/* File retrieval.
- Copyright (C) 1995, 1996, 1997, 1998 Free Software Foundation, Inc.
+ Copyright (C) 1995, 1996, 1997, 1998, 2000, 2001 Free Software Foundation, Inc.
-This file is part of Wget.
+This file is part of GNU Wget.
-This program is free software; you can redistribute it and/or modify
+GNU Wget is free software; you can redistribute it and/or modify
it under the terms of the GNU General Public License as published by
-the Free Software Foundation; either version 2 of the License, or
-(at your option) any later version.
+the Free Software Foundation; either version 2 of the License, or (at
+your option) any later version.
-This program is distributed in the hope that it will be useful,
+GNU Wget is distributed in the hope that it will be useful,
but WITHOUT ANY WARRANTY; without even the implied warranty of
MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
GNU General Public License for more details.
You should have received a copy of the GNU General Public License
-along with this program; if not, write to the Free Software
-Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA. */
+along with Wget; if not, write to the Free Software
+Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.
+
+In addition, as a special exception, the Free Software Foundation
+gives permission to link the code of its release of Wget with the
+OpenSSL project's "OpenSSL" library (or with modified versions of it
+that use the same license as the "OpenSSL" library), and distribute
+the linked executables. You must obey the GNU General Public License
+in all respects for all of the code used other than "OpenSSL". If you
+modify this file, you may extend this exception to your version of the
+file, but you are not obligated to do so. If you do not wish to do
+so, delete this exception statement from your version. */
#include <config.h>
#else
# include <strings.h>
#endif /* HAVE_STRING_H */
-#include <ctype.h>
#include <assert.h>
#include "wget.h"
#include "utils.h"
#include "retr.h"
+#include "progress.h"
#include "url.h"
#include "recur.h"
#include "ftp.h"
#include "host.h"
#include "connect.h"
+#include "hash.h"
+#include "convert.h"
-/* Internal variables used by the timer. */
-static long internal_secs, internal_msecs;
+#ifdef HAVE_SSL
+# include "gen_sslfunc.h" /* for ssl_iread */
+#endif
+
+#ifndef errno
+extern int errno;
+#endif
+
+/* Total size of downloaded files. Used to enforce quota. */
+LARGE_INT total_downloaded_bytes;
-void logflush PARAMS ((void));
+/* If non-NULL, the stream to which output should be written. This
+ stream is initialized when `-O' is used. */
+FILE *output_stream;
-/* From http.c. */
-uerr_t http_loop PARAMS ((struct urlinfo *, char **, int *));
+/* Whether output_document is a regular file we can manipulate,
+ i.e. not `-' or a device file. */
+int output_stream_regular;
\f
-/* Flags for show_progress(). */
-enum spflags { SP_NONE, SP_INIT, SP_FINISH };
+static struct {
+ long chunk_bytes;
+ double chunk_start;
+ double sleep_adjust;
+} limit_data;
+
+static void
+limit_bandwidth_reset (void)
+{
+ limit_data.chunk_bytes = 0;
+ limit_data.chunk_start = 0;
+}
+
+/* Limit the bandwidth by pausing the download for an amount of time.
+ BYTES is the number of bytes received from the network, and TIMER
+ is the timer that started at the beginning of download. */
+
+static void
+limit_bandwidth (long bytes, struct wget_timer *timer)
+{
+ double delta_t = wtimer_read (timer) - limit_data.chunk_start;
+ double expected;
+
+ limit_data.chunk_bytes += bytes;
+
+ /* Calculate the amount of time we expect downloading the chunk
+ should take. If in reality it took less time, sleep to
+ compensate for the difference. */
+ expected = 1000.0 * limit_data.chunk_bytes / opt.limit_rate;
+
+ if (expected > delta_t)
+ {
+ double slp = expected - delta_t + limit_data.sleep_adjust;
+ double t0, t1;
+ if (slp < 200)
+ {
+ DEBUGP (("deferring a %.2f ms sleep (%ld/%.2f).\n",
+ slp, limit_data.chunk_bytes, delta_t));
+ return;
+ }
+ DEBUGP (("\nsleeping %.2f ms for %ld bytes, adjust %.2f ms\n",
+ slp, limit_data.chunk_bytes, limit_data.sleep_adjust));
+
+ t0 = wtimer_read (timer);
+ xsleep (slp / 1000);
+ wtimer_update (timer);
+ t1 = wtimer_read (timer);
+
+ /* Due to scheduling, we probably slept slightly longer (or
+ shorter) than desired. Calculate the difference between the
+ desired and the actual sleep, and adjust the next sleep by
+ that amount. */
+ limit_data.sleep_adjust = slp - (t1 - t0);
+ }
+
+ limit_data.chunk_bytes = 0;
+ limit_data.chunk_start = wtimer_read (timer);
+}
+
+#ifndef MIN
+# define MIN(i, j) ((i) <= (j) ? (i) : (j))
+#endif
+
+/* Write data in BUF to OUT. However, if *SKIP is non-zero, skip that
+ amount of data and decrease SKIP. Increment *TOTAL by the amount
+ of data written. */
+
+static int
+write_data (FILE *out, const char *buf, int bufsize, long *skip,
+ long *written)
+{
+ if (!out)
+ return 1;
+ if (*skip > bufsize)
+ {
+ *skip -= bufsize;
+ return 1;
+ }
+ if (*skip)
+ {
+ buf += *skip;
+ bufsize -= *skip;
+ *skip = 0;
+ if (bufsize == 0)
+ return 1;
+ }
-static int show_progress PARAMS ((long, long, enum spflags));
+ fwrite (buf, 1, bufsize, out);
+ *written += bufsize;
-/* Reads the contents of file descriptor FD, until it is closed, or a
- read error occurs. The data is read in 8K chunks, and stored to
- stream fp, which should have been open for writing. If BUF is
- non-NULL and its file descriptor is equal to FD, flush RBUF first.
