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 | |
Denis Vlasenko | a552eeb | 2006-09-26 09:22:12 +0000 | [diff] [blame] | 9 | #include <getopt.h> /* for struct option */ |
Denis Vlasenko | b6adbf1 | 2007-05-26 19:00:18 +0000 | [diff] [blame] | 10 | #include "libbb.h" |
Denis Vlasenko | a552eeb | 2006-09-26 09:22:12 +0000 | [diff] [blame] | 11 | |
Eric Andersen | 79757c9 | 2001-04-05 21:45:54 +0000 | [diff] [blame] | 12 | struct host_info { |
Denis Vlasenko | 96e9d3c | 2006-10-07 14:28:55 +0000 | [diff] [blame] | 13 | // May be used if we ever will want to free() all xstrdup()s... |
| 14 | /* char *allocated; */ |
Denis Vlasenko | 818322b | 2007-09-24 18:27:04 +0000 | [diff] [blame] | 15 | const char *path; |
| 16 | const char *user; |
| 17 | char *host; |
| 18 | int port; |
| 19 | smallint is_ftp; |
Eric Andersen | 79757c9 | 2001-04-05 21:45:54 +0000 | [diff] [blame] | 20 | }; |
| 21 | |
Denis Vlasenko | 7710563 | 2007-09-24 15:04:00 +0000 | [diff] [blame] | 22 | |
| 23 | /* Globals (can be accessed from signal handlers) */ |
| 24 | struct globals { |
| 25 | off_t content_len; /* Content-length of the file */ |
| 26 | off_t beg_range; /* Range at which continue begins */ |
| 27 | #if ENABLE_FEATURE_WGET_STATUSBAR |
| 28 | off_t lastsize; |
| 29 | off_t totalsize; |
| 30 | off_t transferred; /* Number of bytes transferred so far */ |
| 31 | const char *curfile; /* Name of current file being transferred */ |
| 32 | unsigned lastupdate_sec; |
| 33 | unsigned start_sec; |
| 34 | #endif |
Denis Vlasenko | a7ce207 | 2007-09-24 18:25:08 +0000 | [diff] [blame] | 35 | smallint chunked; /* chunked transfer encoding */ |
Denis Vlasenko | 7710563 | 2007-09-24 15:04:00 +0000 | [diff] [blame] | 36 | }; |
| 37 | #define G (*(struct globals*)&bb_common_bufsiz1) |
| 38 | struct BUG_G_too_big { |
| 39 | char BUG_G_too_big[sizeof(G) <= COMMON_BUFSIZE ? 1 : -1]; |
| 40 | }; |
| 41 | #define content_len (G.content_len ) |
| 42 | #define beg_range (G.beg_range ) |
| 43 | #define lastsize (G.lastsize ) |
| 44 | #define totalsize (G.totalsize ) |
| 45 | #define transferred (G.transferred ) |
| 46 | #define curfile (G.curfile ) |
| 47 | #define lastupdate_sec (G.lastupdate_sec ) |
| 48 | #define start_sec (G.start_sec ) |
| 49 | #define chunked (G.chunked ) |
| 50 | #define INIT_G() do { } while (0) |
| 51 | |
| 52 | |
Denis Vlasenko | 9cade08 | 2006-11-21 10:43:02 +0000 | [diff] [blame] | 53 | #if ENABLE_FEATURE_WGET_STATUSBAR |
Rob Landley | 19a3940 | 2006-06-13 17:10:26 +0000 | [diff] [blame] | 54 | enum { |
Denis Vlasenko | f8aa109 | 2006-10-01 10:58:54 +0000 | [diff] [blame] | 55 | STALLTIME = 5 /* Seconds when xfer considered "stalled" */ |
Rob Landley | 19a3940 | 2006-06-13 17:10:26 +0000 | [diff] [blame] | 56 | }; |
Denis Vlasenko | 47ddd01 | 2007-09-24 18:24:17 +0000 | [diff] [blame] | 57 | |
| 58 | static int getttywidth(void) |
| 59 | { |
| 60 | int width; |
| 61 | get_terminal_width_height(0, &width, NULL); |
| 62 | return width; |
| 63 | } |
| 64 | |
Denis Vlasenko | 47ddd01 | 2007-09-24 18:24:17 +0000 | [diff] [blame] | 65 | static void progressmeter(int flag) |
| 66 | { |
Denis Vlasenko | a7ce207 | 2007-09-24 18:25:08 +0000 | [diff] [blame] | 67 | /* We can be called from signal handler */ |
| 68 | int save_errno = errno; |
Denis Vlasenko | 47ddd01 | 2007-09-24 18:24:17 +0000 | [diff] [blame] | 69 | off_t abbrevsize; |
| 70 | unsigned since_last_update, elapsed; |
| 71 | unsigned ratio; |
| 72 | int barlength, i; |
| 73 | |
| 74 | if (flag == -1) { /* first call to progressmeter */ |
| 75 | start_sec = monotonic_sec(); |
| 76 | lastupdate_sec = start_sec; |
| 77 | lastsize = 0; |
| 78 | totalsize = content_len + beg_range; /* as content_len changes.. */ |
| 79 | } |
| 80 | |
| 81 | ratio = 100; |
| 82 | if (totalsize != 0 && !chunked) { |
| 83 | /* long long helps to have it working even if !LFS */ |
| 84 | ratio = (unsigned) (100ULL * (transferred+beg_range) / totalsize); |
| 85 | if (ratio > 100) ratio = 100; |
| 86 | } |
| 87 | |
| 88 | fprintf(stderr, "\r%-20.20s%4d%% ", curfile, ratio); |
| 89 | |
| 90 | barlength = getttywidth() - 49; |
| 91 | if (barlength > 0) { |
| 92 | /* god bless gcc for variable arrays :) */ |
| 93 | i = barlength * ratio / 100; |
| 94 | { |
| 95 | char buf[i+1]; |
| 96 | memset(buf, '*', i); |
| 97 | buf[i] = '\0'; |
| 98 | fprintf(stderr, "|%s%*s|", buf, barlength - i, ""); |
| 99 | } |
| 100 | } |
| 101 | i = 0; |
| 102 | abbrevsize = transferred + beg_range; |
| 103 | while (abbrevsize >= 100000) { |
| 104 | i++; |
| 105 | abbrevsize >>= 10; |
| 106 | } |
| 107 | /* see http://en.wikipedia.org/wiki/Tera */ |
| 108 | fprintf(stderr, "%6d%c ", (int)abbrevsize, " kMGTPEZY"[i]); |
| 109 | |
| 110 | // Nuts! Ain't it easier to update progress meter ONLY when we transferred++? |
Denis Vlasenko | 47ddd01 | 2007-09-24 18:24:17 +0000 | [diff] [blame] | 111 | |
| 112 | elapsed = monotonic_sec(); |
| 113 | since_last_update = elapsed - lastupdate_sec; |
| 114 | if (transferred > lastsize) { |
| 115 | lastupdate_sec = elapsed; |
| 116 | lastsize = transferred; |
| 117 | if (since_last_update >= STALLTIME) { |
| 118 | /* We "cut off" these seconds from elapsed time |
| 119 | * by adjusting start time */ |
| 120 | start_sec += since_last_update; |
| 121 | } |
| 122 | since_last_update = 0; /* we are un-stalled now */ |
| 123 | } |
| 124 | elapsed -= start_sec; /* now it's "elapsed since start" */ |
| 125 | |
| 126 | if (since_last_update >= STALLTIME) { |
| 127 | fprintf(stderr, " - stalled -"); |
| 128 | } else { |
| 129 | off_t to_download = totalsize - beg_range; |
| 130 | if (transferred <= 0 || (int)elapsed <= 0 || transferred > to_download || chunked) { |
| 131 | fprintf(stderr, "--:--:-- ETA"); |
| 132 | } else { |
| 133 | /* to_download / (transferred/elapsed) - elapsed: */ |
| 134 | int eta = (int) ((unsigned long long)to_download*elapsed/transferred - elapsed); |
| 135 | /* (long long helps to have working ETA even if !LFS) */ |
| 136 | i = eta % 3600; |
| 137 | fprintf(stderr, "%02d:%02d:%02d ETA", eta / 3600, i / 60, i % 60); |
| 138 | } |
| 139 | } |
| 140 | |
