1 | /* vi: set sw=4 ts=4: */
|
---|
2 | /*
|
---|
3 | * wget - retrieve a file using HTTP or FTP
|
---|
4 | *
|
---|
5 | * Chip Rosenthal Covad Communications <chip@laserlink.net>
|
---|
6 | * Licensed under GPLv2, see file LICENSE in this source tree.
|
---|
7 | *
|
---|
8 | * Copyright (C) 2010 Bradley M. Kuhn <bkuhn@ebb.org>
|
---|
9 | * Kuhn's copyrights are licensed GPLv2-or-later. File as a whole remains GPLv2.
|
---|
10 | */
|
---|
11 |
|
---|
12 | //config:config WGET
|
---|
13 | //config: bool "wget"
|
---|
14 | //config: default y
|
---|
15 | //config: help
|
---|
16 | //config: wget is a utility for non-interactive download of files from HTTP
|
---|
17 | //config: and FTP servers.
|
---|
18 | //config:
|
---|
19 | //config:config FEATURE_WGET_STATUSBAR
|
---|
20 | //config: bool "Enable a nifty process meter (+2k)"
|
---|
21 | //config: default y
|
---|
22 | //config: depends on WGET
|
---|
23 | //config: help
|
---|
24 | //config: Enable the transfer progress bar for wget transfers.
|
---|
25 | //config:
|
---|
26 | //config:config FEATURE_WGET_AUTHENTICATION
|
---|
27 | //config: bool "Enable HTTP authentication"
|
---|
28 | //config: default y
|
---|
29 | //config: depends on WGET
|
---|
30 | //config: help
|
---|
31 | //config: Support authenticated HTTP transfers.
|
---|
32 | //config:
|
---|
33 | //config:config FEATURE_WGET_LONG_OPTIONS
|
---|
34 | //config: bool "Enable long options"
|
---|
35 | //config: default y
|
---|
36 | //config: depends on WGET && LONG_OPTS
|
---|
37 | //config: help
|
---|
38 | //config: Support long options for the wget applet.
|
---|
39 | //config:
|
---|
40 | //config:config FEATURE_WGET_TIMEOUT
|
---|
41 | //config: bool "Enable timeout option -T SEC"
|
---|
42 | //config: default y
|
---|
43 | //config: depends on WGET
|
---|
44 | //config: help
|
---|
45 | //config: Supports network read and connect timeouts for wget,
|
---|
46 | //config: so that wget will give up and timeout, through the -T
|
---|
47 | //config: command line option.
|
---|
48 | //config:
|
---|
49 | //config: Currently only connect and network data read timeout are
|
---|
50 | //config: supported (i.e., timeout is not applied to the DNS query). When
|
---|
51 | //config: FEATURE_WGET_LONG_OPTIONS is also enabled, the --timeout option
|
---|
52 | //config: will work in addition to -T.
|
---|
53 | //config:
|
---|
54 | //config:config FEATURE_WGET_OPENSSL
|
---|
55 | //config: bool "Try to connect to HTTPS using openssl"
|
---|
56 | //config: default y
|
---|
57 | //config: depends on WGET
|
---|
58 | //config: help
|
---|
59 | //config: Choose how wget establishes SSL connection for https:// URLs.
|
---|
60 | //config:
|
---|
61 | //config: Busybox itself contains no SSL code. wget will spawn
|
---|
62 | //config: a helper program to talk over HTTPS.
|
---|
63 | //config:
|
---|
64 | //config: OpenSSL has a simple SSL client for debug purposes.
|
---|
65 | //config: If you select "openssl" helper, wget will effectively call
|
---|
66 | //config: "openssl s_client -quiet -connect IP:443 2>/dev/null"
|
---|
67 | //config: and pipe its data through it.
|
---|
68 | //config: Note inconvenient API: host resolution is done twice,
|
---|
69 | //config: and there is no guarantee openssl's idea of IPv6 address
|
---|
70 | //config: format is the same as ours.
|
---|
71 | //config: Another problem is that s_client prints debug information
|
---|
72 | //config: to stderr, and it needs to be suppressed. This means
|
---|
73 | //config: all error messages get suppressed too.
|
---|
74 | //config: openssl is also a big binary, often dynamically linked
|
---|
75 | //config: against ~15 libraries.
|
---|
76 | //config:
|
---|
77 | //config:config FEATURE_WGET_SSL_HELPER
|
---|
78 | //config: bool "Try to connect to HTTPS using ssl_helper"
|
---|
79 | //config: default y
|
---|
80 | //config: depends on WGET
|
---|
81 | //config: help
|
---|
82 | //config: Choose how wget establishes SSL connection for https:// URLs.
|
---|
83 | //config:
|
---|
84 | //config: Busybox itself contains no SSL code. wget will spawn
|
---|
85 | //config: a helper program to talk over HTTPS.
|
---|
86 | //config:
|
---|
87 | //config: ssl_helper is a tool which can be built statically
|
---|
88 | //config: from busybox sources against a small embedded SSL library.
|
---|
89 | //config: Please see networking/ssl_helper/README.
|
---|
90 | //config: It does not require double host resolution and emits
|
---|
91 | //config: error messages to stderr.
|
---|
92 | //config:
|
---|
93 | //config: Precompiled static binary may be available at
|
---|
94 | //config: http://busybox.net/downloads/binaries/
|
---|
95 |
|
---|
96 | //applet:IF_WGET(APPLET(wget, BB_DIR_USR_BIN, BB_SUID_DROP))
|
---|
97 |
|
---|
98 | //kbuild:lib-$(CONFIG_WGET) += wget.o
|
---|
99 |
|
---|
100 | //usage:#define wget_trivial_usage
|
---|
101 | //usage: IF_FEATURE_WGET_LONG_OPTIONS(
|
---|
102 | //usage: "[-c|--continue] [-s|--spider] [-q|--quiet] [-O|--output-document FILE]\n"
|
---|
103 | //usage: " [--header 'header: value'] [-Y|--proxy on/off] [-P DIR]\n"
|
---|
104 | /* Since we ignore these opts, we don't show them in --help */
|
---|
105 | /* //usage: " [--no-check-certificate] [--no-cache] [--passive-ftp] [-t TRIES]" */
|
---|
106 | /* //usage: " [-nv] [-nc] [-nH] [-np]" */
|
---|
107 | //usage: " [-U|--user-agent AGENT]" IF_FEATURE_WGET_TIMEOUT(" [-T SEC]") " URL..."
|
---|
108 | //usage: )
|
---|
109 | //usage: IF_NOT_FEATURE_WGET_LONG_OPTIONS(
|
---|
110 | //usage: "[-csq] [-O FILE] [-Y on/off] [-P DIR] [-U AGENT]"
|
---|
111 | //usage: IF_FEATURE_WGET_TIMEOUT(" [-T SEC]") " URL..."
|
---|
112 | //usage: )
|
---|
113 | //usage:#define wget_full_usage "\n\n"
|
---|
114 | //usage: "Retrieve files via HTTP or FTP\n"
|
---|
115 | //usage: "\n -s Spider mode - only check file existence"
|
---|
116 | //usage: "\n -c Continue retrieval of aborted transfer"
|
---|
117 | //usage: "\n -q Quiet"
|
---|
118 | //usage: "\n -P DIR Save to DIR (default .)"