- This function will *not* use the rbuf_* functions!
+ /* Immediately flush the downloaded data. This should not hinder
+ performance: fast downloads will arrive in large 16K chunks
+ (which stdio would write out immediately anyway), and slow
+ downloads wouldn't be limited by disk speed. */
+ fflush (out);
+ return !ferror (out);
+}
- The EXPECTED argument is passed to show_progress() unchanged, but
- otherwise ignored.
+/* Read the contents of file descriptor FD until it the connection
+ terminates or a read error occurs. The data is read in portions of
+ up to 16K and written to OUT as it arrives. If opt.verbose is set,
+ the progress is shown.
- If opt.verbose is set, the progress is also shown. RESTVAL
- represents a value from which to start downloading (which will be
- shown accordingly). If RESTVAL is non-zero, the stream should have
- been open for appending.
+ TOREAD is the amount of data expected to arrive, normally only used
+ by the progress gauge.
- The function exits and returns codes of 0, -1 and -2 if the
- connection was closed, there was a read error, or if it could not
- write to the output stream, respectively.
+ STARTPOS is the position from which the download starts, used by
+ the progress gauge. If QTYREAD is non-NULL, the value it points to
+ is incremented by the amount of data read from the network. If
+ QTYWRITTEN is non-NULL, the value it points to is incremented by
+ the amount of data written to disk. The time it took to download
+ the data (in milliseconds) is stored to ELAPSED.
+
+ The function exits and returns the amount of data read. In case of
+ error while reading data, -1 is returned. In case of error while
+ writing data, -2 is returned. */
- IMPORTANT: The function flushes the contents of the buffer in
- rbuf_flush() before actually reading from fd. If you wish to read
- from fd immediately, flush or discard the buffer. */
int
-get_contents (int fd, FILE *fp, long *len, long restval, long expected,
- struct rbuf *rbuf)
+fd_read_body (int fd, FILE *out, long toread, long startpos,
+ long *qtyread, long *qtywritten, double *elapsed, int flags)
{
- int res;
- static char c[8192];
+ int ret = 0;
+
+ static char dlbuf[16384];
+ int dlbufsize = sizeof (dlbuf);
+
+ struct wget_timer *timer = NULL;
+ double last_successful_read_tm = 0;
+
+ /* The progress gauge, set according to the user preferences. */
+ void *progress = NULL;
+
+ /* Non-zero if the progress gauge is interactive, i.e. if it can
+ continually update the display. When true, smaller timeout
+ values are used so that the gauge can update the display when
+ data arrives slowly. */
+ int progress_interactive = 0;
+
+ int exact = flags & rb_read_exactly;
+ long skip = 0;
+
+ /* How much data we've read/written. */
+ long sum_read = 0;
+ long sum_written = 0;
+
+ if (flags & rb_skip_startpos)
+ skip = startpos;
- *len = restval;
if (opt.verbose)
- show_progress (restval, expected, SP_INIT);
- if (rbuf && RBUF_FD (rbuf) == fd)
{
- while ((res = rbuf_flush (rbuf, c, sizeof (c))) != 0)
+ /* If we're skipping STARTPOS bytes, pass 0 as the INITIAL
+ argument to progress_create because the indicator doesn't
+ (yet) know about "skipping" data. */
+ progress = progress_create (skip ? 0 : startpos, startpos + toread);
+ progress_interactive = progress_interactive_p (progress);
+ }
+
+ if (opt.limit_rate)
+ limit_bandwidth_reset ();
+
+ /* A timer is needed for tracking progress, for throttling, and for
+ tracking elapsed time. If either of these are requested, start
+ the timer. */
+ if (progress || opt.limit_rate || elapsed)
+ {
+ timer = wtimer_new ();
+ last_successful_read_tm = 0;
+ }
+
+ /* Use a smaller buffer for low requested bandwidths. For example,
+ with --limit-rate=2k, it doesn't make sense to slurp in 16K of
+ data and then sleep for 8s. With buffer size equal to the limit,
+ we never have to sleep for more than one second. */
+ if (opt.limit_rate && opt.limit_rate < dlbufsize)
+ dlbufsize = opt.limit_rate;
+
+ /* Read from FD while there is data to read. Normally toread==0
+ means that it is unknown how much data is to arrive. However, if
+ EXACT is set, then toread==0 means what it says: that no data
+ should be read. */
+ while (!exact || (sum_read < toread))
+ {
+ int rdsize = exact ? MIN (toread - sum_read, dlbufsize) : dlbufsize;