Denis Vlasenko | a7ce207 | 2007-09-24 18:25:08 +0000 | [diff] [blame] | 141 | if (flag == 0) { |
| 142 | /* last call to progressmeter */ |
| 143 | alarm(0); |
Denis Vlasenko | 47ddd01 | 2007-09-24 18:24:17 +0000 | [diff] [blame] | 144 | transferred = 0; |
| 145 | putc('\n', stderr); |
Denis Vlasenko | a7ce207 | 2007-09-24 18:25:08 +0000 | [diff] [blame] | 146 | } else { |
| 147 | if (flag == -1) { |
| 148 | /* first call to progressmeter */ |
| 149 | struct sigaction sa; |
| 150 | sa.sa_handler = progressmeter; |
| 151 | sigemptyset(&sa.sa_mask); |
| 152 | sa.sa_flags = SA_RESTART; |
| 153 | sigaction(SIGALRM, &sa, NULL); |
| 154 | } |
| 155 | alarm(1); |
Denis Vlasenko | 47ddd01 | 2007-09-24 18:24:17 +0000 | [diff] [blame] | 156 | } |
Denis Vlasenko | a7ce207 | 2007-09-24 18:25:08 +0000 | [diff] [blame] | 157 | |
| 158 | errno = save_errno; |
Denis Vlasenko | 47ddd01 | 2007-09-24 18:24:17 +0000 | [diff] [blame] | 159 | } |
| 160 | /* Original copyright notice which applies to the CONFIG_FEATURE_WGET_STATUSBAR stuff, |
| 161 | * much of which was blatantly stolen from openssh. */ |
| 162 | /*- |
| 163 | * Copyright (c) 1992, 1993 |
| 164 | * The Regents of the University of California. All rights reserved. |
| 165 | * |
| 166 | * Redistribution and use in source and binary forms, with or without |
| 167 | * modification, are permitted provided that the following conditions |
| 168 | * are met: |
| 169 | * 1. Redistributions of source code must retain the above copyright |
| 170 | * notice, this list of conditions and the following disclaimer. |
| 171 | * 2. Redistributions in binary form must reproduce the above copyright |
| 172 | * notice, this list of conditions and the following disclaimer in the |
| 173 | * documentation and/or other materials provided with the distribution. |
| 174 | * |
| 175 | * 3. <BSD Advertising Clause omitted per the July 22, 1999 licensing change |
| 176 | * ftp://ftp.cs.berkeley.edu/pub/4bsd/README.Impt.License.Change> |
| 177 | * |
| 178 | * 4. Neither the name of the University nor the names of its contributors |
| 179 | * may be used to endorse or promote products derived from this software |
| 180 | * without specific prior written permission. |
| 181 | * |
| 182 | * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND |
| 183 | * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE |
| 184 | * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE |
| 185 | * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE |
| 186 | * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL |
| 187 | * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS |
| 188 | * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) |
| 189 | * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT |
| 190 | * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY |
| 191 | * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF |
| 192 | * SUCH DAMAGE. |
| 193 | * |
| 194 | */ |
| 195 | #else /* FEATURE_WGET_STATUSBAR */ |
| 196 | |
| 197 | static ALWAYS_INLINE void progressmeter(int flag) { } |
| 198 | |
Eric Andersen | b520e08 | 2000-10-03 00:21:45 +0000 | [diff] [blame] | 199 | #endif |
Eric Andersen | c7bda1c | 2004-03-15 08:29:22 +0000 | [diff] [blame] | 200 | |
Denis Vlasenko | 47ddd01 | 2007-09-24 18:24:17 +0000 | [diff] [blame] | 201 | |
Denis Vlasenko | 12d2129 | 2007-06-27 21:40:07 +0000 | [diff] [blame] | 202 | /* Read NMEMB bytes into PTR from STREAM. Returns the number of bytes read, |
| 203 | * and a short count if an eof or non-interrupt error is encountered. */ |
| 204 | static size_t safe_fread(void *ptr, size_t nmemb, FILE *stream) |
Matt Kraai | 854125f | 2001-05-09 19:15:46 +0000 | [diff] [blame] | 205 | { |
Denis Vlasenko | 12d2129 | 2007-06-27 21:40:07 +0000 | [diff] [blame] | 206 | size_t ret; |
| 207 | char *p = (char*)ptr; |
Matt Kraai | 854125f | 2001-05-09 19:15:46 +0000 | [diff] [blame] | 208 | |
| 209 | do { |
| 210 | clearerr(stream); |
Denis Vlasenko | 12d2129 | 2007-06-27 21:40:07 +0000 | [diff] [blame] | 211 | ret = fread(p, 1, nmemb, stream); |
| 212 | p += ret; |
| 213 | nmemb -= ret; |
| 214 | } while (nmemb && ferror(stream) && errno == EINTR); |
Matt Kraai | 854125f | 2001-05-09 19:15:46 +0000 | [diff] [blame] | 215 | |
Denis Vlasenko | 12d2129 | 2007-06-27 21:40:07 +0000 | [diff] [blame] | 216 | return p - (char*)ptr; |
Matt Kraai | 854125f | 2001-05-09 19:15:46 +0000 | [diff] [blame] | 217 | } |
| 218 | |
Denis Vlasenko | 12d2129 | 2007-06-27 21:40:07 +0000 | [diff] [blame] | 219 | /* Read a line or SIZE-1 bytes into S, whichever is less, from STREAM. |
Matt Kraai | 854125f | 2001-05-09 19:15:46 +0000 | [diff] [blame] | 220 | * Returns S, or NULL if an eof or non-interrupt error is encountered. */ |
| 221 | static char *safe_fgets(char *s, int size, FILE *stream) |
| 222 | { |
| 223 | char *ret; |
| 224 | |
| 225 | do { |
| 226 | clearerr(stream); |
| 227 | ret = fgets(s, size, stream); |
| 228 | } while (ret == NULL && ferror(stream) && errno == EINTR); |
| 229 | |
| 230 | return ret; |
| 231 | } |
| 232 | |
Denis Vlasenko | 9cade08 | 2006-11-21 10:43:02 +0000 | [diff] [blame] | 233 | #if ENABLE_FEATURE_WGET_AUTHENTICATION |
Denis Vlasenko | 12d2129 | 2007-06-27 21:40:07 +0000 | [diff] [blame] | 234 | /* Base64-encode character string. buf is assumed to be char buf[512]. */ |
| 235 | static char *base64enc_512(char buf[512], const char *str) |
Denis Vlasenko | 3526a13 | 2006-09-09 12:20:57 +0000 | [diff] [blame] | 236 | { |
Denis Vlasenko | 12d2129 | 2007-06-27 21:40:07 +0000 | [diff] [blame] | 237 | unsigned len = strlen(str); |
| 238 | if (len > 512/4*3 - 10) /* paranoia */ |
| 239 | len = 512/4*3 - 10; |
| 240 | bb_uuencode(buf, str, len, bb_uuenc_tbl_base64); |
Rob Landley | 76ef08c | 2006-06-13 16:44:26 +0000 | [diff] [blame] | 241 | return buf; |
Eric Andersen | 79757c9 | 2001-04-05 21:45:54 +0000 | [diff] [blame] | 242 | } |
| 243 | #endif |
| 244 | |
Denis Vlasenko | 47ddd01 | 2007-09-24 18:24:17 +0000 | [diff] [blame] | 245 | |
| 246 | static FILE *open_socket(len_and_sockaddr *lsa) |
| 247 | { |
| 248 | FILE *fp; |
| 249 | |
| 250 | /* glibc 2.4 seems to try seeking on it - ??! */ |
| 251 | /* hopefully it understands what ESPIPE means... */ |
