freebsd-dev/gnu/usr.bin/rcs/lib/rcslex.c

/* lexical analysis of RCS files */

/******************************************************************************
 *                     Lexical Analysis.
 *                     hashtable, Lexinit, nextlex, getlex, getkey,
 *                     getid, getnum, readstring, printstring, savestring,
 *                     checkid, fatserror, error, faterror, warn, diagnose
 *                     Testprogram: define LEXDB
 ******************************************************************************
 */

/* Copyright 1982, 1988, 1989 Walter Tichy
   Copyright 1990, 1991, 1992, 1993, 1994, 1995 Paul Eggert
   Distributed under license by the Free Software Foundation, Inc.

This file is part of RCS.

RCS is free software; you can redistribute it and/or modify
it under the terms of the GNU General Public License as published by
the Free Software Foundation; either version 2, or (at your option)
any later version.

RCS is distributed in the hope that it will be useful,
but WITHOUT ANY WARRANTY; without even the implied warranty of
MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
GNU General Public License for more details.

You should have received a copy of the GNU General Public License
along with RCS; see the file COPYING.
If not, write to the Free Software Foundation,
59 Temple Place - Suite 330, Boston, MA 02111-1307, USA.

Report problems and direct all questions to:

    rcs-bugs@cs.purdue.edu

*/



/*
 * Revision 5.19  1995/06/16 06:19:24  eggert
 * Update FSF address.
 *
 * Revision 5.18  1995/06/01 16:23:43  eggert
 * (map_fd_deallocate,mmap_deallocate,read_deallocate,nothing_to_deallocate):
 * New functions.
 * (Iclose): If large_memory and maps_memory, use them to deallocate mapping.
 * (fd2RILE): Use map_fd if available.
 * If one mapping method fails, try the next instead of giving up;
 * if they all fail, fall back on ordinary read.
 * Work around bug: root mmap over NFS succeeds, but accessing dumps core.
 * Use MAP_FAILED macro for mmap failure, and `char *' instead of caddr_t.
 * (advise_access): Use madvise only if this instance used mmap.
 * (Iopen): Use fdSafer to get safer file descriptor.
 * (aflush): Moved here from rcsedit.c.
 *
 * Revision 5.17  1994/03/20 04:52:58  eggert
 * Don't worry if madvise fails.  Add Orewind.  Remove lint.
 *
 * Revision 5.16  1993/11/09 17:55:29  eggert
 * Fix `label: }' typo.
 *
 * Revision 5.15  1993/11/03 17:42:27  eggert
 * Improve quality of diagnostics by putting file names in them more often.
 * Don't discard ignored phrases.
 *
 * Revision 5.14  1992/07/28  16:12:44  eggert
 * Identifiers may now start with a digit and (unless they are symbolic names)
 * may contain `.'.  Avoid `unsigned'.  Statement macro names now end in _.
 *
 * Revision 5.13  1992/02/17  23:02:27  eggert
 * Work around NFS mmap SIGBUS problem.
 *
 * Revision 5.12  1992/01/06  02:42:34  eggert
 * Use OPEN_O_BINARY if mode contains 'b'.
 *
 * Revision 5.11  1991/11/03  03:30:44  eggert
 * Fix porting bug to ancient hosts lacking vfprintf.
 *
 * Revision 5.10  1991/10/07  17:32:46  eggert
 * Support piece tables even if !has_mmap.
 *
 * Revision 5.9  1991/09/24  00:28:42  eggert
 * Don't export errsay().
 *
 * Revision 5.8  1991/08/19  03:13:55  eggert
 * Add eoflex(), mmap support.  Tune.
 *
 * Revision 5.7  1991/04/21  11:58:26  eggert
 * Add MS-DOS support.
 *
 * Revision 5.6  1991/02/25  07:12:42  eggert
 * Work around fputs bug.  strsave -> str_save (DG/UX name clash)
 *
 * Revision 5.5  1990/12/04  05:18:47  eggert
 * Use -I for prompts and -q for diagnostics.
 *
 * Revision 5.4  1990/11/19  20:05:28  hammer
 * no longer gives warning about unknown keywords if -q is specified
 *
 * Revision 5.3  1990/11/01  05:03:48  eggert
 * When ignoring unknown phrases, copy them to the output RCS file.
 *
 * Revision 5.2  1990/09/04  08:02:27  eggert
 * Count RCS lines better.
 *
 * Revision 5.1  1990/08/29  07:14:03  eggert
 * Work around buggy compilers with defective argument promotion.
 *
 * Revision 5.0  1990/08/22  08:12:55  eggert
 * Remove compile-time limits; use malloc instead.
 * Report errno-related errors with perror().
 * Ansify and Posixate.  Add support for ISO 8859.
 * Use better hash function.
 *
 * Revision 4.6  89/05/01  15:13:07  narten
 * changed copyright header to reflect current distribution rules
 *
 * Revision 4.5  88/08/28  15:01:12  eggert
 * Don't loop when writing error messages to a full filesystem.
 * Flush stderr/stdout when mixing output.
 * Yield exit status compatible with diff(1).
 * Shrink stdio code size; allow cc -R; remove lint.
 *
 * Revision 4.4  87/12/18  11:44:47  narten
 * fixed to use "varargs" in "fprintf"; this is required if it is to
 * work on a SPARC machine such as a Sun-4
 *
 * Revision 4.3  87/10/18  10:37:18  narten
 * Updating version numbers. Changes relative to 1.1 actually relative
 * to version 4.1
 *
 * Revision 1.3  87/09/24  14:00:17  narten
 * Sources now pass through lint (if you ignore printf/sprintf/fprintf
 * warnings)
 *
 * Revision 1.2  87/03/27  14:22:33  jenkins
 * Port to suns
 *
 * Revision 4.1  83/03/25  18:12:51  wft
 * Only changed $Header to $Id.
 *
 * Revision 3.3  82/12/10  16:22:37  wft
 * Improved error messages, changed exit status on error to 1.
 *
 * Revision 3.2  82/11/28  21:27:10  wft
 * Renamed ctab to map and included EOFILE; ctab is now a macro in rcsbase.h.
 * Added fflsbuf(), fputs(), and fprintf(), which abort the RCS operations
 * properly in case there is an IO-error (e.g., file system full).
 *
 * Revision 3.1  82/10/11  19:43:56  wft
 * removed unused label out:;
 * made sure all calls to getc() return into an integer, not a char.
 */