+ double tmout = opt.read_timeout;
+ if (progress_interactive)
{
- if (fwrite (c, sizeof (char), res, fp) < res)
- return -2;
- if (opt.verbose)
+ /* For interactive progress gauges, always specify a ~1s
+ timeout, so that the gauge can be updated regularly even
+ when the data arrives very slowly or stalls. */
+ tmout = 0.95;
+ if (opt.read_timeout)
{
- if (show_progress (res, expected, SP_NONE))
- fflush (fp);
+ double waittm;
+ waittm = (wtimer_read (timer) - last_successful_read_tm) / 1000;
+ if (waittm + tmout > opt.read_timeout)
+ {
+ /* Don't let total idle time exceed read timeout. */
+ tmout = opt.read_timeout - waittm;
+ if (tmout < 0)
+ {
+ /* We've already exceeded the timeout. */
+ ret = -1, errno = ETIMEDOUT;
+ break;
+ }
+ }
}
- *len += res;
}
- }
- /* Read from fd while there is available data. */
- do
- {
- res = iread (fd, c, sizeof (c));
- if (res > 0)
+ ret = fd_read (fd, dlbuf, rdsize, tmout);
+
+ if (ret == 0 || (ret < 0 && errno != ETIMEDOUT))
+ break; /* read error */
+ else if (ret < 0)
+ ret = 0; /* read timeout */
+
+ if (progress || opt.limit_rate)
{
- if (fwrite (c, sizeof (char), res, fp) < res)
- return -2;
- if (opt.verbose)
+ wtimer_update (timer);
+ if (ret > 0)
+ last_successful_read_tm = wtimer_read (timer);
+ }
+
+ if (ret > 0)
+ {
+ sum_read += ret;
+ if (!write_data (out, dlbuf, ret, &skip, &sum_written))
{
- if (show_progress (res, expected, SP_NONE))
- fflush (fp);
+ ret = -2;
+ goto out;
}
- *len += res;
}
- } while (res > 0);
- if (res < -1)
- res = -1;
- if (opt.verbose)
- show_progress (0, expected, SP_FINISH);
- return res;
-}
-static void
-print_percentage (long bytes, long expected)
-{
- int percentage = (int)(100.0 * bytes / expected);
- logprintf (LOG_VERBOSE, " [%3d%%]", percentage);
+ if (opt.limit_rate)
+ limit_bandwidth (ret, timer);
+
+ if (progress)
+ progress_update (progress, ret, wtimer_read (timer));
+#ifdef WINDOWS
+ if (toread > 0)
+ ws_percenttitle (100.0 *
+ (startpos + sum_read) / (startpos + toread));
+#endif
+ }
+ if (ret < -1)
+ ret = -1;
+
+ out:
+ if (progress)
+ progress_finish (progress, wtimer_read (timer));
+
+ if (elapsed)
+ *elapsed = wtimer_read (timer);
+ if (timer)
+ wtimer_delete (timer);
+
+ if (qtyread)
+ *qtyread += sum_read;
+ if (qtywritten)
+ *qtywritten += sum_written;
+
+ return ret;
}
+\f
+/* Read a hunk of data from FD, up until a terminator. The terminator
+ is whatever the TERMINATOR function determines it to be; for
+ example, it can be a line of data, or the head of an HTTP response.
+ The function returns the data read allocated with malloc.
-/* Show the dotted progress report of file loading. Called with
- length and a flag to tell it whether to reset or not. It keeps the
- offset information in static local variables.
+ In case of error, NULL is returned. In case of EOF and no data
+ read, NULL is returned and errno set to 0. In case of EOF with
+ data having been read, the data is returned, but it will
+ (obviously) not contain the terminator.
- Return value: 1 or 0, designating whether any dots have been drawn.
+ The idea is to be able to read a line of input, or otherwise a hunk
+ of text, such as the head of an HTTP request, without crossing the
+ boundary, so that the next call to fd_read etc. reads the data
+ after the hunk. To achieve that, this function does the following:
- If the init argument is set, the routine will initialize.
+ 1. Peek at available data.
- If the res is non-zero, res/line_bytes lines are skipped
- (meaning the appropriate number ok kilobytes), and the number of
- "dots" fitting on the first line are drawn as ','. */
-static int
-show_progress (long res, long expected, enum spflags flags)
+ 2. Determine whether the peeked data, along with the previously
+ read data, includes the terminator.
+
+ 2a. If yes, read the data until the end of the terminator, and
+ exit.
+
+ 2b. If no, read the peeked data and goto 1.
+
+ The function is careful to assume as little as possible about the
+ implementation of peeking. For example, every peek is followed by
+ a read. If the read returns a different amount of data, the
+ process is retried until all data arrives safely.
+
+ BUFSIZE is the size of the initial buffer expected to read all the
+ data in the typical case.