|
---|
119 | //usage: IF_FEATURE_WGET_TIMEOUT(
|
---|
120 | //usage: "\n -T SEC Network read timeout is SEC seconds"
|
---|
121 | //usage: )
|
---|
122 | //usage: "\n -O FILE Save to FILE ('-' for stdout)"
|
---|
123 | //usage: "\n -U STR Use STR for User-Agent header"
|
---|
124 | //usage: "\n -Y Use proxy ('on' or 'off')"
|
---|
125 |
|
---|
126 | #include "libbb.h"
|
---|
127 |
|
---|
128 | #if 0
|
---|
129 | # define log_io(...) bb_error_msg(__VA_ARGS__)
|
---|
130 | # define SENDFMT(fp, fmt, ...) \
|
---|
131 | do { \
|
---|
132 | log_io("> " fmt, ##__VA_ARGS__); \
|
---|
133 | fprintf(fp, fmt, ##__VA_ARGS__); \
|
---|
134 | } while (0);
|
---|
135 | #else
|
---|
136 | # define log_io(...) ((void)0)
|
---|
137 | # define SENDFMT(fp, fmt, ...) fprintf(fp, fmt, ##__VA_ARGS__)
|
---|
138 | #endif
|
---|
139 |
|
---|
140 |
|
---|
141 | struct host_info {
|
---|
142 | char *allocated;
|
---|
143 | const char *path;
|
---|
144 | char *user;
|
---|
145 | const char *protocol;
|
---|
146 | char *host;
|
---|
147 | int port;
|
---|
148 | };
|
---|
149 | static const char P_FTP[] ALIGN1 = "ftp";
|
---|
150 | static const char P_HTTP[] ALIGN1 = "http";
|
---|
151 | #if ENABLE_FEATURE_WGET_OPENSSL || ENABLE_FEATURE_WGET_SSL_HELPER
|
---|
152 | static const char P_HTTPS[] ALIGN1 = "https";
|
---|
153 | #endif
|
---|
154 |
|
---|
155 | #if ENABLE_FEATURE_WGET_LONG_OPTIONS
|
---|
156 | /* User-specified headers prevent using our corresponding built-in headers. */
|
---|
157 | enum {
|
---|
158 | HDR_HOST = (1<<0),
|
---|
159 | HDR_USER_AGENT = (1<<1),
|
---|
160 | HDR_RANGE = (1<<2),
|
---|
161 | HDR_AUTH = (1<<3) * ENABLE_FEATURE_WGET_AUTHENTICATION,
|
---|
162 | HDR_PROXY_AUTH = (1<<4) * ENABLE_FEATURE_WGET_AUTHENTICATION,
|
---|
163 | };
|
---|
164 | static const char wget_user_headers[] ALIGN1 =
|
---|
165 | "Host:\0"
|
---|
166 | "User-Agent:\0"
|
---|
167 | "Range:\0"
|
---|
168 | # if ENABLE_FEATURE_WGET_AUTHENTICATION
|
---|
169 | "Authorization:\0"
|
---|
170 | "Proxy-Authorization:\0"
|
---|
171 | # endif
|
---|
172 | ;
|
---|
173 | # define USR_HEADER_HOST (G.user_headers & HDR_HOST)
|
---|
174 | # define USR_HEADER_USER_AGENT (G.user_headers & HDR_USER_AGENT)
|
---|
175 | # define USR_HEADER_RANGE (G.user_headers & HDR_RANGE)
|
---|
176 | # define USR_HEADER_AUTH (G.user_headers & HDR_AUTH)
|
---|
177 | # define USR_HEADER_PROXY_AUTH (G.user_headers & HDR_PROXY_AUTH)
|
---|
178 | #else /* No long options, no user-headers :( */
|
---|
179 | # define USR_HEADER_HOST 0
|
---|
180 | # define USR_HEADER_USER_AGENT 0
|
---|
181 | # define USR_HEADER_RANGE 0
|
---|
182 | # define USR_HEADER_AUTH 0
|
---|
183 | # define USR_HEADER_PROXY_AUTH 0
|
---|
184 | #endif
|
---|
185 |
|
---|
186 | /* Globals */
|
---|
187 | struct globals {
|
---|
188 | off_t content_len; /* Content-length of the file */
|
---|
189 | off_t beg_range; /* Range at which continue begins */
|
---|
190 | #if ENABLE_FEATURE_WGET_STATUSBAR
|
---|
191 | off_t transferred; /* Number of bytes transferred so far */
|
---|
192 | const char *curfile; /* Name of current file being transferred */
|
---|
193 | bb_progress_t pmt;
|
---|
194 | #endif
|
---|
195 | char *dir_prefix;
|
---|
196 | #if ENABLE_FEATURE_WGET_LONG_OPTIONS
|
---|
197 | char *post_data;
|
---|
198 | char *extra_headers;
|
---|
199 | unsigned char user_headers; /* Headers mentioned by the user */
|
---|
200 | #endif
|
---|
201 | char *fname_out; /* where to direct output (-O) */
|
---|
202 | const char *proxy_flag; /* Use proxies if env vars are set */
|
---|
203 | const char *user_agent; /* "User-Agent" header field */
|
---|
204 | #if ENABLE_FEATURE_WGET_TIMEOUT
|
---|
205 | unsigned timeout_seconds;
|
---|
206 | bool die_if_timed_out;
|
---|
207 | #endif
|
---|
208 | int output_fd;
|
---|
209 | int o_flags;
|
---|
210 | smallint chunked; /* chunked transfer encoding */
|
---|
211 | smallint got_clen; /* got content-length: from server */
|
---|
212 | /* Local downloads do benefit from big buffer.
|
---|
213 | * With 512 byte buffer, it was measured to be
|
---|
214 | * an order of magnitude slower than with big one.
|
---|
215 | */
|
---|
216 | uint64_t just_to_align_next_member;
|
---|
217 | char wget_buf[CONFIG_FEATURE_COPYBUF_KB*1024];
|
---|
218 | } FIX_ALIASING;
|
---|
219 | #define G (*ptr_to_globals)
|
---|
220 | #define INIT_G() do { \
|
---|
221 | SET_PTR_TO_GLOBALS(xzalloc(sizeof(G))); \
|
---|
222 | } while (0)
|
---|
223 | #define FINI_G() do { \
|
---|
224 | FREE_PTR_TO_GLOBALS(); \
|
---|
225 | } while (0)
|
---|
226 |
|
---|
227 |
|
---|
228 | /* Must match option string! */
|
---|
229 | enum {
|
---|
230 | WGET_OPT_CONTINUE = (1 << 0),
|
---|
231 | WGET_OPT_SPIDER = (1 << 1),
|
---|
232 | WGET_OPT_QUIET = (1 << 2),
|
---|
233 | WGET_OPT_OUTNAME = (1 << 3),
|
---|
234 | WGET_OPT_PREFIX = (1 << 4),
|
---|
235 | WGET_OPT_PROXY = (1 << 5),
|
---|
236 | WGET_OPT_USER_AGENT = (1 << 6),
|
---|
237 | WGET_OPT_NETWORK_READ_TIMEOUT = (1 << 7),
|
---|
238 | WGET_OPT_RETRIES = (1 << 8),
|
---|
239 | WGET_OPT_PASSIVE = (1 << 9),
|
---|
240 | WGET_OPT_HEADER = (1 << 10) * ENABLE_FEATURE_WGET_LONG_OPTIONS,
|
---|
241 | WGET_OPT_POST_DATA = (1 << 11) * ENABLE_FEATURE_WGET_LONG_OPTIONS,
|
---|
242 | };
|
---|
243 |
|
---|
244 | enum {
|
---|
245 | PROGRESS_START = -1,
|
---|
246 | PROGRESS_END = 0,
|
---|
247 | PROGRESS_BUMP = 1,
|
---|
248 | };
|
---|
249 | #if ENABLE_FEATURE_WGET_STATUSBAR
|
---|
250 | static void progress_meter(int flag)
|
---|
251 | {
|
---|
252 | if (option_mask32 & WGET_OPT_QUIET)
|
---|
253 | return;
|
---|
254 |
|
---|
255 | if (flag == PROGRESS_START)
|
---|
256 | bb_progress_init(&G.pmt, G.curfile);
|
---|
257 |
|
---|
258 | bb_progress_update(&G.pmt,
|
---|
259 | G.beg_range,
|
---|
260 | G.transferred,
|
---|
261 | (G.chunked || !G.got_clen) ? 0 : G.beg_range + G.transferred + G.content_len
|
---|
262 | );
|
---|
263 |
|
---|
264 | if (flag == PROGRESS_END) {
|
---|
265 | bb_progress_free(&G.pmt);
|
---|
266 | bb_putchar_stderr('\n');
|
---|
267 | G.transferred = 0;
|
---|
268 | }
|
---|
269 | }
|
---|
270 | #else
|
---|
271 | static ALWAYS_INLINE void progress_meter(int flag UNUSED_PARAM) { }
|
---|
272 | #endif
|
---|
273 |
|
---|
274 |
|
---|
275 | /* IPv6 knows scoped address types i.e. link and site local addresses. Link
|
---|
276 | * local addresses can have a scope identifier to specify the
|
---|
277 | * interface/link an address is valid on (e.g. fe80::1%eth0). This scope
|
---|
278 | * identifier is only valid on a single node.
|
---|
279 | *
|
---|
280 | * RFC 4007 says that the scope identifier MUST NOT be sent across the wire,
|
---|
281 | * unless all nodes agree on the semantic. Apache e.g. regards zone identifiers
|
---|
282 | * in the Host header as invalid requests, see
|
---|
283 | * https://issues.apache.org/bugzilla/show_bug.cgi?id=35122
|
---|
284 | */
|
---|
285 | static void strip_ipv6_scope_id(char *host)
|
---|
286 | {
|
---|
287 | char *scope, *cp;
|
---|
288 |
|
---|
289 | /* bbox wget actually handles IPv6 addresses without [], like
|
---|
290 | * wget "http://::1/xxx", but this is not standard.