/*
#define LEXDB
*/
/* version LEXDB is for testing the lexical analyzer. The testprogram
 * reads a stream of lexemes, enters the revision numbers into the
 * hashtable, and prints the recognized tokens. Keywords are recognized
 * as identifiers.
 */



#include "rcsbase.h"

libId(lexId, "$Id$")

static char *checkidentifier P((char*,int,int));
static void errsay P((char const*));
static void fatsay P((char const*));
static void lookup P((char const*));
static void startsay P((const char*,const char*));
static void warnsay P((char const*));

static struct hshentry *nexthsh;  /*pointer to next hash entry, set by lookup*/

enum tokens     nexttok;    /*next token, set by nextlex                    */

int             hshenter;   /*if true, next suitable lexeme will be entered */
                            /*into the symbol table. Handle with care.      */
int             nextc;      /*next input character, initialized by Lexinit  */

long		rcsline;    /*current line-number of input		    */
int             nerror;     /*counter for errors                            */
int             quietflag;  /*indicates quiet mode                          */
RILE *		finptr;	    /*input file descriptor			    */

FILE *          frewrite;   /*file descriptor for echoing input             */

FILE *		foutptr;    /* copy of frewrite, but 0 to suppress echo  */

static struct buf tokbuf;   /* token buffer				    */

char const *    NextString; /* next token				    */

/*
 * Our hash algorithm is h[0] = 0, h[i+1] = 4*h[i] + c,
 * so hshsize should be odd.
 * See B J McKenzie, R Harries & T Bell, Selecting a hashing algorithm,
 * Software--practice & experience 20, 2 (Feb 1990), 209-224.
 */
#ifndef hshsize
#	define hshsize 511
#endif

static struct hshentry *hshtab[hshsize]; /*hashtable			    */

static int ignored_phrases; /* have we ignored phrases in this RCS file? */

    void
warnignore()
{
    if (!ignored_phrases) {
	ignored_phrases = true;
	rcswarn("Unknown phrases like `%s ...;' are present.", NextString);
    }
}



	static void
lookup(str)
	char const *str;
/* Function: Looks up the character string pointed to by str in the
 * hashtable. If the string is not present, a new entry for it is created.
 * In any case, the address of the corresponding hashtable entry is placed
 * into nexthsh.
 */
{
	register unsigned ihash;  /* index into hashtable */
	register char const *sp;
	register struct hshentry *n, **p;

        /* calculate hash code */
	sp = str;
        ihash = 0;
	while (*sp)
		ihash  =  (ihash<<2) + *sp++;
	ihash %= hshsize;

	for (p = &hshtab[ihash];  ;  p = &n->nexthsh)
		if (!(n = *p)) {
			/* empty slot found */
			*p = n = ftalloc(struct hshentry);
			n->num = fstr_save(str);
			n->nexthsh = 0;
#			ifdef LEXDB
				VOID printf("\nEntered: %s at %u ", str, ihash);
#			endif
			break;
		} else if (strcmp(str, n->num) == 0)
			/* match found */
			break;
	nexthsh = n;
	NextString = n->num;
}






	void
Lexinit()
/* Function: Initialization of lexical analyzer:
 * initializes the hashtable,
 * initializes nextc, nexttok if finptr != 0
 */
{       register int            c;

	for (c = hshsize;  0 <= --c;  ) {
		hshtab[c] = 0;
        }

	nerror = 0;
	if (finptr) {
		foutptr = 0;
		hshenter = true;
		ignored_phrases = false;
		rcsline = 1;
		bufrealloc(&tokbuf, 2);
		Iget_(finptr, nextc)
                nextlex();            /*initial token*/
        }
}







	void
nextlex()