+
+ This function should be used as a building block for other
+ functions -- see fd_read_line as a simple example. */
+
+char *
+fd_read_hunk (int fd, hunk_terminator_t hunk_terminator, int bufsize)
{
- static long line_bytes;
- static long offs;
- static int ndot, nrow;
- int any_output = 0;
+ char *hunk = xmalloc (bufsize);
+ int tail = 0; /* tail position in HUNK */
- if (flags == SP_FINISH)
+ while (1)
{
- if (expected)
+ const char *end;
+ int pklen, rdlen, remain;
+
+ /* First, peek at the available data. */
+
+ pklen = fd_peek (fd, hunk + tail, bufsize - 1 - tail, -1);
+ if (pklen < 0)
{
- int dot = ndot;
- char *tmpstr = (char *)alloca (2 * opt.dots_in_line + 1);
- char *tmpp = tmpstr;
- for (; dot < opt.dots_in_line; dot++)
+ xfree (hunk);
+ return NULL;
+ }
+ end = hunk_terminator (hunk, tail, pklen);
+ if (end)
+ {
+ /* The data contains the terminator: we'll drain the data up
+ to the end of the terminator. */
+ remain = end - (hunk + tail);
+ if (remain == 0)
+ {
+ /* No more data needs to be read. */
+ hunk[tail] = '\0';
+ return hunk;
+ }
+ if (bufsize - 1 < tail + remain)
{
- if (!(dot % opt.dot_spacing))
- *tmpp++ = ' ';
- *tmpp++ = ' ';
+ bufsize = tail + remain + 1;
+ hunk = xrealloc (hunk, bufsize);
}
- *tmpp = '\0';
- logputs (LOG_VERBOSE, tmpstr);
- print_percentage (nrow * line_bytes + ndot * opt.dot_bytes + offs,
- expected);
}
- logputs (LOG_VERBOSE, "\n\n");
- return 0;
- }
+ else
+ /* No terminator: simply read the data we know is (or should
+ be) available. */
+ remain = pklen;
- /* Temporarily disable flushing. */
- opt.no_flush = 1;
- /* init set means initialization. If res is set, it also means that
- the retrieval is *not* done from the beginning. The part that
- was already retrieved is not shown again. */
- if (flags == SP_INIT)
- {
- /* Generic initialization of static variables. */
- offs = 0L;
- ndot = nrow = 0;
- line_bytes = (long)opt.dots_in_line * opt.dot_bytes;
- if (res)
+ /* Now, read the data. Note that we make no assumptions about
+ how much data we'll get. (Some TCP stacks are notorious for
+ read returning less data than the previous MSG_PEEK.) */
+
+ rdlen = fd_read (fd, hunk + tail, remain, 0);
+ if (rdlen < 0)
+ {
+ xfree_null (hunk);
+ return NULL;
+ }
+ tail += rdlen;
+ hunk[tail] = '\0';
+
+ if (rdlen == 0)
{
- if (res >= line_bytes)
+ if (tail == 0)
{
- nrow = res / line_bytes;
- res %= line_bytes;
- logprintf (LOG_VERBOSE,
- _("\n [ skipping %dK ]"),
- (int) ((nrow * line_bytes) / 1024));
- ndot = 0;
+ /* EOF without anything having been read */
+ xfree (hunk);
+ errno = 0;
+ return NULL;
}
+ else
+ /* EOF seen: return the data we've read. */
+ return hunk;
}
- logprintf (LOG_VERBOSE, "\n%5ldK ->", nrow * line_bytes / 1024);
- }
- /* Offset gets incremented by current value. */
- offs += res;
- /* While offset is >= opt.dot_bytes, print dots, taking care to
- precede every 50th dot with a status message. */
- for (; offs >= opt.dot_bytes; offs -= opt.dot_bytes)
- {
- if (!(ndot % opt.dot_spacing))
- logputs (LOG_VERBOSE, " ");
- any_output = 1;
- logputs (LOG_VERBOSE, flags == SP_INIT ? "," : ".");
- ++ndot;
- if (ndot == opt.dots_in_line)
+ if (end && rdlen == remain)
+ /* The terminator was seen and the remaining data drained --
+ we got what we came for. */
+ return hunk;
+
+ /* Keep looping until all the data arrives. */
+
+ if (tail == bufsize - 1)
{
- ndot = 0;
- ++nrow;
- if (expected)
- print_percentage (nrow * line_bytes, expected);
- logprintf (LOG_VERBOSE, "\n%5ldK ->", nrow * line_bytes / 1024);
+ bufsize <<= 1;
+ hunk = xrealloc (hunk, bufsize);
}
}
- /* Reenable flushing. */
- opt.no_flush = 0;
- if (any_output)
- /* Force flush. #### Oh, what a kludge! */
- logflush ();
- return any_output;
}
-\f
-/* Reset the internal timer. */
-void
-reset_timer (void)
+
+static const char *
+line_terminator (const char *hunk, int oldlen, int peeklen)
{
-#ifdef HAVE_GETTIMEOFDAY
- struct timeval t;
- gettimeofday (&t, NULL);
- internal_secs = t.tv_sec;
- internal_msecs = t.tv_usec / 1000;
-#else
- internal_secs = time (NULL);
- internal_msecs = 0;
-#endif
+ const char *p = memchr (hunk + oldlen, '\n', peeklen);
+ if (p)
+ /* p+1 because we want the line to include '\n' */
+ return p + 1;
+ return NULL;
}
-/* Return the time elapsed from the last call to reset_timer(), in
- milliseconds. */
-long
-elapsed_time (void)
+/* Read one line from FD and return it. The line is allocated using
+ malloc.
