* Licensed under GPLv2, see file LICENSE in this source tree.
*
* Copyright (C) 2010 Bradley M. Kuhn <bkuhn@ebb.org>
- * Kuhn's copyrights are licensed GPLv2-or-later. File as a whole remains GPLv2.
+ * Kuhn's copyrights are licensed GPLv2-or-later. File as a whole remains GPLv2.
*/
+
+//config:config WGET
+//config: bool "wget (35 kb)"
+//config: default y
+//config: help
+//config: wget is a utility for non-interactive download of files from HTTP
+//config: and FTP servers.
+//config:
+//config:config FEATURE_WGET_LONG_OPTIONS
+//config: bool "Enable long options"
+//config: default y
+//config: depends on WGET && LONG_OPTS
+//config:
+//config:config FEATURE_WGET_STATUSBAR
+//config: bool "Enable progress bar (+2k)"
+//config: default y
+//config: depends on WGET
+//config:
+//config:config FEATURE_WGET_AUTHENTICATION
+//config: bool "Enable HTTP authentication"
+//config: default y
+//config: depends on WGET
+//config: help
+//config: Support authenticated HTTP transfers.
+//config:
+//config:config FEATURE_WGET_TIMEOUT
+//config: bool "Enable timeout option -T SEC"
+//config: default y
+//config: depends on WGET
+//config: help
+//config: Supports network read and connect timeouts for wget,
+//config: so that wget will give up and timeout, through the -T
+//config: command line option.
+//config:
+//config: Currently only connect and network data read timeout are
+//config: supported (i.e., timeout is not applied to the DNS query). When
+//config: FEATURE_WGET_LONG_OPTIONS is also enabled, the --timeout option
+//config: will work in addition to -T.
+//config:
+//config:config FEATURE_WGET_HTTPS
+//config: bool "Support HTTPS using internal TLS code"
+//config: default y
+//config: depends on WGET
+//config: select TLS
+//config: help
+//config: wget will use internal TLS code to connect to https:// URLs.
+//config: Note:
+//config: On NOMMU machines, ssl_helper applet should be available
+//config: in the $PATH for this to work. Make sure to select that applet.
+//config:
+//config: Note: currently, TLS code only makes TLS I/O work, it
+//config: does *not* check that the peer is who it claims to be, etc.
+//config: IOW: it uses peer-supplied public keys to establish encryption
+//config: and signing keys, then encrypts and signs outgoing data and
+//config: decrypts incoming data.
+//config: It does not check signature hashes on the incoming data:
+//config: this means that attackers manipulating TCP packets can
+//config: send altered data and we unknowingly receive garbage.
+//config: (This check might be relatively easy to add).
+//config: It does not check public key's certificate:
+//config: this means that the peer may be an attacker impersonating
+//config: the server we think we are talking to.
+//config:
+//config: If you think this is unacceptable, consider this. As more and more
+//config: servers switch to HTTPS-only operation, without such "crippled"
+//config: TLS code it is *impossible* to simply download a kernel source
+//config: from kernel.org. Which can in real world translate into
+//config: "my small automatic tooling to build cross-compilers from sources
+//config: no longer works, I need to additionally keep a local copy
+//config: of ~4 megabyte source tarball of a SSL library and ~2 megabyte
+//config: source of wget, need to compile and built both before I can
+//config: download anything. All this despite the fact that the build
+//config: is done in a QEMU sandbox on a machine with absolutely nothing
+//config: worth stealing, so I don't care if someone would go to a lot
+//config: of trouble to intercept my HTTPS download to send me an altered
+//config: kernel tarball".
+//config:
+//config: If you still think this is unacceptable, send patches.
+//config:
+//config: If you still think this is unacceptable, do not want to send
+//config: patches, but do want to waste bandwidth expaining how wrong
+//config: it is, you will be ignored.
+//config:
+//config:config FEATURE_WGET_OPENSSL
+//config: bool "Try to connect to HTTPS using openssl"
+//config: default y
+//config: depends on WGET
+//config: help
+//config: Try to use openssl to handle HTTPS.
+//config:
+//config: OpenSSL has a simple SSL client for debug purposes.
+//config: If you select this option, wget will effectively run:
+//config: "openssl s_client -quiet -connect hostname:443
+//config: -servername hostname 2>/dev/null" and pipe its data
+//config: through it. -servername is not used if hostname is numeric.
+//config: Note inconvenient API: host resolution is done twice,
+//config: and there is no guarantee openssl's idea of IPv6 address
+//config: format is the same as ours.
+//config: Another problem is that s_client prints debug information
+//config: to stderr, and it needs to be suppressed. This means
+//config: all error messages get suppressed too.
+//config: openssl is also a big binary, often dynamically linked
+//config: against ~15 libraries.
+//config:
+//config: If openssl can't be executed, internal TLS code will be used
+//config: (if you enabled it); if openssl can be executed but fails later,
+//config: wget can't detect this, and download will fail.
+
+//applet:IF_WGET(APPLET(wget, BB_DIR_USR_BIN, BB_SUID_DROP))
+
+//kbuild:lib-$(CONFIG_WGET) += wget.o
+
+//usage:#define wget_trivial_usage
+//usage: IF_FEATURE_WGET_LONG_OPTIONS(
+//usage: "[-c|--continue] [--spider] [-q|--quiet] [-O|--output-document FILE]\n"
+//usage: " [--header 'header: value'] [-Y|--proxy on/off] [-P DIR]\n"
+/* Since we ignore these opts, we don't show them in --help */
+/* //usage: " [--no-check-certificate] [--no-cache] [--passive-ftp] [-t TRIES]" */
+/* //usage: " [-nv] [-nc] [-nH] [-np]" */
+//usage: " [-S|--server-response] [-U|--user-agent AGENT]" IF_FEATURE_WGET_TIMEOUT(" [-T SEC]") " URL..."
+//usage: )
+//usage: IF_NOT_FEATURE_WGET_LONG_OPTIONS(
+//usage: "[-cq] [-O FILE] [-Y on/off] [-P DIR] [-S] [-U AGENT]"
+//usage: IF_FEATURE_WGET_TIMEOUT(" [-T SEC]") " URL..."
+//usage: )
+//usage:#define wget_full_usage "\n\n"
+//usage: "Retrieve files via HTTP or FTP\n"
+//usage: IF_FEATURE_WGET_LONG_OPTIONS(
+//usage: "\n --spider Only check URL existence: $? is 0 if exists"
+//usage: )
+//usage: "\n -c Continue retrieval of aborted transfer"
+//usage: "\n -q Quiet"
+//usage: "\n -P DIR Save to DIR (default .)"
+//usage: "\n -S Show server response"
+//usage: IF_FEATURE_WGET_TIMEOUT(
+//usage: "\n -T SEC Network read timeout is SEC seconds"
+//usage: )
+//usage: "\n -O FILE Save to FILE ('-' for stdout)"
+//usage: "\n -U STR Use STR for User-Agent header"
+//usage: "\n -Y on/off Use proxy"
+
#include "libbb.h"
+#if 0
+# define log_io(...) bb_error_msg(__VA_ARGS__)
+# define SENDFMT(fp, fmt, ...) \
+ do { \
+ log_io("> " fmt, ##__VA_ARGS__); \
+ fprintf(fp, fmt, ##__VA_ARGS__); \
+ } while (0);
+#else
+# define log_io(...) ((void)0)
+# define SENDFMT(fp, fmt, ...) fprintf(fp, fmt, ##__VA_ARGS__)
+#endif
+
+
+#define SSL_SUPPORTED (ENABLE_FEATURE_WGET_OPENSSL || ENABLE_FEATURE_WGET_HTTPS)
+
struct host_info {
- // May be used if we ever will want to free() all xstrdup()s...
