Eric Andersen | 9670083 | 2000-09-04 15:15:55 +0000 | [diff] [blame] | 1 | /* vi: set sw=4 ts=4: */ |
| 2 | /* |
Eric Andersen | 79757c9 | 2001-04-05 21:45:54 +0000 | [diff] [blame] | 3 | * wget - retrieve a file using HTTP or FTP |
Eric Andersen | 9670083 | 2000-09-04 15:15:55 +0000 | [diff] [blame] | 4 | * |
Eric Andersen | 4e573f4 | 2000-11-14 23:29:24 +0000 | [diff] [blame] | 5 | * Chip Rosenthal Covad Communications <chip@laserlink.net> |
Eric Andersen | b520e08 | 2000-10-03 00:21:45 +0000 | [diff] [blame] | 6 | * |
Eric Andersen | 9670083 | 2000-09-04 15:15:55 +0000 | [diff] [blame] | 7 | */ |
| 8 | |
Eric Andersen | 9670083 | 2000-09-04 15:15:55 +0000 | [diff] [blame] | 9 | #include <stdio.h> |
Eric Andersen | dff9d54 | 2001-01-26 02:04:49 +0000 | [diff] [blame] | 10 | #include <errno.h> |
Eric Andersen | 9670083 | 2000-09-04 15:15:55 +0000 | [diff] [blame] | 11 | #include <stdlib.h> |
| 12 | #include <unistd.h> |
| 13 | #include <ctype.h> |
| 14 | #include <string.h> |
Eric Andersen | b520e08 | 2000-10-03 00:21:45 +0000 | [diff] [blame] | 15 | #include <unistd.h> |
| 16 | #include <signal.h> |
| 17 | #include <sys/ioctl.h> |
Eric Andersen | 9670083 | 2000-09-04 15:15:55 +0000 | [diff] [blame] | 18 | |
Eric Andersen | b520e08 | 2000-10-03 00:21:45 +0000 | [diff] [blame] | 19 | #include <sys/time.h> |
Eric Andersen | 9670083 | 2000-09-04 15:15:55 +0000 | [diff] [blame] | 20 | #include <sys/types.h> |
| 21 | #include <sys/stat.h> |
| 22 | #include <sys/socket.h> |
| 23 | #include <netinet/in.h> |
| 24 | #include <arpa/inet.h> |
| 25 | #include <netdb.h> |
| 26 | |
Eric Andersen | 50ae310 | 2001-05-15 17:51:37 +0000 | [diff] [blame] | 27 | #include <getopt.h> |
| 28 | |
Eric Andersen | cbe31da | 2001-02-20 06:14:08 +0000 | [diff] [blame] | 29 | #include "busybox.h" |
| 30 | |
Eric Andersen | 79757c9 | 2001-04-05 21:45:54 +0000 | [diff] [blame] | 31 | struct host_info { |
| 32 | char *host; |
| 33 | int port; |
| 34 | char *path; |
| 35 | int is_ftp; |
| 36 | char *user; |
| 37 | }; |
| 38 | |
| 39 | static void parse_url(char *url, struct host_info *h); |
Glenn L McGrath | ffccf6e | 2003-12-20 01:47:18 +0000 | [diff] [blame] | 40 | static FILE *open_socket(struct sockaddr_in *s_in); |
Eric Andersen | 3e6ff90 | 2001-03-09 21:24:12 +0000 | [diff] [blame] | 41 | static char *gethdr(char *buf, size_t bufsiz, FILE *fp, int *istrunc); |
Eric Andersen | 79757c9 | 2001-04-05 21:45:54 +0000 | [diff] [blame] | 42 | static int ftpcmd(char *s1, char *s2, FILE *fp, char *buf); |
Eric Andersen | 9670083 | 2000-09-04 15:15:55 +0000 | [diff] [blame] | 43 | |
Eric Andersen | b520e08 | 2000-10-03 00:21:45 +0000 | [diff] [blame] | 44 | /* Globals (can be accessed from signal handlers */ |
| 45 | static off_t filesize = 0; /* content-length of the file */ |
Eric Andersen | 6d7fa43 | 2001-04-10 18:17:05 +0000 | [diff] [blame] | 46 | static int chunked = 0; /* chunked transfer encoding */ |
Eric Andersen | bdfd0d7 | 2001-10-24 05:00:29 +0000 | [diff] [blame] | 47 | #ifdef CONFIG_FEATURE_WGET_STATUSBAR |
Eric Andersen | 20aab26 | 2001-07-19 22:28:02 +0000 | [diff] [blame] | 48 | static void progressmeter(int flag); |
Eric Andersen | b520e08 | 2000-10-03 00:21:45 +0000 | [diff] [blame] | 49 | static char *curfile; /* Name of current file being transferred. */ |
| 50 | static struct timeval start; /* Time a transfer started. */ |
Mark Whitley | 30ac01c | 2001-04-17 18:13:16 +0000 | [diff] [blame] | 51 | static volatile unsigned long statbytes = 0; /* Number of bytes transferred so far. */ |
Eric Andersen | b520e08 | 2000-10-03 00:21:45 +0000 | [diff] [blame] | 52 | /* For progressmeter() -- number of seconds before xfer considered "stalled" */ |
Mark Whitley | 59ab025 | 2001-01-23 22:30:04 +0000 | [diff] [blame] | 53 | static const int STALLTIME = 5; |
Eric Andersen | b520e08 | 2000-10-03 00:21:45 +0000 | [diff] [blame] | 54 | #endif |
Eric Andersen | c7bda1c | 2004-03-15 08:29:22 +0000 | [diff] [blame] | 55 | |
Eric Andersen | 3e6ff90 | 2001-03-09 21:24:12 +0000 | [diff] [blame] | 56 | static void close_and_delete_outfile(FILE* output, char *fname_out, int do_continue) |
Eric Andersen | 7d69701 | 2001-01-24 20:28:35 +0000 | [diff] [blame] | 57 | { |
| 58 | if (output != stdout && do_continue==0) { |
| 59 | fclose(output); |
| 60 | unlink(fname_out); |
| 61 | } |
| 62 | } |
Eric Andersen | 9670083 | 2000-09-04 15:15:55 +0000 | [diff] [blame] | 63 | |
Matt Kraai | 854125f | 2001-05-09 19:15:46 +0000 | [diff] [blame] | 64 | /* Read NMEMB elements of SIZE bytes into PTR from STREAM. Returns the |
| 65 | * number of elements read, and a short count if an eof or non-interrupt |
| 66 | * error is encountered. */ |
| 67 | static size_t safe_fread(void *ptr, size_t size, size_t nmemb, FILE *stream) |
| 68 | { |
| 69 | size_t ret = 0; |
| 70 | |
| 71 | do { |
| 72 | clearerr(stream); |
| 73 | ret += fread((char *)ptr + (ret * size), size, nmemb - ret, stream); |
| 74 | } while (ret < nmemb && ferror(stream) && errno == EINTR); |
| 75 | |
| 76 | return ret; |
| 77 | } |
| 78 | |
| 79 | /* Write NMEMB elements of SIZE bytes from PTR to STREAM. Returns the |
| 80 | * number of elements written, and a short count if an eof or non-interrupt |
| 81 | * error is encountered. */ |
| 82 | static size_t safe_fwrite(void *ptr, size_t size, size_t nmemb, FILE *stream) |
| 83 | { |
| 84 | size_t ret = 0; |
| 85 | |
| 86 | do { |
| 87 | clearerr(stream); |
| 88 | ret += fwrite((char *)ptr + (ret * size), size, nmemb - ret, stream); |
| 89 | } while (ret < nmemb && ferror(stream) && errno == EINTR); |
| 90 | |
| 91 | return ret; |
| 92 | } |
| 93 | |
| 94 | /* Read a line or SIZE - 1 bytes into S, whichever is less, from STREAM. |
| 95 | * Returns S, or NULL if an eof or non-interrupt error is encountered. */ |
| 96 | static char *safe_fgets(char *s, int size, FILE *stream) |
| 97 | { |
| 98 | char *ret; |
| 99 | |
| 100 | do { |
| 101 | clearerr(stream); |
| 102 | ret = fgets(s, size, stream); |
| 103 | } while (ret == NULL && ferror(stream) && errno == EINTR); |
| 104 | |
| 105 | return ret; |
| 106 | } |
| 107 | |
Eric Andersen | 79757c9 | 2001-04-05 21:45:54 +0000 | [diff] [blame] | 108 | #define close_delete_and_die(s...) { \ |
| 109 | close_and_delete_outfile(output, fname_out, do_continue); \ |
Manuel Novoa III | cad5364 | 2003-03-19 09:13:01 +0000 | [diff] [blame] | 110 | bb_error_msg_and_die(s); } |