+
+ If an error occurs, or if no data can be read, NULL is returned.
+ In the former case errno indicates the error condition, and in the
+ latter case, errno is NULL. */
+
+char *
+fd_read_line (int fd)
{
-#ifdef HAVE_GETTIMEOFDAY
- struct timeval t;
- gettimeofday (&t, NULL);
- return ((t.tv_sec - internal_secs) * 1000
- + (t.tv_usec / 1000 - internal_msecs));
-#else
- return 1000 * ((long)time (NULL) - internal_secs);
-#endif
+ return fd_read_hunk (fd, line_terminator, 128);
}
-
-/* Print out the appropriate download rate. Appropriate means that if
- rate is > 1024 bytes per second, kilobytes are used, and if rate >
- 1024 * 1024 bps, megabytes are used. */
+\f
+/* Return a printed representation of the download rate, as
+ appropriate for the speed. If PAD is non-zero, strings will be
+ padded to the width of 7 characters (xxxx.xx). */
char *
-rate (long bytes, long msecs)
+retr_rate (long bytes, double msecs, int pad)
+{
+ static char res[20];
+ static char *rate_names[] = {"B/s", "KB/s", "MB/s", "GB/s" };
+ int units = 0;
+
+ double dlrate = calc_rate (bytes, msecs, &units);
+ sprintf (res, pad ? "%7.2f %s" : "%.2f %s", dlrate, rate_names[units]);
+
+ return res;
+}
+
+/* Calculate the download rate and trim it as appropriate for the
+ speed. Appropriate means that if rate is greater than 1K/s,
+ kilobytes are used, and if rate is greater than 1MB/s, megabytes
+ are used.
+
+ UNITS is zero for B/s, one for KB/s, two for MB/s, and three for
+ GB/s. */
+double
+calc_rate (long bytes, double msecs, int *units)
{
- static char res[15];
double dlrate;
- if (!msecs)
- ++msecs;
+ assert (msecs >= 0);
+ assert (bytes >= 0);
+
+ if (msecs == 0)
+ /* If elapsed time is exactly zero, it means we're under the
+ granularity of the timer. This often happens on systems that
+ use time() for the timer. */
+ msecs = wtimer_granularity ();
+
dlrate = (double)1000 * bytes / msecs;
- /* #### Should these strings be translatable? */
if (dlrate < 1024.0)
- sprintf (res, "%.2f B/s", dlrate);
+ *units = 0;
else if (dlrate < 1024.0 * 1024.0)
- sprintf (res, "%.2f KB/s", dlrate / 1024.0);
+ *units = 1, dlrate /= 1024.0;
+ else if (dlrate < 1024.0 * 1024.0 * 1024.0)
+ *units = 2, dlrate /= (1024.0 * 1024.0);
else
- sprintf (res, "%.2f MB/s", dlrate / (1024.0 * 1024.0));
- return res;
+ /* Maybe someone will need this, one day. */
+ *units = 3, dlrate /= (1024.0 * 1024.0 * 1024.0);
+
+ return dlrate;
}
\f
-#define USE_PROXY_P(u) (opt.use_proxy && getproxy((u)->proto) \
- && no_proxy_match((u)->host, \
- (const char **)opt.no_proxy))
+/* Maximum number of allowed redirections. 20 was chosen as a
+ "reasonable" value, which is low enough to not cause havoc, yet
+ high enough to guarantee that normal retrievals will not be hurt by
+ the check. */
+
+#define MAX_REDIRECTIONS 20
+
+#define SUSPEND_POST_DATA do { \
+ post_data_suspended = 1; \
+ saved_post_data = opt.post_data; \
+ saved_post_file_name = opt.post_file_name; \
+ opt.post_data = NULL; \
+ opt.post_file_name = NULL; \
+} while (0)
+
+#define RESTORE_POST_DATA do { \
+ if (post_data_suspended) \
+ { \
+ opt.post_data = saved_post_data; \
+ opt.post_file_name = saved_post_file_name; \
+ post_data_suspended = 0; \
+ } \
+} while (0)
+
+static char *getproxy PARAMS ((struct url *));
/* Retrieve the given URL. Decides which loop to call -- HTTP, FTP,
- or simply copy it with file:// (#### the latter not yet
- implemented!). */
+ FTP, proxy, etc. */
+
+/* #### This function should be rewritten so it doesn't return from
+ multiple points. */
+
uerr_t
retrieve_url (const char *origurl, char **file, char **newloc,
const char *refurl, int *dt)
{
uerr_t result;
char *url;
- int location_changed, already_redirected, dummy;
- int local_use_proxy;