| 252 | fp = fdopen(xconnect_stream(lsa), "r+"); |
| 253 | if (fp == NULL) |
| 254 | bb_perror_msg_and_die("fdopen"); |
| 255 | |
| 256 | return fp; |
| 257 | } |
| 258 | |
| 259 | |
| 260 | static int ftpcmd(const char *s1, const char *s2, FILE *fp, char *buf) |
| 261 | { |
| 262 | int result; |
| 263 | if (s1) { |
| 264 | if (!s2) s2 = ""; |
| 265 | fprintf(fp, "%s%s\r\n", s1, s2); |
| 266 | fflush(fp); |
| 267 | } |
| 268 | |
| 269 | do { |
| 270 | char *buf_ptr; |
| 271 | |
| 272 | if (fgets(buf, 510, fp) == NULL) { |
| 273 | bb_perror_msg_and_die("error getting response"); |
| 274 | } |
| 275 | buf_ptr = strstr(buf, "\r\n"); |
| 276 | if (buf_ptr) { |
| 277 | *buf_ptr = '\0'; |
| 278 | } |
| 279 | } while (!isdigit(buf[0]) || buf[3] != ' '); |
| 280 | |
| 281 | buf[3] = '\0'; |
| 282 | result = xatoi_u(buf); |
| 283 | buf[3] = ' '; |
| 284 | return result; |
| 285 | } |
| 286 | |
| 287 | |
| 288 | static void parse_url(char *src_url, struct host_info *h) |
| 289 | { |
| 290 | char *url, *p, *sp; |
| 291 | |
| 292 | /* h->allocated = */ url = xstrdup(src_url); |
| 293 | |
| 294 | if (strncmp(url, "http://", 7) == 0) { |
| 295 | h->port = bb_lookup_port("http", "tcp", 80); |
| 296 | h->host = url + 7; |
| 297 | h->is_ftp = 0; |
| 298 | } else if (strncmp(url, "ftp://", 6) == 0) { |
| 299 | h->port = bb_lookup_port("ftp", "tcp", 21); |
| 300 | h->host = url + 6; |
| 301 | h->is_ftp = 1; |
| 302 | } else |
| 303 | bb_error_msg_and_die("not an http or ftp url: %s", url); |
| 304 | |
| 305 | // FYI: |
| 306 | // "Real" wget 'http://busybox.net?var=a/b' sends this request: |
| 307 | // 'GET /?var=a/b HTTP 1.0' |
| 308 | // and saves 'index.html?var=a%2Fb' (we save 'b') |
| 309 | // wget 'http://busybox.net?login=john@doe': |
| 310 | // request: 'GET /?login=john@doe HTTP/1.0' |
| 311 | // saves: 'index.html?login=john@doe' (we save '?login=john@doe') |
| 312 | // wget 'http://busybox.net#test/test': |
| 313 | // request: 'GET / HTTP/1.0' |
| 314 | // saves: 'index.html' (we save 'test') |
| 315 | // |
| 316 | // We also don't add unique .N suffix if file exists... |
| 317 | sp = strchr(h->host, '/'); |
| 318 | p = strchr(h->host, '?'); if (!sp || (p && sp > p)) sp = p; |
| 319 | p = strchr(h->host, '#'); if (!sp || (p && sp > p)) sp = p; |
| 320 | if (!sp) { |
Denis Vlasenko | 818322b | 2007-09-24 18:27:04 +0000 | [diff] [blame] | 321 | h->path = ""; |
Denis Vlasenko | 47ddd01 | 2007-09-24 18:24:17 +0000 | [diff] [blame] | 322 | } else if (*sp == '/') { |
| 323 | *sp = '\0'; |
| 324 | h->path = sp + 1; |
| 325 | } else { // '#' or '?' |
| 326 | // http://busybox.net?login=john@doe is a valid URL |
| 327 | // memmove converts to: |
| 328 | // http:/busybox.nett?login=john@doe... |
Denis Vlasenko | 818322b | 2007-09-24 18:27:04 +0000 | [diff] [blame] | 329 | memmove(h->host - 1, h->host, sp - h->host); |
Denis Vlasenko | 47ddd01 | 2007-09-24 18:24:17 +0000 | [diff] [blame] | 330 | h->host--; |
| 331 | sp[-1] = '\0'; |
| 332 | h->path = sp; |
| 333 | } |
| 334 | |
| 335 | sp = strrchr(h->host, '@'); |
| 336 | h->user = NULL; |
| 337 | if (sp != NULL) { |
| 338 | h->user = h->host; |
| 339 | *sp = '\0'; |
| 340 | h->host = sp + 1; |
| 341 | } |
| 342 | |
| 343 | sp = h->host; |
| 344 | } |
| 345 | |
| 346 | |
| 347 | static char *gethdr(char *buf, size_t bufsiz, FILE *fp /*, int *istrunc*/) |
| 348 | { |
| 349 | char *s, *hdrval; |
| 350 | int c; |
| 351 | |
| 352 | /* *istrunc = 0; */ |
| 353 | |
| 354 | /* retrieve header line */ |
| 355 | if (fgets(buf, bufsiz, fp) == NULL) |
| 356 | return NULL; |
| 357 | |
| 358 | /* see if we are at the end of the headers */ |
| 359 | for (s = buf; *s == '\r'; ++s) |
| 360 | continue; |
| 361 | if (*s == '\n') |
| 362 | return NULL; |
| 363 | |
| 364 | /* convert the header name to lower case */ |
| 365 | for (s = buf; isalnum(*s) || *s == '-' || *s == '.'; ++s) |
| 366 | *s = tolower(*s); |
| 367 | |
| 368 | /* verify we are at the end of the header name */ |
| 369 | if (*s != ':') |
| 370 | bb_error_msg_and_die("bad header line: %s", buf); |
| 371 | |
| 372 | /* locate the start of the header value */ |
| 373 | *s++ = '\0'; |
| 374 | hdrval = skip_whitespace(s); |
| 375 | |
| 376 | /* locate the end of header */ |
| 377 | while (*s && *s != '\r' && *s != '\n') |
| 378 | ++s; |
| 379 | |
| 380 | /* end of header found */ |
| 381 | if (*s) { |
| 382 | *s = '\0'; |
| 383 | return hdrval; |
| 384 | } |
| 385 | |
| 386 | /* Rats! The buffer isn't big enough to hold the entire header value. */ |
| 387 | while (c = getc(fp), c != EOF && c != '\n') |
| 388 | continue; |
| 389 | /* *istrunc = 1; */ |
| 390 | return hdrval; |
| 391 | } |
| 392 | |
| 393 | |
Denis Vlasenko | 06af216 | 2007-02-03 17:28:39 +0000 | [diff] [blame] | 394 | int wget_main(int argc, char **argv); |
Eric Andersen | 9670083 | 2000-09-04 15:15:55 +0000 | [diff] [blame] | 395 | int wget_main(int argc, char **argv) |
| 396 | { |
Denis Vlasenko | c8400a2 | 2006-10-25 00:33:44 +0000 | [diff] [blame] | 397 | char buf[512]; |
Eric Andersen | 79757c9 | 2001-04-05 21:45:54 +0000 | [diff] [blame] | 398 | struct host_info server, target; |
Denis Vlasenko | 6536a9b | 2007-01-12 10:35:23 +0000 | [diff] [blame] | 399 | len_and_sockaddr *lsa; |
Denis Vlasenko | 06783a5 | 2007-09-24 13:51:54 +0000 | [diff] [blame] | 400 | int status; |
Denis Vlasenko | c8400a2 | 2006-10-25 00:33:44 +0000 | [diff] [blame] | 401 | int port; |
| 402 | int try = 5; |
| 403 | unsigned opt; |
Bernhard Reutner-Fischer | 7e8a53a | 2007-04-10 09:37:29 +0000 | [diff] [blame] | 404 | char *str; |
Denis Vlasenko | c8400a2 | 2006-10-25 00:33:44 +0000 | [diff] [blame] | 405 | char *proxy = 0; |
| 406 | char *dir_prefix = NULL; |
| 407 | #if ENABLE_FEATURE_WGET_LONG_OPTIONS |
| 408 | char *extra_headers = NULL; |
Glenn L McGrath | 514aeab | 2003-12-19 12:08:56 +0000 | [diff] [blame] | 409 | llist_t *headers_llist = NULL; |
Denis Vlasenko | c8400a2 | 2006-10-25 00:33:44 +0000 | [diff] [blame] | 410 | #endif |
Denis Vlasenko | a552eeb | 2006-09-26 09:22:12 +0000 | [diff] [blame] | 411 | FILE *sfp = NULL; /* socket to web/ftp server */ |