/* Function: Reads the next token and sets nexttok to the next token code.
 * Only if hshenter is set, a revision number is entered into the
 * hashtable and a pointer to it is placed into nexthsh.
 * This is useful for avoiding that dates are placed into the hashtable.
 * For ID's and NUM's, NextString is set to the character string.
 * Assumption: nextc contains the next character.
 */
{       register c;
	declarecache;
	register FILE *frew;
        register char * sp;
	char const *limit;
        register enum tokens d;
	register RILE *fin;

	fin=finptr; frew=foutptr;
	setupcache(fin); cache(fin);
	c = nextc;

	for (;;) { switch ((d = ctab[c])) {

	default:
		fatserror("unknown character `%c'", c);
		/*NOTREACHED*/

        case NEWLN:
		++rcsline;
#               ifdef LEXDB
		afputc('\n',stdout);
#               endif
                /* Note: falls into next case */

        case SPACE:
		GETC_(frew, c)
		continue;

	case IDCHAR:
	case LETTER:
	case Letter:
		d = ID;
		/* fall into */
	case DIGIT:
	case PERIOD:
		sp = tokbuf.string;
		limit = sp + tokbuf.size;
		*sp++ = c;
		for (;;) {
			GETC_(frew, c)
			switch (ctab[c]) {
			    case IDCHAR:
			    case LETTER:
			    case Letter:
				d = ID;
				/* fall into */
			    case DIGIT:
			    case PERIOD:
				*sp++ = c;
				if (limit <= sp)
					sp = bufenlarge(&tokbuf, &limit);
				continue;

			    default:
				break;
			}
			break;
                }
		*sp = 0;
		if (d == DIGIT  ||  d == PERIOD) {
			d = NUM;
			if (hshenter) {
				lookup(tokbuf.string);
				break;
			}
		}
		NextString = fstr_save(tokbuf.string);
		break;

        case SBEGIN: /* long string */
		d = STRING;
                /* note: only the initial SBEGIN has been read*/
                /* read the string, and reset nextc afterwards*/
		break;

	case COLON:
	case SEMI:
		GETC_(frew, c)
		break;
	} break; }
	nextc = c;
	nexttok = d;
	uncache(fin);
}

	int
eoflex()
/*
 * Yield true if we look ahead to the end of the input, false otherwise.
 * nextc becomes undefined at end of file.
 */
{
	register int c;
	declarecache;
	register FILE *fout;
	register RILE *fin;

	c = nextc;
	fin = finptr;
	fout = foutptr;
	setupcache(fin); cache(fin);

	for (;;) {
		switch (ctab[c]) {
			default:
				nextc = c;
				uncache(fin);
				return false;

			case NEWLN:
				++rcsline;
				/* fall into */
			case SPACE:
				cachegeteof_(c, {uncache(fin);return true;})
				break;
		}
		if (fout)
			aputc_(c, fout)
	}
}


int getlex(token)
enum tokens token;
/* Function: Checks if nexttok is the same as token. If so,
 * advances the input by calling nextlex and returns true.
 * otherwise returns false.
 * Doesn't work for strings and keywords; loses the character string for ids.
 */
{
        if (nexttok==token) {
                nextlex();
                return(true);
        } else  return(false);
}

	int
getkeyopt(key)
	char const *key;
/* Function: If the current token is a keyword identical to key,
 * advances the input by calling nextlex and returns true;
 * otherwise returns false.
 */
{
	if (nexttok==ID  &&  strcmp(key,NextString) == 0) {
		 /* match found */
		 ffree1(NextString);
		 nextlex();
		 return(true);
        }
        return(false);
}

	void
getkey(key)
	char const *key;
/* Check that the current input token is a keyword identical to key,
 * and advance the input by calling nextlex.
 */
{
	if (!getkeyopt(key))
		fatserror("missing '%s' keyword", key);
}

	void
getkeystring(key)
	char const *key;
/* Check that the current input token is a keyword identical to key,
 * and advance the input by calling nextlex; then look ahead for a string.
 */
{
	getkey(key);
	if (nexttok != STRING)
		fatserror("missing string after '%s' keyword", key);
}


	char const *
getid()
/* Function: Checks if nexttok is an identifier. If so,
 * advances the input by calling nextlex and returns a pointer
 * to the identifier; otherwise returns 0.
 * Treats keywords as identifiers.
 */
{
	register char const *name;
        if (nexttok==ID) {
                name = NextString;
                nextlex();
                return name;
	} else
		return 0;
}


struct hshentry * getnum()
/* Function: Checks if nexttok is a number. If so,
 * advances the input by calling nextlex and returns a pointer
 * to the hashtable entry.  Otherwise returns 0.
 * Doesn't work if hshenter is false.
 */
{
        register struct hshentry * num;
        if (nexttok==NUM) {
                num=nexthsh;
                nextlex();
                return num;
	} else
		return 0;
}