+ int location_changed, dummy;
char *mynewloc, *proxy;
- struct urlinfo *u;
+ struct url *u, *proxy_url;
+ int up_error_code; /* url parse error code */
+ char *local_file;
+ int redirection_count = 0;
+ int post_data_suspended = 0;
+ char *saved_post_data = NULL;
+ char *saved_post_file_name = NULL;
- /* If dt is NULL, just ignore it. */
+ /* If dt is NULL, use local storage. */
if (!dt)
- dt = &dummy;
+ {
+ dt = &dummy;
+ dummy = 0;
+ }
url = xstrdup (origurl);
if (newloc)
*newloc = NULL;
if (file)
*file = NULL;
- already_redirected = 0;
-
- again:
- u = newurl ();
- /* Parse the URL. RFC2068 requires `Location' to contain an
- absoluteURI, but many sites break this requirement. #### We
- should be liberal and accept a relative location, too. */
- result = parseurl (url, u, already_redirected);
- if (result != URLOK)
- {
- freeurl (u, 1);
- logprintf (LOG_NOTQUIET, "%s: %s.\n", url, uerrmsg (result));
- return result;
- }
- /* Set the referer. */
- if (refurl)
- u->referer = xstrdup (refurl);
- else
+ u = url_parse (url, &up_error_code);
+ if (!u)
{
- if (opt.referer)
- u->referer = xstrdup (opt.referer);
- else
- u->referer = NULL;
+ logprintf (LOG_NOTQUIET, "%s: %s.\n", url, url_error (up_error_code));
+ xfree (url);
+ return URLERROR;
}
- local_use_proxy = USE_PROXY_P (u);
- if (local_use_proxy)
+ if (!refurl)
+ refurl = opt.referer;
+
+ redirected:
+
+ result = NOCONERROR;
+ mynewloc = NULL;
+ local_file = NULL;
+ proxy_url = NULL;
+
+ proxy = getproxy (u);
+ if (proxy)
{
- struct urlinfo *pu = newurl ();
-
- /* Copy the original URL to new location. */
- memcpy (pu, u, sizeof (*u));
- pu->proxy = NULL; /* A minor correction :) */
- /* Initialize u to nil. */
- memset (u, 0, sizeof (*u));
- u->proxy = pu;
- /* Get the appropriate proxy server, appropriate for the
- current protocol. */
- proxy = getproxy (pu->proto);
- if (!proxy)
+ /* Parse the proxy URL. */
+ proxy_url = url_parse (proxy, &up_error_code);
+ if (!proxy_url)
{
- logputs (LOG_NOTQUIET, _("Could not find proxy host.\n"));
- freeurl (u, 1);
+ logprintf (LOG_NOTQUIET, _("Error parsing proxy URL %s: %s.\n"),
+ proxy, url_error (up_error_code));
+ xfree (url);
+ RESTORE_POST_DATA;
return PROXERR;
}
- /* Parse the proxy URL. */
- result = parseurl (proxy, u, 0);
- if (result != URLOK || u->proto != URLHTTP)
+ if (proxy_url->scheme != SCHEME_HTTP && proxy_url->scheme != u->scheme)
{
- if (u->proto == URLHTTP)
- logprintf (LOG_NOTQUIET, "Proxy %s: %s.\n", proxy, uerrmsg(result));
- else
- logprintf (LOG_NOTQUIET, _("Proxy %s: Must be HTTP.\n"), proxy);
- freeurl (u, 1);
+ logprintf (LOG_NOTQUIET, _("Error in proxy URL %s: Must be HTTP.\n"), proxy);
+ url_free (proxy_url);
+ xfree (url);
+ RESTORE_POST_DATA;
return PROXERR;
}
- u->proto = URLHTTP;
}
- assert (u->proto != URLFILE); /* #### Implement me! */
- mynewloc = NULL;
-
- if (u->proto == URLHTTP)
- result = http_loop (u, &mynewloc, dt);
- else if (u->proto == URLFTP)
+ if (u->scheme == SCHEME_HTTP
+#ifdef HAVE_SSL
+ || u->scheme == SCHEME_HTTPS
+#endif
+ || (proxy_url && proxy_url->scheme == SCHEME_HTTP))
+ {
+ result = http_loop (u, &mynewloc, &local_file, refurl, dt, proxy_url);
+ }
+ else if (u->scheme == SCHEME_FTP)
{
/* If this is a redirection, we must not allow recursive FTP
retrieval, so we save recursion to oldrec, and restore it
later. */
int oldrec = opt.recursive;
- if (already_redirected)
+ if (redirection_count)
opt.recursive = 0;
- result = ftp_loop (u, dt);
+ result = ftp_loop (u, dt, proxy_url);
opt.recursive = oldrec;
+
/* There is a possibility of having HTTP being redirected to
FTP. In these cases we must decide whether the text is HTML
- according to the suffix. The HTML suffixes are `.html' and
- `.htm', case-insensitive.