| 412 | FILE *dfp = NULL; /* socket to ftp server (data) */ |
| 413 | char *fname_out = NULL; /* where to direct output (-O) */ |
Denis Vlasenko | 4661117 | 2007-08-06 15:43:17 +0000 | [diff] [blame] | 414 | bool got_clen = 0; /* got content-length: from server */ |
Denis Vlasenko | a94554d | 2006-09-23 17:49:09 +0000 | [diff] [blame] | 415 | int output_fd = -1; |
Denis Vlasenko | 4661117 | 2007-08-06 15:43:17 +0000 | [diff] [blame] | 416 | bool use_proxy = 1; /* Use proxies if env vars are set */ |
Denis Vlasenko | 96e9d3c | 2006-10-07 14:28:55 +0000 | [diff] [blame] | 417 | const char *proxy_flag = "on"; /* Use proxies if env vars are set */ |
Bernhard Reutner-Fischer | 7e8a53a | 2007-04-10 09:37:29 +0000 | [diff] [blame] | 418 | const char *user_agent = "Wget";/* "User-Agent" header field */ |
Denis Vlasenko | 7710563 | 2007-09-24 15:04:00 +0000 | [diff] [blame] | 419 | |
Denis Vlasenko | 6ca409e | 2007-08-12 20:58:27 +0000 | [diff] [blame] | 420 | static const char keywords[] ALIGN1 = |
Denis Vlasenko | 990d0f6 | 2007-07-24 15:54:42 +0000 | [diff] [blame] | 421 | "content-length\0""transfer-encoding\0""chunked\0""location\0"; |
Bernhard Reutner-Fischer | 7e8a53a | 2007-04-10 09:37:29 +0000 | [diff] [blame] | 422 | enum { |
| 423 | KEY_content_length = 1, KEY_transfer_encoding, KEY_chunked, KEY_location |
| 424 | }; |
Denis Vlasenko | c8400a2 | 2006-10-25 00:33:44 +0000 | [diff] [blame] | 425 | enum { |
| 426 | WGET_OPT_CONTINUE = 0x1, |
Bernhard Reutner-Fischer | 2e75dcc | 2007-04-05 10:31:47 +0000 | [diff] [blame] | 427 | WGET_OPT_SPIDER = 0x2, |
| 428 | WGET_OPT_QUIET = 0x4, |
| 429 | WGET_OPT_OUTNAME = 0x8, |
| 430 | WGET_OPT_PREFIX = 0x10, |
| 431 | WGET_OPT_PROXY = 0x20, |
| 432 | WGET_OPT_USER_AGENT = 0x40, |
| 433 | WGET_OPT_PASSIVE = 0x80, |
| 434 | WGET_OPT_HEADER = 0x100, |
Denis Vlasenko | c8400a2 | 2006-10-25 00:33:44 +0000 | [diff] [blame] | 435 | }; |
Bernhard Reutner-Fischer | 289e86a | 2006-08-20 20:01:24 +0000 | [diff] [blame] | 436 | #if ENABLE_FEATURE_WGET_LONG_OPTIONS |
Denis Vlasenko | 6ca409e | 2007-08-12 20:58:27 +0000 | [diff] [blame] | 437 | static const char wget_longopts[] ALIGN1 = |
Denis Vlasenko | bdc88fd | 2007-07-23 17:14:14 +0000 | [diff] [blame] | 438 | /* name, has_arg, val */ |
| 439 | "continue\0" No_argument "c" |
| 440 | "spider\0" No_argument "s" |
| 441 | "quiet\0" No_argument "q" |
| 442 | "output-document\0" Required_argument "O" |
| 443 | "directory-prefix\0" Required_argument "P" |
| 444 | "proxy\0" Required_argument "Y" |
| 445 | "user-agent\0" Required_argument "U" |
| 446 | "passive-ftp\0" No_argument "\xff" |
| 447 | "header\0" Required_argument "\xfe" |
Denis Vlasenko | 990d0f6 | 2007-07-24 15:54:42 +0000 | [diff] [blame] | 448 | ; |
Denis Vlasenko | 7710563 | 2007-09-24 15:04:00 +0000 | [diff] [blame] | 449 | #endif |
| 450 | |
| 451 | INIT_G(); |
| 452 | |
| 453 | #if ENABLE_FEATURE_WGET_LONG_OPTIONS |
Denis Vlasenko | bdc88fd | 2007-07-23 17:14:14 +0000 | [diff] [blame] | 454 | applet_long_options = wget_longopts; |
Bernhard Reutner-Fischer | 8d3a6f7 | 2006-05-31 14:11:38 +0000 | [diff] [blame] | 455 | #endif |
Bernhard Reutner-Fischer | 7e8a53a | 2007-04-10 09:37:29 +0000 | [diff] [blame] | 456 | /* server.allocated = target.allocated = NULL; */ |
Denis Vlasenko | c8400a2 | 2006-10-25 00:33:44 +0000 | [diff] [blame] | 457 | opt_complementary = "-1" USE_FEATURE_WGET_LONG_OPTIONS(":\xfe::"); |
Denis Vlasenko | fe7cd64 | 2007-08-18 15:32:12 +0000 | [diff] [blame] | 458 | opt = getopt32(argv, "csqO:P:Y:U:", |
Denis Vlasenko | c8400a2 | 2006-10-25 00:33:44 +0000 | [diff] [blame] | 459 | &fname_out, &dir_prefix, |
| 460 | &proxy_flag, &user_agent |
| 461 | USE_FEATURE_WGET_LONG_OPTIONS(, &headers_llist) |
| 462 | ); |
Glenn L McGrath | f1c4b11 | 2004-02-22 00:27:34 +0000 | [diff] [blame] | 463 | if (strcmp(proxy_flag, "off") == 0) { |
Bernhard Reutner-Fischer | 7e8a53a | 2007-04-10 09:37:29 +0000 | [diff] [blame] | 464 | /* Use the proxy if necessary */ |
Glenn L McGrath | f1c4b11 | 2004-02-22 00:27:34 +0000 | [diff] [blame] | 465 | use_proxy = 0; |
Glenn L McGrath | 514aeab | 2003-12-19 12:08:56 +0000 | [diff] [blame] | 466 | } |
Denis Vlasenko | c8400a2 | 2006-10-25 00:33:44 +0000 | [diff] [blame] | 467 | #if ENABLE_FEATURE_WGET_LONG_OPTIONS |
Denis Vlasenko | 7534e08 | 2006-10-23 23:21:58 +0000 | [diff] [blame] | 468 | if (headers_llist) { |
| 469 | int size = 1; |
| 470 | char *cp; |
Denis Vlasenko | 8d9f495 | 2007-04-08 15:08:42 +0000 | [diff] [blame] | 471 | llist_t *ll = headers_llist; |
Denis Vlasenko | 7534e08 | 2006-10-23 23:21:58 +0000 | [diff] [blame] | 472 | while (ll) { |
| 473 | size += strlen(ll->data) + 2; |
| 474 | ll = ll->link; |
| 475 | } |
| 476 | extra_headers = cp = xmalloc(size); |
Glenn L McGrath | 514aeab | 2003-12-19 12:08:56 +0000 | [diff] [blame] | 477 | while (headers_llist) { |
Denis Vlasenko | 7534e08 | 2006-10-23 23:21:58 +0000 | [diff] [blame] | 478 | cp += sprintf(cp, "%s\r\n", headers_llist->data); |
Glenn L McGrath | 514aeab | 2003-12-19 12:08:56 +0000 | [diff] [blame] | 479 | headers_llist = headers_llist->link; |
Eric Andersen | 9670083 | 2000-09-04 15:15:55 +0000 | [diff] [blame] | 480 | } |
| 481 | } |
Denis Vlasenko | c8400a2 | 2006-10-25 00:33:44 +0000 | [diff] [blame] | 482 | #endif |
Tim Riker | c1ef7bd | 2006-01-25 00:08:53 +0000 | [diff] [blame] | 483 | |
Eric Andersen | 79757c9 | 2001-04-05 21:45:54 +0000 | [diff] [blame] | 484 | parse_url(argv[optind], &target); |
| 485 | server.host = target.host; |
| 486 | server.port = target.port; |
| 487 | |
Bernhard Reutner-Fischer | 7e8a53a | 2007-04-10 09:37:29 +0000 | [diff] [blame] | 488 | /* Use the proxy if necessary */ |
Glenn L McGrath | f1c4b11 | 2004-02-22 00:27:34 +0000 | [diff] [blame] | 489 | if (use_proxy) { |
Robert Griebl | d776011 | 2002-05-14 23:36:45 +0000 | [diff] [blame] | 490 | proxy = getenv(target.is_ftp ? "ftp_proxy" : "http_proxy"); |
Glenn L McGrath | f1c4b11 | 2004-02-22 00:27:34 +0000 | [diff] [blame] | 491 | if (proxy && *proxy) { |
Denis Vlasenko | 96e9d3c | 2006-10-07 14:28:55 +0000 | [diff] [blame] | 492 | parse_url(proxy, &server); |
Glenn L McGrath | f1c4b11 | 2004-02-22 00:27:34 +0000 | [diff] [blame] | 493 | } else { |