	struct cbuf
getphrases(key)
	char const *key;
/*
* Get a series of phrases that do not start with KEY.  Yield resulting buffer.
* Stop when the next phrase starts with a token that is not an identifier,
* or is KEY.  Copy input to foutptr if it is set.  Unlike ignorephrases(),
* this routine assumes nextlex() has already been invoked before we start.
*/
{
    declarecache;
    register int c;
    register char const *kn;
    struct cbuf r;
    register RILE *fin;
    register FILE *frew;
#   if large_memory
#	define savech_(c) ;
#   else
	register char *p;
	char const *limit;
	struct buf b;
#	define savech_(c) {if (limit<=p)p=bufenlarge(&b,&limit); *p++ =(c);}
#   endif

    if (nexttok!=ID  ||  strcmp(NextString,key) == 0)
	clear_buf(&r);
    else {
	warnignore();
	fin = finptr;
	frew = foutptr;
	setupcache(fin); cache(fin);
#	if large_memory
	    r.string = (char const*)cacheptr() - strlen(NextString) - 1;
#	else
	    bufautobegin(&b);
	    bufscpy(&b, NextString);
	    p = b.string + strlen(b.string);
	    limit = b.string + b.size;
#	endif
	ffree1(NextString);
	c = nextc;
	for (;;) {
	    for (;;) {
		savech_(c)
		switch (ctab[c]) {
		    default:
			fatserror("unknown character `%c'", c);
			/*NOTREACHED*/
		    case NEWLN:
			++rcsline;
			/* fall into */
		    case COLON: case DIGIT: case LETTER: case Letter:
		    case PERIOD: case SPACE:
			GETC_(frew, c)
			continue;
		    case SBEGIN: /* long string */
			for (;;) {
			    for (;;) {
				GETC_(frew, c)
				savech_(c)
				switch (c) {
				    case '\n':
					++rcsline;
					/* fall into */
				    default:
					continue;

				    case SDELIM:
					break;
				}
				break;
			    }
			    GETC_(frew, c)
			    if (c != SDELIM)
				break;
			    savech_(c)
			}
			continue;
		    case SEMI:
			cacheget_(c)
			if (ctab[c] == NEWLN) {
			    if (frew)
				aputc_(c, frew)
			    ++rcsline;
			    savech_(c)
			    cacheget_(c)
			}
#			if large_memory
			    r.size = (char const*)cacheptr() - 1 - r.string;
#			endif
			for (;;) {
			    switch (ctab[c]) {
				case NEWLN:
					++rcsline;
					/* fall into */
				case SPACE:
					cacheget_(c)
					continue;

				default: break;
			    }
			    break;
			}
			if (frew)
			    aputc_(c, frew)
			break;
		}
		break;
	    }
	    if (ctab[c] == Letter) {
		    for (kn = key;  c && *kn==c;  kn++)
			GETC_(frew, c)
		    if (!*kn)
			switch (ctab[c]) {
			    case DIGIT: case LETTER: case Letter:
			    case IDCHAR: case PERIOD:
				break;
			    default:
				nextc = c;
				NextString = fstr_save(key);
				nexttok = ID;
				uncache(fin);
				goto returnit;
			}
#		    if !large_memory
			{
			    register char const *ki;
			    for (ki=key; ki<kn; )
				savech_(*ki++)
			}
#		    endif
	    } else {
		    nextc = c;
		    uncache(fin);
		    nextlex();
		    break;
	    }
	}
    returnit:;
#	if !large_memory
	    return bufremember(&b, (size_t)(p - b.string));
#	endif
    }
    return r;
}


	void
readstring()
/* skip over characters until terminating single SDELIM        */
/* If foutptr is set, copy every character read to foutptr.    */
/* Does not advance nextlex at the end.                        */
{       register c;
	declarecache;
	register FILE *frew;
	register RILE *fin;
	fin=finptr; frew=foutptr;
	setupcache(fin); cache(fin);
	for (;;) {
		GETC_(frew, c)
		switch (c) {
		    case '\n':
			++rcsline;
			break;

		    case SDELIM:
			GETC_(frew, c)
			if (c != SDELIM) {
				/* end of string */
				nextc = c;
				uncache(fin);
				return;
			}
			break;
		}
	}
}


	void
printstring()
/* Function: copy a string to stdout, until terminated with a single SDELIM.
 * Does not advance nextlex at the end.
 */
{
        register c;
	declarecache;
	register FILE *fout;
	register RILE *fin;
	fin=finptr;
	fout = stdout;
	setupcache(fin); cache(fin);
	for (;;) {
		cacheget_(c)
		switch (c) {
		    case '\n':
			++rcsline;
			break;
		    case SDELIM:
			cacheget_(c)
			if (c != SDELIM) {
                                nextc=c;
				uncache(fin);
                                return;
                        }
			break;
                }
		aputc_(c,fout)
        }
}



