Home | History | Annotate | Line # | Download | only in lib
load_http.c revision 1.1 
      1 /*	$NetBSD: load_http.c,v 1.1 2012/03/23 21:20:09 christos Exp $	*/
      2 
      3 /*
      4  * Copyright (C) 2010 by Darren Reed.
      5  *
      6  * See the IPFILTER.LICENCE file for details on licencing.
      7  *
      8  * Id: load_http.c,v 1.5.2.3 2012/01/26 05:44:26 darren_r Exp
      9  */
     10 
     11 #include "ipf.h"
     12 #include <ctype.h>
     13 
     14 /*
     15  * Because the URL can be included twice into the buffer, once as the
     16  * full path for the "GET" and once as the "Host:", the buffer it is
     17  * put in needs to be larger than 512*2 to make room for the supporting
     18  * text. Why not just use snprintf and truncate? The warning about the
     19  * URL being too long tells you something is wrong and does not fetch
     20  * any data - just truncating the URL (with snprintf, etc) and sending
     21  * that to the server is allowing an unknown and unintentioned action
     22  * to happen.
     23  */
     24 #define	MAX_URL_LEN	512
     25 #define	LOAD_BUFSIZE	(MAX_URL_LEN * 2 + 128)
     26 
     27 /*
     28  * Format expected is one addres per line, at the start of each line.
     29  */
     30 alist_t *
     31 load_http(char *url)
     32 {
     33 	int fd, len, left, port, endhdr, removed, linenum = 0;
     34 	char *s, *t, *u, buffer[LOAD_BUFSIZE], *myurl;
     35 	alist_t *a, *rtop, *rbot;
     36 
     37 	/*
     38 	 * More than this would just be absurd.
     39 	 */
     40 	if (strlen(url) > MAX_URL_LEN) {
     41 		fprintf(stderr, "load_http has a URL > %d bytes?!\n",
     42 			MAX_URL_LEN);
     43 		return NULL;
     44 	}
     45 
     46 	fd = -1;
     47 	rtop = NULL;
     48 	rbot = NULL;
     49 
     50 	sprintf(buffer, "GET %s HTTP/1.0\r\n", url);
     51 
     52 	myurl = strdup(url);
     53 	if (myurl == NULL)
     54 		goto done;
     55 
     56 	s = myurl + 7;			/* http:// */
     57 	t = strchr(s, '/');
     58 	if (t == NULL) {
     59 		fprintf(stderr, "load_http has a malformed URL '%s'\n", url);
     60 		free(myurl);
     61 		return NULL;
     62 	}
     63 	*t++ = '\0';
     64 
     65 	/*
     66 	 * 10 is the length of 'Host: \r\n\r\n' below.
     67 	 */
     68 	if (strlen(s) + strlen(buffer) + 10 > sizeof(buffer)) {
     69 		fprintf(stderr, "load_http has a malformed URL '%s'\n", url);
     70 		free(myurl);
     71 		return NULL;
     72 	}
     73 
     74 	u = strchr(s, '@');
     75 	if (u != NULL)
     76 		s = u + 1;		/* AUTH */
     77 
     78 	sprintf(buffer + strlen(buffer), "Host: %s\r\n\r\n", s);
     79 
     80 	u = strchr(s, ':');
     81 	if (u != NULL) {
     82 		*u++ = '\0';
     83 		port = atoi(u);
     84 		if (port < 0 || port > 65535)
     85 			goto done;
     86 	} else {
     87 		port = 80;
     88 	}
     89 
     90 
     91 	fd = connecttcp(s, port);
     92 	if (fd == -1)
     93 		goto done;
     94 
     95 
     96 	len = strlen(buffer);
     97 	if (write(fd, buffer, len) != len) {
     98 		close(fd);
     99 		goto done;
    100 	}
    101 
    102 	s = buffer;
    103 	endhdr = 0;
    104 	left = sizeof(buffer) - 1;
    105 
    106 	while ((len = read(fd, s, left)) > 0) {
    107 		s[len] = '\0';
    108 		left -= len;
    109 		s += len;
    110 
    111 		if (endhdr >= 0) {
    112 			if (endhdr == 0) {
    113 				t = strchr(buffer, ' ');
    114 				if (t == NULL)
    115 					continue;
    116 				t++;
    117 				if (*t != '2')
    118 					break;
    119 			}
    120 
    121 			u = buffer;
    122 			while ((t = strchr(u, '\r')) != NULL) {
    123 				if (t == u) {
    124 					if (*(t + 1) == '\n') {
    125 						u = t + 2;
    126 						endhdr = -1;
    127 						break;
    128 					} else
    129 						t++;
    130 				} else if (*(t + 1) == '\n') {
    131 					endhdr++;
    132 					u = t + 2;
    133 				} else
    134 					u = t + 1;
    135 			}
    136 			if (endhdr >= 0)
    137 				continue;
    138 			removed = (u - buffer) + 1;
    139 			memmove(buffer, u, (sizeof(buffer) - left) - removed);
    140 			s -= removed;
    141 			left += removed;
    142 		}
    143 
    144 		do {
    145 			t = strchr(buffer, '\n');
    146 			if (t == NULL)
    147 				break;
    148 
    149 			linenum++;
    150 			*t = '\0';
    151 
    152 			/*
    153 			 * Remove comment and continue to the next line if
    154 			 * the comment is at the start of the line.
    155 			 */
    156 			u = strchr(buffer, '#');
    157 			if (u != NULL) {
    158 				*u = '\0';
    159 				if (u == buffer);
    160 					continue;
    161 			}
    162 
    163 			/*
    164 			 * Trim off tailing white spaces, will include \r
    165 			 */
    166 			for (u = t - 1; (u >= buffer) && ISSPACE(*u); u--)
    167 				*u = '\0';
    168 
    169 			a = alist_new(AF_UNSPEC, buffer);
    170 			if (a != NULL) {
    171 				if (rbot != NULL)
    172 					rbot->al_next = a;
    173 				else
    174 					rtop = a;
    175 				rbot = a;
    176 			} else {
    177 				fprintf(stderr,
    178 					"%s:%d unrecognised content:%s\n",
    179 					url, linenum, buffer);
    180 			}
    181 
    182 			t++;
    183 			removed = t - buffer;
    184 			memmove(buffer, t, sizeof(buffer) - left - removed);
    185 			s -= removed;
    186 			left += removed;
    187 
    188 		} while (1);
    189 	}
    190 
    191 done:
    192 	if (myurl != NULL)
    193 		free(myurl);
    194 	if (fd != -1)
    195 		close(fd);
    196 	return rtop;
    197 }
    198