| 494 | use_proxy = 0; |
| 495 | } |
Robert Griebl | d776011 | 2002-05-14 23:36:45 +0000 | [diff] [blame] | 496 | } |
Eric Andersen | c7bda1c | 2004-03-15 08:29:22 +0000 | [diff] [blame] | 497 | |
Denis Vlasenko | 818322b | 2007-09-24 18:27:04 +0000 | [diff] [blame] | 498 | /* Guess an output filename, if there was no -O FILE */ |
Eric Andersen | 29edd00 | 2000-12-09 16:55:35 +0000 | [diff] [blame] | 499 | if (!fname_out) { |
Denis Vlasenko | 818322b | 2007-09-24 18:27:04 +0000 | [diff] [blame] | 500 | fname_out = bb_get_last_path_component_nostrip(target.path); |
| 501 | /* handle "wget http://kernel.org//" */ |
| 502 | if (fname_out[0] == '/' || !fname_out[0]) |
Denis Vlasenko | b6aae0f | 2007-01-29 22:51:25 +0000 | [diff] [blame] | 503 | fname_out = (char*)"index.html"; |
Denis Vlasenko | 818322b | 2007-09-24 18:27:04 +0000 | [diff] [blame] | 504 | /* -P DIR is considered only if there was no -O FILE */ |
| 505 | if (dir_prefix) |
Matt Kraai | 0382eb8 | 2001-07-19 19:13:55 +0000 | [diff] [blame] | 506 | fname_out = concat_path_file(dir_prefix, fname_out); |
Eric Andersen | 29edd00 | 2000-12-09 16:55:35 +0000 | [diff] [blame] | 507 | } |
Denis Vlasenko | 818322b | 2007-09-24 18:27:04 +0000 | [diff] [blame] | 508 | #if ENABLE_FEATURE_WGET_STATUSBAR |
| 509 | curfile = bb_get_last_path_component_nostrip(fname_out); |
| 510 | #endif |
| 511 | |
Denis Vlasenko | 4e4662c | 2006-11-23 13:10:23 +0000 | [diff] [blame] | 512 | /* Impossible? |
Denis Vlasenko | a552eeb | 2006-09-26 09:22:12 +0000 | [diff] [blame] | 513 | if ((opt & WGET_OPT_CONTINUE) && !fname_out) |
Denis Vlasenko | 4e4662c | 2006-11-23 13:10:23 +0000 | [diff] [blame] | 514 | 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] | 515 | |
Bernhard Reutner-Fischer | 7e8a53a | 2007-04-10 09:37:29 +0000 | [diff] [blame] | 516 | /* Determine where to start transfer */ |
Denis Vlasenko | 9f73944 | 2006-12-16 23:49:13 +0000 | [diff] [blame] | 517 | if (LONE_DASH(fname_out)) { |
Denis Vlasenko | a94554d | 2006-09-23 17:49:09 +0000 | [diff] [blame] | 518 | output_fd = 1; |
Denis Vlasenko | a552eeb | 2006-09-26 09:22:12 +0000 | [diff] [blame] | 519 | opt &= ~WGET_OPT_CONTINUE; |
Denis Vlasenko | 4e4662c | 2006-11-23 13:10:23 +0000 | [diff] [blame] | 520 | } |
| 521 | if (opt & WGET_OPT_CONTINUE) { |
Denis Vlasenko | 7039a66 | 2006-10-08 17:54:47 +0000 | [diff] [blame] | 522 | output_fd = open(fname_out, O_WRONLY); |
Denis Vlasenko | a94554d | 2006-09-23 17:49:09 +0000 | [diff] [blame] | 523 | if (output_fd >= 0) { |
Denis Vlasenko | ea62077 | 2006-10-14 02:23:43 +0000 | [diff] [blame] | 524 | beg_range = xlseek(output_fd, 0, SEEK_END); |
Denis Vlasenko | a94554d | 2006-09-23 17:49:09 +0000 | [diff] [blame] | 525 | } |
| 526 | /* File doesn't exist. We do not create file here yet. |
| 527 | We are not sure it exists on remove side */ |
Eric Andersen | 9670083 | 2000-09-04 15:15:55 +0000 | [diff] [blame] | 528 | } |
| 529 | |
Eric Andersen | e6dc439 | 2003-10-31 09:31:46 +0000 | [diff] [blame] | 530 | /* We want to do exactly _one_ DNS lookup, since some |
| 531 | * sites (i.e. ftp.us.debian.org) use round-robin DNS |
| 532 | * and we want to connect to only one IP... */ |
Denis Vlasenko | 42823d5 | 2007-02-04 02:39:08 +0000 | [diff] [blame] | 533 | lsa = xhost2sockaddr(server.host, server.port); |
Denis Vlasenko | a552eeb | 2006-09-26 09:22:12 +0000 | [diff] [blame] | 534 | if (!(opt & WGET_OPT_QUIET)) { |
Denis Vlasenko | 85629f0 | 2007-01-22 09:36:41 +0000 | [diff] [blame] | 535 | fprintf(stderr, "Connecting to %s (%s)\n", server.host, |
Denis Vlasenko | a27a11b | 2007-08-18 14:16:39 +0000 | [diff] [blame] | 536 | xmalloc_sockaddr2dotted(&lsa->sa)); |
Denis Vlasenko | 85629f0 | 2007-01-22 09:36:41 +0000 | [diff] [blame] | 537 | /* We leak result of xmalloc_sockaddr2dotted */ |
Eric Andersen | e6dc439 | 2003-10-31 09:31:46 +0000 | [diff] [blame] | 538 | } |
| 539 | |
Glenn L McGrath | f1c4b11 | 2004-02-22 00:27:34 +0000 | [diff] [blame] | 540 | if (use_proxy || !target.is_ftp) { |
Eric Andersen | 79757c9 | 2001-04-05 21:45:54 +0000 | [diff] [blame] | 541 | /* |
| 542 | * HTTP session |
| 543 | */ |
| 544 | do { |
Denis Vlasenko | a7ce207 | 2007-09-24 18:25:08 +0000 | [diff] [blame] | 545 | got_clen = 0; |
| 546 | chunked = 0; |
Glenn L McGrath | e7bdfcc | 2003-08-28 22:03:19 +0000 | [diff] [blame] | 547 | |
Denis Vlasenko | 3526a13 | 2006-09-09 12:20:57 +0000 | [diff] [blame] | 548 | if (!--try) |
Denis Vlasenko | a94554d | 2006-09-23 17:49:09 +0000 | [diff] [blame] | 549 | bb_error_msg_and_die("too many redirections"); |
Eric Andersen | 7d69701 | 2001-01-24 20:28:35 +0000 | [diff] [blame] | 550 | |
Bernhard Reutner-Fischer | 7e8a53a | 2007-04-10 09:37:29 +0000 | [diff] [blame] | 551 | /* Open socket to http server */ |
Eric Andersen | 79757c9 | 2001-04-05 21:45:54 +0000 | [diff] [blame] | 552 | if (sfp) fclose(sfp); |
Denis Vlasenko | 6536a9b | 2007-01-12 10:35:23 +0000 | [diff] [blame] | 553 | sfp = open_socket(lsa); |
Eric Andersen | c7bda1c | 2004-03-15 08:29:22 +0000 | [diff] [blame] | 554 | |
Bernhard Reutner-Fischer | 7e8a53a | 2007-04-10 09:37:29 +0000 | [diff] [blame] | 555 | /* Send HTTP request. */ |
Glenn L McGrath | f1c4b11 | 2004-02-22 00:27:34 +0000 | [diff] [blame] | 556 | if (use_proxy) { |
Denis Vlasenko | 6536a9b | 2007-01-12 10:35:23 +0000 | [diff] [blame] | 557 | fprintf(sfp, "GET %stp://%s/%s HTTP/1.1\r\n", |
Eric Andersen | 79757c9 | 2001-04-05 21:45:54 +0000 | [diff] [blame] | 558 | target.is_ftp ? "f" : "ht", target.host, |
Denis Vlasenko | 6536a9b | 2007-01-12 10:35:23 +0000 | [diff] [blame] | 559 | target.path); |
Eric Andersen | 79757c9 | 2001-04-05 21:45:54 +0000 | [diff] [blame] | 560 | } else { |
Eric Andersen | 6d7fa43 | 2001-04-10 18:17:05 +0000 | [diff] [blame] | 561 | fprintf(sfp, "GET /%s HTTP/1.1\r\n", target.path); |
Eric Andersen | 79757c9 | 2001-04-05 21:45:54 +0000 | [diff] [blame] | 562 | } |
Eric Andersen | 9670083 | 2000-09-04 15:15:55 +0000 | [diff] [blame] | 563 | |
Denis Vlasenko | 6536a9b | 2007-01-12 10:35:23 +0000 | [diff] [blame] | 564 | fprintf(sfp, "Host: %s\r\nUser-Agent: %s\r\n", |