|
---|
291 | * To save code, _here_ we do not support it. */
|
---|
292 |
|
---|
293 | if (host[0] != '[')
|
---|
294 | return; /* not IPv6 */
|
---|
295 |
|
---|
296 | scope = strchr(host, '%');
|
---|
297 | if (!scope)
|
---|
298 | return;
|
---|
299 |
|
---|
300 | /* Remove the IPv6 zone identifier from the host address */
|
---|
301 | cp = strchr(host, ']');
|
---|
302 | if (!cp || (cp[1] != ':' && cp[1] != '\0')) {
|
---|
303 | /* malformed address (not "[xx]:nn" or "[xx]") */
|
---|
304 | return;
|
---|
305 | }
|
---|
306 |
|
---|
307 | /* cp points to "]...", scope points to "%eth0]..." */
|
---|
308 | overlapping_strcpy(scope, cp);
|
---|
309 | }
|
---|
310 |
|
---|
311 | #if ENABLE_FEATURE_WGET_AUTHENTICATION
|
---|
312 | /* Base64-encode character string. */
|
---|
313 | static char *base64enc(const char *str)
|
---|
314 | {
|
---|
315 | unsigned len = strlen(str);
|
---|
316 | if (len > sizeof(G.wget_buf)/4*3 - 10) /* paranoia */
|
---|
317 | len = sizeof(G.wget_buf)/4*3 - 10;
|
---|
318 | bb_uuencode(G.wget_buf, str, len, bb_uuenc_tbl_base64);
|
---|
319 | return G.wget_buf;
|
---|
320 | }
|
---|
321 | #endif
|
---|
322 |
|
---|
323 | static char* sanitize_string(char *s)
|
---|
324 | {
|
---|
325 | unsigned char *p = (void *) s;
|
---|
326 | while (*p >= ' ')
|
---|
327 | p++;
|
---|
328 | *p = '\0';
|
---|
329 | return s;
|
---|
330 | }
|
---|
331 |
|
---|
332 | #if ENABLE_FEATURE_WGET_TIMEOUT
|
---|
333 | static void alarm_handler(int sig UNUSED_PARAM)
|
---|
334 | {
|
---|
335 | /* This is theoretically unsafe (uses stdio and malloc in signal handler) */
|
---|
336 | if (G.die_if_timed_out)
|
---|
337 | bb_error_msg_and_die("download timed out");
|
---|
338 | }
|
---|
339 | static void set_alarm(void)
|
---|
340 | {
|
---|
341 | if (G.timeout_seconds) {
|
---|
342 | alarm(G.timeout_seconds);
|
---|
343 | G.die_if_timed_out = 1;
|
---|
344 | }
|
---|
345 | }
|
---|
346 | # define clear_alarm() ((void)(G.die_if_timed_out = 0))
|
---|
347 | #else
|
---|
348 | # define set_alarm() ((void)0)
|
---|
349 | # define clear_alarm() ((void)0)
|
---|
350 | #endif
|
---|
351 |
|
---|
352 | static FILE *open_socket(len_and_sockaddr *lsa)
|
---|
353 | {
|
---|
354 | int fd;
|
---|
355 | FILE *fp;
|
---|
356 |
|
---|
357 | set_alarm();
|
---|
358 | fd = xconnect_stream(lsa);
|
---|
359 | clear_alarm();
|
---|
360 |
|
---|
361 | /* glibc 2.4 seems to try seeking on it - ??! */
|
---|
362 | /* hopefully it understands what ESPIPE means... */
|
---|
363 | fp = fdopen(fd, "r+");
|
---|
364 | if (!fp)
|
---|
365 | bb_perror_msg_and_die(bb_msg_memory_exhausted);
|
---|
366 |
|
---|
367 | return fp;
|
---|
368 | }
|
---|
369 |
|
---|
370 | /* Returns '\n' if it was seen, else '\0'. Trims at first '\r' or '\n' */
|
---|
371 | static char fgets_and_trim(FILE *fp)
|
---|
372 | {
|
---|
373 | char c;
|
---|
374 | char *buf_ptr;
|
---|
375 |
|
---|
376 | set_alarm();
|
---|
377 | if (fgets(G.wget_buf, sizeof(G.wget_buf) - 1, fp) == NULL)
|
---|
378 | bb_perror_msg_and_die("error getting response");
|
---|
379 | clear_alarm();
|
---|
380 |
|
---|
381 | buf_ptr = strchrnul(G.wget_buf, '\n');
|
---|
382 | c = *buf_ptr;
|
---|
383 | *buf_ptr = '\0';
|
---|
384 | buf_ptr = strchrnul(G.wget_buf, '\r');
|
---|
385 | *buf_ptr = '\0';
|
---|
386 |
|
---|
387 | log_io("< %s", G.wget_buf);
|
---|
388 |
|
---|
389 | return c;
|
---|
390 | }
|
---|
391 |
|
---|
392 | static int ftpcmd(const char *s1, const char *s2, FILE *fp)
|
---|
393 | {
|
---|
394 | int result;
|
---|
395 | if (s1) {
|
---|
396 | if (!s2)
|
---|
397 | s2 = "";
|
---|
398 | fprintf(fp, "%s%s\r\n", s1, s2);
|
---|
399 | fflush(fp);
|
---|
400 | log_io("> %s%s", s1, s2);
|
---|
401 | }
|
---|
402 |
|
---|
403 | do {
|
---|
404 | fgets_and_trim(fp);
|
---|
405 | } while (!isdigit(G.wget_buf[0]) || G.wget_buf[3] != ' ');
|
---|
406 |
|
---|
407 | G.wget_buf[3] = '\0';
|
---|
408 | result = xatoi_positive(G.wget_buf);
|
---|
409 | G.wget_buf[3] = ' ';
|
---|
410 | return result;
|
---|
411 | }
|
---|
412 |
|
---|
413 | static void parse_url(const char *src_url, struct host_info *h)
|
---|
414 | {
|
---|
415 | char *url, *p, *sp;
|
---|
416 |
|
---|
417 | free(h->allocated);
|
---|
418 | h->allocated = url = xstrdup(src_url);
|
---|
419 |
|
---|
420 | h->protocol = P_FTP;
|
---|
421 | p = strstr(url, "://");
|
---|
422 | if (p) {
|
---|
423 | *p = '\0';
|
---|
424 | h->host = p + 3;
|
---|
425 | if (strcmp(url, P_FTP) == 0) {
|
---|
426 | h->port = bb_lookup_port(P_FTP, "tcp", 21);
|
---|
427 | } else
|
---|
428 | #if ENABLE_FEATURE_WGET_OPENSSL || ENABLE_FEATURE_WGET_SSL_HELPER
|
---|
429 | if (strcmp(url, P_HTTPS) == 0) {
|
---|
430 | h->port = bb_lookup_port(P_HTTPS, "tcp", 443);
|
---|
431 | h->protocol = P_HTTPS;
|
---|
432 | } else
|
---|
433 | #endif
|
---|
434 | if (strcmp(url, P_HTTP) == 0) {
|
---|
435 | http:
|
---|
436 | h->port = bb_lookup_port(P_HTTP, "tcp", 80);
|
---|
437 | h->protocol = P_HTTP;
|
---|
438 | } else {
|
---|
439 | *p = ':';
|
---|
440 | bb_error_msg_and_die("not an http or ftp url: %s", sanitize_string(url));
|
---|
441 | }
|
---|
442 | } else {
|
---|
443 | // GNU wget is user-friendly and falls back to http://
|
---|
444 | h->host = url;
|
---|
445 | goto http;
|
---|
446 | }
|
---|
447 |
|
---|
448 | // FYI:
|
---|
449 | // "Real" wget 'http://busybox.net?var=a/b' sends this request:
|
---|
450 | // 'GET /?var=a/b HTTP 1.0'
|
---|
451 | // and saves 'index.html?var=a%2Fb' (we save 'b')
|
---|
452 | // wget 'http://busybox.net?login=john@doe':
|
---|
453 | // request: 'GET /?login=john@doe HTTP/1.0'
|
---|
454 | // saves: 'index.html?login=john@doe' (we save '?login=john@doe')
|
---|
455 | // wget 'http://busybox.net#test/test':
|
---|
456 | // request: 'GET / HTTP/1.0'
|
---|
457 | // saves: 'index.html' (we save 'test')
|
---|
458 | //
|
---|
459 | // We also don't add unique .N suffix if file exists...
|
---|
460 | sp = strchr(h->host, '/');
|
---|
461 | p = strchr(h->host, '?'); if (!sp || (p && sp > p)) sp = p;
|
---|
462 | p = strchr(h->host, '#'); if (!sp || (p && sp > p)) sp = p;
|
---|
463 | if (!sp) {
|
---|
464 | h->path = "";
|
---|
465 | } else if (*sp == '/') {
|
---|
466 | *sp = '\0';
|
---|
467 | h->path = sp + 1;
|
---|
468 | } else { // '#' or '?'
|
---|
469 | // http://busybox.net?login=john@doe is a valid URL
|
---|
470 | // memmove converts to:
|
---|
471 | // http:/busybox.nett?login=john@doe...
|
---|
472 | memmove(h->host - 1, h->host, sp - h->host);
|
---|
473 | h->host--;
|
---|
474 | sp[-1] = '\0';
|
---|
475 | h->path = sp;
|
---|
476 | }
|
---|
477 |
|
---|
478 | sp = strrchr(h->host, '@');
|
---|
479 | if (sp != NULL) {
|
---|
480 | // URL-decode "user:password" string before base64-encoding:
|
---|
481 | // wget http://test:my%20pass@example.com should send
|
---|
482 | // Authorization: Basic dGVzdDpteSBwYXNz
|
---|
483 | // which decodes to "test:my pass".
|
---|
484 | // Standard wget and curl do this too.
|
---|
485 | *sp = '\0';
|
---|
486 | free(h->user);
|
---|
487 | h->user = xstrdup(percent_decode_in_place(h->host, /*strict:*/ 0));
|
---|
488 | h->host = sp + 1;
|
---|
489 | }
|
---|
490 | /* else: h->user remains NULL, or as set by original request
|
---|
491 | * before redirect (if we are here after a redirect).