Eric Andersen | 79757c9 | 2001-04-05 21:45:54 +0000 | [diff] [blame] | 111 | |
| 112 | |
Eric Andersen | bdfd0d7 | 2001-10-24 05:00:29 +0000 | [diff] [blame] | 113 | #ifdef CONFIG_FEATURE_WGET_AUTHENTICATION |
Eric Andersen | 79757c9 | 2001-04-05 21:45:54 +0000 | [diff] [blame] | 114 | /* |
| 115 | * Base64-encode character string |
| 116 | * oops... isn't something similar in uuencode.c? |
Eric Andersen | 14f5c8d | 2005-04-16 19:39:00 +0000 | [diff] [blame] | 117 | * XXX: It would be better to use already existing code |
Eric Andersen | 79757c9 | 2001-04-05 21:45:54 +0000 | [diff] [blame] | 118 | */ |
Eric Andersen | 14f5c8d | 2005-04-16 19:39:00 +0000 | [diff] [blame] | 119 | static char *base64enc(unsigned char *p, char *buf, int len) { |
Eric Andersen | 79757c9 | 2001-04-05 21:45:54 +0000 | [diff] [blame] | 120 | |
Tim Riker | c1ef7bd | 2006-01-25 00:08:53 +0000 | [diff] [blame] | 121 | char al[] = "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz" |
| 122 | "0123456789+/"; |
Eric Andersen | 79757c9 | 2001-04-05 21:45:54 +0000 | [diff] [blame] | 123 | char *s = buf; |
| 124 | |
Tim Riker | c1ef7bd | 2006-01-25 00:08:53 +0000 | [diff] [blame] | 125 | while(*p) { |
Eric Andersen | 79757c9 | 2001-04-05 21:45:54 +0000 | [diff] [blame] | 126 | if (s >= buf+len-4) |
Manuel Novoa III | cad5364 | 2003-03-19 09:13:01 +0000 | [diff] [blame] | 127 | bb_error_msg_and_die("buffer overflow"); |
Tim Riker | c1ef7bd | 2006-01-25 00:08:53 +0000 | [diff] [blame] | 128 | *(s++) = al[(*p >> 2) & 0x3F]; |
| 129 | *(s++) = al[((*p << 4) & 0x30) | ((*(p+1) >> 4) & 0x0F)]; |
| 130 | *s = *(s+1) = '='; |
| 131 | *(s+2) = 0; |
| 132 | if (! *(++p)) break; |
| 133 | *(s++) = al[((*p << 2) & 0x3C) | ((*(p+1) >> 6) & 0x03)]; |
| 134 | if (! *(++p)) break; |
| 135 | *(s++) = al[*(p++) & 0x3F]; |
| 136 | } |
Eric Andersen | 79757c9 | 2001-04-05 21:45:54 +0000 | [diff] [blame] | 137 | |
| 138 | return buf; |
| 139 | } |
| 140 | #endif |
| 141 | |
Glenn L McGrath | 514aeab | 2003-12-19 12:08:56 +0000 | [diff] [blame] | 142 | #define WGET_OPT_CONTINUE 1 |
| 143 | #define WGET_OPT_QUIET 2 |
| 144 | #define WGET_OPT_PASSIVE 4 |
| 145 | #define WGET_OPT_OUTNAME 8 |
| 146 | #define WGET_OPT_HEADER 16 |
| 147 | #define WGET_OPT_PREFIX 32 |
| 148 | #define WGET_OPT_PROXY 64 |
| 149 | |
| 150 | static const struct option wget_long_options[] = { |
| 151 | { "continue", 0, NULL, 'c' }, |
| 152 | { "quiet", 0, NULL, 'q' }, |
| 153 | { "passive-ftp", 0, NULL, 139 }, |
| 154 | { "output-document", 1, NULL, 'O' }, |
| 155 | { "header", 1, NULL, 131 }, |
| 156 | { "directory-prefix",1, NULL, 'P' }, |
| 157 | { "proxy", 1, NULL, 'Y' }, |
| 158 | { 0, 0, 0, 0 } |
| 159 | }; |
| 160 | |
Eric Andersen | 9670083 | 2000-09-04 15:15:55 +0000 | [diff] [blame] | 161 | int wget_main(int argc, char **argv) |
| 162 | { |
Eric Andersen | 79757c9 | 2001-04-05 21:45:54 +0000 | [diff] [blame] | 163 | int n, try=5, status; |
Glenn L McGrath | 514aeab | 2003-12-19 12:08:56 +0000 | [diff] [blame] | 164 | unsigned long opt; |
Eric Andersen | 79757c9 | 2001-04-05 21:45:54 +0000 | [diff] [blame] | 165 | int port; |
Robert Griebl | d776011 | 2002-05-14 23:36:45 +0000 | [diff] [blame] | 166 | char *proxy = 0; |
Eric Andersen | 8071c02 | 2001-06-21 19:45:06 +0000 | [diff] [blame] | 167 | char *dir_prefix=NULL; |
Eric Andersen | 29edd00 | 2000-12-09 16:55:35 +0000 | [diff] [blame] | 168 | char *s, buf[512]; |
| 169 | struct stat sbuf; |
Eric Andersen | 50ae310 | 2001-05-15 17:51:37 +0000 | [diff] [blame] | 170 | char extra_headers[1024]; |
| 171 | char *extra_headers_ptr = extra_headers; |
| 172 | int extra_headers_left = sizeof(extra_headers); |
Eric Andersen | 79757c9 | 2001-04-05 21:45:54 +0000 | [diff] [blame] | 173 | struct host_info server, target; |
Eric Andersen | e6dc439 | 2003-10-31 09:31:46 +0000 | [diff] [blame] | 174 | struct sockaddr_in s_in; |
Glenn L McGrath | 514aeab | 2003-12-19 12:08:56 +0000 | [diff] [blame] | 175 | llist_t *headers_llist = NULL; |
Eric Andersen | 79757c9 | 2001-04-05 21:45:54 +0000 | [diff] [blame] | 176 | |
"Vladimir N. Oleynik" | f704b27 | 2005-10-14 09:56:52 +0000 | [diff] [blame] | 177 | FILE *sfp = NULL; /* socket to web/ftp server */ |
| 178 | FILE *dfp = NULL; /* socket to ftp server (data) */ |
| 179 | char *fname_out = NULL; /* where to direct output (-O) */ |
| 180 | int do_continue = 0; /* continue a prev transfer (-c) */ |
| 181 | long beg_range = 0L; /* range at which continue begins */ |
| 182 | int got_clen = 0; /* got content-length: from server */ |
| 183 | FILE *output; /* socket to web server */ |
| 184 | int quiet_flag = FALSE; /* Be verry, verry quiet... */ |
| 185 | int use_proxy = 1; /* Use proxies if env vars are set */ |
Glenn L McGrath | 514aeab | 2003-12-19 12:08:56 +0000 | [diff] [blame] | 186 | char *proxy_flag = "on"; /* Use proxies if env vars are set */ |
Eric Andersen | 29edd00 | 2000-12-09 16:55:35 +0000 | [diff] [blame] | 187 | |
Eric Andersen | 9670083 | 2000-09-04 15:15:55 +0000 | [diff] [blame] | 188 | /* |
| 189 | * Crack command line. |
| 190 | */ |
"Vladimir N. Oleynik" | f704b27 | 2005-10-14 09:56:52 +0000 | [diff] [blame] | 191 | bb_opt_complementally = "-1:\203::"; |
Glenn L McGrath | 514aeab | 2003-12-19 12:08:56 +0000 | [diff] [blame] | 192 | bb_applet_long_options = wget_long_options; |
Tim Riker | c1ef7bd | 2006-01-25 00:08:53 +0000 | [diff] [blame] | 193 | opt = bb_getopt_ulflags(argc, argv, "cq\213O:\203:P:Y:", |
| 194 | &fname_out, &headers_llist, |
"Vladimir N. Oleynik" | f704b27 | 2005-10-14 09:56:52 +0000 | [diff] [blame] | 195 | &dir_prefix, &proxy_flag); |
Glenn L McGrath | 514aeab | 2003-12-19 12:08:56 +0000 | [diff] [blame] | 196 | if (opt & WGET_OPT_CONTINUE) { |
| 197 | ++do_continue; |
| 198 | } |
| 199 | if (opt & WGET_OPT_QUIET) { |
| 200 | quiet_flag = TRUE; |
| 201 | } |
Glenn L McGrath | f1c4b11 | 2004-02-22 00:27:34 +0000 | [diff] [blame] | 202 | if (strcmp(proxy_flag, "off") == 0) { |
Glenn L McGrath | 514aeab | 2003-12-19 12:08:56 +0000 | [diff] [blame] | 203 | /* Use the proxy if necessary. */ |
Glenn L McGrath | f1c4b11 | 2004-02-22 00:27:34 +0000 | [diff] [blame] | 204 | use_proxy = 0; |
Glenn L McGrath | 514aeab | 2003-12-19 12:08:56 +0000 | [diff] [blame] | 205 | } |
| 206 | if (opt & WGET_OPT_HEADER) { |
| 207 | while (headers_llist) { |
| 208 | int arglen = strlen(headers_llist->data); |