| 565 | target.host, user_agent); |
Eric Andersen | 79757c9 | 2001-04-05 21:45:54 +0000 | [diff] [blame] | 566 | |
Denis Vlasenko | 9cade08 | 2006-11-21 10:43:02 +0000 | [diff] [blame] | 567 | #if ENABLE_FEATURE_WGET_AUTHENTICATION |
Eric Andersen | 79757c9 | 2001-04-05 21:45:54 +0000 | [diff] [blame] | 568 | if (target.user) { |
Denis Vlasenko | 12d2129 | 2007-06-27 21:40:07 +0000 | [diff] [blame] | 569 | fprintf(sfp, "Proxy-Authorization: Basic %s\r\n"+6, |
| 570 | base64enc_512(buf, target.user)); |
Eric Andersen | 79757c9 | 2001-04-05 21:45:54 +0000 | [diff] [blame] | 571 | } |
Glenn L McGrath | f1c4b11 | 2004-02-22 00:27:34 +0000 | [diff] [blame] | 572 | if (use_proxy && server.user) { |
Eric Andersen | 79757c9 | 2001-04-05 21:45:54 +0000 | [diff] [blame] | 573 | fprintf(sfp, "Proxy-Authorization: Basic %s\r\n", |
Denis Vlasenko | 12d2129 | 2007-06-27 21:40:07 +0000 | [diff] [blame] | 574 | base64enc_512(buf, server.user)); |
Eric Andersen | 79757c9 | 2001-04-05 21:45:54 +0000 | [diff] [blame] | 575 | } |
| 576 | #endif |
| 577 | |
Denis Vlasenko | a94554d | 2006-09-23 17:49:09 +0000 | [diff] [blame] | 578 | if (beg_range) |
Denis Vlasenko | cf30cc8 | 2006-11-24 14:53:18 +0000 | [diff] [blame] | 579 | fprintf(sfp, "Range: bytes=%"OFF_FMT"d-\r\n", beg_range); |
Denis Vlasenko | c8400a2 | 2006-10-25 00:33:44 +0000 | [diff] [blame] | 580 | #if ENABLE_FEATURE_WGET_LONG_OPTIONS |
Denis Vlasenko | 7534e08 | 2006-10-23 23:21:58 +0000 | [diff] [blame] | 581 | if (extra_headers) |
| 582 | fputs(extra_headers, sfp); |
Denis Vlasenko | c8400a2 | 2006-10-25 00:33:44 +0000 | [diff] [blame] | 583 | #endif |
Denis Vlasenko | 7534e08 | 2006-10-23 23:21:58 +0000 | [diff] [blame] | 584 | fprintf(sfp, "Connection: close\r\n\r\n"); |
Eric Andersen | 79757c9 | 2001-04-05 21:45:54 +0000 | [diff] [blame] | 585 | |
| 586 | /* |
Tim Riker | c1ef7bd | 2006-01-25 00:08:53 +0000 | [diff] [blame] | 587 | * Retrieve HTTP response line and check for "200" status code. |
| 588 | */ |
Denis Vlasenko | 023b57d | 2006-10-15 17:05:55 +0000 | [diff] [blame] | 589 | read_response: |
Glenn L McGrath | e7bdfcc | 2003-08-28 22:03:19 +0000 | [diff] [blame] | 590 | if (fgets(buf, sizeof(buf), sfp) == NULL) |
Denis Vlasenko | a94554d | 2006-09-23 17:49:09 +0000 | [diff] [blame] | 591 | bb_error_msg_and_die("no response from server"); |
Eric Andersen | c7bda1c | 2004-03-15 08:29:22 +0000 | [diff] [blame] | 592 | |
Bernhard Reutner-Fischer | 7e8a53a | 2007-04-10 09:37:29 +0000 | [diff] [blame] | 593 | str = buf; |
| 594 | str = skip_non_whitespace(str); |
| 595 | str = skip_whitespace(str); |
Denis Vlasenko | 023b57d | 2006-10-15 17:05:55 +0000 | [diff] [blame] | 596 | // FIXME: no error check |
| 597 | // xatou wouldn't work: "200 OK" |
Bernhard Reutner-Fischer | 7e8a53a | 2007-04-10 09:37:29 +0000 | [diff] [blame] | 598 | status = atoi(str); |
Denis Vlasenko | 023b57d | 2006-10-15 17:05:55 +0000 | [diff] [blame] | 599 | switch (status) { |
| 600 | case 0: |
| 601 | case 100: |
Denis Vlasenko | 06783a5 | 2007-09-24 13:51:54 +0000 | [diff] [blame] | 602 | while (gethdr(buf, sizeof(buf), sfp /*, &n*/) != NULL) |
Denis Vlasenko | 023b57d | 2006-10-15 17:05:55 +0000 | [diff] [blame] | 603 | /* eat all remaining headers */; |
| 604 | goto read_response; |
| 605 | case 200: |
| 606 | break; |
| 607 | case 300: /* redirection */ |
| 608 | case 301: |
| 609 | case 302: |
| 610 | case 303: |
| 611 | break; |
| 612 | case 206: |
| 613 | if (beg_range) |
Eric Andersen | 79757c9 | 2001-04-05 21:45:54 +0000 | [diff] [blame] | 614 | break; |
Denis Vlasenko | 023b57d | 2006-10-15 17:05:55 +0000 | [diff] [blame] | 615 | /*FALLTHRU*/ |
| 616 | default: |
Denis Vlasenko | 067e3f0 | 2006-11-10 23:25:53 +0000 | [diff] [blame] | 617 | /* Show first line only and kill any ESC tricks */ |
| 618 | buf[strcspn(buf, "\n\r\x1b")] = '\0'; |
| 619 | bb_error_msg_and_die("server returned error: %s", buf); |
Eric Andersen | 79757c9 | 2001-04-05 21:45:54 +0000 | [diff] [blame] | 620 | } |
Eric Andersen | c7bda1c | 2004-03-15 08:29:22 +0000 | [diff] [blame] | 621 | |
Eric Andersen | 79757c9 | 2001-04-05 21:45:54 +0000 | [diff] [blame] | 622 | /* |
| 623 | * Retrieve HTTP headers. |
| 624 | */ |
Denis Vlasenko | 06783a5 | 2007-09-24 13:51:54 +0000 | [diff] [blame] | 625 | while ((str = gethdr(buf, sizeof(buf), sfp /*, &n*/)) != NULL) { |
Bernhard Reutner-Fischer | 7e8a53a | 2007-04-10 09:37:29 +0000 | [diff] [blame] | 626 | /* gethdr did already convert the "FOO:" string to lowercase */ |
Denis Vlasenko | 990d0f6 | 2007-07-24 15:54:42 +0000 | [diff] [blame] | 627 | smalluint key = index_in_strings(keywords, *&buf) + 1; |
Bernhard Reutner-Fischer | 7e8a53a | 2007-04-10 09:37:29 +0000 | [diff] [blame] | 628 | if (key == KEY_content_length) { |
| 629 | content_len = BB_STRTOOFF(str, NULL, 10); |
Denis Vlasenko | d686a04 | 2006-11-27 14:43:21 +0000 | [diff] [blame] | 630 | if (errno || content_len < 0) { |
Bernhard Reutner-Fischer | 7e8a53a | 2007-04-10 09:37:29 +0000 | [diff] [blame] | 631 | bb_error_msg_and_die("content-length %s is garbage", str); |
Eric Andersen | 2479445 | 2004-03-06 22:11:45 +0000 | [diff] [blame] | 632 | } |
Eric Andersen | 79757c9 | 2001-04-05 21:45:54 +0000 | [diff] [blame] | 633 | got_clen = 1; |
| 634 | continue; |
| 635 | } |
Bernhard Reutner-Fischer | 7e8a53a | 2007-04-10 09:37:29 +0000 | [diff] [blame] | 636 | if (key == KEY_transfer_encoding) { |
Denis Vlasenko | 990d0f6 | 2007-07-24 15:54:42 +0000 | [diff] [blame] | 637 | if (index_in_strings(keywords, str_tolower(str)) + 1 != KEY_chunked) |
Denis Vlasenko | 7710563 | 2007-09-24 15:04:00 +0000 | [diff] [blame] | 638 | bb_error_msg_and_die("transfer encoding '%s' is not supported", str); |
Denis Vlasenko | 023b57d | 2006-10-15 17:05:55 +0000 | [diff] [blame] | 639 | chunked = got_clen = 1; |
Eric Andersen | 6d7fa43 | 2001-04-10 18:17:05 +0000 | [diff] [blame] | 640 | } |
Bernhard Reutner-Fischer | 7e8a53a | 2007-04-10 09:37:29 +0000 | [diff] [blame] | 641 | if (key == KEY_location) { |
| 642 | if (str[0] == '/') |
Denis Vlasenko | 96e9d3c | 2006-10-07 14:28:55 +0000 | [diff] [blame] | 643 | /* free(target.allocated); */ |
