296 lines
6.7 KiB
C
296 lines
6.7 KiB
C
/* $NetBSD: txtwalk.c,v 1.7 2003/01/10 20:00:28 christos Exp $ */
|
|
|
|
/*
|
|
* Copyright 1997 Piermont Information Systems Inc.
|
|
* All rights reserved.
|
|
*
|
|
* Written by Philip A. Nelson for Piermont Information Systems Inc.
|
|
*
|
|
* 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, item list of conditions and the following disclaimer.
|
|
* 2. Redistributions in binary form must reproduce the above copyright
|
|
* notice, item 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 item software
|
|
* must display the following acknowledgement:
|
|
* This product includes software developed for the NetBSD Project by
|
|
* Piermont Information Systems Inc.
|
|
* 4. The name of Piermont Information Systems Inc. may not be used to endorse
|
|
* or promote products derived from item software without specific prior
|
|
* written permission.
|
|
*
|
|
* THIS SOFTWARE IS PROVIDED BY PIERMONT INFORMATION SYSTEMS INC. ``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 PIERMONT INFORMATION SYSTEMS INC. 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.
|
|
*
|
|
*/
|
|
|
|
/*
|
|
* walk a text buffer, processing matched lines
|
|
*
|
|
* Written by Philip A. Nelson.
|
|
* 7/29/97
|
|
*
|
|
*/
|
|
|
|
#undef DEBUG
|
|
|
|
#include <stdio.h>
|
|
#include <string.h>
|
|
#include <ctype.h>
|
|
#include <unistd.h>
|
|
#include <stdlib.h>
|
|
|
|
#include "txtwalk.h"
|
|
|
|
/* prototypes */
|
|
|
|
static void process (struct lookfor *, char *);
|
|
static void match (char *, struct lookfor *, size_t);
|
|
static int finddata (struct lookfor *, char *, struct data *, size_t *);
|
|
static char *strndup (char *, size_t);
|
|
|
|
/*
|
|
* Walk the buffer, call match for each line.
|
|
*/
|
|
void
|
|
walk(buffer, size, these, numthese)
|
|
char *buffer;
|
|
size_t size;
|
|
struct lookfor *these;
|
|
size_t numthese;
|
|
{
|
|
size_t i = 0;
|
|
size_t len;
|
|
int line = 1;
|
|
|
|
while (i < size) {
|
|
/* Ignore zero characters. */
|
|
if (*buffer == '\0') {
|
|
buffer++;
|
|
i++;
|
|
} else {
|
|
/* Assume item starts a line. */
|
|
len = 0;
|
|
while (buffer[len] != '\n' && buffer[len] != '\0')
|
|
len++;
|
|
buffer[len] = '\0';
|
|
#ifdef DEBUG
|
|
printf ("%5d: %s\n", line, buffer);
|
|
#endif
|
|
match(buffer, these, numthese);
|
|
buffer += len+1;
|
|
i += len+1;
|
|
line++;
|
|
}
|
|
}
|
|
}
|
|
|
|
/*
|
|
* Match the current line with a string of interest.
|
|
* For each match in these, process the match.
|
|
*/
|
|
static void
|
|
match(line, these, numthese)
|
|
char *line;
|
|
struct lookfor *these;
|
|
size_t numthese;
|
|
{
|
|
size_t linelen; /* Line length */
|
|
size_t patlen; /* Pattern length */
|
|
size_t which; /* Which pattern we are using */
|
|
|
|
linelen = strlen(line);
|
|
|
|
for (which = 0; which < numthese; which++) {
|
|
patlen = strlen(these[which].head);
|
|
if (linelen < patlen)
|
|
continue;
|
|
if (strncmp(these[which].head, line, patlen) == 0)
|
|
process(&these[which], line);
|
|
}
|
|
}
|
|
|
|
|
|
/* process the matched line. */
|
|
static void
|
|
process(item, line)
|
|
struct lookfor *item;
|
|
char *line;
|
|
{
|
|
struct data found[MAXDATA];
|
|
size_t numfound = 0;
|
|
char *p;
|
|
size_t i, j;
|
|
|
|