|
---|
492 | */
|
---|
493 | }
|
---|
494 |
|
---|
495 | static char *gethdr(FILE *fp)
|
---|
496 | {
|
---|
497 | char *s, *hdrval;
|
---|
498 | int c;
|
---|
499 |
|
---|
500 | /* retrieve header line */
|
---|
501 | c = fgets_and_trim(fp);
|
---|
502 |
|
---|
503 | /* end of the headers? */
|
---|
504 | if (G.wget_buf[0] == '\0')
|
---|
505 | return NULL;
|
---|
506 |
|
---|
507 | /* convert the header name to lower case */
|
---|
508 | for (s = G.wget_buf; isalnum(*s) || *s == '-' || *s == '.' || *s == '_'; ++s) {
|
---|
509 | /*
|
---|
510 | * No-op for 20-3f and 60-7f. "0-9a-z-." are in these ranges.
|
---|
511 | * 40-5f range ("@A-Z[\]^_") maps to 60-7f.
|
---|
512 | * "A-Z" maps to "a-z".
|
---|
513 | * "@[\]" can't occur in header names.
|
---|
514 | * "^_" maps to "~,DEL" (which is wrong).
|
---|
515 | * "^" was never seen yet, "_" was seen from web.archive.org
|
---|
516 | * (x-archive-orig-x_commoncrawl_Signature: HEXSTRING).
|
---|
517 | */
|
---|
518 | *s |= 0x20;
|
---|
519 | }
|
---|
520 |
|
---|
521 | /* verify we are at the end of the header name */
|
---|
522 | if (*s != ':')
|
---|
523 | bb_error_msg_and_die("bad header line: %s", sanitize_string(G.wget_buf));
|
---|
524 |
|
---|
525 | /* locate the start of the header value */
|
---|
526 | *s++ = '\0';
|
---|
527 | hdrval = skip_whitespace(s);
|
---|
528 |
|
---|
529 | if (c != '\n') {
|
---|
530 | /* Rats! The buffer isn't big enough to hold the entire header value */
|
---|
531 | while (c = getc(fp), c != EOF && c != '\n')
|
---|
532 | continue;
|
---|
533 | }
|
---|
534 |
|
---|
535 | return hdrval;
|
---|
536 | }
|
---|
537 |
|
---|
538 | static void reset_beg_range_to_zero(void)
|
---|
539 | {
|
---|
540 | bb_error_msg("restart failed");
|
---|
541 | G.beg_range = 0;
|
---|
542 | xlseek(G.output_fd, 0, SEEK_SET);
|
---|
543 | /* Done at the end instead: */
|
---|
544 | /* ftruncate(G.output_fd, 0); */
|
---|
545 | }
|
---|
546 |
|
---|
547 | static FILE* prepare_ftp_session(FILE **dfpp, struct host_info *target, len_and_sockaddr *lsa)
|
---|
548 | {
|
---|
549 | FILE *sfp;
|
---|
550 | char *str;
|
---|
551 | int port;
|
---|
552 |
|
---|
553 | if (!target->user)
|
---|
554 | target->user = xstrdup("anonymous:busybox@");
|
---|
555 |
|
---|
556 | sfp = open_socket(lsa);
|
---|
557 | if (ftpcmd(NULL, NULL, sfp) != 220)
|
---|
558 | bb_error_msg_and_die("%s", sanitize_string(G.wget_buf + 4));
|
---|
559 |
|
---|
560 | /*
|
---|
561 | * Splitting username:password pair,
|
---|
562 | * trying to log in
|
---|
563 | */
|
---|
564 | str = strchr(target->user, ':');
|
---|
565 | if (str)
|
---|
566 | *str++ = '\0';
|
---|
567 | switch (ftpcmd("USER ", target->user, sfp)) {
|
---|
568 | case 230:
|
---|
569 | break;
|
---|
570 | case 331:
|
---|
571 | if (ftpcmd("PASS ", str, sfp) == 230)
|
---|
572 | break;
|
---|
573 | /* fall through (failed login) */
|
---|
574 | default:
|
---|
575 | bb_error_msg_and_die("ftp login: %s", sanitize_string(G.wget_buf + 4));
|
---|
576 | }
|
---|
577 |
|
---|
578 | ftpcmd("TYPE I", NULL, sfp);
|
---|
579 |
|
---|
580 | /*
|
---|
581 | * Querying file size
|
---|
582 | */
|
---|
583 | if (ftpcmd("SIZE ", target->path, sfp) == 213) {
|
---|
584 | G.content_len = BB_STRTOOFF(G.wget_buf + 4, NULL, 10);
|
---|
585 | if (G.content_len < 0 || errno) {
|
---|
586 | bb_error_msg_and_die("SIZE value is garbage");
|
---|
587 | }
|
---|
588 | G.got_clen = 1;
|
---|
589 | }
|
---|
590 |
|
---|
591 | /*
|
---|
592 | * Entering passive mode
|
---|
593 | */
|
---|
594 | if (ftpcmd("PASV", NULL, sfp) != 227) {
|
---|
595 | pasv_error:
|
---|
596 | bb_error_msg_and_die("bad response to %s: %s", "PASV", sanitize_string(G.wget_buf));
|
---|
597 | }
|
---|
598 | // Response is "227 garbageN1,N2,N3,N4,P1,P2[)garbage]
|
---|
599 | // Server's IP is N1.N2.N3.N4 (we ignore it)
|
---|
600 | // Server's port for data connection is P1*256+P2
|
---|
601 | str = strrchr(G.wget_buf, ')');
|
---|
602 | if (str) str[0] = '\0';
|
---|
603 | str = strrchr(G.wget_buf, ',');
|
---|
604 | if (!str) goto pasv_error;
|
---|
605 | port = xatou_range(str+1, 0, 255);
|
---|
606 | *str = '\0';
|
---|
607 | str = strrchr(G.wget_buf, ',');
|
---|
608 | if (!str) goto pasv_error;
|
---|
609 | port += xatou_range(str+1, 0, 255) * 256;
|
---|
610 | set_nport(&lsa->u.sa, htons(port));
|
---|
611 |
|
---|
612 | *dfpp = open_socket(lsa);
|
---|
613 |
|
---|
614 | if (G.beg_range != 0) {
|
---|
615 | sprintf(G.wget_buf, "REST %"OFF_FMT"u", G.beg_range);
|
---|
616 | if (ftpcmd(G.wget_buf, NULL, sfp) == 350)
|
---|
617 | G.content_len -= G.beg_range;
|
---|
618 | else
|
---|
619 | reset_beg_range_to_zero();
|
---|
620 | }
|
---|
621 |
|
---|
622 | if (ftpcmd("RETR ", target->path, sfp) > 150)
|
---|
623 | bb_error_msg_and_die("bad response to %s: %s", "RETR", sanitize_string(G.wget_buf));
|
---|
624 |
|
---|
625 | return sfp;
|
---|
626 | }
|
---|
627 |
|
---|
628 | #if ENABLE_FEATURE_WGET_OPENSSL
|
---|
629 | static int spawn_https_helper_openssl(const char *host, unsigned port)
|
---|
630 | {
|
---|
631 | char *allocated = NULL;
|
---|
632 | int sp[2];
|
---|
633 | int pid;
|
---|
634 | IF_FEATURE_WGET_SSL_HELPER(volatile int child_failed = 0;)
|
---|
635 |
|
---|
636 | if (socketpair(AF_UNIX, SOCK_STREAM, 0, sp) != 0)
|
---|
637 | /* Kernel can have AF_UNIX support disabled */
|
---|
638 | bb_perror_msg_and_die("socketpair");
|
---|
639 |
|
---|
640 | if (!strchr(host, ':'))
|
---|
641 | host = allocated = xasprintf("%s:%u", host, port);
|
---|
642 |
|
---|
643 | fflush_all();
|
---|
644 | pid = xvfork();
|
---|
645 | if (pid == 0) {
|
---|
646 | /* Child */
|
---|
647 | char *argv[6];
|
---|
648 |
|
---|
649 | close(sp[0]);
|
---|
650 | xmove_fd(sp[1], 0);
|
---|
651 | xdup2(0, 1);
|
---|
652 | /*
|
---|
653 | * openssl s_client -quiet -connect www.kernel.org:443 2>/dev/null
|
---|
654 | * It prints some debug stuff on stderr, don't know how to suppress it.