Bernhard Reutner-Fischer | 7e8a53a | 2007-04-10 09:37:29 +0000 | [diff] [blame] | 644 | target.path = /* target.allocated = */ xstrdup(str+1); |
Eric Andersen | 79757c9 | 2001-04-05 21:45:54 +0000 | [diff] [blame] | 645 | else { |
Bernhard Reutner-Fischer | 7e8a53a | 2007-04-10 09:37:29 +0000 | [diff] [blame] | 646 | parse_url(str, &target); |
Glenn L McGrath | f1c4b11 | 2004-02-22 00:27:34 +0000 | [diff] [blame] | 647 | if (use_proxy == 0) { |
Eric Andersen | 79757c9 | 2001-04-05 21:45:54 +0000 | [diff] [blame] | 648 | server.host = target.host; |
| 649 | server.port = target.port; |
| 650 | } |
Denis Vlasenko | 6536a9b | 2007-01-12 10:35:23 +0000 | [diff] [blame] | 651 | free(lsa); |
Denis Vlasenko | 42823d5 | 2007-02-04 02:39:08 +0000 | [diff] [blame] | 652 | lsa = xhost2sockaddr(server.host, server.port); |
Glenn L McGrath | 58a2e0e | 2004-01-17 23:07:14 +0000 | [diff] [blame] | 653 | break; |
Eric Andersen | 79757c9 | 2001-04-05 21:45:54 +0000 | [diff] [blame] | 654 | } |
| 655 | } |
| 656 | } |
Denis Vlasenko | 3469c18 | 2006-12-16 22:19:47 +0000 | [diff] [blame] | 657 | } while (status >= 300); |
Eric Andersen | c7bda1c | 2004-03-15 08:29:22 +0000 | [diff] [blame] | 658 | |
Eric Andersen | 79757c9 | 2001-04-05 21:45:54 +0000 | [diff] [blame] | 659 | dfp = sfp; |
Denis Vlasenko | 96e9d3c | 2006-10-07 14:28:55 +0000 | [diff] [blame] | 660 | |
| 661 | } else { |
| 662 | |
Eric Andersen | 79757c9 | 2001-04-05 21:45:54 +0000 | [diff] [blame] | 663 | /* |
| 664 | * FTP session |
| 665 | */ |
Denis Vlasenko | 3526a13 | 2006-09-09 12:20:57 +0000 | [diff] [blame] | 666 | if (!target.user) |
Rob Landley | d921b2e | 2006-08-03 15:41:12 +0000 | [diff] [blame] | 667 | target.user = xstrdup("anonymous:busybox@"); |
Eric Andersen | 79757c9 | 2001-04-05 21:45:54 +0000 | [diff] [blame] | 668 | |
Denis Vlasenko | 6536a9b | 2007-01-12 10:35:23 +0000 | [diff] [blame] | 669 | sfp = open_socket(lsa); |
Eric Andersen | 79757c9 | 2001-04-05 21:45:54 +0000 | [diff] [blame] | 670 | if (ftpcmd(NULL, NULL, sfp, buf) != 220) |
Denis Vlasenko | a94554d | 2006-09-23 17:49:09 +0000 | [diff] [blame] | 671 | bb_error_msg_and_die("%s", buf+4); |
Eric Andersen | 79757c9 | 2001-04-05 21:45:54 +0000 | [diff] [blame] | 672 | |
Eric Andersen | c7bda1c | 2004-03-15 08:29:22 +0000 | [diff] [blame] | 673 | /* |
Eric Andersen | 79757c9 | 2001-04-05 21:45:54 +0000 | [diff] [blame] | 674 | * Splitting username:password pair, |
| 675 | * trying to log in |
| 676 | */ |
Bernhard Reutner-Fischer | 7e8a53a | 2007-04-10 09:37:29 +0000 | [diff] [blame] | 677 | str = strchr(target.user, ':'); |
| 678 | if (str) |
| 679 | *(str++) = '\0'; |
Denis Vlasenko | c16bd21 | 2006-09-27 19:51:06 +0000 | [diff] [blame] | 680 | switch (ftpcmd("USER ", target.user, sfp, buf)) { |
Denis Vlasenko | 023b57d | 2006-10-15 17:05:55 +0000 | [diff] [blame] | 681 | case 230: |
| 682 | break; |
| 683 | case 331: |
Bernhard Reutner-Fischer | 7e8a53a | 2007-04-10 09:37:29 +0000 | [diff] [blame] | 684 | if (ftpcmd("PASS ", str, sfp, buf) == 230) |
Eric Andersen | b520e08 | 2000-10-03 00:21:45 +0000 | [diff] [blame] | 685 | break; |
Denis Vlasenko | 023b57d | 2006-10-15 17:05:55 +0000 | [diff] [blame] | 686 | /* FALLTHRU (failed login) */ |
| 687 | default: |
| 688 | bb_error_msg_and_die("ftp login: %s", buf+4); |
Eric Andersen | 79757c9 | 2001-04-05 21:45:54 +0000 | [diff] [blame] | 689 | } |
Eric Andersen | c7bda1c | 2004-03-15 08:29:22 +0000 | [diff] [blame] | 690 | |
Eric Andersen | 79757c9 | 2001-04-05 21:45:54 +0000 | [diff] [blame] | 691 | ftpcmd("TYPE I", NULL, sfp, buf); |
Eric Andersen | c7bda1c | 2004-03-15 08:29:22 +0000 | [diff] [blame] | 692 | |
Eric Andersen | 79757c9 | 2001-04-05 21:45:54 +0000 | [diff] [blame] | 693 | /* |
| 694 | * Querying file size |
| 695 | */ |
Rob Landley | af12cb3 | 2006-06-27 18:41:03 +0000 | [diff] [blame] | 696 | if (ftpcmd("SIZE ", target.path, sfp, buf) == 213) { |
Denis Vlasenko | d686a04 | 2006-11-27 14:43:21 +0000 | [diff] [blame] | 697 | content_len = BB_STRTOOFF(buf+4, NULL, 10); |
| 698 | if (errno || content_len < 0) { |
Denis Vlasenko | a94554d | 2006-09-23 17:49:09 +0000 | [diff] [blame] | 699 | bb_error_msg_and_die("SIZE value is garbage"); |
Eric Andersen | 2479445 | 2004-03-06 22:11:45 +0000 | [diff] [blame] | 700 | } |
Eric Andersen | 9670083 | 2000-09-04 15:15:55 +0000 | [diff] [blame] | 701 | got_clen = 1; |
Eric Andersen | 9670083 | 2000-09-04 15:15:55 +0000 | [diff] [blame] | 702 | } |
Eric Andersen | c7bda1c | 2004-03-15 08:29:22 +0000 | [diff] [blame] | 703 | |
Eric Andersen | 79757c9 | 2001-04-05 21:45:54 +0000 | [diff] [blame] | 704 | /* |
| 705 | * Entering passive mode |
| 706 | */ |
Denis Vlasenko | 023b57d | 2006-10-15 17:05:55 +0000 | [diff] [blame] | 707 | if (ftpcmd("PASV", NULL, sfp, buf) != 227) { |
| 708 | pasv_error: |
| 709 | bb_error_msg_and_die("bad response to %s: %s", "PASV", buf); |
| 710 | } |
Denis Vlasenko | f8c8bb1 | 2006-11-21 19:10:26 +0000 | [diff] [blame] | 711 | // Response is "227 garbageN1,N2,N3,N4,P1,P2[)garbage] |
Denis Vlasenko | 023b57d | 2006-10-15 17:05:55 +0000 | [diff] [blame] | 712 | // Server's IP is N1.N2.N3.N4 (we ignore it) |
| 713 | // Server's port for data connection is P1*256+P2 |
Bernhard Reutner-Fischer | 7e8a53a | 2007-04-10 09:37:29 +0000 | [diff] [blame] | 714 | str = strrchr(buf, ')'); |
| 715 | if (str) str[0] = '\0'; |
| 716 | str = strrchr(buf, ','); |
| 717 | if (!str) goto pasv_error; |
| 718 | port = xatou_range(str+1, 0, 255); |
| 719 | *str = '\0'; |
| 720 | str = strrchr(buf, ','); |
| 721 | if (!str) goto pasv_error; |
| 722 | port += xatou_range(str+1, 0, 255) * 256; |
Denis Vlasenko | 5d68724 | 2007-01-12 20:59:31 +0000 | [diff] [blame] | 723 | set_nport(lsa, htons(port)); |
Denis Vlasenko | 6536a9b | 2007-01-12 10:35:23 +0000 | [diff] [blame] | 724 | dfp = open_socket(lsa); |
Eric Andersen | 79757c9 | 2001-04-05 21:45:54 +0000 | [diff] [blame] | 725 | |
Denis Vlasenko | a94554d | 2006-09-23 17:49:09 +0000 | [diff] [blame] | 726 | if (beg_range) { |
Denis Vlasenko | cf30cc8 | 2006-11-24 14:53:18 +0000 | [diff] [blame] | 727 | sprintf(buf, "REST %"OFF_FMT"d", beg_range); |