| 209 | if (extra_headers_left - arglen - 2 <= 0) |
| 210 | bb_error_msg_and_die("extra_headers buffer too small(need %i)", extra_headers_left - arglen); |
| 211 | strcpy(extra_headers_ptr, headers_llist->data); |
| 212 | extra_headers_ptr += arglen; |
| 213 | extra_headers_left -= ( arglen + 2 ); |
| 214 | *extra_headers_ptr++ = '\r'; |
| 215 | *extra_headers_ptr++ = '\n'; |
| 216 | *(extra_headers_ptr + 1) = 0; |
| 217 | headers_llist = headers_llist->link; |
Eric Andersen | 9670083 | 2000-09-04 15:15:55 +0000 | [diff] [blame] | 218 | } |
| 219 | } |
Tim Riker | c1ef7bd | 2006-01-25 00:08:53 +0000 | [diff] [blame] | 220 | |
Eric Andersen | 79757c9 | 2001-04-05 21:45:54 +0000 | [diff] [blame] | 221 | parse_url(argv[optind], &target); |
| 222 | server.host = target.host; |
| 223 | server.port = target.port; |
| 224 | |
Eric Andersen | 9670083 | 2000-09-04 15:15:55 +0000 | [diff] [blame] | 225 | /* |
Eric Andersen | f3b2b52 | 2000-12-07 22:42:11 +0000 | [diff] [blame] | 226 | * Use the proxy if necessary. |
Eric Andersen | 9670083 | 2000-09-04 15:15:55 +0000 | [diff] [blame] | 227 | */ |
Glenn L McGrath | f1c4b11 | 2004-02-22 00:27:34 +0000 | [diff] [blame] | 228 | if (use_proxy) { |
Robert Griebl | d776011 | 2002-05-14 23:36:45 +0000 | [diff] [blame] | 229 | proxy = getenv(target.is_ftp ? "ftp_proxy" : "http_proxy"); |
Glenn L McGrath | f1c4b11 | 2004-02-22 00:27:34 +0000 | [diff] [blame] | 230 | if (proxy && *proxy) { |
Manuel Novoa III | cad5364 | 2003-03-19 09:13:01 +0000 | [diff] [blame] | 231 | parse_url(bb_xstrdup(proxy), &server); |
Glenn L McGrath | f1c4b11 | 2004-02-22 00:27:34 +0000 | [diff] [blame] | 232 | } else { |
| 233 | use_proxy = 0; |
| 234 | } |
Robert Griebl | d776011 | 2002-05-14 23:36:45 +0000 | [diff] [blame] | 235 | } |
Eric Andersen | c7bda1c | 2004-03-15 08:29:22 +0000 | [diff] [blame] | 236 | |
Eric Andersen | 29edd00 | 2000-12-09 16:55:35 +0000 | [diff] [blame] | 237 | /* Guess an output filename */ |
| 238 | if (!fname_out) { |
Eric Andersen | 751750e | 2004-10-08 08:27:40 +0000 | [diff] [blame] | 239 | // Dirty hack. Needed because bb_get_last_path_component |
| 240 | // will destroy trailing / by storing '\0' in last byte! |
| 241 | if(target.path[strlen(target.path)-1]!='/') { |
| 242 | fname_out = |
Eric Andersen | bdfd0d7 | 2001-10-24 05:00:29 +0000 | [diff] [blame] | 243 | #ifdef CONFIG_FEATURE_WGET_STATUSBAR |
Eric Andersen | 751750e | 2004-10-08 08:27:40 +0000 | [diff] [blame] | 244 | curfile = |
Eric Andersen | 29edd00 | 2000-12-09 16:55:35 +0000 | [diff] [blame] | 245 | #endif |
Eric Andersen | 751750e | 2004-10-08 08:27:40 +0000 | [diff] [blame] | 246 | bb_get_last_path_component(target.path); |
| 247 | } |
Eric Andersen | 29edd00 | 2000-12-09 16:55:35 +0000 | [diff] [blame] | 248 | if (fname_out==NULL || strlen(fname_out)<1) { |
Eric Andersen | c7bda1c | 2004-03-15 08:29:22 +0000 | [diff] [blame] | 249 | fname_out = |
Eric Andersen | bdfd0d7 | 2001-10-24 05:00:29 +0000 | [diff] [blame] | 250 | #ifdef CONFIG_FEATURE_WGET_STATUSBAR |
Eric Andersen | c7bda1c | 2004-03-15 08:29:22 +0000 | [diff] [blame] | 251 | curfile = |
Eric Andersen | 29edd00 | 2000-12-09 16:55:35 +0000 | [diff] [blame] | 252 | #endif |
| 253 | "index.html"; |
| 254 | } |
Matt Kraai | 0382eb8 | 2001-07-19 19:13:55 +0000 | [diff] [blame] | 255 | if (dir_prefix != NULL) |
| 256 | fname_out = concat_path_file(dir_prefix, fname_out); |
Eric Andersen | bdfd0d7 | 2001-10-24 05:00:29 +0000 | [diff] [blame] | 257 | #ifdef CONFIG_FEATURE_WGET_STATUSBAR |
Eric Andersen | 29edd00 | 2000-12-09 16:55:35 +0000 | [diff] [blame] | 258 | } else { |
Manuel Novoa III | cad5364 | 2003-03-19 09:13:01 +0000 | [diff] [blame] | 259 | curfile = bb_get_last_path_component(fname_out); |
Eric Andersen | 29edd00 | 2000-12-09 16:55:35 +0000 | [diff] [blame] | 260 | #endif |
| 261 | } |
Eric Andersen | 7d69701 | 2001-01-24 20:28:35 +0000 | [diff] [blame] | 262 | if (do_continue && !fname_out) |
Manuel Novoa III | cad5364 | 2003-03-19 09:13:01 +0000 | [diff] [blame] | 263 | bb_error_msg_and_die("cannot specify continue (-c) without a filename (-O)"); |
Eric Andersen | 29edd00 | 2000-12-09 16:55:35 +0000 | [diff] [blame] | 264 | |
Eric Andersen | 9670083 | 2000-09-04 15:15:55 +0000 | [diff] [blame] | 265 | |
| 266 | /* |
Eric Andersen | 29edd00 | 2000-12-09 16:55:35 +0000 | [diff] [blame] | 267 | * Open the output file stream. |
Eric Andersen | 9670083 | 2000-09-04 15:15:55 +0000 | [diff] [blame] | 268 | */ |
Matt Kraai | 65317ea | 2001-04-11 20:03:01 +0000 | [diff] [blame] | 269 | if (strcmp(fname_out, "-") == 0) { |
Randolph Chung | 02553a2 | 2000-12-07 03:53:47 +0000 | [diff] [blame] | 270 | output = stdout; |
Eric Andersen | 95a349f | 2001-05-13 00:55:54 +0000 | [diff] [blame] | 271 | quiet_flag = TRUE; |
Matt Kraai | 65317ea | 2001-04-11 20:03:01 +0000 | [diff] [blame] | 272 | } else { |
Manuel Novoa III | cad5364 | 2003-03-19 09:13:01 +0000 | [diff] [blame] | 273 | output = bb_xfopen(fname_out, (do_continue ? "a" : "w")); |
Eric Andersen | 9670083 | 2000-09-04 15:15:55 +0000 | [diff] [blame] | 274 | } |
| 275 | |
| 276 | /* |
| 277 | * Determine where to start transfer. |
| 278 | */ |
| 279 | if (do_continue) { |
Eric Andersen | b520e08 | 2000-10-03 00:21:45 +0000 | [diff] [blame] | 280 | if (fstat(fileno(output), &sbuf) < 0) |
Manuel Novoa III | cad5364 | 2003-03-19 09:13:01 +0000 | [diff] [blame] | 281 | bb_perror_msg_and_die("fstat()"); |
Eric Andersen | 9670083 | 2000-09-04 15:15:55 +0000 | [diff] [blame] | 282 | if (sbuf.st_size > 0) |
| 283 | beg_range = sbuf.st_size; |
| 284 | else |
| 285 | do_continue = 0; |
| 286 | } |
| 287 | |
Eric Andersen | e6dc439 | 2003-10-31 09:31:46 +0000 | [diff] [blame] | 288 | /* We want to do exactly _one_ DNS lookup, since some |
| 289 | * sites (i.e. ftp.us.debian.org) use round-robin DNS |
| 290 | * and we want to connect to only one IP... */ |
Glenn L McGrath | ffccf6e | 2003-12-20 01:47:18 +0000 | [diff] [blame] | 291 | bb_lookup_host(&s_in, server.host); |
| 292 | s_in.sin_port = server.port; |
Eric Andersen | e6dc439 | 2003-10-31 09:31:46 +0000 | [diff] [blame] | 293 | if (quiet_flag==FALSE) { |
| 294 | fprintf(stdout, "Connecting to %s[%s]:%d\n", |
Glenn L McGrath | ffccf6e | 2003-12-20 01:47:18 +0000 | [diff] [blame] | 295 | server.host, inet_ntoa(s_in.sin_addr), ntohs(server.port)); |