- /* char *allocated; */
+ char *allocated;
const char *path;
- const char *user;
+ char *user;
+ const char *protocol;
char *host;
int port;
- smallint is_ftp;
};
+static const char P_FTP[] ALIGN1 = "ftp";
+static const char P_HTTP[] ALIGN1 = "http";
+#if SSL_SUPPORTED
+static const char P_HTTPS[] ALIGN1 = "https";
+#endif
+#if ENABLE_FEATURE_WGET_LONG_OPTIONS
+/* User-specified headers prevent using our corresponding built-in headers. */
+enum {
+ HDR_HOST = (1<<0),
+ HDR_USER_AGENT = (1<<1),
+ HDR_RANGE = (1<<2),
+ HDR_AUTH = (1<<3) * ENABLE_FEATURE_WGET_AUTHENTICATION,
+ HDR_PROXY_AUTH = (1<<4) * ENABLE_FEATURE_WGET_AUTHENTICATION,
+};
+static const char wget_user_headers[] ALIGN1 =
+ "Host:\0"
+ "User-Agent:\0"
+ "Range:\0"
+# if ENABLE_FEATURE_WGET_AUTHENTICATION
+ "Authorization:\0"
+ "Proxy-Authorization:\0"
+# endif
+ ;
+# define USR_HEADER_HOST (G.user_headers & HDR_HOST)
+# define USR_HEADER_USER_AGENT (G.user_headers & HDR_USER_AGENT)
+# define USR_HEADER_RANGE (G.user_headers & HDR_RANGE)
+# define USR_HEADER_AUTH (G.user_headers & HDR_AUTH)
+# define USR_HEADER_PROXY_AUTH (G.user_headers & HDR_PROXY_AUTH)
+#else /* No long options, no user-headers :( */
+# define USR_HEADER_HOST 0
+# define USR_HEADER_USER_AGENT 0
+# define USR_HEADER_RANGE 0
+# define USR_HEADER_AUTH 0
+# define USR_HEADER_PROXY_AUTH 0
+#endif
/* Globals */
struct globals {
const char *curfile; /* Name of current file being transferred */
bb_progress_t pmt;
#endif
+ char *dir_prefix;
+#if ENABLE_FEATURE_WGET_LONG_OPTIONS
+ char *post_data;
+ char *extra_headers;
+ unsigned char user_headers; /* Headers mentioned by the user */
+#endif
+ char *fname_out; /* where to direct output (-O) */
+ const char *proxy_flag; /* Use proxies if env vars are set */
+ const char *user_agent; /* "User-Agent" header field */
#if ENABLE_FEATURE_WGET_TIMEOUT
unsigned timeout_seconds;
+ bool die_if_timed_out;
#endif
+ int output_fd;
+ int o_flags;
smallint chunked; /* chunked transfer encoding */
smallint got_clen; /* got content-length: from server */
+ /* Local downloads do benefit from big buffer.
+ * With 512 byte buffer, it was measured to be
+ * an order of magnitude slower than with big one.
+ */
+ uint64_t just_to_align_next_member;
+ char wget_buf[CONFIG_FEATURE_COPYBUF_KB*1024];
} FIX_ALIASING;
-#define G (*(struct globals*)&bb_common_bufsiz1)
-struct BUG_G_too_big {
- char BUG_G_too_big[sizeof(G) <= COMMON_BUFSIZE ? 1 : -1];
-};
+#define G (*ptr_to_globals)
#define INIT_G() do { \
- IF_FEATURE_WGET_TIMEOUT(G.timeout_seconds = 900;) \
+ SET_PTR_TO_GLOBALS(xzalloc(sizeof(G))); \
+} while (0)
+#define FINI_G() do { \
+ FREE_PTR_TO_GLOBALS(); \
} while (0)
/* Must match option string! */
enum {
WGET_OPT_CONTINUE = (1 << 0),
- WGET_OPT_SPIDER = (1 << 1),
- WGET_OPT_QUIET = (1 << 2),
+ WGET_OPT_QUIET = (1 << 1),
+ WGET_OPT_SERVER_RESPONSE = (1 << 2),
WGET_OPT_OUTNAME = (1 << 3),
WGET_OPT_PREFIX = (1 << 4),
WGET_OPT_PROXY = (1 << 5),
WGET_OPT_USER_AGENT = (1 << 6),
WGET_OPT_NETWORK_READ_TIMEOUT = (1 << 7),
WGET_OPT_RETRIES = (1 << 8),
- WGET_OPT_PASSIVE = (1 << 9),
+ WGET_OPT_nsomething = (1 << 9),
WGET_OPT_HEADER = (1 << 10) * ENABLE_FEATURE_WGET_LONG_OPTIONS,
WGET_OPT_POST_DATA = (1 << 11) * ENABLE_FEATURE_WGET_LONG_OPTIONS,
+ WGET_OPT_SPIDER = (1 << 12) * ENABLE_FEATURE_WGET_LONG_OPTIONS,
};
enum {
return;
if (flag == PROGRESS_START)
- bb_progress_init(&G.pmt);
+ bb_progress_init(&G.pmt, G.curfile);
- bb_progress_update(&G.pmt, G.curfile, G.beg_range, G.transferred,
- G.chunked ? 0 : G.beg_range + G.transferred + G.content_len);
+ bb_progress_update(&G.pmt,
+ G.beg_range,
+ G.transferred,
+ (G.chunked || !G.got_clen) ? 0 : G.beg_range + G.transferred + G.content_len
+ );
if (flag == PROGRESS_END) {
+ bb_progress_free(&G.pmt);
bb_putchar_stderr('\n');
G.transferred = 0;
}
overlapping_strcpy(scope, cp);
}
-/* Read NMEMB bytes into PTR from STREAM. Returns the number of bytes read,
- * and a short count if an eof or non-interrupt error is encountered. */
-static size_t safe_fread(void *ptr, size_t nmemb, FILE *stream)
-{
- size_t ret;
- char *p = (char*)ptr;
-
- do {
- clearerr(stream);
- errno = 0;
- ret = fread(p, 1, nmemb, stream);
- p += ret;
- nmemb -= ret;
- } while (nmemb && ferror(stream) && errno == EINTR);
-
- return p - (char*)ptr;
-}
-
-/* Read a line or SIZE-1 bytes into S, whichever is less, from STREAM.
- * Returns S, or NULL if an eof or non-interrupt error is encountered. */
-static char *safe_fgets(char *s, int size, FILE *stream)
-{
- char *ret;
-
- do {
- clearerr(stream);
- errno = 0;
- ret = fgets(s, size, stream);
- } while (ret == NULL && ferror(stream) && errno == EINTR);
-
- return ret;
-}
-
#if ENABLE_FEATURE_WGET_AUTHENTICATION
-/* Base64-encode character string. buf is assumed to be char buf[512]. */
-static char *base64enc_512(char buf[512], const char *str)
+/* Base64-encode character string. */
+static char *base64enc(const char *str)
{
unsigned len = strlen(str);
- if (len > 512/4*3 - 10) /* paranoia */
- len = 512/4*3 - 10;
- bb_uuencode(buf, str, len, bb_uuenc_tbl_base64);
- return buf;
+ if (len > sizeof(G.wget_buf)/4*3 - 10) /* paranoia */
+ len = sizeof(G.wget_buf)/4*3 - 10;
+ bb_uuencode(G.wget_buf, str, len, bb_uuenc_tbl_base64);
+ return G.wget_buf;
}
#endif
return s;
}
+#if ENABLE_FEATURE_WGET_TIMEOUT
+static void alarm_handler(int sig UNUSED_PARAM)
+{
+ /* This is theoretically unsafe (uses stdio and malloc in signal handler) */
+ if (G.die_if_timed_out)
+ bb_error_msg_and_die("download timed out");
+}
+static void set_alarm(void)
+{
+ if (G.timeout_seconds) {
+ alarm(G.timeout_seconds);
+ G.die_if_timed_out = 1;
+ }
+}
+# define clear_alarm() ((void)(G.die_if_timed_out = 0))
+#else
+# define set_alarm() ((void)0)
+# define clear_alarm() ((void)0)
+#endif
+
+#if ENABLE_FEATURE_WGET_OPENSSL
+/*
+ * is_ip_address() attempts to verify whether or not a string
+ * contains an IPv4 or IPv6 address (vs. an FQDN). The result
+ * of inet_pton() can be used to determine this.