	struct cbuf
savestring(target)
	struct buf *target;
/* Copies a string terminated with SDELIM from file finptr to buffer target.
 * Double SDELIM is replaced with SDELIM.
 * If foutptr is set, the string is also copied unchanged to foutptr.
 * Does not advance nextlex at the end.
 * Yield a copy of *TARGET, except with exact length.
 */
{
        register c;
	declarecache;
	register FILE *frew;
	register char *tp;
	register RILE *fin;
	char const *limit;
	struct cbuf r;

	fin=finptr; frew=foutptr;
	setupcache(fin); cache(fin);
	tp = target->string;  limit = tp + target->size;
	for (;;) {
		GETC_(frew, c)
		switch (c) {
		    case '\n':
			++rcsline;
			break;
		    case SDELIM:
			GETC_(frew, c)
			if (c != SDELIM) {
                                /* end of string */
                                nextc=c;
				r.string = target->string;
				r.size = tp - r.string;
				uncache(fin);
				return r;
                        }
			break;
                }
		if (tp == limit)
			tp = bufenlarge(target, &limit);
		*tp++ = c;
        }
}


	static char *
checkidentifier(id, delimiter, dotok)
	register char *id;
	int delimiter;
	register int dotok;
/*   Function:  check whether the string starting at id is an   */
/*		identifier and return a pointer to the delimiter*/
/*		after the identifier.  White space, delim and 0 */
/*              are legal delimiters.  Aborts the program if not*/
/*              a legal identifier. Useful for checking commands*/
/*		If !delim, the only delimiter is 0.		*/
/*		Allow '.' in identifier only if DOTOK is set.   */
{
        register char    *temp;
	register char c;
	register char delim = delimiter;
	int isid = false;

	temp = id;
	for (;;  id++) {
		switch (ctab[(unsigned char)(c = *id)]) {
			case IDCHAR:
			case LETTER:
			case Letter:
				isid = true;
				continue;

			case DIGIT:
				continue;

			case PERIOD:
				if (dotok)
					continue;
				break;

			default:
				break;
		}
		break;
	}
	if (	 ! isid
	    ||	 (c  &&  (!delim || (c!=delim && c!=' ' && c!='\t' && c!='\n')))
	) {
                /* append \0 to end of id before error message */
		while ((c = *id) && c!=' ' && c!='\t' && c!='\n' && c!=delim)
		    id++;
                *id = '\0';
		faterror("invalid %s `%s'",
			dotok ? "identifier" : "symbol", temp
		);
	}
	return id;
}

	char *
checkid(id, delimiter)
	char *id;
	int delimiter;
{
	return checkidentifier(id, delimiter, true);
}

	char *
checksym(sym, delimiter)
	char *sym;
	int delimiter;
{
	return checkidentifier(sym, delimiter, false);
}

	void
checksid(id)
	char *id;
/* Check whether the string ID is an identifier.  */
{
	VOID checkid(id, 0);
}

	void
checkssym(sym)
	char *sym;
{
	VOID checksym(sym, 0);
}


#if !large_memory
#   define Iclose(f) fclose(f)
#else
# if !maps_memory
    static int Iclose P((RILE *));
	static int
    Iclose(f)
	register RILE *f;
    {
	tfree(f->base);
	f->base = 0;
	return fclose(f->stream);
    }
# else
    static int Iclose P((RILE *));
	static int
    Iclose(f)
	register RILE *f;
    {
	(* f->deallocate) (f);
	f->base = 0;
	return close(f->fd);
    }

#   if has_map_fd
	static void map_fd_deallocate P((RILE *));
	    static void
	map_fd_deallocate(f)
	    register RILE *f;
	{
	    if (vm_deallocate(
		task_self(),
		(vm_address_t) f->base,
		(vm_size_t) (f->lim - f->base)
	    ) != KERN_SUCCESS)
		efaterror("vm_deallocate");
	}
#   endif
#   if has_mmap
	static void mmap_deallocate P((RILE *));
	    static void
	mmap_deallocate(f)
	    register RILE *f;
	{
	    if (munmap((char *) f->base, (size_t) (f->lim - f->base)) != 0)
		efaterror("munmap");
	}
#   endif
    static void read_deallocate P((RILE *));
	static void
    read_deallocate(f)
	RILE *f;
    {
	tfree(f->base);
    }

    static void nothing_to_deallocate P((RILE *));
	static void
    nothing_to_deallocate(f)
	RILE *f;
    {
    }
# endif
#endif


#if large_memory && maps_memory
	static RILE *fd2_RILE P((int,char const*,struct stat*));
	static RILE *
fd2_RILE(fd, name, status)
#else
	static RILE *fd2RILE P((int,char const*,char const*,struct stat*));
	static RILE *
fd2RILE(fd, name, type, status)
	char const *type;
#endif
	int fd;
	char const *name;
	register struct stat *status;
{
	struct stat st;

	if (!status)
		status = &st;
	if (fstat(fd, status) != 0)
		efaterror(name);
	if (!S_ISREG(status->st_mode)) {
		error("`%s' is not a regular file", name);
		VOID close(fd);
		errno = EINVAL;
		return 0;
	} else {