Eric Andersen | e6dc439 | 2003-10-31 09:31:46 +0000 | [diff] [blame] | 296 | } |
| 297 | |
Glenn L McGrath | f1c4b11 | 2004-02-22 00:27:34 +0000 | [diff] [blame] | 298 | if (use_proxy || !target.is_ftp) { |
Eric Andersen | 79757c9 | 2001-04-05 21:45:54 +0000 | [diff] [blame] | 299 | /* |
| 300 | * HTTP session |
| 301 | */ |
| 302 | do { |
Glenn L McGrath | e7bdfcc | 2003-08-28 22:03:19 +0000 | [diff] [blame] | 303 | got_clen = chunked = 0; |
| 304 | |
Eric Andersen | 79757c9 | 2001-04-05 21:45:54 +0000 | [diff] [blame] | 305 | if (! --try) |
| 306 | close_delete_and_die("too many redirections"); |
Eric Andersen | 7d69701 | 2001-01-24 20:28:35 +0000 | [diff] [blame] | 307 | |
Eric Andersen | 79757c9 | 2001-04-05 21:45:54 +0000 | [diff] [blame] | 308 | /* |
| 309 | * Open socket to http server |
| 310 | */ |
| 311 | if (sfp) fclose(sfp); |
Glenn L McGrath | ffccf6e | 2003-12-20 01:47:18 +0000 | [diff] [blame] | 312 | sfp = open_socket(&s_in); |
Eric Andersen | c7bda1c | 2004-03-15 08:29:22 +0000 | [diff] [blame] | 313 | |
Eric Andersen | 79757c9 | 2001-04-05 21:45:54 +0000 | [diff] [blame] | 314 | /* |
| 315 | * Send HTTP request. |
| 316 | */ |
Glenn L McGrath | f1c4b11 | 2004-02-22 00:27:34 +0000 | [diff] [blame] | 317 | if (use_proxy) { |
Glenn L McGrath | cc20ebc | 2003-09-10 23:52:15 +0000 | [diff] [blame] | 318 | const char *format = "GET %stp://%s:%d/%s HTTP/1.1\r\n"; |
| 319 | #ifdef CONFIG_FEATURE_WGET_IP6_LITERAL |
| 320 | if (strchr (target.host, ':')) |
| 321 | format = "GET %stp://[%s]:%d/%s HTTP/1.1\r\n"; |
| 322 | #endif |
| 323 | fprintf(sfp, format, |
Eric Andersen | 79757c9 | 2001-04-05 21:45:54 +0000 | [diff] [blame] | 324 | target.is_ftp ? "f" : "ht", target.host, |
Glenn L McGrath | 24cb17f | 2004-01-31 08:08:57 +0000 | [diff] [blame] | 325 | ntohs(target.port), target.path); |
Eric Andersen | 79757c9 | 2001-04-05 21:45:54 +0000 | [diff] [blame] | 326 | } else { |
Eric Andersen | 6d7fa43 | 2001-04-10 18:17:05 +0000 | [diff] [blame] | 327 | fprintf(sfp, "GET /%s HTTP/1.1\r\n", target.path); |
Eric Andersen | 79757c9 | 2001-04-05 21:45:54 +0000 | [diff] [blame] | 328 | } |
Eric Andersen | 9670083 | 2000-09-04 15:15:55 +0000 | [diff] [blame] | 329 | |
Eric Andersen | 79757c9 | 2001-04-05 21:45:54 +0000 | [diff] [blame] | 330 | fprintf(sfp, "Host: %s\r\nUser-Agent: Wget\r\n", target.host); |
| 331 | |
Eric Andersen | bdfd0d7 | 2001-10-24 05:00:29 +0000 | [diff] [blame] | 332 | #ifdef CONFIG_FEATURE_WGET_AUTHENTICATION |
Eric Andersen | 79757c9 | 2001-04-05 21:45:54 +0000 | [diff] [blame] | 333 | if (target.user) { |
| 334 | fprintf(sfp, "Authorization: Basic %s\r\n", |
Eric Andersen | 0cb6f35 | 2006-01-30 22:30:41 +0000 | [diff] [blame^] | 335 | base64enc((unsigned char*)target.user, buf, sizeof(buf))); |
Eric Andersen | 79757c9 | 2001-04-05 21:45:54 +0000 | [diff] [blame] | 336 | } |
Glenn L McGrath | f1c4b11 | 2004-02-22 00:27:34 +0000 | [diff] [blame] | 337 | if (use_proxy && server.user) { |
Eric Andersen | 79757c9 | 2001-04-05 21:45:54 +0000 | [diff] [blame] | 338 | fprintf(sfp, "Proxy-Authorization: Basic %s\r\n", |
Eric Andersen | 0cb6f35 | 2006-01-30 22:30:41 +0000 | [diff] [blame^] | 339 | base64enc((unsigned char*)server.user, buf, sizeof(buf))); |
Eric Andersen | 79757c9 | 2001-04-05 21:45:54 +0000 | [diff] [blame] | 340 | } |
| 341 | #endif |
| 342 | |
Eric Andersen | b520e08 | 2000-10-03 00:21:45 +0000 | [diff] [blame] | 343 | if (do_continue) |
Eric Andersen | 79757c9 | 2001-04-05 21:45:54 +0000 | [diff] [blame] | 344 | fprintf(sfp, "Range: bytes=%ld-\r\n", beg_range); |
Eric Andersen | 50ae310 | 2001-05-15 17:51:37 +0000 | [diff] [blame] | 345 | if(extra_headers_left < sizeof(extra_headers)) |
Eric Andersen | 9abfe85 | 2001-05-15 20:11:49 +0000 | [diff] [blame] | 346 | fputs(extra_headers,sfp); |
Eric Andersen | 79757c9 | 2001-04-05 21:45:54 +0000 | [diff] [blame] | 347 | fprintf(sfp,"Connection: close\r\n\r\n"); |
| 348 | |
| 349 | /* |
Tim Riker | c1ef7bd | 2006-01-25 00:08:53 +0000 | [diff] [blame] | 350 | * Retrieve HTTP response line and check for "200" status code. |
| 351 | */ |
Glenn L McGrath | e7bdfcc | 2003-08-28 22:03:19 +0000 | [diff] [blame] | 352 | read_response: |
| 353 | if (fgets(buf, sizeof(buf), sfp) == NULL) |
Eric Andersen | 79757c9 | 2001-04-05 21:45:54 +0000 | [diff] [blame] | 354 | close_delete_and_die("no response from server"); |
Eric Andersen | c7bda1c | 2004-03-15 08:29:22 +0000 | [diff] [blame] | 355 | |
Eric Andersen | 79757c9 | 2001-04-05 21:45:54 +0000 | [diff] [blame] | 356 | for (s = buf ; *s != '\0' && !isspace(*s) ; ++s) |
| 357 | ; |
| 358 | for ( ; isspace(*s) ; ++s) |
| 359 | ; |
| 360 | switch (status = atoi(s)) { |
| 361 | case 0: |
Eric Andersen | 6d7fa43 | 2001-04-10 18:17:05 +0000 | [diff] [blame] | 362 | case 100: |
| 363 | while (gethdr(buf, sizeof(buf), sfp, &n) != NULL); |
| 364 | goto read_response; |
Eric Andersen | 79757c9 | 2001-04-05 21:45:54 +0000 | [diff] [blame] | 365 | case 200: |
| 366 | if (do_continue && output != stdout) |
| 367 | output = freopen(fname_out, "w", output); |
| 368 | do_continue = 0; |
| 369 | break; |
| 370 | case 300: /* redirection */ |
| 371 | case 301: |
| 372 | case 302: |
| 373 | case 303: |
| 374 | break; |
| 375 | case 206: |
| 376 | if (do_continue) |
| 377 | break; |
| 378 | /*FALLTHRU*/ |
| 379 | default: |
| 380 | chomp(buf); |
| 381 | close_delete_and_die("server returned error %d: %s", atoi(s), buf); |
| 382 | } |
Eric Andersen | c7bda1c | 2004-03-15 08:29:22 +0000 | [diff] [blame] | 383 | |
Eric Andersen | 79757c9 | 2001-04-05 21:45:54 +0000 | [diff] [blame] | 384 | /* |
| 385 | * Retrieve HTTP headers. |
| 386 | */ |
| 387 | while ((s = gethdr(buf, sizeof(buf), sfp, &n)) != NULL) { |
| 388 | if (strcasecmp(buf, "content-length") == 0) { |
Eric Andersen | 2479445 | 2004-03-06 22:11:45 +0000 | [diff] [blame] | 389 | unsigned long value; |
| 390 | if (safe_strtoul(s, &value)) { |
| 391 | close_delete_and_die("content-length %s is garbage", s); |
| 392 | } |
| 393 | filesize = value; |
Eric Andersen | 79757c9 | 2001-04-05 21:45:54 +0000 | [diff] [blame] | 394 | got_clen = 1; |
| 395 | continue; |
| 396 | } |
Eric Andersen | 6d7fa43 | 2001-04-10 18:17:05 +0000 | [diff] [blame] | 397 | if (strcasecmp(buf, "transfer-encoding") == 0) { |
| 398 | if (strcasecmp(s, "chunked") == 0) { |
| 399 | chunked = got_clen = 1; |
| 400 | } else { |
