rcslex.c

早期freebsd实现

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

/* Copyright (C) 1982, 1988, 1989 Walter Tichy
   Copyright 1990, 1991 by 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, 675 Mass Ave, Cambridge, MA 02139, USA.

Report problems and direct all questions to:

    [e-mail]

*/



/* $Log: rcslex.c,v $
 * 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: rcslex.c,v 5.11 1991/11/03 03:30:44 eggert Exp $")

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  */

unsigned 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|quietflag)) {
	ignored_phrases = true;
	warn("Unknown phrases like `%s ...;' are in the RCS file.", 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 = nil;
#			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] = nil;
        }

	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 DIGIT:
		sp = tokbuf.string;
		limit = sp + tokbuf.size;
		*sp++ = c;
		for (;;) {
			GETC(frew, c);
			if ((d=ctab[c])!=DIGIT && d!=PERIOD)
				break;
                        *sp++ = c;         /* 1.2. and 1.2 are different */
			if (limit <= sp)
				sp = bufenlarge(&tokbuf, &limit);
                }
		*sp = 0;
		if (hshenter)
			lookup(tokbuf.string);
		else
			NextString = fstr_save(tokbuf.string);
		d = NUM;
		break;


        case LETTER:
	case Letter:
		sp = tokbuf.string;
		limit = sp + tokbuf.size;
		*sp++ = c;
		for (;;) {
			GETC(frew, c);
			if ((d=ctab[c])!=LETTER && d!=Letter && d!=DIGIT && d!=IDCHAR)
				break;
                        *sp++ = c;
			if (limit <= sp)
				sp = bufenlarge(&tokbuf, &limit);
                }
		*sp = 0;
		NextString = fstr_save(tokbuf.string);
		d = ID;  /* may be ID or keyword */
		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 nil.
 * Treats keywords as identifiers.
 */
{
	register char const *name;
        if (nexttok==ID) {
                name = NextString;
                nextlex();
                return name;
        } else  return nil;
}


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 nil.
 * Doesn't work if hshenter is false.
 */
{
        register struct hshentry * num;
        if (nexttok==NUM) {
                num=nexthsh;
                nextlex();
                return num;
        } else  return nil;
}

	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.
 * Assume !foutptr.
 */
{
    declarecache;
    register int c;
    register char *p;
    char const *limit;
    register char const *ki, *kn;
    struct cbuf r;
    struct buf b;
    register RILE *fin;

    if (nexttok!=ID  ||  strcmp(NextString,key) == 0) {
	r.string = 0;
	r.size = 0;
	return r;
    } else {
	warnignore();
	fin = finptr;
	setupcache(fin); cache(fin);
	bufautobegin(&b);
	bufscpy(&b, NextString);
	ffree1(NextString);
	p = b.string + strlen(b.string);
	limit = b.string + b.size;
	c = nextc;
	for (;;) {
	    for (;;) {
		if (limit <= p)
		    p = bufenlarge(&b, &limit);
		*p++ = 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:
			cacheget(c);
			continue;
		    case SBEGIN: /* long string */
			for (;;) {
			    for (;;) {
				if (limit <= p)
				    p = bufenlarge(&b, &limit);
				cacheget(c);
				*p++ = c;
				switch (c) {
				    case '\n':
					++rcsline;
					/* fall into */
				    default:
					continue;

				    case SDELIM:
					break;
				}
				break;
			    }
			    cacheget(c);
			    if (c != SDELIM)
				break;
			    if (limit <= p)
				p = bufenlarge(&b, &limit);
			    *p++ = c;
			}
			continue;
		    case SEMI:
			cacheget(c);
			if (ctab[c] == NEWLN) {
			    ++rcsline;
			    if (limit <= p)
				p = bufenlarge(&b, &limit);
			    *p++ = c;
			    cacheget(c);
			}
			for (;;) {
			    switch (ctab[c]) {
				case NEWLN:
					++rcsline;
					/* fall into */
				case SPACE:
					cacheget(c);
					continue;

				default: break;
			    }
			    break;
			}
			break;
		}
		break;
	    }
	    switch (ctab[c]) {
		case LETTER:
		case Letter:
		    for (kn = key;  c && *kn==c;  kn++)
			cacheget(c);
		    if (!*kn)
			switch (ctab[c]) {
			    case DIGIT: case LETTER: case Letter:
				break;
			    default:
				nextc = c;
				NextString = fstr_save(key);
				nexttok = ID;
				uncache(fin);
				goto returnit;
			}
		    for (ki=key; ki<kn; ) {
			if (limit <= p)
			    p = bufenlarge(&b, &limit);
			*p++ = *ki++;
		    }
		    break;

		default:
		    nextc = c;
		    uncache(fin);
		    nextlex();
		    goto returnit;
	    }
	}
    returnit:
	return bufremember(&b, (size_t)(p - b.string));
    }
}


	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 (;;) {