    *token.e++ = inp_next();
    *token.e = '\0';

    lexer_symbol lsym;
    bool next_unary;

    switch (token.e[-1]) {

    /* INDENT OFF */
    case '(':
    case '[':	lsym = lsym_lparen_or_lbracket;	next_unary = true;	break;
    case ')':
    case ']':	lsym = lsym_rparen_or_rbracket;	next_unary = false;	break;
    case '?':	lsym = lsym_question;		next_unary = true;	break;
    case ':':	lsym = lsym_colon;		next_unary = true;	break;
    case ';':	lsym = lsym_semicolon;		next_unary = true;	break;
    case '{':	lsym = lsym_lbrace;		next_unary = true;	break;
    case '}':	lsym = lsym_rbrace;		next_unary = true;	break;
    case ',':	lsym = lsym_comma;		next_unary = true;	break;
    case '.':	lsym = lsym_period;		next_unary = false;	break;
    /* INDENT ON */

    case '\n':
	/* if data has been exhausted, the '\n' is a dummy. */
	lsym = had_eof ? lsym_eof : lsym_newline;
	next_unary = ps.next_unary;
	ps.next_col_1 = true;
	break;

    case '\f':
	lsym = lsym_form_feed;
	next_unary = ps.next_unary;
	ps.next_col_1 = true;
	break;

    case '#':
	lsym = lsym_preprocessing;
	next_unary = ps.next_unary;
	break;

    case '\'':
    case '"':
	lex_char_or_string();
	lsym = lsym_word;
	next_unary = false;
	break;

    case '-':
    case '+':
	lsym = ps.next_unary ? lsym_unary_op : lsym_binary_op;
	next_unary = true;

	if (inp_peek() == token.e[-1]) {	/* '++' or '--' */
	    *token.e++ = inp_next();
	    if (ps.prev_token == lsym_word ||
		    ps.prev_token == lsym_rparen_or_rbracket) {
		lsym = ps.next_unary ? lsym_unary_op : lsym_postfix_op;
		next_unary = false;
	    }

	} else if (inp_peek() == '=') {	/* '+=' or '-=' */
	    *token.e++ = inp_next();

	} else if (inp_peek() == '>') {	/* '->' */
	    *token.e++ = inp_next();
	    lsym = lsym_unary_op;
	    next_unary = false;
	    ps.want_blank = false;
	}
	break;

    case '=':
	if (ps.init_or_struct)
	    ps.block_init = true;
	if (inp_peek() == '=') {	/* == */
	    *token.e++ = inp_next();
	    *token.e = '\0';
	}
	lsym = lsym_binary_op;
	next_unary = true;
	break;

    case '>':
    case '<':
    case '!':			/* ops like <, <<, <=, !=, etc */
	if (inp_peek() == '>' || inp_peek() == '<' || inp_peek() == '=')
	    *token.e++ = inp_next();
	if (inp_peek() == '=')
	    *token.e++ = inp_next();
	lsym = ps.next_unary ? lsym_unary_op : lsym_binary_op;
	next_unary = true;
	break;

    case '*':
	if (is_asterisk_unary()) {
	    lex_asterisk_unary();
	    lsym = lsym_unary_op;
	    next_unary = true;
	} else {
	    if (inp_peek() == '=')
		*token.e++ = inp_next();
	    lsym = lsym_binary_op;
	    next_unary = true;
	}
	break;

    default:
	if (token.e[-1] == '/' && (inp_peek() == '*' || inp_peek() == '/')) {
	    *token.e++ = inp_next();
	    lsym = lsym_comment;
	    next_unary = ps.next_unary;
	    break;
	}

	/* handle '||', '&&', etc., and also things as in 'int *****i' */
	while (token.e[-1] == inp_peek() || inp_peek() == '=')
	    token_add_char(inp_next());

	lsym = ps.next_unary ? lsym_unary_op : lsym_binary_op;
	next_unary = true;
    }

    if (ps.in_enum == in_enum_enum || ps.in_enum == in_enum_type)
	ps.in_enum = lsym == lsym_lbrace ? in_enum_brace : in_enum_no;
    if (lsym == lsym_rbrace)
	ps.in_enum = in_enum_no;

    ps.next_unary = next_unary;

    check_size_token(1);
    *token.e = '\0';

    return lexi_end(lsym);
}

void
register_typename(const char *name)
{
    if (typenames.len >= typenames.cap) {
	typenames.cap = 16 + 2 * typenames.cap;
	typenames.items = xrealloc(typenames.items,
	    sizeof(typenames.items[0]) * typenames.cap);
    }

    int pos = bsearch_typenames(name);
    if (pos >= 0)
	return;			/* already in the list */

    pos = -(pos + 1);
    memmove(typenames.items + pos + 1, typenames.items + pos,
	sizeof(typenames.items[0]) * (typenames.len++ - (unsigned)pos));
    typenames.items[pos] = xstrdup(name);
}