Denis Vlasenko | a94554d | 2006-09-23 17:49:09 +0000 | [diff] [blame] | 728 | if (ftpcmd(buf, NULL, sfp, buf) == 350) |
Denis Vlasenko | f8aa109 | 2006-10-01 10:58:54 +0000 | [diff] [blame] | 729 | content_len -= beg_range; |
Eric Andersen | 9670083 | 2000-09-04 15:15:55 +0000 | [diff] [blame] | 730 | } |
Eric Andersen | c7bda1c | 2004-03-15 08:29:22 +0000 | [diff] [blame] | 731 | |
Rob Landley | af12cb3 | 2006-06-27 18:41:03 +0000 | [diff] [blame] | 732 | if (ftpcmd("RETR ", target.path, sfp, buf) > 150) |
Denis Vlasenko | 7710563 | 2007-09-24 15:04:00 +0000 | [diff] [blame] | 733 | bb_error_msg_and_die("bad response to %s: %s", "RETR", buf); |
Eric Andersen | 9670083 | 2000-09-04 15:15:55 +0000 | [diff] [blame] | 734 | } |
Denis Vlasenko | 7710563 | 2007-09-24 15:04:00 +0000 | [diff] [blame] | 735 | |
Bernhard Reutner-Fischer | 2e75dcc | 2007-04-05 10:31:47 +0000 | [diff] [blame] | 736 | if (opt & WGET_OPT_SPIDER) { |
| 737 | if (ENABLE_FEATURE_CLEAN_UP) |
| 738 | fclose(sfp); |
Denis Vlasenko | 7710563 | 2007-09-24 15:04:00 +0000 | [diff] [blame] | 739 | return EXIT_SUCCESS; |
Bernhard Reutner-Fischer | 2e75dcc | 2007-04-05 10:31:47 +0000 | [diff] [blame] | 740 | } |
Eric Andersen | 79757c9 | 2001-04-05 21:45:54 +0000 | [diff] [blame] | 741 | |
Eric Andersen | 9670083 | 2000-09-04 15:15:55 +0000 | [diff] [blame] | 742 | /* |
Eric Andersen | 79757c9 | 2001-04-05 21:45:54 +0000 | [diff] [blame] | 743 | * Retrieve file |
Eric Andersen | 9670083 | 2000-09-04 15:15:55 +0000 | [diff] [blame] | 744 | */ |
Rob Landley | 19a3940 | 2006-06-13 17:10:26 +0000 | [diff] [blame] | 745 | |
Denis Vlasenko | f8aa109 | 2006-10-01 10:58:54 +0000 | [diff] [blame] | 746 | /* Do it before progressmeter (want to have nice error message) */ |
| 747 | if (output_fd < 0) |
Denis Vlasenko | cf749bc | 2006-11-26 15:45:17 +0000 | [diff] [blame] | 748 | output_fd = xopen(fname_out, |
| 749 | O_WRONLY|O_CREAT|O_EXCL|O_TRUNC); |
Denis Vlasenko | f8aa109 | 2006-10-01 10:58:54 +0000 | [diff] [blame] | 750 | |
Denis Vlasenko | a552eeb | 2006-09-26 09:22:12 +0000 | [diff] [blame] | 751 | if (!(opt & WGET_OPT_QUIET)) |
Glenn L McGrath | 1bca5ed | 2000-12-09 08:12:06 +0000 | [diff] [blame] | 752 | progressmeter(-1); |
Rob Landley | 19a3940 | 2006-06-13 17:10:26 +0000 | [diff] [blame] | 753 | |
Denis Vlasenko | 7710563 | 2007-09-24 15:04:00 +0000 | [diff] [blame] | 754 | if (chunked) |
| 755 | goto get_clen; |
| 756 | |
Denis Vlasenko | 06783a5 | 2007-09-24 13:51:54 +0000 | [diff] [blame] | 757 | /* Loops only if chunked */ |
| 758 | while (1) { |
Denis Vlasenko | f8aa109 | 2006-10-01 10:58:54 +0000 | [diff] [blame] | 759 | while (content_len > 0 || !got_clen) { |
Denis Vlasenko | 06783a5 | 2007-09-24 13:51:54 +0000 | [diff] [blame] | 760 | int n; |
Denis Vlasenko | 3526a13 | 2006-09-09 12:20:57 +0000 | [diff] [blame] | 761 | unsigned rdsz = sizeof(buf); |
Denis Vlasenko | 06783a5 | 2007-09-24 13:51:54 +0000 | [diff] [blame] | 762 | |
Denis Vlasenko | f8aa109 | 2006-10-01 10:58:54 +0000 | [diff] [blame] | 763 | if (content_len < sizeof(buf) && (chunked || got_clen)) |
| 764 | rdsz = (unsigned)content_len; |
Denis Vlasenko | 12d2129 | 2007-06-27 21:40:07 +0000 | [diff] [blame] | 765 | n = safe_fread(buf, rdsz, dfp); |
Denis Vlasenko | 06783a5 | 2007-09-24 13:51:54 +0000 | [diff] [blame] | 766 | if (n <= 0) { |
| 767 | if (ferror(dfp)) { |
| 768 | /* perror will not work: ferror doesn't set errno */ |
| 769 | bb_error_msg_and_die(bb_msg_read_error); |
| 770 | } |
Denis Vlasenko | 3526a13 | 2006-09-09 12:20:57 +0000 | [diff] [blame] | 771 | break; |
Glenn L McGrath | 83e4a5b | 2003-08-28 21:55:22 +0000 | [diff] [blame] | 772 | } |
Denis Vlasenko | 06783a5 | 2007-09-24 13:51:54 +0000 | [diff] [blame] | 773 | xwrite(output_fd, buf, n); |
Denis Vlasenko | 9cade08 | 2006-11-21 10:43:02 +0000 | [diff] [blame] | 774 | #if ENABLE_FEATURE_WGET_STATUSBAR |
Rob Landley | 19a3940 | 2006-06-13 17:10:26 +0000 | [diff] [blame] | 775 | transferred += n; |
Eric Andersen | b520e08 | 2000-10-03 00:21:45 +0000 | [diff] [blame] | 776 | #endif |
Denis Vlasenko | 06783a5 | 2007-09-24 13:51:54 +0000 | [diff] [blame] | 777 | if (got_clen) |
Denis Vlasenko | f8aa109 | 2006-10-01 10:58:54 +0000 | [diff] [blame] | 778 | content_len -= n; |
Glenn L McGrath | 83e4a5b | 2003-08-28 21:55:22 +0000 | [diff] [blame] | 779 | } |
Eric Andersen | 79757c9 | 2001-04-05 21:45:54 +0000 | [diff] [blame] | 780 | |
Denis Vlasenko | 06783a5 | 2007-09-24 13:51:54 +0000 | [diff] [blame] | 781 | if (!chunked) |
| 782 | break; |
Eric Andersen | 6d7fa43 | 2001-04-10 18:17:05 +0000 | [diff] [blame] | 783 | |
Denis Vlasenko | 06783a5 | 2007-09-24 13:51:54 +0000 | [diff] [blame] | 784 | safe_fgets(buf, sizeof(buf), dfp); /* This is a newline */ |
Denis Vlasenko | 7710563 | 2007-09-24 15:04:00 +0000 | [diff] [blame] | 785 | get_clen: |
Denis Vlasenko | 06783a5 | 2007-09-24 13:51:54 +0000 | [diff] [blame] | 786 | safe_fgets(buf, sizeof(buf), dfp); |
| 787 | content_len = STRTOOFF(buf, NULL, 16); |
| 788 | /* FIXME: error check? */ |
| 789 | if (content_len == 0) |
| 790 | break; /* all done! */ |
| 791 | } |
Rob Landley | 19a3940 | 2006-06-13 17:10:26 +0000 | [diff] [blame] | 792 | |
Denis Vlasenko | a552eeb | 2006-09-26 09:22:12 +0000 | [diff] [blame] | 793 | if (!(opt & WGET_OPT_QUIET)) |
Denis Vlasenko | a7ce207 | 2007-09-24 18:25:08 +0000 | [diff] [blame] | 794 | progressmeter(0); |
Rob Landley | 19a3940 | 2006-06-13 17:10:26 +0000 | [diff] [blame] | 795 | |
Glenn L McGrath | f1c4b11 | 2004-02-22 00:27:34 +0000 | [diff] [blame] | 796 | if ((use_proxy == 0) && target.is_ftp) { |
Eric Andersen | 79757c9 | 2001-04-05 21:45:54 +0000 | [diff] [blame] | 797 | fclose(dfp); |
| 798 | if (ftpcmd(NULL, NULL, sfp, buf) != 226) |
Manuel Novoa III | cad5364 | 2003-03-19 09:13:01 +0000 | [diff] [blame] | 799 | bb_error_msg_and_die("ftp error: %s", buf+4); |
Eric Andersen | 79757c9 | 2001-04-05 21:45:54 +0000 | [diff] [blame] | 800 | ftpcmd("QUIT", NULL, sfp, buf); |
| 801 | } |
Denis Vlasenko | 7710563 | 2007-09-24 15:04:00 +0000 | [diff] [blame] | 802 | |
| 803 | return EXIT_SUCCESS; |
Eric Andersen | 9670083 | 2000-09-04 15:15:55 +0000 | [diff] [blame] | 804 | } |