mirror of
https://git.FreeBSD.org/src.git
synced 2024-12-21 11:13:30 +00:00
76d00450e6
vi-mode, removal of clause 3, cleanups and the export of the tokenization functions. Not included: config.h, filecomplete.{c,h}
446 lines
9.6 KiB
C
446 lines
9.6 KiB
C
/*-
|
|
* Copyright (c) 1992, 1993
|
|
* The Regents of the University of California. All rights reserved.
|
|
*
|
|
* This code is derived from software contributed to Berkeley by
|
|
* Christos Zoulas of Cornell University.
|
|
*
|
|
* 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. 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.
|
|
*
|
|
* $NetBSD: tokenizer.c,v 1.14 2003/12/05 13:37:48 lukem Exp $
|
|
*/
|
|
|
|
#if !defined(lint) && !defined(SCCSID)
|
|
static char sccsid[] = "@(#)tokenizer.c 8.1 (Berkeley) 6/4/93";
|
|
#endif /* not lint && not SCCSID */
|
|
#include <sys/cdefs.h>
|
|
__FBSDID("$FreeBSD$");
|
|
|
|
/*
|
|
* tokenize.c: Bourne shell like tokenizer
|
|
*/
|
|
#include "sys.h"
|
|
#include <string.h>
|
|
#include <stdlib.h>
|
|
#include "histedit.h"
|
|
|
|
typedef enum {
|
|
Q_none, Q_single, Q_double, Q_one, Q_doubleone
|
|
} quote_t;
|
|
|
|
#define IFS "\t \n"
|
|
|
|
#define TOK_KEEP 1
|
|
#define TOK_EAT 2
|
|
|
|
#define WINCR 20
|
|
#define AINCR 10
|
|
|
|
#define tok_strdup(a) strdup(a)
|
|
#define tok_malloc(a) malloc(a)
|
|
#define tok_free(a) free(a)
|
|
#define tok_realloc(a, b) realloc(a, b)
|
|
|
|
|
|
struct tokenizer {
|
|
char *ifs; /* In field separator */
|
|
int argc, amax; /* Current and maximum number of args */
|
|
char **argv; /* Argument list */
|
|
char *wptr, *wmax; /* Space and limit on the word buffer */
|
|
char *wstart; /* Beginning of next word */
|
|
char *wspace; /* Space of word buffer */
|
|
quote_t quote; /* Quoting state */
|
|
int flags; /* flags; */
|
|
};
|
|
|
|
|
|
private void tok_finish(Tokenizer *);
|
|
|
|
|
|
/* tok_finish():
|
|
* Finish a word in the tokenizer.
|
|
*/
|
|
private void
|
|
tok_finish(Tokenizer *tok)
|
|
{
|
|
|
|
*tok->wptr = '\0';
|
|
if ((tok->flags & TOK_KEEP) || tok->wptr != tok->wstart) {
|
|
tok->argv[tok->argc++] = tok->wstart;
|
|
tok->argv[tok->argc] = NULL;
|
|
tok->wstart = ++tok->wptr;
|
|
}
|
|
tok->flags &= ~TOK_KEEP;
|
|
}
|
|
|
|
|
|
/* tok_init():
|
|
* Initialize the tokenizer
|
|
*/
|
|
public Tokenizer *
|
|
tok_init(const char *ifs)
|
|
{
|
|
Tokenizer *tok = (Tokenizer *) tok_malloc(sizeof(Tokenizer));
|
|
|
|
if (tok == NULL)
|
|
return NULL;
|
|
tok->ifs = tok_strdup(ifs ? ifs : IFS);
|
|
if (tok->ifs == NULL) {
|
|
tok_free((ptr_t)tok);
|
|
return NULL;
|
|
}
|
|
tok->argc = 0;
|
|
tok->amax = AINCR;
|
|
tok->argv = (char **) tok_malloc(sizeof(char *) * tok->amax);
|
|
if (tok->argv == NULL) {
|
|
tok_free((ptr_t)tok->ifs);
|
|
tok_free((ptr_t)tok);
|
|
return NULL;
|
|
}
|
|
tok->argv[0] = NULL;
|
|
tok->wspace = (char *) tok_malloc(WINCR);
|
|
if (tok->wspace == NULL) {
|
|
tok_free((ptr_t)tok->argv);
|
|
tok_free((ptr_t)tok->ifs);
|
|
tok_free((ptr_t)tok);
|
|
return NULL;
|
|
}
|
|
tok->wmax = tok->wspace + WINCR;
|
|
tok->wstart = tok->wspace;
|
|
tok->wptr = tok->wspace;
|
|
tok->flags = 0;
|
|
tok->quote = Q_none;