+ *
+ * TODO add proper error checking when inet_pton() returns -1
+ * (some form of system error has occurred, and errno is set)
+ */
+static int is_ip_address(const char *string)
+{
+ struct sockaddr_in sa;
+
+ int result = inet_pton(AF_INET, string, &(sa.sin_addr));
+# if ENABLE_FEATURE_IPV6
+ if (result == 0) {
+ struct sockaddr_in6 sa6;
+ result = inet_pton(AF_INET6, string, &(sa6.sin6_addr));
+ }
+# endif
+ return (result == 1);
+}
+#endif
+
static FILE *open_socket(len_and_sockaddr *lsa)
{
+ int fd;
FILE *fp;
+ set_alarm();
+ fd = xconnect_stream(lsa);
+ clear_alarm();
+
/* glibc 2.4 seems to try seeking on it - ??! */
/* hopefully it understands what ESPIPE means... */
- fp = fdopen(xconnect_stream(lsa), "r+");
- if (fp == NULL)
- bb_perror_msg_and_die("fdopen");
+ fp = fdopen(fd, "r+");
+ if (!fp)
+ bb_perror_msg_and_die(bb_msg_memory_exhausted);
return fp;
}
-static int ftpcmd(const char *s1, const char *s2, FILE *fp, char *buf)
+/* Returns '\n' if it was seen, else '\0'. Trims at first '\r' or '\n' */
+static char fgets_and_trim(FILE *fp, const char *fmt)
+{
+ char c;
+ char *buf_ptr;
+
+ set_alarm();
+ if (fgets(G.wget_buf, sizeof(G.wget_buf) - 1, fp) == NULL)
+ bb_perror_msg_and_die("error getting response");
+ clear_alarm();
+
+ buf_ptr = strchrnul(G.wget_buf, '\n');
+ c = *buf_ptr;
+ *buf_ptr = '\0';
+ buf_ptr = strchrnul(G.wget_buf, '\r');
+ *buf_ptr = '\0';
+
+ log_io("< %s", G.wget_buf);
+
+ if (fmt && (option_mask32 & WGET_OPT_SERVER_RESPONSE))
+ fprintf(stderr, fmt, G.wget_buf);
+
+ return c;
+}
+
+static int ftpcmd(const char *s1, const char *s2, FILE *fp)
{
int result;
if (s1) {
- if (!s2) s2 = "";
+ if (!s2)
+ s2 = "";
fprintf(fp, "%s%s\r\n", s1, s2);
+ /* With --server-response, wget also shows its ftp commands */
+ if (option_mask32 & WGET_OPT_SERVER_RESPONSE)
+ fprintf(stderr, "--> %s%s\n\n", s1, s2);
fflush(fp);
+ log_io("> %s%s", s1, s2);
}
do {
- char *buf_ptr;
+ fgets_and_trim(fp, "%s\n");
+ } while (!isdigit(G.wget_buf[0]) || G.wget_buf[3] != ' ');
- if (fgets(buf, 510, fp) == NULL) {
- bb_perror_msg_and_die("error getting response");
- }
- buf_ptr = strstr(buf, "\r\n");
- if (buf_ptr) {
- *buf_ptr = '\0';
- }
- } while (!isdigit(buf[0]) || buf[3] != ' ');
-
- buf[3] = '\0';
- result = xatoi_positive(buf);
- buf[3] = ' ';
+ G.wget_buf[3] = '\0';
+ result = xatoi_positive(G.wget_buf);
+ G.wget_buf[3] = ' ';
return result;
}
-static void parse_url(char *src_url, struct host_info *h)
+static void parse_url(const char *src_url, struct host_info *h)
{
char *url, *p, *sp;
- /* h->allocated = */ url = xstrdup(src_url);
+ free(h->allocated);
+ h->allocated = url = xstrdup(src_url);
- if (strncmp(url, "http://", 7) == 0) {
- h->port = bb_lookup_port("http", "tcp", 80);
- h->host = url + 7;
- h->is_ftp = 0;
- } else if (strncmp(url, "ftp://", 6) == 0) {
- h->port = bb_lookup_port("ftp", "tcp", 21);
- h->host = url + 6;
- h->is_ftp = 1;
- } else
- bb_error_msg_and_die("not an http or ftp url: %s", sanitize_string(url));
+ h->protocol = P_FTP;
+ p = strstr(url, "://");
+ if (p) {
+ *p = '\0';
+ h->host = p + 3;
+ if (strcmp(url, P_FTP) == 0) {
+ h->port = bb_lookup_port(P_FTP, "tcp", 21);
+ } else
+#if SSL_SUPPORTED
+ if (strcmp(url, P_HTTPS) == 0) {
+ h->port = bb_lookup_port(P_HTTPS, "tcp", 443);
+ h->protocol = P_HTTPS;
+ } else
+#endif
+ if (strcmp(url, P_HTTP) == 0) {
+ http:
+ h->port = bb_lookup_port(P_HTTP, "tcp", 80);
+ h->protocol = P_HTTP;
+ } else {
+ *p = ':';
+ bb_error_msg_and_die("not an http or ftp url: %s", sanitize_string(url));
+ }
+ } else {
+ // GNU wget is user-friendly and falls back to http://
+ h->host = url;
+ goto http;
+ }
// FYI:
// "Real" wget 'http://busybox.net?var=a/b' sends this request:
- // 'GET /?var=a/b HTTP 1.0'
+ // 'GET /?var=a/b HTTP/1.0'
// and saves 'index.html?var=a%2Fb' (we save 'b')
// wget 'http://busybox.net?login=john@doe':
// request: 'GET /?login=john@doe HTTP/1.0'
h->path = sp;
}
- // We used to set h->user to NULL here, but this interferes
- // with handling of code 302 ("object was moved")
-
sp = strrchr(h->host, '@');
if (sp != NULL) {
- h->user = h->host;
+ // URL-decode "user:password" string before base64-encoding:
+ // wget http://test:my%20pass@example.com should send
+ // Authorization: Basic dGVzdDpteSBwYXNz
+ // which decodes to "test:my pass".
+ // Standard wget and curl do this too.
*sp = '\0';
+ free(h->user);
+ h->user = xstrdup(percent_decode_in_place(h->host, /*strict:*/ 0));
h->host = sp + 1;
}
-
- sp = h->host;
+ /* else: h->user remains NULL, or as set by original request
+ * before redirect (if we are here after a redirect).
+ */
}
-static char *gethdr(char *buf, size_t bufsiz, FILE *fp /*, int *istrunc*/)
+static char *gethdr(FILE *fp)
{
char *s, *hdrval;
int c;
- /* *istrunc = 0; */
-
/* retrieve header line */
- if (fgets(buf, bufsiz, fp) == NULL)
- return NULL;
+ c = fgets_and_trim(fp, " %s\n");
- /* see if we are at the end of the headers */
- for (s = buf; *s == '\r'; ++s)
- continue;
- if (*s == '\n')
+ /* end of the headers? */
+ if (G.wget_buf[0] == '\0')
return NULL;
/* convert the header name to lower case */
- for (s = buf; isalnum(*s) || *s == '-' || *s == '.'; ++s) {
- /* tolower for "A-Z", no-op for "0-9a-z-." */
- *s = (*s | 0x20);
+ for (s = G.wget_buf; isalnum(*s) || *s == '-' || *s == '.' || *s == '_'; ++s) {
+ /*
+ * No-op for 20-3f and 60-7f. "0-9a-z-." are in these ranges.
+ * 40-5f range ("@A-Z[\]^_") maps to 60-7f.
+ * "A-Z" maps to "a-z".
+ * "@[\]" can't occur in header names.
+ * "^_" maps to "~,DEL" (which is wrong).
+ * "^" was never seen yet, "_" was seen from web.archive.org
+ * (x-archive-orig-x_commoncrawl_Signature: HEXSTRING).
+ */
+ *s |= 0x20;
}
/* verify we are at the end of the header name */
if (*s != ':')
- bb_error_msg_and_die("bad header line: %s", sanitize_string(buf));
+ bb_error_msg_and_die("bad header line: %s", sanitize_string(G.wget_buf));
/* locate the start of the header value */
*s++ = '\0';
hdrval = skip_whitespace(s);
- /* locate the end of header */
- while (*s && *s != '\r' && *s != '\n')
- ++s;
-
- /* end of header found */
- if (*s) {
- *s = '\0';
- return hdrval;
+ if (c != '\n') {
+ /* Rats! The buffer isn't big enough to hold the entire header value */
+ while (c = getc(fp), c != EOF && c != '\n')
+ continue;
}
- /* Rats! The buffer isn't big enough to hold the entire header value */
- while (c = getc(fp), c != EOF && c != '\n')
- continue;
- /* *istrunc = 1; */
return hdrval;
}
-#if ENABLE_FEATURE_WGET_LONG_OPTIONS
-static char *URL_escape(const char *str)
+static void reset_beg_range_to_zero(void)
{
- /* URL encode, see RFC 2396 */
- char *dst;
- char *res = dst = xmalloc(strlen(str) * 3 + 1);
- unsigned char c;
-
- while (1) {
- c = *str++;
- if (c == '\0'
- /* || strchr("!&'()*-.=_~", c) - more code */
- || c == '!'