|
---|
655 | * Work around by dev-nulling stderr. We lose all error messages :(
|
---|
656 | */
|
---|
657 | xmove_fd(2, 3);
|
---|
658 | xopen("/dev/null", O_RDWR);
|
---|
659 | argv[0] = (char*)"openssl";
|
---|
660 | argv[1] = (char*)"s_client";
|
---|
661 | argv[2] = (char*)"-quiet";
|
---|
662 | argv[3] = (char*)"-connect";
|
---|
663 | argv[4] = (char*)host;
|
---|
664 | argv[5] = NULL;
|
---|
665 | BB_EXECVP(argv[0], argv);
|
---|
666 | xmove_fd(3, 2);
|
---|
667 | # if ENABLE_FEATURE_WGET_SSL_HELPER
|
---|
668 | child_failed = 1;
|
---|
669 | xfunc_die();
|
---|
670 | # else
|
---|
671 | bb_perror_msg_and_die("can't execute '%s'", argv[0]);
|
---|
672 | # endif
|
---|
673 | /* notreached */
|
---|
674 | }
|
---|
675 |
|
---|
676 | /* Parent */
|
---|
677 | free(allocated);
|
---|
678 | close(sp[1]);
|
---|
679 | # if ENABLE_FEATURE_WGET_SSL_HELPER
|
---|
680 | if (child_failed) {
|
---|
681 | close(sp[0]);
|
---|
682 | return -1;
|
---|
683 | }
|
---|
684 | # endif
|
---|
685 | return sp[0];
|
---|
686 | }
|
---|
687 | #endif
|
---|
688 |
|
---|
689 | /* See networking/ssl_helper/README how to build one */
|
---|
690 | #if ENABLE_FEATURE_WGET_SSL_HELPER
|
---|
691 | static void spawn_https_helper_small(int network_fd)
|
---|
692 | {
|
---|
693 | int sp[2];
|
---|
694 | int pid;
|
---|
695 |
|
---|
696 | if (socketpair(AF_UNIX, SOCK_STREAM, 0, sp) != 0)
|
---|
697 | /* Kernel can have AF_UNIX support disabled */
|
---|
698 | bb_perror_msg_and_die("socketpair");
|
---|
699 |
|
---|
700 | pid = BB_MMU ? xfork() : xvfork();
|
---|
701 | if (pid == 0) {
|
---|
702 | /* Child */
|
---|
703 | char *argv[3];
|
---|
704 |
|
---|
705 | close(sp[0]);
|
---|
706 | xmove_fd(sp[1], 0);
|
---|
707 | xdup2(0, 1);
|
---|
708 | xmove_fd(network_fd, 3);
|
---|
709 | /*
|
---|
710 | * A simple ssl/tls helper
|
---|
711 | */
|
---|
712 | argv[0] = (char*)"ssl_helper";
|
---|
713 | argv[1] = (char*)"-d3";
|
---|
714 | argv[2] = NULL;
|
---|
715 | BB_EXECVP(argv[0], argv);
|
---|
716 | bb_perror_msg_and_die("can't execute '%s'", argv[0]);
|
---|
717 | /* notreached */
|
---|
718 | }
|
---|
719 |
|
---|
720 | /* Parent */
|
---|
721 | close(sp[1]);
|
---|
722 | xmove_fd(sp[0], network_fd);
|
---|
723 | }
|
---|
724 | #endif
|
---|
725 |
|
---|
726 | static void NOINLINE retrieve_file_data(FILE *dfp)
|
---|
727 | {
|
---|
728 | #if ENABLE_FEATURE_WGET_STATUSBAR || ENABLE_FEATURE_WGET_TIMEOUT
|
---|
729 | # if ENABLE_FEATURE_WGET_TIMEOUT
|
---|
730 | unsigned second_cnt = G.timeout_seconds;
|
---|
731 | # endif
|
---|
732 | struct pollfd polldata;
|
---|
733 |
|
---|
734 | polldata.fd = fileno(dfp);
|
---|
735 | polldata.events = POLLIN | POLLPRI;
|
---|
736 | #endif
|
---|
737 | progress_meter(PROGRESS_START);
|
---|
738 |
|
---|
739 | if (G.chunked)
|
---|
740 | goto get_clen;
|
---|
741 |
|
---|
742 | /* Loops only if chunked */
|
---|
743 | while (1) {
|
---|
744 |
|
---|
745 | #if ENABLE_FEATURE_WGET_STATUSBAR || ENABLE_FEATURE_WGET_TIMEOUT
|
---|
746 | /* Must use nonblocking I/O, otherwise fread will loop
|
---|
747 | * and *block* until it reads full buffer,
|
---|
748 | * which messes up progress bar and/or timeout logic.
|
---|
749 | * Because of nonblocking I/O, we need to dance
|
---|
750 | * very carefully around EAGAIN. See explanation at
|
---|
751 | * clearerr() calls.
|
---|
752 | */
|
---|
753 | ndelay_on(polldata.fd);
|
---|
754 | #endif
|
---|
755 | while (1) {
|
---|
756 | int n;
|
---|
757 | unsigned rdsz;
|
---|
758 |
|
---|
759 | #if ENABLE_FEATURE_WGET_STATUSBAR || ENABLE_FEATURE_WGET_TIMEOUT
|
---|
760 | /* fread internally uses read loop, which in our case
|
---|
761 | * is usually exited when we get EAGAIN.
|
---|
762 | * In this case, libc sets error marker on the stream.
|
---|
763 | * Need to clear it before next fread to avoid possible
|
---|
764 | * rare false positive ferror below. Rare because usually
|
---|
765 | * fread gets more than zero bytes, and we don't fall
|
---|
766 | * into if (n <= 0) ...
|
---|
767 | */
|
---|
768 | clearerr(dfp);
|
---|
769 | #endif
|
---|
770 | errno = 0;
|
---|
771 | rdsz = sizeof(G.wget_buf);
|
---|
772 | if (G.got_clen) {
|
---|
773 | if (G.content_len < (off_t)sizeof(G.wget_buf)) {
|
---|
774 | if ((int)G.content_len <= 0)
|
---|
775 | break;
|
---|
776 | rdsz = (unsigned)G.content_len;
|
---|
777 | }
|
---|
778 | }
|
---|
779 | n = fread(G.wget_buf, 1, rdsz, dfp);
|
---|
780 |
|
---|
781 | if (n > 0) {
|
---|
782 | xwrite(G.output_fd, G.wget_buf, n);
|
---|
783 | #if ENABLE_FEATURE_WGET_STATUSBAR
|
---|
784 | G.transferred += n;
|
---|
785 | #endif
|
---|
786 | if (G.got_clen) {
|
---|
787 | G.content_len -= n;
|
---|
788 | if (G.content_len == 0)
|
---|
789 | break;
|
---|
790 | }
|
---|
791 | #if ENABLE_FEATURE_WGET_TIMEOUT
|
---|
792 | second_cnt = G.timeout_seconds;
|
---|
793 | #endif
|
---|
794 | goto bump;
|
---|
795 | }
|
---|
796 |
|
---|
797 | /* n <= 0.
|
---|
798 | * man fread:
|
---|
799 | * If error occurs, or EOF is reached, the return value
|
---|
800 | * is a short item count (or zero).
|
---|
801 | * fread does not distinguish between EOF and error.
|
---|
802 | */
|
---|
803 | if (errno != EAGAIN) {
|
---|
804 | if (ferror(dfp)) {
|
---|
805 | progress_meter(PROGRESS_END);
|
---|
806 | bb_perror_msg_and_die(bb_msg_read_error);
|
---|
807 | }
|
---|
808 | break; /* EOF, not error */
|
---|
809 | }
|
---|
810 |
|
---|
811 | #if ENABLE_FEATURE_WGET_STATUSBAR || ENABLE_FEATURE_WGET_TIMEOUT
|
---|
812 | /* It was EAGAIN. There is no data. Wait up to one second
|
---|
813 | * then abort if timed out, or update the bar and try reading again.
|
---|
814 | */
|
---|
815 | if (safe_poll(&polldata, 1, 1000) == 0) {
|
---|
816 | # if ENABLE_FEATURE_WGET_TIMEOUT
|
---|
817 | if (second_cnt != 0 && --second_cnt == 0) {
|
---|
818 | progress_meter(PROGRESS_END);
|
---|
819 | bb_error_msg_and_die("download timed out");
|
---|
820 | }
|
---|
821 | # endif
|
---|
822 | /* We used to loop back to poll here,
|
---|
823 | * but there is no great harm in letting fread
|
---|
824 | * to try reading anyway.
|
---|
825 | */
|
---|
826 | }
|
---|
827 | #endif
|
---|
828 | bump:
|
---|
829 | /* Need to do it _every_ second for "stalled" indicator
|
---|
830 | * to be shown properly.
|
---|
831 | */
|
---|
832 | progress_meter(PROGRESS_BUMP);
|
---|
833 | } /* while (reading data) */
|
---|
834 |
|
---|
835 | #if ENABLE_FEATURE_WGET_STATUSBAR || ENABLE_FEATURE_WGET_TIMEOUT
|
---|
836 | clearerr(dfp);
|
---|
837 | ndelay_off(polldata.fd); /* else fgets can get very unhappy */
|
---|
838 | #endif
|
---|
839 | if (!G.chunked)
|
---|
840 | break;
|
---|
841 |
|
---|
842 | fgets_and_trim(dfp); /* Eat empty line */
|
---|
843 | get_clen:
|
---|
844 | fgets_and_trim(dfp);
|
---|
845 | G.content_len = STRTOOFF(G.wget_buf, NULL, 16);
|
---|
846 | /* FIXME: error check? */
|
---|
847 | if (G.content_len == 0)
|
---|
848 | break; /* all done! */
|
---|
849 | G.got_clen = 1;
|
---|
850 | /*
|
---|
851 | * Note that fgets may result in some data being buffered in dfp.
|
---|
852 | * We loop back to fread, which will retrieve this data.