|
|
|
|
return (tok);
|
|
}
|
|
|
|
|
|
/* tok_reset():
|
|
* Reset the tokenizer
|
|
*/
|
|
public void
|
|
tok_reset(Tokenizer *tok)
|
|
{
|
|
|
|
tok->argc = 0;
|
|
tok->wstart = tok->wspace;
|
|
tok->wptr = tok->wspace;
|
|
tok->flags = 0;
|
|
tok->quote = Q_none;
|
|
}
|
|
|
|
|
|
/* tok_end():
|
|
* Clean up
|
|
*/
|
|
public void
|
|
tok_end(Tokenizer *tok)
|
|
{
|
|
|
|
tok_free((ptr_t) tok->ifs);
|
|
tok_free((ptr_t) tok->wspace);
|
|
tok_free((ptr_t) tok->argv);
|
|
tok_free((ptr_t) tok);
|
|
}
|
|
|
|
|
|
|
|
/* tok_line():
|
|
* Bourne shell (sh(1)) like tokenizing
|
|
* Arguments:
|
|
* tok current tokenizer state (setup with tok_init())
|
|
* line line to parse
|
|
* Returns:
|
|
* -1 Internal error
|
|
* 3 Quoted return
|
|
* 2 Unmatched double quote
|
|
* 1 Unmatched single quote
|
|
* 0 Ok
|
|
* Modifies (if return value is 0):
|
|
* argc number of arguments
|
|
* argv argument array
|
|
* cursorc if !NULL, argv element containing cursor
|
|
* cursorv if !NULL, offset in argv[cursorc] of cursor
|
|
*/
|
|
public int
|
|
tok_line(Tokenizer *tok, const LineInfo *line,
|
|
int *argc, const char ***argv, int *cursorc, int *cursoro)
|
|
{
|
|
const char *ptr;
|
|
int cc, co;
|
|
|
|
cc = co = -1;
|
|
ptr = line->buffer;
|
|
for (ptr = line->buffer; ;ptr++) {
|
|
if (ptr >= line->lastchar)
|
|
ptr = "";
|
|
if (ptr == line->cursor) {
|
|
cc = tok->argc;
|
|
co = tok->wptr - tok->wstart;
|
|
}
|
|
switch (*ptr) {
|
|
case '\'':
|
|
tok->flags |= TOK_KEEP;
|
|
tok->flags &= ~TOK_EAT;
|
|
switch (tok->quote) {
|
|
case Q_none:
|
|
tok->quote = Q_single; /* Enter single quote
|
|
* mode */
|
|
break;
|
|
|
|
case Q_single: /* Exit single quote mode */
|
|
tok->quote = Q_none;
|
|
break;
|
|
|
|
case Q_one: /* Quote this ' */
|
|
tok->quote = Q_none;
|
|
*tok->wptr++ = *ptr;
|
|
break;
|
|
|
|
case Q_double: /* Stay in double quote mode */
|
|
*tok->wptr++ = *ptr;
|
|
break;
|
|
|
|
case Q_doubleone: /* Quote this ' */
|
|
tok->quote = Q_double;
|
|
*tok->wptr++ = *ptr;
|
|
break;
|
|
|
|
default:
|
|
return (-1);
|
|
}
|
|
break;
|
|
|
|
case '"':
|
|
tok->flags &= ~TOK_EAT;
|
|
tok->flags |= TOK_KEEP;
|
|
switch (tok->quote) {
|
|
case Q_none: /* Enter double quote mode */
|
|
tok->quote = Q_double;
|
|
break;
|
|
|
|
case Q_double: /* Exit double quote mode */
|
|
tok->quote = Q_none;
|
|
break;
|
|
|
|
case Q_one: /* Quote this " */
|
|
tok->quote = Q_none;
|
|
*tok->wptr++ = *ptr;
|
|
break;
|
|
|
|
case Q_single: /* Stay in single quote mode */
|
|
*tok->wptr++ = *ptr;
|
|
break;
|
|
|
|
case Q_doubleone: /* Quote this " */
|
|
tok->quote = Q_double;
|
|
*tok->wptr++ = *ptr;
|
|
break;
|
|
|
|
default:
|
|
return (-1);
|
|
}
|
|
break;
|
|
|
|
case '\\':
|
|
tok->flags |= TOK_KEEP;
|
|
tok->flags &= ~TOK_EAT;
|
|
switch (tok->quote) {
|
|
case Q_none: /* Quote next character */
|
|
tok->quote = Q_one;
|
|
break;
|
|
|
|
case Q_double: /* Quote next character */
|
|
tok->quote = Q_doubleone;
|
|
break;
|
|
|
|
case Q_one: /* Quote this, restore state */
|
|
*tok->wptr++ = *ptr;
|
|
tok->quote = Q_none;
|
|
break;
|
|
|
|
case Q_single: /* Stay in single quote mode */
|
|
*tok->wptr++ = *ptr;
