retr.c

一个从网络上自动下载文件的自由工具

/* File retrieval.
   Copyright (C) 1996, 1997, 1998, 1999, 2000, 2001, 2002, 2003,
   2004, 2005, 2006, 2007, 2008 Free Software Foundation, Inc.

This file is part of GNU Wget.

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 3 of the License, or (at
your option) any later version.

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 Wget.  If not, see <http://www.gnu.org/licenses/>.

Additional permission under GNU GPL version 3 section 7

If you modify this program, or any covered work, by linking or
combining it with the OpenSSL project's OpenSSL library (or a
modified version of that library), containing parts covered by the
terms of the OpenSSL or SSLeay licenses, the Free Software Foundation
grants you additional permission to convey the resulting work.
Corresponding Source for a non-source form of such a combination
shall include the source code for the parts of OpenSSL used as well
as that of the covered work.  */

#include <config.h>

#include <stdio.h>
#include <stdlib.h>
#ifdef HAVE_UNISTD_H
# include <unistd.h>
#endif /* HAVE_UNISTD_H */
#include <errno.h>
#include <string.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 "http.h"
#include "host.h"
#include "connect.h"
#include "hash.h"
#include "convert.h"
#include "ptimer.h"

/* Total size of downloaded files.  Used to enforce quota.  */
SUM_SIZE_INT total_downloaded_bytes;

/* Total download time in seconds. */
double total_download_time;

/* If non-NULL, the stream to which output should be written.  This
   stream is initialized when `-O' is used.  */
FILE *output_stream;

/* Whether output_document is a regular file we can manipulate,
   i.e. not `-' or a device file. */
bool output_stream_regular;

static struct {
  wgint chunk_bytes;
  double chunk_start;
  double sleep_adjust;
} limit_data;

static void
limit_bandwidth_reset (void)
{
  xzero (limit_data);
}

/* 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 (wgint bytes, struct ptimer *timer)
{
  double delta_t = ptimer_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 = (double) 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 < 0.2)
        {
          DEBUGP (("deferring a %.2f ms sleep (%s/%.2f).\n",
                   slp * 1000, number_to_static_string (limit_data.chunk_bytes),
                   delta_t));
          return;
        }
      DEBUGP (("\nsleeping %.2f ms for %s bytes, adjust %.2f ms\n",
               slp * 1000, number_to_static_string (limit_data.chunk_bytes),
               limit_data.sleep_adjust));

      t0 = ptimer_read (timer);
      xsleep (slp);
      t1 = ptimer_measure (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);
      /* If sleep_adjust is very large, it's likely due to suspension
         and not clock inaccuracy.  Don't enforce those.  */
      if (limit_data.sleep_adjust > 0.5)
        limit_data.sleep_adjust = 0.5;
      else if (limit_data.sleep_adjust < -0.5)
        limit_data.sleep_adjust = -0.5;
    }

  limit_data.chunk_bytes = 0;
  limit_data.chunk_start = ptimer_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, wgint *skip,
            wgint *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;
    }

  fwrite (buf, 1, bufsize, out);
  *written += bufsize;

  /* 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);
}

/* 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.

   TOREAD is the amount of data expected to arrive, normally only used
   by the progress gauge.

   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 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.  */

int
fd_read_body (int fd, FILE *out, wgint toread, wgint startpos,
              wgint *qtyread, wgint *qtywritten, double *elapsed, int flags)
{
  int ret = 0;

  static char dlbuf[16384];
  int dlbufsize = sizeof (dlbuf);

  struct ptimer *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. */
  bool progress_interactive = false;

  bool exact = !!(flags & rb_read_exactly);
  wgint skip = 0;

  /* How much data we've read/written.  */
  wgint sum_read = 0;
  wgint sum_written = 0;

  if (flags & rb_skip_startpos)
    skip = startpos;

  if (opt.verbose)
    {
      /* 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 = ptimer_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)
        {
          /* 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)
            {
              double waittm;
              waittm = ptimer_read (timer) - last_successful_read_tm;
              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;
                    }
                }
            }
        }
      ret = fd_read (fd, dlbuf, rdsize, tmout);

      if (progress_interactive && ret < 0 && errno == ETIMEDOUT)
        ret = 0;                /* interactive timeout, handled above */
      else if (ret <= 0)
        break;                  /* EOF or read error */

      if (progress || opt.limit_rate)
        {
          ptimer_measure (timer);
          if (ret > 0)
            last_successful_read_tm = ptimer_read (timer);
        }

      if (ret > 0)
        {
          sum_read += ret;
          if (!write_data (out, dlbuf, ret, &skip, &sum_written))
            {
              ret = -2;
              goto out;
            }
        }

      if (opt.limit_rate)
        limit_bandwidth (ret, timer);

      if (progress)
        progress_update (progress, ret, ptimer_read (timer));
#ifdef WINDOWS
      if (toread > 0 && !opt.quiet)
        ws_percenttitle (100.0 *
                         (startpos + sum_read) / (startpos + toread));
#endif
    }
  if (ret < -1)
    ret = -1;

 out:
  if (progress)
    progress_finish (progress, ptimer_read (timer));

  if (elapsed)
    *elapsed = ptimer_read (timer);
  if (timer)
    ptimer_destroy (timer);

  if (qtyread)
    *qtyread += sum_read;
  if (qtywritten)
    *qtywritten += sum_written;

  return ret;
}

/* Read a hunk of data from FD, up until a terminator.  The hunk is
   limited by whatever the TERMINATOR callback chooses as its
   terminator.  For example, if terminator stops at newline, the hunk
   will consist of a line of data; if terminator stops at two
   newlines, it can be used to read the head of an HTTP response.
   Upon determining the boundary, the function returns the data (up to
   the terminator) in malloc-allocated storage.

   In case of read error, NULL is returned.  In case of EOF and no
   data read, NULL is returned and errno set to 0.  In case of having
   read some data, but encountering EOF before seeing the terminator,