Eric Andersen | 79757c9 | 2001-04-05 21:45:54 +0000 | [diff] [blame] | 401 | close_delete_and_die("server wants to do %s transfer encoding", s); |
Eric Andersen | 6d7fa43 | 2001-04-10 18:17:05 +0000 | [diff] [blame] | 402 | } |
| 403 | } |
Eric Andersen | 79757c9 | 2001-04-05 21:45:54 +0000 | [diff] [blame] | 404 | if (strcasecmp(buf, "location") == 0) { |
| 405 | if (s[0] == '/') |
Manuel Novoa III | cad5364 | 2003-03-19 09:13:01 +0000 | [diff] [blame] | 406 | target.path = bb_xstrdup(s+1); |
Eric Andersen | 79757c9 | 2001-04-05 21:45:54 +0000 | [diff] [blame] | 407 | else { |
Manuel Novoa III | cad5364 | 2003-03-19 09:13:01 +0000 | [diff] [blame] | 408 | parse_url(bb_xstrdup(s), &target); |
Glenn L McGrath | f1c4b11 | 2004-02-22 00:27:34 +0000 | [diff] [blame] | 409 | if (use_proxy == 0) { |
Eric Andersen | 79757c9 | 2001-04-05 21:45:54 +0000 | [diff] [blame] | 410 | server.host = target.host; |
| 411 | server.port = target.port; |
| 412 | } |
Glenn L McGrath | 58a2e0e | 2004-01-17 23:07:14 +0000 | [diff] [blame] | 413 | bb_lookup_host(&s_in, server.host); |
| 414 | s_in.sin_port = server.port; |
| 415 | break; |
Eric Andersen | 79757c9 | 2001-04-05 21:45:54 +0000 | [diff] [blame] | 416 | } |
| 417 | } |
| 418 | } |
| 419 | } while(status >= 300); |
Eric Andersen | c7bda1c | 2004-03-15 08:29:22 +0000 | [diff] [blame] | 420 | |
Eric Andersen | 79757c9 | 2001-04-05 21:45:54 +0000 | [diff] [blame] | 421 | dfp = sfp; |
| 422 | } |
| 423 | else |
| 424 | { |
| 425 | /* |
| 426 | * FTP session |
| 427 | */ |
| 428 | if (! target.user) |
Manuel Novoa III | cad5364 | 2003-03-19 09:13:01 +0000 | [diff] [blame] | 429 | target.user = bb_xstrdup("anonymous:busybox@"); |
Eric Andersen | 79757c9 | 2001-04-05 21:45:54 +0000 | [diff] [blame] | 430 | |
Glenn L McGrath | ffccf6e | 2003-12-20 01:47:18 +0000 | [diff] [blame] | 431 | sfp = open_socket(&s_in); |
Eric Andersen | 79757c9 | 2001-04-05 21:45:54 +0000 | [diff] [blame] | 432 | if (ftpcmd(NULL, NULL, sfp, buf) != 220) |
| 433 | close_delete_and_die("%s", buf+4); |
| 434 | |
Eric Andersen | c7bda1c | 2004-03-15 08:29:22 +0000 | [diff] [blame] | 435 | /* |
Eric Andersen | 79757c9 | 2001-04-05 21:45:54 +0000 | [diff] [blame] | 436 | * Splitting username:password pair, |
| 437 | * trying to log in |
| 438 | */ |
| 439 | s = strchr(target.user, ':'); |
| 440 | if (s) |
| 441 | *(s++) = '\0'; |
| 442 | switch(ftpcmd("USER ", target.user, sfp, buf)) { |
| 443 | case 230: |
Eric Andersen | b520e08 | 2000-10-03 00:21:45 +0000 | [diff] [blame] | 444 | break; |
Eric Andersen | 79757c9 | 2001-04-05 21:45:54 +0000 | [diff] [blame] | 445 | case 331: |
| 446 | if (ftpcmd("PASS ", s, sfp, buf) == 230) |
| 447 | break; |
| 448 | /* FALLTHRU (failed login) */ |
| 449 | default: |
| 450 | close_delete_and_die("ftp login: %s", buf+4); |
| 451 | } |
Eric Andersen | c7bda1c | 2004-03-15 08:29:22 +0000 | [diff] [blame] | 452 | |
Eric Andersen | 79757c9 | 2001-04-05 21:45:54 +0000 | [diff] [blame] | 453 | ftpcmd("CDUP", NULL, sfp, buf); |
| 454 | ftpcmd("TYPE I", NULL, sfp, buf); |
Eric Andersen | c7bda1c | 2004-03-15 08:29:22 +0000 | [diff] [blame] | 455 | |
Eric Andersen | 79757c9 | 2001-04-05 21:45:54 +0000 | [diff] [blame] | 456 | /* |
| 457 | * Querying file size |
| 458 | */ |
| 459 | if (ftpcmd("SIZE /", target.path, sfp, buf) == 213) { |
Eric Andersen | 2479445 | 2004-03-06 22:11:45 +0000 | [diff] [blame] | 460 | unsigned long value; |
| 461 | if (safe_strtoul(buf+4, &value)) { |
| 462 | close_delete_and_die("SIZE value is garbage"); |
| 463 | } |
| 464 | filesize = value; |
Eric Andersen | 9670083 | 2000-09-04 15:15:55 +0000 | [diff] [blame] | 465 | got_clen = 1; |
Eric Andersen | 9670083 | 2000-09-04 15:15:55 +0000 | [diff] [blame] | 466 | } |
Eric Andersen | c7bda1c | 2004-03-15 08:29:22 +0000 | [diff] [blame] | 467 | |
Eric Andersen | 79757c9 | 2001-04-05 21:45:54 +0000 | [diff] [blame] | 468 | /* |
| 469 | * Entering passive mode |
| 470 | */ |
| 471 | if (ftpcmd("PASV", NULL, sfp, buf) != 227) |
| 472 | close_delete_and_die("PASV: %s", buf+4); |
| 473 | s = strrchr(buf, ','); |
| 474 | *s = 0; |
| 475 | port = atoi(s+1); |
| 476 | s = strrchr(buf, ','); |
| 477 | port += atoi(s+1) * 256; |
Glenn L McGrath | ffccf6e | 2003-12-20 01:47:18 +0000 | [diff] [blame] | 478 | s_in.sin_port = htons(port); |
| 479 | dfp = open_socket(&s_in); |
Eric Andersen | 79757c9 | 2001-04-05 21:45:54 +0000 | [diff] [blame] | 480 | |
| 481 | if (do_continue) { |
| 482 | sprintf(buf, "REST %ld", beg_range); |
| 483 | if (ftpcmd(buf, NULL, sfp, buf) != 350) { |
| 484 | if (output != stdout) |
| 485 | output = freopen(fname_out, "w", output); |
| 486 | do_continue = 0; |
| 487 | } else |
| 488 | filesize -= beg_range; |
Eric Andersen | 9670083 | 2000-09-04 15:15:55 +0000 | [diff] [blame] | 489 | } |
Eric Andersen | c7bda1c | 2004-03-15 08:29:22 +0000 | [diff] [blame] | 490 | |
Eric Andersen | 79757c9 | 2001-04-05 21:45:54 +0000 | [diff] [blame] | 491 | if (ftpcmd("RETR /", target.path, sfp, buf) > 150) |
| 492 | close_delete_and_die("RETR: %s", buf+4); |
| 493 | |
Eric Andersen | 9670083 | 2000-09-04 15:15:55 +0000 | [diff] [blame] | 494 | } |
| 495 | |
Eric Andersen | 79757c9 | 2001-04-05 21:45:54 +0000 | [diff] [blame] | 496 | |
Eric Andersen | 9670083 | 2000-09-04 15:15:55 +0000 | [diff] [blame] | 497 | /* |
Eric Andersen | 79757c9 | 2001-04-05 21:45:54 +0000 | [diff] [blame] | 498 | * Retrieve file |
Eric Andersen | 9670083 | 2000-09-04 15:15:55 +0000 | [diff] [blame] | 499 | */ |
Eric Andersen | 6d7fa43 | 2001-04-10 18:17:05 +0000 | [diff] [blame] | 500 | if (chunked) { |
| 501 | fgets(buf, sizeof(buf), dfp); |
| 502 | filesize = strtol(buf, (char **) NULL, 16); |
| 503 | } |
Eric Andersen | bdfd0d7 | 2001-10-24 05:00:29 +0000 | [diff] [blame] | 504 | #ifdef CONFIG_FEATURE_WGET_STATUSBAR |
Glenn L McGrath | 1bca5ed | 2000-12-09 08:12:06 +0000 | [diff] [blame] | 505 | if (quiet_flag==FALSE) |
| 506 | progressmeter(-1); |
Eric Andersen | b520e08 | 2000-10-03 00:21:45 +0000 | [diff] [blame] | 507 | #endif |
Mark Whitley | 30ac01c | 2001-04-17 18:13:16 +0000 | [diff] [blame] | 508 | do { |
Glenn L McGrath | 2336597 | 2003-08-29 06:25:04 +0000 | [diff] [blame] | 509 | while ((filesize > 0 || !got_clen) && (n = safe_fread(buf, 1, ((chunked || got_clen) && (filesize < sizeof(buf)) ? filesize : sizeof(buf)), dfp)) > 0) { |
Glenn L McGrath | 83e4a5b | 2003-08-28 21:55:22 +0000 | [diff] [blame] | 510 | if (safe_fwrite(buf, 1, n, output) != n) { |