|
|
break;
|
|
|
|
case Q_doubleone: /* Quote this \ */
|
|
tok->quote = Q_double;
|
|
*tok->wptr++ = *ptr;
|
|
break;
|
|
|
|
default:
|
|
return (-1);
|
|
}
|
|
break;
|
|
|
|
case '\n':
|
|
tok->flags &= ~TOK_EAT;
|
|
switch (tok->quote) {
|
|
case Q_none:
|
|
goto tok_line_outok;
|
|
|
|
case Q_single:
|
|
case Q_double:
|
|
*tok->wptr++ = *ptr; /* Add the return */
|
|
break;
|
|
|
|
case Q_doubleone: /* Back to double, eat the '\n' */
|
|
tok->flags |= TOK_EAT;
|
|
tok->quote = Q_double;
|
|
break;
|
|
|
|
case Q_one: /* No quote, more eat the '\n' */
|
|
tok->flags |= TOK_EAT;
|
|
tok->quote = Q_none;
|
|
break;
|
|
|
|
default:
|
|
return (0);
|
|
}
|
|
break;
|
|
|
|
case '\0':
|
|
switch (tok->quote) {
|
|
case Q_none:
|
|
/* Finish word and return */
|
|
if (tok->flags & TOK_EAT) {
|
|
tok->flags &= ~TOK_EAT;
|
|
return (3);
|
|
}
|
|
goto tok_line_outok;
|
|
|
|
case Q_single:
|
|
return (1);
|
|
|
|
case Q_double:
|
|
return (2);
|
|
|
|
case Q_doubleone:
|
|
tok->quote = Q_double;
|
|
*tok->wptr++ = *ptr;
|
|
break;
|
|
|
|
case Q_one:
|
|
tok->quote = Q_none;
|
|
*tok->wptr++ = *ptr;
|
|
break;
|
|
|
|
default:
|
|
return (-1);
|
|
}
|
|
break;
|
|
|
|
default:
|
|
tok->flags &= ~TOK_EAT;
|
|
switch (tok->quote) {
|
|
case Q_none:
|
|
if (strchr(tok->ifs, *ptr) != NULL)
|
|
tok_finish(tok);
|
|
else
|
|
*tok->wptr++ = *ptr;
|
|
break;
|
|
|
|
case Q_single:
|
|
case Q_double:
|
|
*tok->wptr++ = *ptr;
|
|
break;
|
|
|
|
|
|
case Q_doubleone:
|
|
*tok->wptr++ = '\\';
|
|
tok->quote = Q_double;
|
|
*tok->wptr++ = *ptr;
|
|
break;
|
|
|
|
case Q_one:
|
|
tok->quote = Q_none;
|
|
*tok->wptr++ = *ptr;
|
|
break;
|
|
|
|
default:
|
|
return (-1);
|
|
|
|
}
|
|
break;
|
|
}
|
|
|
|
if (tok->wptr >= tok->wmax - 4) {
|
|
size_t size = tok->wmax - tok->wspace + WINCR;
|
|
char *s = (char *) tok_realloc(tok->wspace, size);
|
|
if (s == NULL)
|
|
return (-1);
|
|
|
|
if (s != tok->wspace) {
|
|
int i;
|
|
for (i = 0; i < tok->argc; i++) {
|
|
tok->argv[i] =
|
|
(tok->argv[i] - tok->wspace) + s;
|
|
}
|
|
tok->wptr = (tok->wptr - tok->wspace) + s;
|
|
tok->wstart = (tok->wstart - tok->wspace) + s;
|
|
tok->wspace = s;
|
|
}
|
|
tok->wmax = s + size;
|
|
}
|
|
if (tok->argc >= tok->amax - 4) {
|
|
char **p;
|
|
tok->amax += AINCR;
|
|
p = (char **) tok_realloc(tok->argv,
|
|
tok->amax * sizeof(char *));
|
|
if (p == NULL)
|
|
return (-1);
|
|
tok->argv = p;
|
|
}
|
|
}
|
|
tok_line_outok:
|
|
if (cc == -1 && co == -1) {
|
|
cc = tok->argc;
|
|
co = tok->wptr - tok->wstart;
|
|
}
|
|
if (cursorc != NULL)
|
|
*cursorc = cc;
|
|
if (cursoro != NULL)
|
|
*cursoro = co;
|
|
tok_finish(tok);
|
|
*argv = (const char **)tok->argv;
|
|
*argc = tok->argc;
|
|
return (0);
|
|
}
|
|
|
|
/* tok_str():
|
|
* Simpler version of tok_line, taking a NUL terminated line
|
|
* and splitting into words, ignoring cursor state.
|
|
*/
|
|
public int
|
|
tok_str(Tokenizer *tok, const char *line, int *argc, const char ***argv)
|
|
{
|
|
LineInfo li;
|
|
|
|
memset(&li, 0, sizeof(li));
|
|
li.buffer = line;
|
|
li.cursor = li.lastchar = strchr(line, '\0');
|
|
return (tok_line(tok, &li, argc, argv, NULL, NULL));
|
|
}
|