|
---|
853 | * Also note that code has to be arranged so that fread
|
---|
854 | * is done _before_ one-second poll wait - poll doesn't know
|
---|
855 | * about stdio buffering and can result in spurious one second waits!
|
---|
856 | */
|
---|
857 | }
|
---|
858 |
|
---|
859 | /* If -c failed, we restart from the beginning,
|
---|
860 | * but we do not truncate file then, we do it only now, at the end.
|
---|
861 | * This lets user to ^C if his 99% complete 10 GB file download
|
---|
862 | * failed to restart *without* losing the almost complete file.
|
---|
863 | */
|
---|
864 | {
|
---|
865 | off_t pos = lseek(G.output_fd, 0, SEEK_CUR);
|
---|
866 | if (pos != (off_t)-1)
|
---|
867 | ftruncate(G.output_fd, pos);
|
---|
868 | }
|
---|
869 |
|
---|
870 | /* Draw full bar and free its resources */
|
---|
871 | G.chunked = 0; /* makes it show 100% even for chunked download */
|
---|
872 | G.got_clen = 1; /* makes it show 100% even for download of (formerly) unknown size */
|
---|
873 | progress_meter(PROGRESS_END);
|
---|
874 | }
|
---|
875 |
|
---|
876 | static void download_one_url(const char *url)
|
---|
877 | {
|
---|
878 | bool use_proxy; /* Use proxies if env vars are set */
|
---|
879 | int redir_limit;
|
---|
880 | len_and_sockaddr *lsa;
|
---|
881 | FILE *sfp; /* socket to web/ftp server */
|
---|
882 | FILE *dfp; /* socket to ftp server (data) */
|
---|
883 | char *proxy = NULL;
|
---|
884 | char *fname_out_alloc;
|
---|
885 | char *redirected_path = NULL;
|
---|
886 | struct host_info server;
|
---|
887 | struct host_info target;
|
---|
888 |
|
---|
889 | server.allocated = NULL;
|
---|
890 | target.allocated = NULL;
|
---|
891 | server.user = NULL;
|
---|
892 | target.user = NULL;
|
---|
893 |
|
---|
894 | parse_url(url, &target);
|
---|
895 |
|
---|
896 | /* Use the proxy if necessary */
|
---|
897 | use_proxy = (strcmp(G.proxy_flag, "off") != 0);
|
---|
898 | if (use_proxy) {
|
---|
899 | proxy = getenv(target.protocol == P_FTP ? "ftp_proxy" : "http_proxy");
|
---|
900 | //FIXME: what if protocol is https? Ok to use http_proxy?
|
---|
901 | use_proxy = (proxy && proxy[0]);
|
---|
902 | if (use_proxy)
|
---|
903 | parse_url(proxy, &server);
|
---|
904 | }
|
---|
905 | if (!use_proxy) {
|
---|
906 | server.port = target.port;
|
---|
907 | if (ENABLE_FEATURE_IPV6) {
|
---|
908 | //free(server.allocated); - can't be non-NULL
|
---|
909 | server.host = server.allocated = xstrdup(target.host);
|
---|
910 | } else {
|
---|
911 | server.host = target.host;
|
---|
912 | }
|
---|
913 | }
|
---|
914 |
|
---|
915 | if (ENABLE_FEATURE_IPV6)
|
---|
916 | strip_ipv6_scope_id(target.host);
|
---|
917 |
|
---|
918 | /* If there was no -O FILE, guess output filename */
|
---|
919 | fname_out_alloc = NULL;
|
---|
920 | if (!(option_mask32 & WGET_OPT_OUTNAME)) {
|
---|
921 | G.fname_out = bb_get_last_path_component_nostrip(target.path);
|
---|
922 | /* handle "wget http://kernel.org//" */
|
---|
923 | if (G.fname_out[0] == '/' || !G.fname_out[0])
|
---|
924 | G.fname_out = (char*)"index.html";
|
---|
925 | /* -P DIR is considered only if there was no -O FILE */
|
---|
926 | if (G.dir_prefix)
|
---|
927 | G.fname_out = fname_out_alloc = concat_path_file(G.dir_prefix, G.fname_out);
|
---|
928 | else {
|
---|
929 | /* redirects may free target.path later, need to make a copy */
|
---|
930 | G.fname_out = fname_out_alloc = xstrdup(G.fname_out);
|
---|
931 | }
|
---|
932 | }
|
---|
933 | #if ENABLE_FEATURE_WGET_STATUSBAR
|
---|
934 | G.curfile = bb_get_last_path_component_nostrip(G.fname_out);
|
---|
935 | #endif
|
---|
936 |
|
---|
937 | /* Determine where to start transfer */
|
---|
938 | G.beg_range = 0;
|
---|
939 | if (option_mask32 & WGET_OPT_CONTINUE) {
|
---|
940 | G.output_fd = open(G.fname_out, O_WRONLY);
|
---|
941 | if (G.output_fd >= 0) {
|
---|
942 | G.beg_range = xlseek(G.output_fd, 0, SEEK_END);
|
---|
943 | }
|
---|
944 | /* File doesn't exist. We do not create file here yet.
|
---|
945 | * We are not sure it exists on remote side */
|
---|
946 | }
|
---|
947 |
|
---|
948 | redir_limit = 5;
|
---|
949 | resolve_lsa:
|
---|
950 | lsa = xhost2sockaddr(server.host, server.port);
|
---|
951 | if (!(option_mask32 & WGET_OPT_QUIET)) {
|
---|
952 | char *s = xmalloc_sockaddr2dotted(&lsa->u.sa);
|
---|
953 | fprintf(stderr, "Connecting to %s (%s)\n", server.host, s);
|
---|
954 | free(s);
|
---|
955 | }
|
---|
956 | establish_session:
|
---|
957 | /*G.content_len = 0; - redundant, got_clen = 0 is enough */
|
---|
958 | G.got_clen = 0;
|
---|
959 | G.chunked = 0;
|
---|
960 | if (use_proxy || target.protocol != P_FTP) {
|
---|
961 | /*
|
---|
962 | * HTTP session
|
---|
963 | */
|
---|
964 | char *str;
|
---|
965 | int status;
|
---|
966 |
|
---|
967 | /* Open socket to http(s) server */
|
---|
968 | #if ENABLE_FEATURE_WGET_OPENSSL
|
---|
969 | /* openssl (and maybe ssl_helper) support is configured */
|
---|
970 | if (target.protocol == P_HTTPS) {
|
---|
971 | /* openssl-based helper
|
---|
972 | * Inconvenient API since we can't give it an open fd
|
---|
973 | */
|
---|
974 | int fd = spawn_https_helper_openssl(server.host, server.port);
|
---|
975 | # if ENABLE_FEATURE_WGET_SSL_HELPER
|
---|
976 | if (fd < 0) { /* no openssl? try ssl_helper */
|
---|
977 | sfp = open_socket(lsa);
|
---|
978 | spawn_https_helper_small(fileno(sfp));
|
---|
979 | goto socket_opened;
|
---|
980 | }
|
---|
981 | # else
|
---|
982 | /* We don't check for exec("openssl") failure in this case */
|
---|
983 | # endif
|
---|
984 | sfp = fdopen(fd, "r+");
|
---|
985 | if (!sfp)
|
---|
986 | bb_perror_msg_and_die(bb_msg_memory_exhausted);
|
---|
987 | goto socket_opened;
|
---|
988 | }
|
---|
989 | sfp = open_socket(lsa);
|
---|
990 | socket_opened:
|
---|
991 | #elif ENABLE_FEATURE_WGET_SSL_HELPER
|
---|
992 | /* Only ssl_helper support is configured */
|
---|
993 | sfp = open_socket(lsa);
|
---|
994 | if (target.protocol == P_HTTPS)
|
---|
995 | spawn_https_helper_small(fileno(sfp));
|
---|
996 | #else
|
---|
997 | /* ssl (https) support is not configured */
|
---|
998 | sfp = open_socket(lsa);
|
---|
999 | #endif
|
---|
1000 | /* Send HTTP request */
|
---|
1001 | if (use_proxy) {
|
---|
1002 | SENDFMT(sfp, "GET %s://%s/%s HTTP/1.1\r\n",
|
---|
1003 | target.protocol, target.host,
|
---|
1004 | target.path);
|
---|
1005 | } else {
|
---|
1006 | SENDFMT(sfp, "%s /%s HTTP/1.1\r\n",
|
---|
1007 | (option_mask32 & WGET_OPT_POST_DATA) ? "POST" : "GET",
|
---|
1008 | target.path);
|
---|
1009 | }
|
---|
1010 | if (!USR_HEADER_HOST)
|
---|
1011 | SENDFMT(sfp, "Host: %s\r\n", target.host);
|
---|
1012 | if (!USR_HEADER_USER_AGENT)
|
---|
1013 | SENDFMT(sfp, "User-Agent: %s\r\n", G.user_agent);
|
---|
1014 |
|
---|
1015 | /* Ask server to close the connection as soon as we are done
|
---|
1016 | * (IOW: we do not intend to send more requests)
|
---|
1017 | */
|
---|
1018 | SENDFMT(sfp, "Connection: close\r\n");
|
---|
1019 |
|
---|
1020 | #if ENABLE_FEATURE_WGET_AUTHENTICATION
|
---|
1021 | if (target.user && !USR_HEADER_AUTH) {
|
---|
1022 | SENDFMT(sfp, "Proxy-Authorization: Basic %s\r\n"+6,
|
---|
1023 | base64enc(target.user));
|
---|
1024 | }
|
---|
1025 | if (use_proxy && server.user && !USR_HEADER_PROXY_AUTH) {
|
---|
1026 | SENDFMT(sfp, "Proxy-Authorization: Basic %s\r\n",
|
---|
1027 | base64enc(server.user));
|
---|
1028 | }
|
---|
1029 | #endif
|
---|
1030 |
|
---|
1031 | if (G.beg_range != 0 && !USR_HEADER_RANGE)
|
---|
1032 | SENDFMT(sfp, "Range: bytes=%"OFF_FMT"u-\r\n", G.beg_range);
|
---|
1033 |
|
---|
1034 | #if ENABLE_FEATURE_WGET_LONG_OPTIONS
|
---|
1035 | if (G.extra_headers) {
|
---|
1036 | log_io(G.extra_headers);
|
---|
1037 | fputs(G.extra_headers, sfp);
|
---|
1038 | }
|
---|
1039 |
|
---|
1040 | if (option_mask32 & WGET_OPT_POST_DATA) {
|
---|
1041 | SENDFMT(sfp,
|
---|
1042 | "Content-Type: application/x-www-form-urlencoded\r\n"
|
---|
1043 | "Content-Length: %u\r\n"
|
---|
1044 | "\r\n"
|
---|
1045 | "%s",
|
---|
1046 | (int) strlen(G.post_data), G.post_data
|
---|
1047 | );
|
---|
1048 | } else
|
---|
1049 | #endif
|
---|
1050 | {
|
---|
1051 | SENDFMT(sfp, "\r\n");
|
---|
1052 | }
|
---|
1053 |
|
---|
1054 | fflush(sfp);
|
---|
1055 |
|
---|
1056 | /*
|
---|
1057 | * Retrieve HTTP response line and check for "200" status code.