-
- #### All of this is, of course, crap. These types should be
- determined through mailcap. */
- if (already_redirected && u->local && (u->proto == URLFTP ))
+ according to the suffix. The HTML suffixes are `.html',
+ `.htm' and a few others, case-insensitive. */
+ if (redirection_count && local_file && u->scheme == SCHEME_FTP)
{
- char *suf = suffix (u->local);
- if (suf && (!strcasecmp (suf, "html") || !strcasecmp (suf, "htm")))
+ if (has_html_suffix_p (local_file))
*dt |= TEXTHTML;
- FREE_MAYBE (suf);
}
}
+
+ if (proxy_url)
+ {
+ url_free (proxy_url);
+ proxy_url = NULL;
+ }
+
location_changed = (result == NEWLOCATION);
if (location_changed)
{
- /* Check for redirection to oneself. */
- if (url_equal (url, mynewloc))
+ char *construced_newloc;
+ struct url *newloc_parsed;
+
+ assert (mynewloc != NULL);
+
+ if (local_file)
+ xfree (local_file);
+
+ /* The HTTP specs only allow absolute URLs to appear in
+ redirects, but a ton of boneheaded webservers and CGIs out
+ there break the rules and use relative URLs, and popular
+ browsers are lenient about this, so wget should be too. */
+ construced_newloc = uri_merge (url, mynewloc);
+ xfree (mynewloc);
+ mynewloc = construced_newloc;
+
+ /* Now, see if this new location makes sense. */
+ newloc_parsed = url_parse (mynewloc, &up_error_code);
+ if (!newloc_parsed)
{
- logprintf (LOG_NOTQUIET, _("%s: Redirection to itself.\n"),
- mynewloc);
+ logprintf (LOG_NOTQUIET, "%s: %s.\n", mynewloc,
+ url_error (up_error_code));
+ url_free (u);
+ xfree (url);
+ xfree (mynewloc);
+ RESTORE_POST_DATA;
+ return result;
+ }
+
+ /* Now mynewloc will become newloc_parsed->url, because if the
+ Location contained relative paths like .././something, we
+ don't want that propagating as url. */
+ xfree (mynewloc);
+ mynewloc = xstrdup (newloc_parsed->url);
+
+ /* Check for max. number of redirections. */
+ if (++redirection_count > MAX_REDIRECTIONS)
+ {
+ logprintf (LOG_NOTQUIET, _("%d redirections exceeded.\n"),
+ MAX_REDIRECTIONS);
+ url_free (newloc_parsed);
+ url_free (u);
+ xfree (url);
+ xfree (mynewloc);
+ RESTORE_POST_DATA;
return WRONGCODE;
}
- if (mynewloc)
+
+ xfree (url);
+ url = mynewloc;
+ url_free (u);
+ u = newloc_parsed;
+
+ /* If we're being redirected from POST, we don't want to POST
+ again. Many requests answer POST with a redirection to an
+ index page; that redirection is clearly a GET. We "suspend"
+ POST data for the duration of the redirections, and restore
+ it when we're done. */
+ if (!post_data_suspended)
+ SUSPEND_POST_DATA;
+
+ goto redirected;
+ }
+
+ if (local_file)
+ {
+ if (*dt & RETROKF)
{
- free (url);
- url = mynewloc;
+ register_download (u->url, local_file);
+ if (redirection_count && 0 != strcmp (origurl, u->url))
+ register_redirection (origurl, u->url);
+ if (*dt & TEXTHTML)
+ register_html (u->url, local_file);
}
- freeurl (u, 1);
- already_redirected = 1;
- goto again;
}
+
if (file)
+ *file = local_file ? local_file : NULL;
+ else
+ xfree_null (local_file);
+
+ url_free (u);
+
+ if (redirection_count)
{
- if (u->local)
- *file = xstrdup (u->local);
+ if (newloc)
+ *newloc = url;
else
- *file = NULL;
+ xfree (url);
}
- freeurl (u, 1);
-
- if (newloc)
- *newloc = url;
else
- free (url);
+ {
+ if (newloc)
+ *newloc = NULL;
+ xfree (url);
+ }
+
+ RESTORE_POST_DATA;
return result;
}
-/* Find the URL-s in the file and call retrieve_url() for each of
+/* Find the URLs in the file and call retrieve_url() for each of
them. If HTML is non-zero, treat the file as HTML, and construct
- the URL-s accordingly.
+ the URLs accordingly.
+
+ If opt.recursive is set, call retrieve_tree() for each file. */
- If opt.recursive is set, call recursive_retrieve() for each file. */
uerr_t
retrieve_from_file (const char *file, int html, int *count)
{
uerr_t status;
- urlpos *url_list, *cur_url;
+ struct urlpos *url_list, *cur_url;
- /* If spider-mode is on, we do not want get_urls_html barfing
- errors on baseless links. */
- url_list = (html ? get_urls_html (file, NULL, opt.spider)
+ url_list = (html ? get_urls_html (file, NULL, NULL)
: get_urls_file (file));
status = RETROK; /* Suppose everything is OK. */
*count = 0; /* Reset the URL count. */
- recursive_reset ();
+
for (cur_url = url_list; cur_url; cur_url = cur_url->next, ++*count)
{
- char *filename, *new_file;
+ char *filename = NULL, *new_file = NULL;
int dt;
- if (opt.quota && opt.downloaded > opt.quota)
+ if (cur_url->ignore_when_downloading)
+ continue;
+
+ if (opt.quota && total_downloaded_bytes > opt.quota)
{
status = QUOTEXC;
break;
}
- status = retrieve_url (cur_url->url, &filename, &new_file, NULL, &dt);
- if (opt.recursive && status == RETROK && (dt & TEXTHTML))
- status = recursive_retrieve (filename, new_file ? new_file
- : cur_url->url);
+ if ((opt.recursive || opt.page_requisites)
+ && cur_url->url->scheme != SCHEME_FTP)
+ status = retrieve_tree (cur_url->url->url);
+ else
+ status = retrieve_url (cur_url->url->url, &filename, &new_file, NULL, &dt);
if (filename && opt.delete_after && file_exists_p (filename))
{
+ DEBUGP (("Removing file due to --delete-after in"
+ " retrieve_from_file():\n"));
logprintf (LOG_VERBOSE, _("Removing %s.\n"), filename);
if (unlink (filename))
logprintf (LOG_NOTQUIET, "unlink: %s\n", strerror (errno));
dt &= ~RETROKF;
}
- FREE_MAYBE (new_file);
- FREE_MAYBE (filename);
+ xfree_null (new_file);
+ xfree_null (filename);
}
/* Free the linked list of URL-s. */
{
logputs (LOG_VERBOSE, (n1 == n2) ? _("Giving up.\n\n") : _("Retrying.\n\n"));
}
+
+/* If opt.wait or opt.waitretry are specified, and if certain
+ conditions are met, sleep the appropriate number of seconds. See
+ the documentation of --wait and --waitretry for more information.