- || c == '&'
- || c == '\''
- || c == '('
- || c == ')'
- || c == '*'
- || c == '-'
- || c == '.'
- || c == '='
- || c == '_'
- || c == '~'
- || (c >= '0' && c <= '9')
- || ((c|0x20) >= 'a' && (c|0x20) <= 'z')
- ) {
- *dst++ = c;
- if (c == '\0')
- return res;
- } else {
- *dst++ = '%';
- *dst++ = bb_hexdigits_upcase[c >> 4];
- *dst++ = bb_hexdigits_upcase[c & 0xf];
- }
- }
+ bb_error_msg("restart failed");
+ G.beg_range = 0;
+ xlseek(G.output_fd, 0, SEEK_SET);
+ /* Done at the end instead: */
+ /* ftruncate(G.output_fd, 0); */
}
-#endif
static FILE* prepare_ftp_session(FILE **dfpp, struct host_info *target, len_and_sockaddr *lsa)
{
- char buf[512];
FILE *sfp;
char *str;
int port;
target->user = xstrdup("anonymous:busybox@");
sfp = open_socket(lsa);
- if (ftpcmd(NULL, NULL, sfp, buf) != 220)
- bb_error_msg_and_die("%s", sanitize_string(buf+4));
+ if (ftpcmd(NULL, NULL, sfp) != 220)
+ bb_error_msg_and_die("%s", sanitize_string(G.wget_buf + 4));
/*
* Splitting username:password pair,
str = strchr(target->user, ':');
if (str)
*str++ = '\0';
- switch (ftpcmd("USER ", target->user, sfp, buf)) {
+ switch (ftpcmd("USER ", target->user, sfp)) {
case 230:
break;
case 331:
- if (ftpcmd("PASS ", str, sfp, buf) == 230)
+ if (ftpcmd("PASS ", str, sfp) == 230)
break;
/* fall through (failed login) */
default:
- bb_error_msg_and_die("ftp login: %s", sanitize_string(buf+4));
+ bb_error_msg_and_die("ftp login: %s", sanitize_string(G.wget_buf + 4));
}
- ftpcmd("TYPE I", NULL, sfp, buf);
+ ftpcmd("TYPE I", NULL, sfp);
/*
* Querying file size
*/
- if (ftpcmd("SIZE ", target->path, sfp, buf) == 213) {
- G.content_len = BB_STRTOOFF(buf+4, NULL, 10);
+ if (ftpcmd("SIZE ", target->path, sfp) == 213) {
+ G.content_len = BB_STRTOOFF(G.wget_buf + 4, NULL, 10);
if (G.content_len < 0 || errno) {
bb_error_msg_and_die("SIZE value is garbage");
}
/*
* Entering passive mode
*/
- if (ftpcmd("PASV", NULL, sfp, buf) != 227) {
+ if (ftpcmd("PASV", NULL, sfp) != 227) {
pasv_error:
- bb_error_msg_and_die("bad response to %s: %s", "PASV", sanitize_string(buf));
+ bb_error_msg_and_die("bad response to %s: %s", "PASV", sanitize_string(G.wget_buf));
}
// Response is "227 garbageN1,N2,N3,N4,P1,P2[)garbage]
// Server's IP is N1.N2.N3.N4 (we ignore it)
// Server's port for data connection is P1*256+P2
- str = strrchr(buf, ')');
+ str = strrchr(G.wget_buf, ')');
if (str) str[0] = '\0';
- str = strrchr(buf, ',');
+ str = strrchr(G.wget_buf, ',');
if (!str) goto pasv_error;
port = xatou_range(str+1, 0, 255);
*str = '\0';
- str = strrchr(buf, ',');
+ str = strrchr(G.wget_buf, ',');
if (!str) goto pasv_error;
port += xatou_range(str+1, 0, 255) * 256;
- set_nport(lsa, htons(port));
+ set_nport(&lsa->u.sa, htons(port));
*dfpp = open_socket(lsa);
- if (G.beg_range) {
- sprintf(buf, "REST %"OFF_FMT"u", G.beg_range);
- if (ftpcmd(buf, NULL, sfp, buf) == 350)
+ if (G.beg_range != 0) {
+ sprintf(G.wget_buf, "REST %"OFF_FMT"u", G.beg_range);
+ if (ftpcmd(G.wget_buf, NULL, sfp) == 350)
G.content_len -= G.beg_range;
+ else
+ reset_beg_range_to_zero();
}
- if (ftpcmd("RETR ", target->path, sfp, buf) > 150)
- bb_error_msg_and_die("bad response to %s: %s", "RETR", sanitize_string(buf));
+ if (ftpcmd("RETR ", target->path, sfp) > 150)
+ bb_error_msg_and_die("bad response to %s: %s", "RETR", sanitize_string(G.wget_buf));
return sfp;
}
-static void NOINLINE retrieve_file_data(FILE *dfp, int output_fd)
+#if ENABLE_FEATURE_WGET_OPENSSL
+static int spawn_https_helper_openssl(const char *host, unsigned port)
+{
+ char *allocated = NULL;
+ char *servername;
+ int sp[2];
+ int pid;
+ IF_FEATURE_WGET_HTTPS(volatile int child_failed = 0;)
+
+ if (socketpair(AF_UNIX, SOCK_STREAM, 0, sp) != 0)
+ /* Kernel can have AF_UNIX support disabled */
+ bb_perror_msg_and_die("socketpair");
+
+ if (!strchr(host, ':'))
+ host = allocated = xasprintf("%s:%u", host, port);
+ servername = xstrdup(host);
+ strrchr(servername, ':')[0] = '\0';
+
+ fflush_all();
+ pid = xvfork();
+ if (pid == 0) {
+ /* Child */
+ char *argv[8];
+
+ close(sp[0]);
+ xmove_fd(sp[1], 0);
+ xdup2(0, 1);
+ /*
+ * openssl s_client -quiet -connect www.kernel.org:443 2>/dev/null
+ * It prints some debug stuff on stderr, don't know how to suppress it.
+ * Work around by dev-nulling stderr. We lose all error messages :(
+ */
+ xmove_fd(2, 3);
+ xopen("/dev/null", O_RDWR);
+ memset(&argv, 0, sizeof(argv));
+ argv[0] = (char*)"openssl";
+ argv[1] = (char*)"s_client";
+ argv[2] = (char*)"-quiet";
+ argv[3] = (char*)"-connect";
+ argv[4] = (char*)host;
+ /*
+ * Per RFC 6066 Section 3, the only permitted values in the
+ * TLS server_name (SNI) field are FQDNs (DNS hostnames).
+ * IPv4 and IPv6 addresses, port numbers are not allowed.
+ */
+ if (!is_ip_address(servername)) {
+ argv[5] = (char*)"-servername";
+ argv[6] = (char*)servername;
+ }
+
+ BB_EXECVP(argv[0], argv);
+ xmove_fd(3, 2);
+# if ENABLE_FEATURE_WGET_HTTPS
+ child_failed = 1;
+ xfunc_die();
+# else
+ bb_perror_msg_and_die("can't execute '%s'", argv[0]);
+# endif
+ /* notreached */
+ }
+
+ /* Parent */
+ free(servername);
+ free(allocated);
+ close(sp[1]);
+# if ENABLE_FEATURE_WGET_HTTPS
+ if (child_failed) {
+ close(sp[0]);
+ return -1;
+ }
+# endif
+ return sp[0];
+}
+#endif
+
+#if ENABLE_FEATURE_WGET_HTTPS
+static void spawn_ssl_client(const char *host, int network_fd)
+{
+ int sp[2];
+ int pid;
+ char *servername, *p;
+
+ servername = xstrdup(host);
+ p = strrchr(servername, ':');
+ if (p) *p = '\0';
+
+ if (socketpair(AF_UNIX, SOCK_STREAM, 0, sp) != 0)
+ /* Kernel can have AF_UNIX support disabled */
+ bb_perror_msg_and_die("socketpair");
+
+ fflush_all();
+ pid = BB_MMU ? xfork() : xvfork();
+ if (pid == 0) {
+ /* Child */
+ close(sp[0]);
+ xmove_fd(sp[1], 0);
+ xdup2(0, 1);
+ if (BB_MMU) {
+ tls_state_t *tls = new_tls_state();
+ tls->ifd = tls->ofd = network_fd;
+ tls_handshake(tls, servername);
+ tls_run_copy_loop(tls);
+ exit(0);
+ } else {
+ char *argv[5];
+ xmove_fd(network_fd, 3);
+ argv[0] = (char*)"ssl_client";
+ argv[1] = (char*)"-s3";
+ //TODO: if (!is_ip_address(servername))...