Manuel Novoa III | cad5364 | 2003-03-19 09:13:01 +0000 | [diff] [blame] | 511 | bb_perror_msg_and_die("write error"); |
Glenn L McGrath | 83e4a5b | 2003-08-28 21:55:22 +0000 | [diff] [blame] | 512 | } |
Eric Andersen | bdfd0d7 | 2001-10-24 05:00:29 +0000 | [diff] [blame] | 513 | #ifdef CONFIG_FEATURE_WGET_STATUSBAR |
Glenn L McGrath | 83e4a5b | 2003-08-28 21:55:22 +0000 | [diff] [blame] | 514 | statbytes+=n; |
Eric Andersen | b520e08 | 2000-10-03 00:21:45 +0000 | [diff] [blame] | 515 | #endif |
Glenn L McGrath | 83e4a5b | 2003-08-28 21:55:22 +0000 | [diff] [blame] | 516 | if (got_clen) { |
| 517 | filesize -= n; |
| 518 | } |
| 519 | } |
Eric Andersen | 79757c9 | 2001-04-05 21:45:54 +0000 | [diff] [blame] | 520 | |
Eric Andersen | 6d7fa43 | 2001-04-10 18:17:05 +0000 | [diff] [blame] | 521 | if (chunked) { |
Matt Kraai | 854125f | 2001-05-09 19:15:46 +0000 | [diff] [blame] | 522 | safe_fgets(buf, sizeof(buf), dfp); /* This is a newline */ |
| 523 | safe_fgets(buf, sizeof(buf), dfp); |
Eric Andersen | 6d7fa43 | 2001-04-10 18:17:05 +0000 | [diff] [blame] | 524 | filesize = strtol(buf, (char **) NULL, 16); |
Glenn L McGrath | 83e4a5b | 2003-08-28 21:55:22 +0000 | [diff] [blame] | 525 | if (filesize==0) { |
| 526 | chunked = 0; /* all done! */ |
| 527 | } |
Eric Andersen | 6d7fa43 | 2001-04-10 18:17:05 +0000 | [diff] [blame] | 528 | } |
| 529 | |
Glenn L McGrath | 83e4a5b | 2003-08-28 21:55:22 +0000 | [diff] [blame] | 530 | if (n == 0 && ferror(dfp)) { |
| 531 | bb_perror_msg_and_die("network read error"); |
| 532 | } |
Eric Andersen | 6d7fa43 | 2001-04-10 18:17:05 +0000 | [diff] [blame] | 533 | } while (chunked); |
Eric Andersen | bdfd0d7 | 2001-10-24 05:00:29 +0000 | [diff] [blame] | 534 | #ifdef CONFIG_FEATURE_WGET_STATUSBAR |
Mark Whitley | 30ac01c | 2001-04-17 18:13:16 +0000 | [diff] [blame] | 535 | if (quiet_flag==FALSE) |
| 536 | progressmeter(1); |
| 537 | #endif |
Glenn L McGrath | f1c4b11 | 2004-02-22 00:27:34 +0000 | [diff] [blame] | 538 | if ((use_proxy == 0) && target.is_ftp) { |
Eric Andersen | 79757c9 | 2001-04-05 21:45:54 +0000 | [diff] [blame] | 539 | fclose(dfp); |
| 540 | if (ftpcmd(NULL, NULL, sfp, buf) != 226) |
Manuel Novoa III | cad5364 | 2003-03-19 09:13:01 +0000 | [diff] [blame] | 541 | bb_error_msg_and_die("ftp error: %s", buf+4); |
Eric Andersen | 79757c9 | 2001-04-05 21:45:54 +0000 | [diff] [blame] | 542 | ftpcmd("QUIT", NULL, sfp, buf); |
| 543 | } |
Eric Andersen | 79757c9 | 2001-04-05 21:45:54 +0000 | [diff] [blame] | 544 | exit(EXIT_SUCCESS); |
Eric Andersen | 9670083 | 2000-09-04 15:15:55 +0000 | [diff] [blame] | 545 | } |
| 546 | |
| 547 | |
Eric Andersen | 79757c9 | 2001-04-05 21:45:54 +0000 | [diff] [blame] | 548 | void parse_url(char *url, struct host_info *h) |
Eric Andersen | 9670083 | 2000-09-04 15:15:55 +0000 | [diff] [blame] | 549 | { |
Glenn L McGrath | cc20ebc | 2003-09-10 23:52:15 +0000 | [diff] [blame] | 550 | char *cp, *sp, *up, *pp; |
Eric Andersen | 9670083 | 2000-09-04 15:15:55 +0000 | [diff] [blame] | 551 | |
Eric Andersen | 79757c9 | 2001-04-05 21:45:54 +0000 | [diff] [blame] | 552 | if (strncmp(url, "http://", 7) == 0) { |
Glenn L McGrath | 036dbaa | 2004-01-17 05:03:31 +0000 | [diff] [blame] | 553 | h->port = bb_lookup_port("http", "tcp", 80); |
Eric Andersen | 79757c9 | 2001-04-05 21:45:54 +0000 | [diff] [blame] | 554 | h->host = url + 7; |
| 555 | h->is_ftp = 0; |
| 556 | } else if (strncmp(url, "ftp://", 6) == 0) { |
Glenn L McGrath | 036dbaa | 2004-01-17 05:03:31 +0000 | [diff] [blame] | 557 | h->port = bb_lookup_port("ftp", "tfp", 21); |
Eric Andersen | 79757c9 | 2001-04-05 21:45:54 +0000 | [diff] [blame] | 558 | h->host = url + 6; |
| 559 | h->is_ftp = 1; |
| 560 | } else |
Manuel Novoa III | cad5364 | 2003-03-19 09:13:01 +0000 | [diff] [blame] | 561 | bb_error_msg_and_die("not an http or ftp url: %s", url); |
Eric Andersen | 9670083 | 2000-09-04 15:15:55 +0000 | [diff] [blame] | 562 | |
Eric Andersen | 79757c9 | 2001-04-05 21:45:54 +0000 | [diff] [blame] | 563 | sp = strchr(h->host, '/'); |
Glenn L McGrath | f1c4b11 | 2004-02-22 00:27:34 +0000 | [diff] [blame] | 564 | if (sp) { |
Matt Kraai | a9711a5 | 2001-01-03 16:15:15 +0000 | [diff] [blame] | 565 | *sp++ = '\0'; |
Eric Andersen | 79757c9 | 2001-04-05 21:45:54 +0000 | [diff] [blame] | 566 | h->path = sp; |
Matt Kraai | a9711a5 | 2001-01-03 16:15:15 +0000 | [diff] [blame] | 567 | } else |
Manuel Novoa III | cad5364 | 2003-03-19 09:13:01 +0000 | [diff] [blame] | 568 | h->path = bb_xstrdup(""); |
Eric Andersen | 79757c9 | 2001-04-05 21:45:54 +0000 | [diff] [blame] | 569 | |
| 570 | up = strrchr(h->host, '@'); |
| 571 | if (up != NULL) { |
| 572 | h->user = h->host; |
| 573 | *up++ = '\0'; |
| 574 | h->host = up; |
| 575 | } else |
| 576 | h->user = NULL; |
| 577 | |
Glenn L McGrath | cc20ebc | 2003-09-10 23:52:15 +0000 | [diff] [blame] | 578 | pp = h->host; |
| 579 | |
| 580 | #ifdef CONFIG_FEATURE_WGET_IP6_LITERAL |
| 581 | if (h->host[0] == '[') { |
| 582 | char *ep; |
| 583 | |
| 584 | ep = h->host + 1; |
Eric Andersen | 6231f09 | 2003-09-11 08:25:11 +0000 | [diff] [blame] | 585 | while (*ep == ':' || isxdigit (*ep)) |
Glenn L McGrath | cc20ebc | 2003-09-10 23:52:15 +0000 | [diff] [blame] | 586 | ep++; |
| 587 | if (*ep == ']') { |
| 588 | h->host++; |
| 589 | *ep = '\0'; |
| 590 | pp = ep + 1; |
| 591 | } |
| 592 | } |
| 593 | #endif |
| 594 | |
| 595 | cp = strchr(pp, ':'); |
Eric Andersen | 79757c9 | 2001-04-05 21:45:54 +0000 | [diff] [blame] | 596 | if (cp != NULL) { |
| 597 | *cp++ = '\0'; |
Glenn L McGrath | f980bd5 | 2003-12-27 00:21:47 +0000 | [diff] [blame] | 598 | h->port = htons(atoi(cp)); |
Eric Andersen | 79757c9 | 2001-04-05 21:45:54 +0000 | [diff] [blame] | 599 | } |
Eric Andersen | 9670083 | 2000-09-04 15:15:55 +0000 | [diff] [blame] | 600 | } |
| 601 | |
| 602 | |
Glenn L McGrath | ffccf6e | 2003-12-20 01:47:18 +0000 | [diff] [blame] | 603 | FILE *open_socket(struct sockaddr_in *s_in) |
Eric Andersen | 9670083 | 2000-09-04 15:15:55 +0000 | [diff] [blame] | 604 | { |
Eric Andersen | 9670083 | 2000-09-04 15:15:55 +0000 | [diff] [blame] | 605 | FILE *fp; |
| 606 | |
Glenn L McGrath | ffccf6e | 2003-12-20 01:47:18 +0000 | [diff] [blame] | 607 | fp = fdopen(xconnect(s_in), "r+"); |
| 608 | if (fp == NULL) |
Manuel Novoa III | cad5364 | 2003-03-19 09:13:01 +0000 | [diff] [blame] | 609 | bb_perror_msg_and_die("fdopen()"); |