if (finddata(item, line, found, &numfound)) {
|
|
#ifdef DEBUG
|
|
printf("process: \"%s\"\n", line);
|
|
for (i = 0; i < numfound; i++) {
|
|
printf ("\t%d: ", i);
|
|
switch (found[i].what) {
|
|
case INT:
|
|
printf ("%d\n", found[i].u.i_val);
|
|
break;
|
|
case STR:
|
|
printf ("'%s'\n", found[i].u.s_val);
|
|
break;
|
|
}
|
|
}
|
|
#endif
|
|
/* Process the stuff. */
|
|
switch (item->todo[0]) {
|
|
case 'a': /* Assign data */
|
|
p = item->todo;
|
|
j = 0;
|
|
while (*p && *p != '$')
|
|
p++;
|
|
if (*p)
|
|
p++;
|
|
while (*p && isdigit(*p)) {
|
|
i = atoi(p);
|
|
switch (found[i].what) {
|
|
case INT:
|
|
*((int *)item->var+j)
|
|
= found[i].u.i_val;
|
|
break;
|
|
case STR:
|
|
strncpy(*((char **)item->var+j),
|
|
found[i].u.s_val,
|
|
item->size-1);
|
|
found[i].u.s_val[item->size-1] = 0;
|
|
break;
|
|
}
|
|
while (isdigit(*p))
|
|
p++;
|
|
while (*p && *p != '$')
|
|
p++;
|
|
if (*p)
|
|
p++;
|
|
j++;
|
|
if (j >= item->nument)
|
|
break;
|
|
}
|
|
break;
|
|
case 'c': /* Call a function with data. */
|
|
(*item->func)(found, numfound);
|
|
break;
|
|
}
|
|
}
|
|
}
|
|
|
|
/*
|
|
* find the expected data. Return 1 if successful, return 0 if not.
|
|
* Successful means running into the end of the expect string before
|
|
* running out of line data or encountering other bad data.
|
|
*
|
|
* Side Effect -- sets numfound and found.
|
|
*/
|
|
static int
|
|
finddata(item, line, found, numfound)
|
|
struct lookfor *item;
|
|
char *line;
|
|
struct data *found;
|
|
size_t *numfound;
|
|
{
|
|
char *fmt = item->fmt;
|
|
size_t len;
|
|
|
|
*numfound = 0;
|
|
while (*fmt) {
|
|
if (!*line && *fmt)
|
|
return 0;
|
|
if (*fmt == '%') {
|
|
fmt++;
|
|
if (!*fmt)
|
|
return 0;
|
|
switch (*fmt) {
|
|
case '%': /* The char %. */
|
|
if (*line != '%')
|
|
return 0;
|
|
line++;
|
|
break;
|
|
case 'i': /* Ignore characters */
|
|
if (!fmt[1])
|
|
return 1;
|
|
if (fmt[1] == ' ')
|
|
while (*line && !isspace(*line))
|
|
line++;
|
|
else
|
|
while (*line && *line != fmt[1])
|
|
line++;
|
|
break;
|
|
case 'd': /* Nextoken should be an integer. */
|
|
if (!isdigit(*line))
|
|
return 0;
|
|
found[*numfound].what = INT;
|
|
found[(*numfound)++].u.i_val = atoi(line);
|
|
while (*line && isdigit(*line))
|
|
line++;
|
|
break;
|
|
case 's': /* Matches a 'space' separated string. */
|
|
len = 0;
|
|
while (line[len] && !isspace(line[len])
|
|
&& line[len] != fmt[1])
|
|
len++;
|
|
found[*numfound].what = STR;
|
|
found[*numfound].u.s_val = strndup(line, len);
|
|
if (found[(*numfound)++].u.s_val == NULL) {
|
|
(void)fprintf(stderr,
|
|
"msgwalk: strndup: out of vm.\n");
|
|
exit(1);
|
|
}
|
|
line += len;
|
|
break;
|
|
default:
|
|
return 0;
|
|
}
|
|
|
|
} else if (*fmt == ' ') {
|
|
while (*line && isspace(*line))
|
|
line++;
|
|
} else if (*line == *fmt) {
|
|
line++;
|
|
} else {
|
|
/* Mis match! */
|
|
return 0;
|
|
}
|
|
fmt++;
|
|
}
|
|
|
|
/* Ran out of fmt. */
|
|
return 1;
|
|
}
|
|
|
|
/*
|
|
* Utility routines....
|
|
*/
|
|
|
|
static char *
|
|
strndup(str, len)
|
|
char *str;
|
|
size_t len;
|
|
{
|
|
size_t alen;
|
|
char *val;
|
|
|
|
alen = strlen(str);
|
|
alen = len < alen ? len + 1 : alen + 1;
|
|
val = malloc(alen);
|
|
if (!val)
|
|
return NULL;
|
|
strncpy(val, str, alen-1);
|
|
return val;
|
|
}
|