+ argv[2] = (char*)"-n";
+ argv[3] = servername;
+ argv[4] = NULL;
+ BB_EXECVP(argv[0], argv);
+ bb_perror_msg_and_die("can't execute '%s'", argv[0]);
+ }
+ /* notreached */
+ }
+
+ /* Parent */
+ free(servername);
+ close(sp[1]);
+ xmove_fd(sp[0], network_fd);
+}
+#endif
+
+static void NOINLINE retrieve_file_data(FILE *dfp)
{
- char buf[512];
#if ENABLE_FEATURE_WGET_STATUSBAR || ENABLE_FEATURE_WGET_TIMEOUT
# if ENABLE_FEATURE_WGET_TIMEOUT
- unsigned second_cnt;
+ unsigned second_cnt = G.timeout_seconds;
# endif
struct pollfd polldata;
polldata.fd = fileno(dfp);
polldata.events = POLLIN | POLLPRI;
- ndelay_on(polldata.fd);
#endif
progress_meter(PROGRESS_START);
/* Loops only if chunked */
while (1) {
+
+#if ENABLE_FEATURE_WGET_STATUSBAR || ENABLE_FEATURE_WGET_TIMEOUT
+ /* Must use nonblocking I/O, otherwise fread will loop
+ * and *block* until it reads full buffer,
+ * which messes up progress bar and/or timeout logic.
+ * Because of nonblocking I/O, we need to dance
+ * very carefully around EAGAIN. See explanation at
+ * clearerr() calls.
+ */
+ ndelay_on(polldata.fd);
+#endif
while (1) {
int n;
unsigned rdsz;
- rdsz = sizeof(buf);
+#if ENABLE_FEATURE_WGET_STATUSBAR || ENABLE_FEATURE_WGET_TIMEOUT
+ /* fread internally uses read loop, which in our case
+ * is usually exited when we get EAGAIN.
+ * In this case, libc sets error marker on the stream.
+ * Need to clear it before next fread to avoid possible
+ * rare false positive ferror below. Rare because usually
+ * fread gets more than zero bytes, and we don't fall
+ * into if (n <= 0) ...
+ */
+ clearerr(dfp);
+#endif
+ errno = 0;
+ rdsz = sizeof(G.wget_buf);
if (G.got_clen) {
- if (G.content_len < (off_t)sizeof(buf)) {
+ if (G.content_len < (off_t)sizeof(G.wget_buf)) {
if ((int)G.content_len <= 0)
break;
rdsz = (unsigned)G.content_len;
}
}
+ n = fread(G.wget_buf, 1, rdsz, dfp);
+
+ if (n > 0) {
+ xwrite(G.output_fd, G.wget_buf, n);
+#if ENABLE_FEATURE_WGET_STATUSBAR
+ G.transferred += n;
+#endif
+ if (G.got_clen) {
+ G.content_len -= n;
+ if (G.content_len == 0)
+ break;
+ }
+#if ENABLE_FEATURE_WGET_TIMEOUT
+ second_cnt = G.timeout_seconds;
+#endif
+ goto bump;
+ }
+
+ /* n <= 0.
+ * man fread:
+ * If error occurs, or EOF is reached, the return value
+ * is a short item count (or zero).
+ * fread does not distinguish between EOF and error.
+ */
+ if (errno != EAGAIN) {
+ if (ferror(dfp)) {
+ progress_meter(PROGRESS_END);
+ bb_perror_msg_and_die(bb_msg_read_error);
+ }
+ break; /* EOF, not error */
+ }
+
#if ENABLE_FEATURE_WGET_STATUSBAR || ENABLE_FEATURE_WGET_TIMEOUT
-# if ENABLE_FEATURE_WGET_TIMEOUT
- second_cnt = G.timeout_seconds;
-# endif
- while (1) {
- if (safe_poll(&polldata, 1, 1000) != 0)
- break; /* error, EOF, or data is available */
+ /* It was EAGAIN. There is no data. Wait up to one second
+ * then abort if timed out, or update the bar and try reading again.
+ */
+ if (safe_poll(&polldata, 1, 1000) == 0) {
# if ENABLE_FEATURE_WGET_TIMEOUT
if (second_cnt != 0 && --second_cnt == 0) {
progress_meter(PROGRESS_END);
- bb_perror_msg_and_die("download timed out");
+ bb_error_msg_and_die("download timed out");
}
# endif
- /* Needed for "stalled" indicator */
- progress_meter(PROGRESS_BUMP);
+ /* We used to loop back to poll here,
+ * but there is no great harm in letting fread
+ * to try reading anyway.
+ */
}
#endif
- n = safe_fread(buf, rdsz, dfp);
- if (n <= 0) {
- if (ferror(dfp)) {
- /* perror will not work: ferror doesn't set errno */
- bb_error_msg_and_die(bb_msg_read_error);
- }
- break;
- }
- xwrite(output_fd, buf, n);
-#if ENABLE_FEATURE_WGET_STATUSBAR
- G.transferred += n;
+ bump:
+ /* Need to do it _every_ second for "stalled" indicator
+ * to be shown properly.
+ */
progress_meter(PROGRESS_BUMP);
-#endif
- if (G.got_clen)
- G.content_len -= n;
- }
+ } /* while (reading data) */
+#if ENABLE_FEATURE_WGET_STATUSBAR || ENABLE_FEATURE_WGET_TIMEOUT
+ clearerr(dfp);
+ ndelay_off(polldata.fd); /* else fgets can get very unhappy */
+#endif
if (!G.chunked)
break;
- safe_fgets(buf, sizeof(buf), dfp); /* This is a newline */
+ fgets_and_trim(dfp, NULL); /* Eat empty line */
get_clen:
- safe_fgets(buf, sizeof(buf), dfp);
- G.content_len = STRTOOFF(buf, NULL, 16);
+ fgets_and_trim(dfp, NULL);
+ G.content_len = STRTOOFF(G.wget_buf, NULL, 16);
/* FIXME: error check? */
if (G.content_len == 0)
break; /* all done! */
G.got_clen = 1;
+ /*
+ * Note that fgets may result in some data being buffered in dfp.
+ * We loop back to fread, which will retrieve this data.
+ * Also note that code has to be arranged so that fread
+ * is done _before_ one-second poll wait - poll doesn't know
+ * about stdio buffering and can result in spurious one second waits!
+ */
+ }
+
+ /* If -c failed, we restart from the beginning,
+ * but we do not truncate file then, we do it only now, at the end.
+ * This lets user to ^C if his 99% complete 10 GB file download
+ * failed to restart *without* losing the almost complete file.