+
+ COUNT is the count of current retrieval, beginning with 1. */
+
+void
+sleep_between_retrievals (int count)
+{
+ static int first_retrieval = 1;
+
+ if (first_retrieval)
+ {
+ /* Don't sleep before the very first retrieval. */
+ first_retrieval = 0;
+ return;
+ }
+
+ if (opt.waitretry && count > 1)
+ {
+ /* If opt.waitretry is specified and this is a retry, wait for
+ COUNT-1 number of seconds, or for opt.waitretry seconds. */
+ if (count <= opt.waitretry)
+ xsleep (count - 1);
+ else
+ xsleep (opt.waitretry);
+ }
+ else if (opt.wait)
+ {
+ if (!opt.random_wait || count > 1)
+ /* If random-wait is not specified, or if we are sleeping
+ between retries of the same download, sleep the fixed
+ interval. */
+ xsleep (opt.wait);
+ else
+ {
+ /* Sleep a random amount of time averaging in opt.wait
+ seconds. The sleeping amount ranges from 0 to
+ opt.wait*2, inclusive. */
+ double waitsecs = 2 * opt.wait * random_float ();
+ DEBUGP (("sleep_between_retrievals: avg=%f,sleep=%f\n",
+ opt.wait, waitsecs));
+ xsleep (waitsecs);
+ }
+ }
+}
+
+/* Free the linked list of urlpos. */
+void
+free_urlpos (struct urlpos *l)
+{
+ while (l)
+ {
+ struct urlpos *next = l->next;
+ if (l->url)
+ url_free (l->url);
+ xfree_null (l->local_name);
+ xfree (l);
+ l = next;
+ }
+}
+
+/* Rotate FNAME opt.backups times */
+void
+rotate_backups(const char *fname)
+{
+ int maxlen = strlen (fname) + 1 + numdigit (opt.backups) + 1;
+ char *from = (char *)alloca (maxlen);
+ char *to = (char *)alloca (maxlen);
+ struct stat sb;
+ int i;
+
+ if (stat (fname, &sb) == 0)
+ if (S_ISREG (sb.st_mode) == 0)
+ return;
+
+ for (i = opt.backups; i > 1; i--)
+ {
+ sprintf (from, "%s.%d", fname, i - 1);
+ sprintf (to, "%s.%d", fname, i);
+ rename (from, to);
+ }
+
+ sprintf (to, "%s.%d", fname, 1);
+ rename(fname, to);
+}
+
+static int no_proxy_match PARAMS ((const char *, const char **));
+
+/* Return the URL of the proxy appropriate for url U. */
+
+static char *
+getproxy (struct url *u)
+{
+ char *proxy = NULL;
+ char *rewritten_url;
+ static char rewritten_storage[1024];
+
+ if (!opt.use_proxy)
+ return NULL;
+ if (!no_proxy_match (u->host, (const char **)opt.no_proxy))
+ return NULL;
+
+ switch (u->scheme)
+ {
+ case SCHEME_HTTP:
+ proxy = opt.http_proxy ? opt.http_proxy : getenv ("http_proxy");
+ break;
+#ifdef HAVE_SSL
+ case SCHEME_HTTPS:
+ proxy = opt.https_proxy ? opt.https_proxy : getenv ("https_proxy");
+ break;
+#endif
+ case SCHEME_FTP:
+ proxy = opt.ftp_proxy ? opt.ftp_proxy : getenv ("ftp_proxy");
+ break;
+ case SCHEME_INVALID:
+ break;
+ }
+ if (!proxy || !*proxy)
+ return NULL;
+
+ /* Handle shorthands. `rewritten_storage' is a kludge to allow
+ getproxy() to return static storage. */
+ rewritten_url = rewrite_shorthand_url (proxy);
+ if (rewritten_url)
+ {
+ strncpy (rewritten_storage, rewritten_url, sizeof (rewritten_storage));
+ rewritten_storage[sizeof (rewritten_storage) - 1] = '\0';
+ proxy = rewritten_storage;
+ }
+
+ return proxy;
+}
+
+/* Should a host be accessed through proxy, concerning no_proxy? */
+int
+no_proxy_match (const char *host, const char **no_proxy)
+{
+ if (!no_proxy)
+ return 1;
+ else
+ return !sufmatch (no_proxy, host);
+}