Eric Andersen | 9670083 | 2000-09-04 15:15:55 +0000 | [diff] [blame] | 610 | |
| 611 | return fp; |
| 612 | } |
| 613 | |
| 614 | |
| 615 | char *gethdr(char *buf, size_t bufsiz, FILE *fp, int *istrunc) |
| 616 | { |
| 617 | char *s, *hdrval; |
| 618 | int c; |
| 619 | |
| 620 | *istrunc = 0; |
| 621 | |
| 622 | /* retrieve header line */ |
| 623 | if (fgets(buf, bufsiz, fp) == NULL) |
| 624 | return NULL; |
| 625 | |
| 626 | /* see if we are at the end of the headers */ |
| 627 | for (s = buf ; *s == '\r' ; ++s) |
| 628 | ; |
| 629 | if (s[0] == '\n') |
| 630 | return NULL; |
| 631 | |
| 632 | /* convert the header name to lower case */ |
| 633 | for (s = buf ; isalnum(*s) || *s == '-' ; ++s) |
| 634 | *s = tolower(*s); |
| 635 | |
| 636 | /* verify we are at the end of the header name */ |
| 637 | if (*s != ':') |
Manuel Novoa III | cad5364 | 2003-03-19 09:13:01 +0000 | [diff] [blame] | 638 | bb_error_msg_and_die("bad header line: %s", buf); |
Eric Andersen | 9670083 | 2000-09-04 15:15:55 +0000 | [diff] [blame] | 639 | |
| 640 | /* locate the start of the header value */ |
| 641 | for (*s++ = '\0' ; *s == ' ' || *s == '\t' ; ++s) |
| 642 | ; |
| 643 | hdrval = s; |
| 644 | |
| 645 | /* locate the end of header */ |
| 646 | while (*s != '\0' && *s != '\r' && *s != '\n') |
| 647 | ++s; |
| 648 | |
| 649 | /* end of header found */ |
| 650 | if (*s != '\0') { |
| 651 | *s = '\0'; |
| 652 | return hdrval; |
| 653 | } |
| 654 | |
Eric Andersen | 5d63884 | 2000-09-14 21:46:30 +0000 | [diff] [blame] | 655 | /* Rats! The buffer isn't big enough to hold the entire header value. */ |
Eric Andersen | 9670083 | 2000-09-04 15:15:55 +0000 | [diff] [blame] | 656 | while (c = getc(fp), c != EOF && c != '\n') |
| 657 | ; |
| 658 | *istrunc = 1; |
| 659 | return hdrval; |
| 660 | } |
| 661 | |
Eric Andersen | 79757c9 | 2001-04-05 21:45:54 +0000 | [diff] [blame] | 662 | static int ftpcmd(char *s1, char *s2, FILE *fp, char *buf) |
| 663 | { |
Eric Andersen | 79757c9 | 2001-04-05 21:45:54 +0000 | [diff] [blame] | 664 | if (s1) { |
| 665 | if (!s2) s2=""; |
Eric Andersen | 3f1cf45 | 2003-03-11 18:03:39 +0000 | [diff] [blame] | 666 | fprintf(fp, "%s%s\r\n", s1, s2); |
Eric Andersen | 79757c9 | 2001-04-05 21:45:54 +0000 | [diff] [blame] | 667 | fflush(fp); |
| 668 | } |
Eric Andersen | c7bda1c | 2004-03-15 08:29:22 +0000 | [diff] [blame] | 669 | |
Eric Andersen | 79757c9 | 2001-04-05 21:45:54 +0000 | [diff] [blame] | 670 | do { |
Glenn L McGrath | 32da885 | 2004-04-08 10:27:11 +0000 | [diff] [blame] | 671 | char *buf_ptr; |
| 672 | |
| 673 | if (fgets(buf, 510, fp) == NULL) { |
Manuel Novoa III | cad5364 | 2003-03-19 09:13:01 +0000 | [diff] [blame] | 674 | bb_perror_msg_and_die("fgets()"); |
Glenn L McGrath | 32da885 | 2004-04-08 10:27:11 +0000 | [diff] [blame] | 675 | } |
| 676 | buf_ptr = strstr(buf, "\r\n"); |
| 677 | if (buf_ptr) { |
| 678 | *buf_ptr = '\0'; |
| 679 | } |
Eric Andersen | 79757c9 | 2001-04-05 21:45:54 +0000 | [diff] [blame] | 680 | } while (! isdigit(buf[0]) || buf[3] != ' '); |
Eric Andersen | c7bda1c | 2004-03-15 08:29:22 +0000 | [diff] [blame] | 681 | |
Eric Andersen | 79757c9 | 2001-04-05 21:45:54 +0000 | [diff] [blame] | 682 | return atoi(buf); |
| 683 | } |
| 684 | |
Eric Andersen | bdfd0d7 | 2001-10-24 05:00:29 +0000 | [diff] [blame] | 685 | #ifdef CONFIG_FEATURE_WGET_STATUSBAR |
Eric Andersen | c7bda1c | 2004-03-15 08:29:22 +0000 | [diff] [blame] | 686 | /* Stuff below is from BSD rcp util.c, as added to openshh. |
Eric Andersen | 4e573f4 | 2000-11-14 23:29:24 +0000 | [diff] [blame] | 687 | * Original copyright notice is retained at the end of this file. |
Eric Andersen | c7bda1c | 2004-03-15 08:29:22 +0000 | [diff] [blame] | 688 | * |
| 689 | */ |
Eric Andersen | b520e08 | 2000-10-03 00:21:45 +0000 | [diff] [blame] | 690 | |
| 691 | |
Eric Andersen | 3e6ff90 | 2001-03-09 21:24:12 +0000 | [diff] [blame] | 692 | static int |
Eric Andersen | b520e08 | 2000-10-03 00:21:45 +0000 | [diff] [blame] | 693 | getttywidth(void) |
| 694 | { |
Eric Andersen | 8efe967 | 2003-09-15 08:33:45 +0000 | [diff] [blame] | 695 | int width=0; |
| 696 | get_terminal_width_height(0, &width, NULL); |
| 697 | return (width); |
Eric Andersen | b520e08 | 2000-10-03 00:21:45 +0000 | [diff] [blame] | 698 | } |
| 699 | |
Eric Andersen | 3e6ff90 | 2001-03-09 21:24:12 +0000 | [diff] [blame] | 700 | static void |
Eric Andersen | b520e08 | 2000-10-03 00:21:45 +0000 | [diff] [blame] | 701 | updateprogressmeter(int ignore) |
| 702 | { |
| 703 | int save_errno = errno; |
| 704 | |
| 705 | progressmeter(0); |
| 706 | errno = save_errno; |
| 707 | } |
| 708 | |
Eric Andersen | 3e6ff90 | 2001-03-09 21:24:12 +0000 | [diff] [blame] | 709 | static void |
Eric Andersen | b520e08 | 2000-10-03 00:21:45 +0000 | [diff] [blame] | 710 | alarmtimer(int wait) |
| 711 | { |
| 712 | struct itimerval itv; |
| 713 | |
| 714 | itv.it_value.tv_sec = wait; |
| 715 | itv.it_value.tv_usec = 0; |
| 716 | itv.it_interval = itv.it_value; |
| 717 | setitimer(ITIMER_REAL, &itv, NULL); |
| 718 | } |
| 719 | |
| 720 | |
Eric Andersen | 3e6ff90 | 2001-03-09 21:24:12 +0000 | [diff] [blame] | 721 | static void |
Eric Andersen | b520e08 | 2000-10-03 00:21:45 +0000 | [diff] [blame] | 722 | progressmeter(int flag) |
| 723 | { |
| 724 | static const char prefixes[] = " KMGTP"; |
| 725 | static struct timeval lastupdate; |
Mark Whitley | 30ac01c | 2001-04-17 18:13:16 +0000 | [diff] [blame] | 726 | static off_t lastsize, totalsize; |
Eric Andersen | b520e08 | 2000-10-03 00:21:45 +0000 | [diff] [blame] | 727 | struct timeval now, td, wait; |
| 728 | off_t cursize, abbrevsize; |
| 729 | double elapsed; |
| 730 | int ratio, barlength, i, remaining; |
| 731 | char buf[256]; |
| 732 | |
| 733 | if (flag == -1) { |
| 734 | (void) gettimeofday(&start, (struct timezone *) 0); |
| 735 | lastupdate = start; |
| 736 | lastsize = 0; |
Mark Whitley | 30ac01c | 2001-04-17 18:13:16 +0000 | [diff] [blame] | 737 | totalsize = filesize; /* as filesize changes.. */ |
Eric Andersen | b520e08 | 2000-10-03 00:21:45 +0000 | [diff] [blame] | 738 | } |
| 739 | |
| 740 | (void) gettimeofday(&now, (struct timezone *) 0); |
| 741 | cursize = statbytes; |
Mark Whitley | 30ac01c | 2001-04-17 18:13:16 +0000 | [diff] [blame] | 742 | if (totalsize != 0 && !chunked) { |
| 743 | ratio = 100.0 * cursize / totalsize; |