+ */
+ {
+ off_t pos = lseek(G.output_fd, 0, SEEK_CUR);
+ if (pos != (off_t)-1)
+ ftruncate(G.output_fd, pos);
}
+ /* Draw full bar and free its resources */
+ G.chunked = 0; /* makes it show 100% even for chunked download */
+ G.got_clen = 1; /* makes it show 100% even for download of (formerly) unknown size */
progress_meter(PROGRESS_END);
}
-int wget_main(int argc, char **argv) MAIN_EXTERNALLY_VISIBLE;
-int wget_main(int argc UNUSED_PARAM, char **argv)
+static void download_one_url(const char *url)
{
- char buf[512];
- struct host_info server, target;
- len_and_sockaddr *lsa;
- unsigned opt;
+ bool use_proxy; /* Use proxies if env vars are set */
int redir_limit;
- char *proxy = NULL;
- char *dir_prefix = NULL;
-#if ENABLE_FEATURE_WGET_LONG_OPTIONS
- char *post_data;
- char *extra_headers = NULL;
- llist_t *headers_llist = NULL;
-#endif
+ len_and_sockaddr *lsa;
FILE *sfp; /* socket to web/ftp server */
FILE *dfp; /* socket to ftp server (data) */
- char *fname_out; /* where to direct output (-O) */
- int output_fd = -1;
- bool use_proxy; /* Use proxies if env vars are set */
- const char *proxy_flag = "on"; /* Use proxies if env vars are set */
- const char *user_agent = "Wget";/* "User-Agent" header field */
-
- static const char keywords[] ALIGN1 =
- "content-length\0""transfer-encoding\0""chunked\0""location\0";
- enum {
- KEY_content_length = 1, KEY_transfer_encoding, KEY_chunked, KEY_location
- };
-#if ENABLE_FEATURE_WGET_LONG_OPTIONS
- static const char wget_longopts[] ALIGN1 =
- /* name, has_arg, val */
- "continue\0" No_argument "c"
- "spider\0" No_argument "s"
- "quiet\0" No_argument "q"
- "output-document\0" Required_argument "O"
- "directory-prefix\0" Required_argument "P"
- "proxy\0" Required_argument "Y"
- "user-agent\0" Required_argument "U"
-#if ENABLE_FEATURE_WGET_TIMEOUT
- "timeout\0" Required_argument "T"
-#endif
- /* Ignored: */
- // "tries\0" Required_argument "t"
- /* Ignored (we always use PASV): */
- "passive-ftp\0" No_argument "\xff"
- "header\0" Required_argument "\xfe"
- "post-data\0" Required_argument "\xfd"
- /* Ignored (we don't do ssl) */
- "no-check-certificate\0" No_argument "\xfc"
- ;
-#endif
-
- INIT_G();
-
-#if ENABLE_FEATURE_WGET_LONG_OPTIONS
- applet_long_options = wget_longopts;
-#endif
- /* server.allocated = target.allocated = NULL; */
- opt_complementary = "-1" IF_FEATURE_WGET_TIMEOUT(":T+") IF_FEATURE_WGET_LONG_OPTIONS(":\xfe::");
- opt = getopt32(argv, "csqO:P:Y:U:T:" /*ignored:*/ "t:",
- &fname_out, &dir_prefix,
- &proxy_flag, &user_agent,
- IF_FEATURE_WGET_TIMEOUT(&G.timeout_seconds) IF_NOT_FEATURE_WGET_TIMEOUT(NULL),
- NULL /* -t RETRIES */
- IF_FEATURE_WGET_LONG_OPTIONS(, &headers_llist)
- IF_FEATURE_WGET_LONG_OPTIONS(, &post_data)
- );
-#if ENABLE_FEATURE_WGET_LONG_OPTIONS
- if (headers_llist) {
- int size = 1;
- char *cp;
- llist_t *ll = headers_llist;
- while (ll) {
- size += strlen(ll->data) + 2;
- ll = ll->link;
- }
- extra_headers = cp = xmalloc(size);
- while (headers_llist) {
- cp += sprintf(cp, "%s\r\n", (char*)llist_pop(&headers_llist));
- }
- }
-#endif
-
- /* TODO: compat issue: should handle "wget URL1 URL2..." */
-
+ char *proxy = NULL;
+ char *fname_out_alloc;
+ char *redirected_path = NULL;
+ struct host_info server;
+ struct host_info target;
+
+ server.allocated = NULL;
+ target.allocated = NULL;
+ server.user = NULL;
target.user = NULL;
- parse_url(argv[optind], &target);
+
+ parse_url(url, &target);
/* Use the proxy if necessary */
- use_proxy = (strcmp(proxy_flag, "off") != 0);
+ use_proxy = (strcmp(G.proxy_flag, "off") != 0);
if (use_proxy) {
- proxy = getenv(target.is_ftp ? "ftp_proxy" : "http_proxy");
- if (proxy && proxy[0]) {
- server.user = NULL;
+ proxy = getenv(target.protocol == P_FTP ? "ftp_proxy" : "http_proxy");
+//FIXME: what if protocol is https? Ok to use http_proxy?
+ use_proxy = (proxy && proxy[0]);
+ if (use_proxy)
parse_url(proxy, &server);
- } else {
- use_proxy = 0;
- }
}
if (!use_proxy) {
server.port = target.port;
if (ENABLE_FEATURE_IPV6) {
- server.host = xstrdup(target.host);
+ //free(server.allocated); - can't be non-NULL
+ server.host = server.allocated = xstrdup(target.host);
} else {
server.host = target.host;
}
if (ENABLE_FEATURE_IPV6)
strip_ipv6_scope_id(target.host);
- /* Guess an output filename, if there was no -O FILE */
- if (!(opt & WGET_OPT_OUTNAME)) {
- fname_out = bb_get_last_path_component_nostrip(target.path);
+ /* If there was no -O FILE, guess output filename */
+ fname_out_alloc = NULL;
+ if (!(option_mask32 & WGET_OPT_OUTNAME)) {
+ G.fname_out = bb_get_last_path_component_nostrip(target.path);
/* handle "wget http://kernel.org//" */
- if (fname_out[0] == '/' || !fname_out[0])
- fname_out = (char*)"index.html";
+ if (G.fname_out[0] == '/' || !G.fname_out[0])
+ G.fname_out = (char*)"index.html";
/* -P DIR is considered only if there was no -O FILE */
- if (dir_prefix)
- fname_out = concat_path_file(dir_prefix, fname_out);
- } else {
- if (LONE_DASH(fname_out)) {
- /* -O - */
- output_fd = 1;
- opt &= ~WGET_OPT_CONTINUE;
+ if (G.dir_prefix)
+ G.fname_out = fname_out_alloc = concat_path_file(G.dir_prefix, G.fname_out);
+ else {
+ /* redirects may free target.path later, need to make a copy */
+ G.fname_out = fname_out_alloc = xstrdup(G.fname_out);
}
}
#if ENABLE_FEATURE_WGET_STATUSBAR
- G.curfile = bb_get_last_path_component_nostrip(fname_out);
+ G.curfile = bb_get_last_path_component_nostrip(G.fname_out);
#endif
- /* Impossible?
- if ((opt & WGET_OPT_CONTINUE) && !fname_out)
- bb_error_msg_and_die("can't specify continue (-c) without a filename (-O)");
- */
-
/* Determine where to start transfer */
- if (opt & WGET_OPT_CONTINUE) {
- output_fd = open(fname_out, O_WRONLY);
- if (output_fd >= 0) {
- G.beg_range = xlseek(output_fd, 0, SEEK_END);
+ G.beg_range = 0;
+ if (option_mask32 & WGET_OPT_CONTINUE) {
+ G.output_fd = open(G.fname_out, O_WRONLY);
+ if (G.output_fd >= 0) {
+ G.beg_range = xlseek(G.output_fd, 0, SEEK_END);
}
/* File doesn't exist. We do not create file here yet.