|
---|
1058 | */
|
---|
1059 | read_response:
|
---|
1060 | fgets_and_trim(sfp);
|
---|
1061 |
|
---|
1062 | str = G.wget_buf;
|
---|
1063 | str = skip_non_whitespace(str);
|
---|
1064 | str = skip_whitespace(str);
|
---|
1065 | // FIXME: no error check
|
---|
1066 | // xatou wouldn't work: "200 OK"
|
---|
1067 | status = atoi(str);
|
---|
1068 | switch (status) {
|
---|
1069 | case 0:
|
---|
1070 | case 100:
|
---|
1071 | while (gethdr(sfp) != NULL)
|
---|
1072 | /* eat all remaining headers */;
|
---|
1073 | goto read_response;
|
---|
1074 | case 200:
|
---|
1075 | /*
|
---|
1076 | Response 204 doesn't say "null file", it says "metadata
|
---|
1077 | has changed but data didn't":
|
---|
1078 |
|
---|
1079 | "10.2.5 204 No Content
|
---|
1080 | The server has fulfilled the request but does not need to return
|
---|
1081 | an entity-body, and might want to return updated metainformation.
|
---|
1082 | The response MAY include new or updated metainformation in the form
|
---|
1083 | of entity-headers, which if present SHOULD be associated with
|
---|
1084 | the requested variant.
|
---|
1085 |
|
---|
1086 | If the client is a user agent, it SHOULD NOT change its document
|
---|
1087 | view from that which caused the request to be sent. This response
|
---|
1088 | is primarily intended to allow input for actions to take place
|
---|
1089 | without causing a change to the user agent's active document view,
|
---|
1090 | although any new or updated metainformation SHOULD be applied
|
---|
1091 | to the document currently in the user agent's active view.
|
---|
1092 |
|
---|
1093 | The 204 response MUST NOT include a message-body, and thus
|
---|
1094 | is always terminated by the first empty line after the header fields."
|
---|
1095 |
|
---|
1096 | However, in real world it was observed that some web servers
|
---|
1097 | (e.g. Boa/0.94.14rc21) simply use code 204 when file size is zero.
|
---|
1098 | */
|
---|
1099 | case 204:
|
---|
1100 | if (G.beg_range != 0) {
|
---|
1101 | /* "Range:..." was not honored by the server.
|
---|
1102 | * Restart download from the beginning.
|
---|
1103 | */
|
---|
1104 | reset_beg_range_to_zero();
|
---|
1105 | }
|
---|
1106 | break;
|
---|
1107 | case 300: /* redirection */
|
---|
1108 | case 301:
|
---|
1109 | case 302:
|
---|
1110 | case 303:
|
---|
1111 | break;
|
---|
1112 | case 206: /* Partial Content */
|
---|
1113 | if (G.beg_range != 0)
|
---|
1114 | /* "Range:..." worked. Good. */
|
---|
1115 | break;
|
---|
1116 | /* Partial Content even though we did not ask for it??? */
|
---|
1117 | /* fall through */
|
---|
1118 | default:
|
---|
1119 | bb_error_msg_and_die("server returned error: %s", sanitize_string(G.wget_buf));
|
---|
1120 | }
|
---|
1121 |
|
---|
1122 | /*
|
---|
1123 | * Retrieve HTTP headers.
|
---|
1124 | */
|
---|
1125 | while ((str = gethdr(sfp)) != NULL) {
|
---|
1126 | static const char keywords[] ALIGN1 =
|
---|
1127 | "content-length\0""transfer-encoding\0""location\0";
|
---|
1128 | enum {
|
---|
1129 | KEY_content_length = 1, KEY_transfer_encoding, KEY_location
|
---|
1130 | };
|
---|
1131 | smalluint key;
|
---|
1132 |
|
---|
1133 | /* gethdr converted "FOO:" string to lowercase */
|
---|
1134 |
|
---|
1135 | /* strip trailing whitespace */
|
---|
1136 | char *s = strchrnul(str, '\0') - 1;
|
---|
1137 | while (s >= str && (*s == ' ' || *s == '\t')) {
|
---|
1138 | *s = '\0';
|
---|
1139 | s--;
|
---|
1140 | }
|
---|
1141 | key = index_in_strings(keywords, G.wget_buf) + 1;
|
---|
1142 | if (key == KEY_content_length) {
|
---|
1143 | G.content_len = BB_STRTOOFF(str, NULL, 10);
|
---|
1144 | if (G.content_len < 0 || errno) {
|
---|
1145 | bb_error_msg_and_die("content-length %s is garbage", sanitize_string(str));
|
---|
1146 | }
|
---|
1147 | G.got_clen = 1;
|
---|
1148 | continue;
|
---|
1149 | }
|
---|
1150 | if (key == KEY_transfer_encoding) {
|
---|
1151 | if (strcmp(str_tolower(str), "chunked") != 0)
|
---|
1152 | bb_error_msg_and_die("transfer encoding '%s' is not supported", sanitize_string(str));
|
---|
1153 | G.chunked = 1;
|
---|
1154 | }
|
---|
1155 | if (key == KEY_location && status >= 300) {
|
---|
1156 | if (--redir_limit == 0)
|
---|
1157 | bb_error_msg_and_die("too many redirections");
|
---|
1158 | fclose(sfp);
|
---|
1159 | if (str[0] == '/') {
|
---|
1160 | free(redirected_path);
|
---|
1161 | target.path = redirected_path = xstrdup(str+1);
|
---|
1162 | /* lsa stays the same: it's on the same server */
|
---|
1163 | } else {
|
---|
1164 | parse_url(str, &target);
|
---|
1165 | if (!use_proxy) {
|
---|
1166 | /* server.user remains untouched */
|
---|
1167 | free(server.allocated);
|
---|
1168 | server.allocated = NULL;
|
---|
1169 | server.host = target.host;
|
---|
1170 | /* strip_ipv6_scope_id(target.host); - no! */
|
---|
1171 | /* we assume remote never gives us IPv6 addr with scope id */
|
---|
1172 | server.port = target.port;
|
---|
1173 | free(lsa);
|
---|
1174 | goto resolve_lsa;
|
---|
1175 | } /* else: lsa stays the same: we use proxy */
|
---|
1176 | }
|
---|
1177 | goto establish_session;
|
---|
1178 | }
|
---|
1179 | }
|
---|
1180 | // if (status >= 300)
|
---|
1181 | // bb_error_msg_and_die("bad redirection (no Location: header from server)");
|
---|
1182 |
|
---|
1183 | /* For HTTP, data is pumped over the same connection */
|
---|
1184 | dfp = sfp;
|
---|
1185 | } else {
|
---|
1186 | /*
|
---|
1187 | * FTP session
|
---|
1188 | */
|
---|
1189 | sfp = prepare_ftp_session(&dfp, &target, lsa);
|
---|
1190 | }
|
---|
1191 |
|
---|
1192 | free(lsa);
|
---|
1193 |
|
---|
1194 | if (!(option_mask32 & WGET_OPT_SPIDER)) {
|
---|
1195 | if (G.output_fd < 0)
|
---|
1196 | G.output_fd = xopen(G.fname_out, G.o_flags);
|
---|
1197 | retrieve_file_data(dfp);
|
---|
1198 | if (!(option_mask32 & WGET_OPT_OUTNAME)) {
|
---|
1199 | xclose(G.output_fd);
|
---|
1200 | G.output_fd = -1;
|
---|
1201 | }
|
---|
1202 | }
|
---|
1203 |
|
---|
1204 | if (dfp != sfp) {
|
---|
1205 | /* It's ftp. Close data connection properly */
|
---|
1206 | fclose(dfp);
|
---|
1207 | if (ftpcmd(NULL, NULL, sfp) != 226)
|
---|
1208 | bb_error_msg_and_die("ftp error: %s", sanitize_string(G.wget_buf + 4));
|
---|
1209 | /* ftpcmd("QUIT", NULL, sfp); - why bother? */
|
---|
1210 | }
|
---|
1211 | fclose(sfp);
|
---|
1212 |
|
---|
1213 | free(server.allocated);
|
---|
1214 | free(target.allocated);
|
---|
1215 | free(server.user);
|
---|
1216 | free(target.user);
|
---|
1217 | free(fname_out_alloc);
|
---|
1218 | free(redirected_path);
|
---|
1219 | }
|
---|
1220 |
|
---|
1221 | int wget_main(int argc, char **argv) MAIN_EXTERNALLY_VISIBLE;
|
---|
1222 | int wget_main(int argc UNUSED_PARAM, char **argv)
|
---|
1223 | {
|
---|
1224 | #if ENABLE_FEATURE_WGET_LONG_OPTIONS
|
---|
1225 | static const char wget_longopts[] ALIGN1 =
|
---|
1226 | /* name, has_arg, val */
|
---|
1227 | "continue\0" No_argument "c"
|
---|
1228 | //FIXME: -s isn't --spider, it's --save-headers!