Eric Andersen | b520e08 | 2000-10-03 00:21:45 +0000 | [diff] [blame] | 744 | ratio = MAX(ratio, 0); |
| 745 | ratio = MIN(ratio, 100); |
| 746 | } else |
| 747 | ratio = 100; |
| 748 | |
| 749 | snprintf(buf, sizeof(buf), "\r%-20.20s %3d%% ", curfile, ratio); |
| 750 | |
| 751 | barlength = getttywidth() - 51; |
| 752 | if (barlength > 0) { |
| 753 | i = barlength * ratio / 100; |
| 754 | snprintf(buf + strlen(buf), sizeof(buf) - strlen(buf), |
| 755 | "|%.*s%*s|", i, |
| 756 | "*****************************************************************************" |
| 757 | "*****************************************************************************", |
| 758 | barlength - i, ""); |
| 759 | } |
| 760 | i = 0; |
| 761 | abbrevsize = cursize; |
| 762 | while (abbrevsize >= 100000 && i < sizeof(prefixes)) { |
| 763 | i++; |
| 764 | abbrevsize >>= 10; |
| 765 | } |
| 766 | snprintf(buf + strlen(buf), sizeof(buf) - strlen(buf), " %5d %c%c ", |
| 767 | (int) abbrevsize, prefixes[i], prefixes[i] == ' ' ? ' ' : |
| 768 | 'B'); |
| 769 | |
| 770 | timersub(&now, &lastupdate, &wait); |
| 771 | if (cursize > lastsize) { |
| 772 | lastupdate = now; |
| 773 | lastsize = cursize; |
| 774 | if (wait.tv_sec >= STALLTIME) { |
| 775 | start.tv_sec += wait.tv_sec; |
| 776 | start.tv_usec += wait.tv_usec; |
| 777 | } |
| 778 | wait.tv_sec = 0; |
| 779 | } |
| 780 | timersub(&now, &start, &td); |
| 781 | elapsed = td.tv_sec + (td.tv_usec / 1000000.0); |
| 782 | |
Mark Whitley | 30ac01c | 2001-04-17 18:13:16 +0000 | [diff] [blame] | 783 | if (wait.tv_sec >= STALLTIME) { |
Eric Andersen | b520e08 | 2000-10-03 00:21:45 +0000 | [diff] [blame] | 784 | snprintf(buf + strlen(buf), sizeof(buf) - strlen(buf), |
| 785 | " - stalled -"); |
Mark Whitley | 30ac01c | 2001-04-17 18:13:16 +0000 | [diff] [blame] | 786 | } else if (statbytes <= 0 || elapsed <= 0.0 || cursize > totalsize || chunked) { |
| 787 | snprintf(buf + strlen(buf), sizeof(buf) - strlen(buf), |
| 788 | " --:-- ETA"); |
Eric Andersen | b520e08 | 2000-10-03 00:21:45 +0000 | [diff] [blame] | 789 | } else { |
Mark Whitley | 30ac01c | 2001-04-17 18:13:16 +0000 | [diff] [blame] | 790 | remaining = (int) (totalsize / (statbytes / elapsed) - elapsed); |
Eric Andersen | b520e08 | 2000-10-03 00:21:45 +0000 | [diff] [blame] | 791 | i = remaining / 3600; |
| 792 | if (i) |
| 793 | snprintf(buf + strlen(buf), sizeof(buf) - strlen(buf), |
| 794 | "%2d:", i); |
| 795 | else |
| 796 | snprintf(buf + strlen(buf), sizeof(buf) - strlen(buf), |
| 797 | " "); |
| 798 | i = remaining % 3600; |
| 799 | snprintf(buf + strlen(buf), sizeof(buf) - strlen(buf), |
| 800 | "%02d:%02d ETA", i / 60, i % 60); |
| 801 | } |
Eric Andersen | 70060d2 | 2004-03-27 10:02:48 +0000 | [diff] [blame] | 802 | write(STDERR_FILENO, buf, strlen(buf)); |
Eric Andersen | b520e08 | 2000-10-03 00:21:45 +0000 | [diff] [blame] | 803 | |
| 804 | if (flag == -1) { |
| 805 | struct sigaction sa; |
| 806 | sa.sa_handler = updateprogressmeter; |
| 807 | sigemptyset(&sa.sa_mask); |
| 808 | sa.sa_flags = SA_RESTART; |
| 809 | sigaction(SIGALRM, &sa, NULL); |
| 810 | alarmtimer(1); |
| 811 | } else if (flag == 1) { |
| 812 | alarmtimer(0); |
| 813 | statbytes = 0; |
Mark Whitley | 30ac01c | 2001-04-17 18:13:16 +0000 | [diff] [blame] | 814 | putc('\n', stderr); |
Eric Andersen | b520e08 | 2000-10-03 00:21:45 +0000 | [diff] [blame] | 815 | } |
| 816 | } |
| 817 | #endif |
Eric Andersen | 4e573f4 | 2000-11-14 23:29:24 +0000 | [diff] [blame] | 818 | |
Eric Andersen | bdfd0d7 | 2001-10-24 05:00:29 +0000 | [diff] [blame] | 819 | /* Original copyright notice which applies to the CONFIG_FEATURE_WGET_STATUSBAR stuff, |
Eric Andersen | aff114c | 2004-04-14 17:51:38 +0000 | [diff] [blame] | 820 | * much of which was blatantly stolen from openssh. */ |
Eric Andersen | c7bda1c | 2004-03-15 08:29:22 +0000 | [diff] [blame] | 821 | |
Eric Andersen | 4e573f4 | 2000-11-14 23:29:24 +0000 | [diff] [blame] | 822 | /*- |
| 823 | * Copyright (c) 1992, 1993 |
| 824 | * The Regents of the University of California. All rights reserved. |
| 825 | * |
| 826 | * Redistribution and use in source and binary forms, with or without |
| 827 | * modification, are permitted provided that the following conditions |
| 828 | * are met: |
| 829 | * 1. Redistributions of source code must retain the above copyright |
| 830 | * notice, this list of conditions and the following disclaimer. |
| 831 | * 2. Redistributions in binary form must reproduce the above copyright |
| 832 | * notice, this list of conditions and the following disclaimer in the |
| 833 | * documentation and/or other materials provided with the distribution. |
| 834 | * |
Eric Andersen | c7bda1c | 2004-03-15 08:29:22 +0000 | [diff] [blame] | 835 | * 3. <BSD Advertising Clause omitted per the July 22, 1999 licensing change |
| 836 | * ftp://ftp.cs.berkeley.edu/pub/4bsd/README.Impt.License.Change> |
Eric Andersen | 4e573f4 | 2000-11-14 23:29:24 +0000 | [diff] [blame] | 837 | * |
| 838 | * 4. Neither the name of the University nor the names of its contributors |
| 839 | * may be used to endorse or promote products derived from this software |
| 840 | * without specific prior written permission. |
| 841 | * |
| 842 | * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND |
| 843 | * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE |
| 844 | * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE |
| 845 | * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE |
| 846 | * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL |
| 847 | * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS |
| 848 | * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) |
| 849 | * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT |
| 850 | * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY |
| 851 | * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF |
| 852 | * SUCH DAMAGE. |
| 853 | * |
Eric Andersen | 751750e | 2004-10-08 08:27:40 +0000 | [diff] [blame] | 854 | * $Id: wget.c,v 1.75 2004/10/08 08:27:40 andersen Exp $ |
Eric Andersen | 4e573f4 | 2000-11-14 23:29:24 +0000 | [diff] [blame] | 855 | */ |
| 856 | |
| 857 | |
| 858 | |
Eric Andersen | 9670083 | 2000-09-04 15:15:55 +0000 | [diff] [blame] | 859 | /* |
| 860 | Local Variables: |
| 861 | c-file-style: "linux" |
| 862 | c-basic-offset: 4 |
| 863 | tab-width: 4 |
| 864 | End: |
| 865 | */ |