- * We are not sure it exists on remove side */
+ * We are not sure it exists on remote side */
}
redir_limit = 5;
resolve_lsa:
lsa = xhost2sockaddr(server.host, server.port);
- if (!(opt & WGET_OPT_QUIET)) {
+ if (!(option_mask32 & WGET_OPT_QUIET)) {
char *s = xmalloc_sockaddr2dotted(&lsa->u.sa);
fprintf(stderr, "Connecting to %s (%s)\n", server.host, s);
free(s);
}
establish_session:
- if (use_proxy || !target.is_ftp) {
+ /*G.content_len = 0; - redundant, got_clen = 0 is enough */
+ G.got_clen = 0;
+ G.chunked = 0;
+ if (use_proxy || target.protocol != P_FTP) {
/*
* HTTP session
*/
char *str;
int status;
- /* Open socket to http server */
+ /* Open socket to http(s) server */
+#if ENABLE_FEATURE_WGET_OPENSSL
+ /* openssl (and maybe internal TLS) support is configured */
+ if (target.protocol == P_HTTPS) {
+ /* openssl-based helper
+ * Inconvenient API since we can't give it an open fd
+ */
+ int fd = spawn_https_helper_openssl(server.host, server.port);
+# if ENABLE_FEATURE_WGET_HTTPS
+ if (fd < 0) { /* no openssl? try internal */
+ sfp = open_socket(lsa);
+ spawn_ssl_client(server.host, fileno(sfp));
+ goto socket_opened;
+ }
+# else
+ /* We don't check for exec("openssl") failure in this case */
+# endif
+ sfp = fdopen(fd, "r+");
+ if (!sfp)
+ bb_perror_msg_and_die(bb_msg_memory_exhausted);
+ goto socket_opened;
+ }
sfp = open_socket(lsa);
-
+ socket_opened:
+#elif ENABLE_FEATURE_WGET_HTTPS
+ /* Only internal TLS support is configured */
+ sfp = open_socket(lsa);
+ if (target.protocol == P_HTTPS)
+ spawn_ssl_client(server.host, fileno(sfp));
+#else
+ /* ssl (https) support is not configured */
+ sfp = open_socket(lsa);
+#endif
/* Send HTTP request */
if (use_proxy) {
- fprintf(sfp, "GET %stp://%s/%s HTTP/1.1\r\n",
- target.is_ftp ? "f" : "ht", target.host,
+ SENDFMT(sfp, "GET %s://%s/%s HTTP/1.1\r\n",
+ target.protocol, target.host,
target.path);
} else {
- if (opt & WGET_OPT_POST_DATA)
- fprintf(sfp, "POST /%s HTTP/1.1\r\n", target.path);
- else
- fprintf(sfp, "GET /%s HTTP/1.1\r\n", target.path);
+ SENDFMT(sfp, "%s /%s HTTP/1.1\r\n",
+ (option_mask32 & WGET_OPT_POST_DATA) ? "POST" : "GET",
+ target.path);
}
+ if (!USR_HEADER_HOST)
+ SENDFMT(sfp, "Host: %s\r\n", target.host);
+ if (!USR_HEADER_USER_AGENT)
+ SENDFMT(sfp, "User-Agent: %s\r\n", G.user_agent);
- fprintf(sfp, "Host: %s\r\nUser-Agent: %s\r\n",
- target.host, user_agent);
+ /* Ask server to close the connection as soon as we are done
+ * (IOW: we do not intend to send more requests)
+ */
+ SENDFMT(sfp, "Connection: close\r\n");
#if ENABLE_FEATURE_WGET_AUTHENTICATION
- if (target.user) {
- fprintf(sfp, "Proxy-Authorization: Basic %s\r\n"+6,
- base64enc_512(buf, target.user));
+ if (target.user && !USR_HEADER_AUTH) {
+ SENDFMT(sfp, "Proxy-Authorization: Basic %s\r\n"+6,
+ base64enc(target.user));
}
- if (use_proxy && server.user) {
- fprintf(sfp, "Proxy-Authorization: Basic %s\r\n",
- base64enc_512(buf, server.user));
+ if (use_proxy && server.user && !USR_HEADER_PROXY_AUTH) {
+ SENDFMT(sfp, "Proxy-Authorization: Basic %s\r\n",
+ base64enc(server.user));
}
#endif
- if (G.beg_range)
- fprintf(sfp, "Range: bytes=%"OFF_FMT"u-\r\n", G.beg_range);
+ if (G.beg_range != 0 && !USR_HEADER_RANGE)
+ SENDFMT(sfp, "Range: bytes=%"OFF_FMT"u-\r\n", G.beg_range);
+
#if ENABLE_FEATURE_WGET_LONG_OPTIONS
- if (extra_headers)
- fputs(extra_headers, sfp);
-
- if (opt & WGET_OPT_POST_DATA) {
- char *estr = URL_escape(post_data);
- fprintf(sfp, "Content-Type: application/x-www-form-urlencoded\r\n");
- fprintf(sfp, "Content-Length: %u\r\n" "\r\n" "%s",
- (int) strlen(estr), estr);
- /*fprintf(sfp, "Connection: Keep-Alive\r\n\r\n");*/
- /*fprintf(sfp, "%s\r\n", estr);*/
- free(estr);
+ if (G.extra_headers) {
+ log_io(G.extra_headers);
+ fputs(G.extra_headers, sfp);
+ }
+
+ if (option_mask32 & WGET_OPT_POST_DATA) {
+ SENDFMT(sfp,
+ "Content-Type: application/x-www-form-urlencoded\r\n"
+ "Content-Length: %u\r\n"
+ "\r\n"
+ "%s",
+ (int) strlen(G.post_data), G.post_data
+ );
} else
#endif
- { /* If "Connection:" is needed, document why */
- fprintf(sfp, /* "Connection: close\r\n" */ "\r\n");
+ {
+ SENDFMT(sfp, "\r\n");
}
fflush(sfp);
+/* Tried doing this unconditionally.
+ * Cloudflare and nginx/1.11.5 are shocked to see SHUT_WR on non-HTTPS.
+ */
+#if SSL_SUPPORTED
+ if (target.protocol == P_HTTPS) {
+ /* If we use SSL helper, keeping our end of the socket open for writing
+ * makes our end (i.e. the same fd!) readable (EAGAIN instead of EOF)
+ * even after child closes its copy of the fd.
+ * This helps:
+ */
+ shutdown(fileno(sfp), SHUT_WR);
+ }
+#endif
+
/*
* Retrieve HTTP response line and check for "200" status code.
*/
read_response:
- if (fgets(buf, sizeof(buf), sfp) == NULL)
- bb_error_msg_and_die("no response from server");
+ fgets_and_trim(sfp, " %s\n");
- str = buf;
+ str = G.wget_buf;
str = skip_non_whitespace(str);
str = skip_whitespace(str);
// FIXME: no error check
switch (status) {
case 0:
case 100:
- while (gethdr(buf, sizeof(buf), sfp /*, &n*/) != NULL)
+ while (gethdr(sfp) != NULL)
/* eat all remaining headers */;
goto read_response;
+
+ /* Success responses */
case 200:
+ /* fall through */
+ case 201: /* 201 Created */
+/* "The request has been fulfilled and resulted in a new resource being created" */
+ /* Standard wget is reported to treat this as success */
+ /* fall through */
+ case 202: /* 202 Accepted */
+/* "The request has been accepted for processing, but the processing has not been completed" */
+ /* Treat as success: fall through */
+ case 203: /* 203 Non-Authoritative Information */
+/* "Use of this response code is not required and is only appropriate when the response would otherwise be 200 (OK)" */
+ /* fall through */
+ case 204: /* 204 No Content */
/*
Response 204 doesn't say "null file", it says "metadata
has changed but data didn't":
However, in real world it was observed that some web servers
(e.g. Boa/0.94.14rc21) simply use code 204 when file size is zero.
*/
- case 204:
+ if (G.beg_range != 0) {
+ /* "Range:..." was not honored by the server.
+ * Restart download from the beginning.
+ */
+ reset_beg_range_to_zero();
+ }
break;
- case 300: /* redirection */
+ /* 205 Reset Content ?? what to do on this ?? */
+
+ case 300: /* redirection */
case 301:
case 302:
case 303:
break;
- case 206:
- if (G.beg_range)
+
+ case 206: /* Partial Content */
+ if (G.beg_range != 0)
+ /* "Range:..." worked. Good. */
break;
+ /* Partial Content even though we did not ask for it??? */
/* fall through */
default:
- bb_error_msg_and_die("server returned error: %s", sanitize_string(buf));
+ bb_error_msg_and_die("server returned error: %s", sanitize_string(G.wget_buf));
}
/*
* Retrieve HTTP headers.