|
---|
1229 | "spider\0" No_argument "s"
|
---|
1230 | "quiet\0" No_argument "q"
|
---|
1231 | "output-document\0" Required_argument "O"
|
---|
1232 | "directory-prefix\0" Required_argument "P"
|
---|
1233 | "proxy\0" Required_argument "Y"
|
---|
1234 | "user-agent\0" Required_argument "U"
|
---|
1235 | IF_FEATURE_WGET_TIMEOUT(
|
---|
1236 | "timeout\0" Required_argument "T")
|
---|
1237 | /* Ignored: */
|
---|
1238 | IF_DESKTOP( "tries\0" Required_argument "t")
|
---|
1239 | "header\0" Required_argument "\xff"
|
---|
1240 | "post-data\0" Required_argument "\xfe"
|
---|
1241 | /* Ignored (we always use PASV): */
|
---|
1242 | IF_DESKTOP( "passive-ftp\0" No_argument "\xf0")
|
---|
1243 | /* Ignored (we don't do ssl) */
|
---|
1244 | IF_DESKTOP( "no-check-certificate\0" No_argument "\xf0")
|
---|
1245 | /* Ignored (we don't support caching) */
|
---|
1246 | IF_DESKTOP( "no-cache\0" No_argument "\xf0")
|
---|
1247 | IF_DESKTOP( "no-verbose\0" No_argument "\xf0")
|
---|
1248 | IF_DESKTOP( "no-clobber\0" No_argument "\xf0")
|
---|
1249 | IF_DESKTOP( "no-host-directories\0" No_argument "\xf0")
|
---|
1250 | IF_DESKTOP( "no-parent\0" No_argument "\xf0")
|
---|
1251 | ;
|
---|
1252 | #endif
|
---|
1253 |
|
---|
1254 | #if ENABLE_FEATURE_WGET_LONG_OPTIONS
|
---|
1255 | llist_t *headers_llist = NULL;
|
---|
1256 | #endif
|
---|
1257 |
|
---|
1258 | INIT_G();
|
---|
1259 |
|
---|
1260 | #if ENABLE_FEATURE_WGET_TIMEOUT
|
---|
1261 | G.timeout_seconds = 900;
|
---|
1262 | signal(SIGALRM, alarm_handler);
|
---|
1263 | #endif
|
---|
1264 | G.proxy_flag = "on"; /* use proxies if env vars are set */
|
---|
1265 | G.user_agent = "Wget"; /* "User-Agent" header field */
|
---|
1266 |
|
---|
1267 | #if ENABLE_FEATURE_WGET_LONG_OPTIONS
|
---|
1268 | applet_long_options = wget_longopts;
|
---|
1269 | #endif
|
---|
1270 | opt_complementary = "-1" /* at least one URL */
|
---|
1271 | IF_FEATURE_WGET_TIMEOUT(":T+") /* -T NUM */
|
---|
1272 | IF_FEATURE_WGET_LONG_OPTIONS(":\xff::"); /* --header is a list */
|
---|
1273 | getopt32(argv, "csqO:P:Y:U:T:"
|
---|
1274 | /*ignored:*/ "t:"
|
---|
1275 | /*ignored:*/ "n::"
|
---|
1276 | /* wget has exactly four -n<letter> opts, all of which we can ignore:
|
---|
1277 | * -nv --no-verbose: be moderately quiet (-q is full quiet)
|
---|
1278 | * -nc --no-clobber: abort if exists, neither download to FILE.n nor overwrite FILE
|
---|
1279 | * -nH --no-host-directories: wget -r http://host/ won't create host/
|
---|
1280 | * -np --no-parent
|
---|
1281 | * "n::" above says that we accept -n[ARG].
|
---|
1282 | * Specifying "n:" would be a bug: "-n ARG" would eat ARG!
|
---|
1283 | */
|
---|
1284 | , &G.fname_out, &G.dir_prefix,
|
---|
1285 | &G.proxy_flag, &G.user_agent,
|
---|
1286 | IF_FEATURE_WGET_TIMEOUT(&G.timeout_seconds) IF_NOT_FEATURE_WGET_TIMEOUT(NULL),
|
---|
1287 | NULL, /* -t RETRIES */
|
---|
1288 | NULL /* -n[ARG] */
|
---|
1289 | IF_FEATURE_WGET_LONG_OPTIONS(, &headers_llist)
|
---|
1290 | IF_FEATURE_WGET_LONG_OPTIONS(, &G.post_data)
|
---|
1291 | );
|
---|
1292 | argv += optind;
|
---|
1293 |
|
---|
1294 | #if ENABLE_FEATURE_WGET_LONG_OPTIONS
|
---|
1295 | if (headers_llist) {
|
---|
1296 | int size = 0;
|
---|
1297 | char *hdr;
|
---|
1298 | llist_t *ll = headers_llist;
|
---|
1299 | while (ll) {
|
---|
1300 | size += strlen(ll->data) + 2;
|
---|
1301 | ll = ll->link;
|
---|
1302 | }
|
---|
1303 | G.extra_headers = hdr = xmalloc(size + 1);
|
---|
1304 | while (headers_llist) {
|
---|
1305 | int bit;
|
---|
1306 | const char *words;
|
---|
1307 |
|
---|
1308 | size = sprintf(hdr, "%s\r\n",
|
---|
1309 | (char*)llist_pop(&headers_llist));
|
---|
1310 | /* a bit like index_in_substrings but don't match full key */
|
---|
1311 | bit = 1;
|
---|
1312 | words = wget_user_headers;
|
---|
1313 | while (*words) {
|
---|
1314 | if (strstr(hdr, words) == hdr) {
|
---|
1315 | G.user_headers |= bit;
|
---|
1316 | break;
|
---|
1317 | }
|
---|
1318 | bit <<= 1;
|
---|
1319 | words += strlen(words) + 1;
|
---|
1320 | }
|
---|
1321 | hdr += size;
|
---|
1322 | }
|
---|
1323 | }
|
---|
1324 | #endif
|
---|
1325 |
|
---|
1326 | G.output_fd = -1;
|
---|
1327 | G.o_flags = O_WRONLY | O_CREAT | O_TRUNC | O_EXCL;
|
---|
1328 | if (G.fname_out) { /* -O FILE ? */
|
---|
1329 | if (LONE_DASH(G.fname_out)) { /* -O - ? */
|
---|
1330 | G.output_fd = 1;
|
---|
1331 | option_mask32 &= ~WGET_OPT_CONTINUE;
|
---|
1332 | }
|
---|
1333 | /* compat with wget: -O FILE can overwrite */
|
---|
1334 | G.o_flags = O_WRONLY | O_CREAT | O_TRUNC;
|
---|
1335 | }
|
---|
1336 |
|
---|
1337 | while (*argv)
|
---|
1338 | download_one_url(*argv++);
|
---|
1339 |
|
---|
1340 | if (G.output_fd >= 0)
|
---|
1341 | xclose(G.output_fd);
|
---|
1342 |
|
---|
1343 | #if ENABLE_FEATURE_CLEAN_UP && ENABLE_FEATURE_WGET_LONG_OPTIONS
|
---|
1344 | free(G.extra_headers);
|
---|
1345 | #endif
|
---|
1346 | FINI_G();
|
---|
1347 |
|
---|
1348 | return EXIT_SUCCESS;
|
---|
1349 | }
|
---|