#	    if !(large_memory && maps_memory)
		FILE *stream;
		if (!(stream = fdopen(fd, type)))
			efaterror(name);
#	    endif

#	    if !large_memory
		return stream;
#	    else
#		define RILES 3
	      {
		static RILE rilebuf[RILES];

		register RILE *f;
		size_t s = status->st_size;

		if (s != status->st_size)
			faterror("%s: too large", name);
		for (f = rilebuf;  f->base;  f++)
			if (f == rilebuf+RILES)
				faterror("too many RILEs");
#		if maps_memory
			f->deallocate = nothing_to_deallocate;
#		endif
		if (!s) {
		    static unsigned char nothing;
		    f->base = &nothing; /* Any nonzero address will do.  */
		} else {
		    f->base = 0;
#		    if has_map_fd
			map_fd(
				fd, (vm_offset_t)0, (vm_address_t*) &f->base,
				TRUE, (vm_size_t)s
			);
			f->deallocate = map_fd_deallocate;
#		    endif
#		    if has_mmap
			if (!f->base) {
			    catchmmapints();
			    f->base = (unsigned char *) mmap(
				(char *)0, s, PROT_READ, MAP_SHARED,
				fd, (off_t)0
			    );
#			    ifndef MAP_FAILED
#			    define MAP_FAILED (-1)
#			    endif
			    if (f->base == (unsigned char *) MAP_FAILED)
				f->base = 0;
			    else {
#				if has_NFS && mmap_signal
				    /*
				    * On many hosts, the superuser
				    * can mmap an NFS file it can't read.
				    * So access the first page now, and print
				    * a nice message if a bus error occurs.
				    */
				    readAccessFilenameBuffer(name, f->base);
#				endif
			    }
			    f->deallocate = mmap_deallocate;
			}
#		    endif
		    if (!f->base) {
			f->base = tnalloc(unsigned char, s);
#			if maps_memory
			{
			    /*
			    * We can't map the file into memory for some reason.
			    * Read it into main memory all at once; this is
			    * the simplest substitute for memory mapping.
			    */
			    char *bufptr = (char *) f->base;
			    size_t bufsiz = s;
			    do {
				ssize_t r = read(fd, bufptr, bufsiz);
				switch (r) {
				    case -1:
					efaterror(name);

				    case 0:
					/* The file must have shrunk!  */
					status->st_size = s -= bufsiz;
					bufsiz = 0;
					break;

				    default:
					bufptr += r;
					bufsiz -= r;
					break;
				}
			    } while (bufsiz);
			    if (lseek(fd, (off_t)0, SEEK_SET) == -1)
				efaterror(name);
			    f->deallocate = read_deallocate;
			}
#			endif
		    }
		}
		f->ptr = f->base;
		f->lim = f->base + s;
		f->fd = fd;
#		if !maps_memory
		    f->readlim = f->base;
		    f->stream = stream;
#		endif
		if_advise_access(s, f, MADV_SEQUENTIAL);
		return f;
	      }
#	    endif
	}
}

#if !maps_memory && large_memory
	int
Igetmore(f)
	register RILE *f;
{
	register fread_type r;
	register size_t s = f->lim - f->readlim;

	if (BUFSIZ < s)
		s = BUFSIZ;
	if (!(r = Fread(f->readlim, sizeof(*f->readlim), s, f->stream))) {
		testIerror(f->stream);
		f->lim = f->readlim;  /* The file might have shrunk!  */
		return 0;
	}
	f->readlim += r;
	return 1;
}
#endif

#if has_madvise && has_mmap && large_memory
	void
advise_access(f, advice)
	register RILE *f;
	int advice;
{
    if (f->deallocate == mmap_deallocate)
	VOID madvise((char *)f->base, (size_t)(f->lim - f->base), advice);
	/* Don't worry if madvise fails; it's only advisory.  */
}
#endif

	RILE *
#if large_memory && maps_memory
I_open(name, status)
#else
Iopen(name, type, status)
	char const *type;
#endif
	char const *name;
	struct stat *status;
/* Open NAME for reading, yield its descriptor, and set *STATUS.  */
{
	int fd = fdSafer(open(name, O_RDONLY
#		if OPEN_O_BINARY
			|  (strchr(type,'b') ? OPEN_O_BINARY : 0)
#		endif
	));

	if (fd < 0)
		return 0;
#	if large_memory && maps_memory
		return fd2_RILE(fd, name, status);
#	else
		return fd2RILE(fd, name, type, status);
#	endif
}


static int Oerrloop;

	void
Oerror()
{
	if (Oerrloop)
		exiterr();
	Oerrloop = true;
	efaterror("output error");
}

void Ieof() { fatserror("unexpected end of file"); }
void Ierror() { efaterror("input error"); }
void testIerror(f) FILE *f; { if (ferror(f)) Ierror(); }
void testOerror(o) FILE *o; { if (ferror(o)) Oerror(); }

void Ifclose(f) RILE *f; { if (f && Iclose(f)!=0) Ierror(); }
void Ofclose(f) FILE *f; { if (f && fclose(f)!=0) Oerror(); }
void Izclose(p) RILE **p; { Ifclose(*p); *p = 0; }
void Ozclose(p) FILE **p; { Ofclose(*p); *p = 0; }