*/
- while ((str = gethdr(buf, sizeof(buf), sfp /*, &n*/)) != NULL) {
- /* gethdr converted "FOO:" string to lowercase */
+ while ((str = gethdr(sfp)) != NULL) {
+ static const char keywords[] ALIGN1 =
+ "content-length\0""transfer-encoding\0""location\0";
+ enum {
+ KEY_content_length = 1, KEY_transfer_encoding, KEY_location
+ };
smalluint key;
+
+ /* gethdr converted "FOO:" string to lowercase */
+
/* strip trailing whitespace */
char *s = strchrnul(str, '\0') - 1;
while (s >= str && (*s == ' ' || *s == '\t')) {
*s = '\0';
s--;
}
- key = index_in_strings(keywords, buf) + 1;
+ key = index_in_strings(keywords, G.wget_buf) + 1;
if (key == KEY_content_length) {
G.content_len = BB_STRTOOFF(str, NULL, 10);
if (G.content_len < 0 || errno) {
continue;
}
if (key == KEY_transfer_encoding) {
- if (index_in_strings(keywords, str_tolower(str)) + 1 != KEY_chunked)
+ if (strcmp(str_tolower(str), "chunked") != 0)
bb_error_msg_and_die("transfer encoding '%s' is not supported", sanitize_string(str));
- G.chunked = G.got_clen = 1;
+ G.chunked = 1;
}
if (key == KEY_location && status >= 300) {
if (--redir_limit == 0)
bb_error_msg_and_die("too many redirections");
fclose(sfp);
- G.got_clen = 0;
- G.chunked = 0;
- if (str[0] == '/')
- /* free(target.allocated); */
- target.path = /* target.allocated = */ xstrdup(str+1);
+ if (str[0] == '/') {
+ free(redirected_path);
+ target.path = redirected_path = xstrdup(str+1);
/* lsa stays the same: it's on the same server */
- else {
+ } else {
parse_url(str, &target);
if (!use_proxy) {
+ /* server.user remains untouched */
+ free(server.allocated);
+ server.allocated = NULL;
server.host = target.host;
/* strip_ipv6_scope_id(target.host); - no! */
/* we assume remote never gives us IPv6 addr with scope id */
/* For HTTP, data is pumped over the same connection */
dfp = sfp;
-
} else {
/*
* FTP session
sfp = prepare_ftp_session(&dfp, &target, lsa);
}
- if (opt & WGET_OPT_SPIDER) {
- if (ENABLE_FEATURE_CLEAN_UP)
- fclose(sfp);
- return EXIT_SUCCESS;
- }
+ free(lsa);
- if (output_fd < 0) {
- int o_flags = O_WRONLY | O_CREAT | O_TRUNC | O_EXCL;
- /* compat with wget: -O FILE can overwrite */
- if (opt & WGET_OPT_OUTNAME)
- o_flags = O_WRONLY | O_CREAT | O_TRUNC;
- output_fd = xopen(fname_out, o_flags);
+ if (!(option_mask32 & WGET_OPT_SPIDER)) {
+ if (G.output_fd < 0)
+ G.output_fd = xopen(G.fname_out, G.o_flags);
+ retrieve_file_data(dfp);
+ if (!(option_mask32 & WGET_OPT_OUTNAME)) {
+ xclose(G.output_fd);
+ G.output_fd = -1;
+ }
}
- retrieve_file_data(dfp, output_fd);
- xclose(output_fd);
-
if (dfp != sfp) {
- /* It's ftp. Close it properly */
+ /* It's ftp. Close data connection properly */
fclose(dfp);
- if (ftpcmd(NULL, NULL, sfp, buf) != 226)
- bb_error_msg_and_die("ftp error: %s", sanitize_string(buf+4));
- /* ftpcmd("QUIT", NULL, sfp, buf); - why bother? */
+ if (ftpcmd(NULL, NULL, sfp) != 226)
+ bb_error_msg_and_die("ftp error: %s", sanitize_string(G.wget_buf + 4));
+ /* ftpcmd("QUIT", NULL, sfp); - why bother? */
+ }
+ fclose(sfp);
+
+ free(server.allocated);
+ free(target.allocated);
+ free(server.user);
+ free(target.user);
+ free(fname_out_alloc);
+ free(redirected_path);
+}
+
+int wget_main(int argc, char **argv) MAIN_EXTERNALLY_VISIBLE;
+int wget_main(int argc UNUSED_PARAM, char **argv)
+{
+#if ENABLE_FEATURE_WGET_LONG_OPTIONS
+ static const char wget_longopts[] ALIGN1 =
+ /* name, has_arg, val */
+ "continue\0" No_argument "c"
+ "quiet\0" No_argument "q"
+ "server-response\0" No_argument "S"
+ "output-document\0" Required_argument "O"
+ "directory-prefix\0" Required_argument "P"
+ "proxy\0" Required_argument "Y"
+ "user-agent\0" Required_argument "U"
+IF_FEATURE_WGET_TIMEOUT(
+ "timeout\0" Required_argument "T")
+ /* Ignored: */
+IF_DESKTOP( "tries\0" Required_argument "t")
+ "header\0" Required_argument "\xff"
+ "post-data\0" Required_argument "\xfe"
+ "spider\0" No_argument "\xfd"
+ /* Ignored (we always use PASV): */
+IF_DESKTOP( "passive-ftp\0" No_argument "\xf0")
+ /* Ignored (we don't do ssl) */
+IF_DESKTOP( "no-check-certificate\0" No_argument "\xf0")
+ /* Ignored (we don't support caching) */
+IF_DESKTOP( "no-cache\0" No_argument "\xf0")
+IF_DESKTOP( "no-verbose\0" No_argument "\xf0")
+IF_DESKTOP( "no-clobber\0" No_argument "\xf0")
+IF_DESKTOP( "no-host-directories\0" No_argument "\xf0")
+IF_DESKTOP( "no-parent\0" No_argument "\xf0")
+ ;
+#endif
+
+#if ENABLE_FEATURE_WGET_LONG_OPTIONS
+ llist_t *headers_llist = NULL;
+#endif
+
+ INIT_G();
+
+#if ENABLE_FEATURE_WGET_TIMEOUT
+ G.timeout_seconds = 900;
+ signal(SIGALRM, alarm_handler);
+#endif
+ G.proxy_flag = "on"; /* use proxies if env vars are set */
+ G.user_agent = "Wget"; /* "User-Agent" header field */
+
+#if ENABLE_FEATURE_WGET_LONG_OPTIONS
+ applet_long_options = wget_longopts;
+#endif
+ opt_complementary = "-1" /* at least one URL */
+ IF_FEATURE_WGET_LONG_OPTIONS(":\xff::"); /* --header is a list */
+ getopt32(argv, "cqSO:P:Y:U:T:+"
+ /*ignored:*/ "t:"
+ /*ignored:*/ "n::"
+ /* wget has exactly four -n<letter> opts, all of which we can ignore:
+ * -nv --no-verbose: be moderately quiet (-q is full quiet)
+ * -nc --no-clobber: abort if exists, neither download to FILE.n nor overwrite FILE
+ * -nH --no-host-directories: wget -r http://host/ won't create host/
+ * -np --no-parent
+ * "n::" above says that we accept -n[ARG].
+ * Specifying "n:" would be a bug: "-n ARG" would eat ARG!
+ */
+ , &G.fname_out, &G.dir_prefix,
+ &G.proxy_flag, &G.user_agent,
+ IF_FEATURE_WGET_TIMEOUT(&G.timeout_seconds) IF_NOT_FEATURE_WGET_TIMEOUT(NULL),
+ NULL, /* -t RETRIES */
+ NULL /* -n[ARG] */
+ IF_FEATURE_WGET_LONG_OPTIONS(, &headers_llist)
+ IF_FEATURE_WGET_LONG_OPTIONS(, &G.post_data)
+ );
+#if 0 /* option bits debug */
+ if (option_mask32 & WGET_OPT_RETRIES) bb_error_msg("-t NUM");
+ if (option_mask32 & WGET_OPT_nsomething) bb_error_msg("-nsomething");
+ if (option_mask32 & WGET_OPT_HEADER) bb_error_msg("--header");
+ if (option_mask32 & WGET_OPT_POST_DATA) bb_error_msg("--post-data");
+ if (option_mask32 & WGET_OPT_SPIDER) bb_error_msg("--spider");
+ exit(0);
+#endif
+ argv += optind;
+
+#if ENABLE_FEATURE_WGET_LONG_OPTIONS
+ if (headers_llist) {
+ int size = 0;
+ char *hdr;
+ llist_t *ll = headers_llist;
+ while (ll) {
+ size += strlen(ll->data) + 2;
+ ll = ll->link;
+ }
+ G.extra_headers = hdr = xmalloc(size + 1);
+ while (headers_llist) {
+ int bit;
+ const char *words;
+
+ size = sprintf(hdr, "%s\r\n",
+ (char*)llist_pop(&headers_llist));
+ /* a bit like index_in_substrings but don't match full key */
+ bit = 1;
+ words = wget_user_headers;
+ while (*words) {
+ if (strstr(hdr, words) == hdr) {
+ G.user_headers |= bit;
+ break;
+ }
+ bit <<= 1;
+ words += strlen(words) + 1;
+ }
+ hdr += size;
+ }
}
+#endif
+
+ G.output_fd = -1;
+ G.o_flags = O_WRONLY | O_CREAT | O_TRUNC | O_EXCL;
+ if (G.fname_out) { /* -O FILE ? */
+ if (LONE_DASH(G.fname_out)) { /* -O - ? */
+ G.output_fd = 1;
+ option_mask32 &= ~WGET_OPT_CONTINUE;
+ }
+ /* compat with wget: -O FILE can overwrite */
+ G.o_flags = O_WRONLY | O_CREAT | O_TRUNC;
+ }
+
+ while (*argv)
+ download_one_url(*argv++);
+
+ if (G.output_fd >= 0)
+ xclose(G.output_fd);
+
+#if ENABLE_FEATURE_CLEAN_UP && ENABLE_FEATURE_WGET_LONG_OPTIONS
+ free(G.extra_headers);
+#endif
+ FINI_G();
return EXIT_SUCCESS;
}
projects / oweals / busybox.git / blobdiff