#if !large_memory
	void
testIeof(f)
	FILE *f;
{
	testIerror(f);
	if (feof(f))
		Ieof();
}
void Irewind(f) FILE *f; { if (fseek(f,0L,SEEK_SET) != 0) Ierror(); }
#endif

void Orewind(f) FILE *f; { if (fseek(f,0L,SEEK_SET) != 0) Oerror(); }

void aflush(f) FILE *f; { if (fflush(f) != 0) Oerror(); }
void eflush() { if (fflush(stderr)!=0 && !Oerrloop) Oerror(); }
void oflush()
{
	if (fflush(workstdout ? workstdout : stdout) != 0  &&  !Oerrloop)
		Oerror();
}

	void
fatcleanup(already_newline)
	int already_newline;
{
	VOID fprintf(stderr, already_newline+"\n%s aborted\n", cmdid);
	exiterr();
}

	static void
startsay(s, t)
	const char *s, *t;
{
	oflush();
	if (s)
	    aprintf(stderr, "%s: %s: %s", cmdid, s, t);
	else
	    aprintf(stderr, "%s: %s", cmdid, t);
}

	static void
fatsay(s)
	char const *s;
{
	startsay(s, "");
}

	static void
errsay(s)
	char const *s;
{
	fatsay(s);
	nerror++;
}

	static void
warnsay(s)
	char const *s;
{
	startsay(s, "warning: ");
}

void eerror(s) char const *s; { enerror(errno,s); }

	void
enerror(e,s)
	int e;
	char const *s;
{
	errsay((char const*)0);
	errno = e;
	perror(s);
	eflush();
}

void efaterror(s) char const *s; { enfaterror(errno,s); }

	void
enfaterror(e,s)
	int e;
	char const *s;
{
	fatsay((char const*)0);
	errno = e;
	perror(s);
	fatcleanup(true);
}

#if has_prototypes
	void
error(char const *format,...)
#else
	/*VARARGS1*/ void error(format, va_alist) char const *format; va_dcl
#endif
/* non-fatal error */
{
	va_list args;
	errsay((char const*)0);
	vararg_start(args, format);
	fvfprintf(stderr, format, args);
	va_end(args);
	afputc('\n',stderr);
	eflush();
}

#if has_prototypes
	void
rcserror(char const *format,...)
#else
	/*VARARGS1*/ void rcserror(format, va_alist) char const *format; va_dcl
#endif
/* non-fatal RCS file error */
{
	va_list args;
	errsay(RCSname);
	vararg_start(args, format);
	fvfprintf(stderr, format, args);
	va_end(args);
	afputc('\n',stderr);
	eflush();
}

#if has_prototypes
	void
workerror(char const *format,...)
#else
	/*VARARGS1*/ void workerror(format, va_alist) char const *format; va_dcl
#endif
/* non-fatal working file error */
{
	va_list args;
	errsay(workname);
	vararg_start(args, format);
	fvfprintf(stderr, format, args);
	va_end(args);
	afputc('\n',stderr);
	eflush();
}

#if has_prototypes
	void
fatserror(char const *format,...)
#else
	/*VARARGS1*/ void
	fatserror(format, va_alist) char const *format; va_dcl
#endif
/* fatal RCS file syntax error */
{
	va_list args;
	oflush();
	VOID fprintf(stderr, "%s: %s:%ld: ", cmdid, RCSname, rcsline);
	vararg_start(args, format);
	fvfprintf(stderr, format, args);
	va_end(args);
	fatcleanup(false);
}

#if has_prototypes
	void
faterror(char const *format,...)
#else
	/*VARARGS1*/ void faterror(format, va_alist)
	char const *format; va_dcl
#endif
/* fatal error, terminates program after cleanup */
{
	va_list args;
	fatsay((char const*)0);
	vararg_start(args, format);
	fvfprintf(stderr, format, args);
	va_end(args);
	fatcleanup(false);
}

#if has_prototypes
	void
rcsfaterror(char const *format,...)
#else
	/*VARARGS1*/ void rcsfaterror(format, va_alist)
	char const *format; va_dcl
#endif
/* fatal RCS file error, terminates program after cleanup */
{
	va_list args;
	fatsay(RCSname);
	vararg_start(args, format);
	fvfprintf(stderr, format, args);
	va_end(args);
	fatcleanup(false);
}

#if has_prototypes
	void
warn(char const *format,...)
#else
	/*VARARGS1*/ void warn(format, va_alist) char const *format; va_dcl
#endif
/* warning */
{
	va_list args;
	if (!quietflag) {
		warnsay((char *)0);
		vararg_start(args, format);
		fvfprintf(stderr, format, args);
		va_end(args);
		afputc('\n', stderr);
		eflush();
	}
}

#if has_prototypes
	void
rcswarn(char const *format,...)
#else
	/*VARARGS1*/ void rcswarn(format, va_alist) char const *format; va_dcl
#endif
/* RCS file warning */
{
	va_list args;
	if (!quietflag) {
		warnsay(RCSname);
		vararg_start(args, format);
		fvfprintf(stderr, format, args);
		va_end(args);
		afputc('\n', stderr);
		eflush();
	}
}

#if has_prototypes
	void
workwarn(char const *format,...)
#else
	/*VARARGS1*/ void workwarn(format, va_alist) char const *format; va_dcl
#endif
/* working file warning */
{
	va_list args;
	if (!quietflag) {
		warnsay(workname);
		vararg_start(args, format);
		fvfprintf(stderr, format, args);
		va_end(args);
		afputc('\n', stderr);
		eflush();
	}
}

	void
redefined(c)
	int c;
{
	warn("redefinition of -%c option", c);
}

#if has_prototypes
	void
diagnose(char const *format,...)
#else
	/*VARARGS1*/ void diagnose(format, va_alist) char const *format; va_dcl
#endif
/* prints a diagnostic message */
/* Unlike the other routines, it does not append a newline. */
/* This lets some callers suppress the newline, and is faster */
/* in implementations that flush stderr just at the end of each printf. */
{
	va_list args;
        if (!quietflag) {
		oflush();
		vararg_start(args, format);
		fvfprintf(stderr, format, args);
		va_end(args);
		eflush();
        }
}



	void
afputc(c, f)
/* afputc(c,f); acts like aputc_(c,f) but is smaller and slower.  */
	int c;
	register FILE *f;
{
	aputc_(c,f)
}


	void
aputs(s, iop)
	char const *s;
	FILE *iop;
/* Function: Put string s on file iop, abort on error.
 */
{
#if has_fputs
	if (fputs(s, iop) < 0)
		Oerror();
#else
	awrite(s, strlen(s), iop);
#endif
}



	void
#if has_prototypes
fvfprintf(FILE *stream, char const *format, va_list args)
#else
	fvfprintf(stream,format,args) FILE *stream; char *format; va_list args;
#endif
/* like vfprintf, except abort program on error */
{
#if has_vfprintf
	if (vfprintf(stream, format, args) < 0)
		Oerror();
#else
#	if has__doprintf
		_doprintf(stream, format, args);
#	else
#	if has__doprnt
		_doprnt(format, args, stream);
#	else
		int *a = (int *)args;
		VOID fprintf(stream, format,
			a[0], a[1], a[2], a[3], a[4],
			a[5], a[6], a[7], a[8], a[9]
		);
#	endif
#	endif
	if (ferror(stream))
		Oerror();
#endif
}

#if has_prototypes
	void
aprintf(FILE *iop, char const *fmt, ...)
#else
	/*VARARGS2*/ void
aprintf(iop, fmt, va_alist)
FILE *iop;
char const *fmt;
va_dcl
#endif
/* Function: formatted output. Same as fprintf in stdio,
 * but aborts program on error
 */
{
	va_list ap;
	vararg_start(ap, fmt);
	fvfprintf(iop, fmt, ap);
	va_end(ap);
}



#ifdef LEXDB
/* test program reading a stream of lexemes and printing the tokens.
 */



	int
main(argc,argv)
int argc; char * argv[];
{
        cmdid="lextest";
        if (argc<2) {
		aputs("No input file\n",stderr);
		exitmain(EXIT_FAILURE);
        }
	if (!(finptr=Iopen(argv[1], FOPEN_R, (struct stat*)0))) {
		faterror("can't open input file %s",argv[1]);
        }
        Lexinit();
	while (!eoflex()) {
        switch (nexttok) {

        case ID:
                VOID printf("ID: %s",NextString);
                break;

        case NUM:
		if (hshenter)
                   VOID printf("NUM: %s, index: %d",nexthsh->num, nexthsh-hshtab);
                else
                   VOID printf("NUM, unentered: %s",NextString);
                hshenter = !hshenter; /*alternate between dates and numbers*/
                break;

        case COLON:
                VOID printf("COLON"); break;

        case SEMI:
                VOID printf("SEMI"); break;

        case STRING:
                readstring();
                VOID printf("STRING"); break;

        case UNKN:
                VOID printf("UNKN"); break;

        default:
                VOID printf("DEFAULT"); break;
        }
        VOID printf(" | ");
        nextlex();
        }
	exitmain(EXIT_SUCCESS);
}

void exiterr() { _exit(EXIT_FAILURE); }


#endif