http.c

wget讓你可以在console介面下

	    {
	      int ext_num = 1;
	      do
		sprintf (*hs->local_file + local_filename_len,
			 ".%d.html", ext_num++);
	      while (file_exists_p (*hs->local_file));
	    }
	  *dt |= ADDED_HTML_EXTENSION;
	}
    }

  if (statcode == HTTP_STATUS_RANGE_NOT_SATISFIABLE)
    {
      /* If `-c' is in use and the file has been fully downloaded (or
	 the remote file has shrunk), Wget effectively requests bytes
	 after the end of file and the server response with 416.  */
      logputs (LOG_VERBOSE, _("\
\n    The file is already fully retrieved; nothing to do.\n\n"));
      /* In case the caller inspects. */
      hs->len = contlen;
      hs->res = 0;
      /* Mark as successfully retrieved. */
      *dt |= RETROKF;
      xfree_null (type);
      CLOSE_INVALIDATE (sock);	/* would be CLOSE_FINISH, but there
				   might be more bytes in the body. */
      return RETRUNNEEDED;
    }
  if ((contrange != 0 && contrange != hs->restval)
      || (H_PARTIAL (statcode) && !contrange))
    {
      /* The Range request was somehow misunderstood by the server.
	 Bail out.  */
      xfree_null (type);
      CLOSE_INVALIDATE (sock);
      return RANGEERR;
    }
  hs->contlen = contlen + contrange;

  if (opt.verbose)
    {
      if (*dt & RETROKF)
	{
	  /* No need to print this output if the body won't be
	     downloaded at all, or if the original server response is
	     printed.  */
	  logputs (LOG_VERBOSE, _("Length: "));
	  if (contlen != -1)
	    {
	      logputs (LOG_VERBOSE, with_thousand_seps (contlen + contrange));
	      if (contlen + contrange >= 1024)
		logprintf (LOG_VERBOSE, " (%s)",
			   human_readable (contlen + contrange));
	      if (contrange)
		{
		  if (contlen >= 1024)
		    logprintf (LOG_VERBOSE, _(", %s (%s) remaining"),
			       with_thousand_seps (contlen),
			       human_readable (contlen));
		  else
		    logprintf (LOG_VERBOSE, _(", %s remaining"),
			       with_thousand_seps (contlen));
		}
	    }
	  else
	    logputs (LOG_VERBOSE,
		     opt.ignore_length ? _("ignored") : _("unspecified"));
	  if (type)
	    logprintf (LOG_VERBOSE, " [%s]\n", escnonprint (type));
	  else
	    logputs (LOG_VERBOSE, "\n");
	}
    }
  xfree_null (type);
  type = NULL;			/* We don't need it any more.  */

  /* Return if we have no intention of further downloading.  */
  if (!(*dt & RETROKF) || head_only)
    {
      /* In case the caller cares to look...  */
      hs->len = 0;
      hs->res = 0;
      xfree_null (type);
      if (head_only)
	/* Pre-1.10 Wget used CLOSE_INVALIDATE here.  Now we trust the
	   servers not to send body in response to a HEAD request.  If
	   you encounter such a server (more likely a broken CGI), use
	   `--no-http-keep-alive'.  */
	CLOSE_FINISH (sock);
      else if (keep_alive && skip_short_body (sock, contlen))
	/* Successfully skipped the body; also keep using the socket. */
	CLOSE_FINISH (sock);
      else
	CLOSE_INVALIDATE (sock);
      return RETRFINISHED;
    }

  /* Open the local file.  */
  if (!output_stream)
    {
      mkalldirs (*hs->local_file);
      if (opt.backups)
	rotate_backups (*hs->local_file);
      if (hs->restval)
	fp = fopen (*hs->local_file, "ab");
      else if (ALLOW_CLOBBER)
	fp = fopen (*hs->local_file, "wb");
      else
	{
	  fp = fopen_excl (*hs->local_file, 1);
	  if (!fp && errno == EEXIST)
	    {
	      /* We cannot just invent a new name and use it (which is
		 what functions like unique_create typically do)
		 because we told the user we'd use this name.
		 Instead, return and retry the download.  */
	      logprintf (LOG_NOTQUIET,
			 _("%s has sprung into existence.\n"),
			 *hs->local_file);
	      CLOSE_INVALIDATE (sock);
	      return FOPEN_EXCL_ERR;
	    }
	}
      if (!fp)
	{
	  logprintf (LOG_NOTQUIET, "%s: %s\n", *hs->local_file, strerror (errno));
	  CLOSE_INVALIDATE (sock);
	  return FOPENERR;
	}
    }
  else
    fp = output_stream;

  /* #### This confuses the timestamping code that checks for file
     size.  Maybe we should save some additional information?  */
  if (opt.save_headers)
    fwrite (head, 1, strlen (head), fp);

  /* Now we no longer need to store the response header. */
  xfree (head);

  /* Download the request body.  */
  flags = 0;
  if (contlen != -1)
    /* If content-length is present, read that much; otherwise, read
       until EOF.  The HTTP spec doesn't require the server to
       actually close the connection when it's done sending data. */
    flags |= rb_read_exactly;
  if (hs->restval > 0 && contrange == 0)
    /* If the server ignored our range request, instruct fd_read_body
       to skip the first RESTVAL bytes of body.  */
    flags |= rb_skip_startpos;
  hs->len = hs->restval;
  hs->rd_size = 0;
  hs->res = fd_read_body (sock, fp, contlen != -1 ? contlen : 0,
			  hs->restval, &hs->rd_size, &hs->len, &hs->dltime,
			  flags);

  if (hs->res >= 0)
    CLOSE_FINISH (sock);
  else
    CLOSE_INVALIDATE (sock);

  {
    /* Close or flush the file.  We have to be careful to check for
       error here.  Checking the result of fwrite() is not enough --
       errors could go unnoticed!  */
    int flush_res;
    if (!output_stream)
      flush_res = fclose (fp);
    else
      flush_res = fflush (fp);
    if (flush_res == EOF)
      hs->res = -2;
  }
  if (hs->res == -2)
    return FWRITEERR;
  return RETRFINISHED;
}

/* The genuine HTTP loop!  This is the part where the retrieval is
   retried, and retried, and retried, and...  */
uerr_t
http_loop (struct url *u, char **newloc, char **local_file, const char *referer,
	   int *dt, struct url *proxy)
{
  int count;
  int use_ts, got_head = 0;	/* time-stamping info */
  char *filename_plus_orig_suffix;
  char *local_filename = NULL;
  char *tms, *locf, *tmrate;
  uerr_t err;
  time_t tml = -1, tmr = -1;	/* local and remote time-stamps */
  wgint local_size = 0;		/* the size of the local file */
  size_t filename_len;
  struct http_stat hstat;	/* HTTP status */
  struct_stat st;
  char *dummy = NULL;

  /* This used to be done in main(), but it's a better idea to do it
     here so that we don't go through the hoops if we're just using
     FTP or whatever. */
  if (opt.cookies)
    {
      if (!wget_cookie_jar)
	wget_cookie_jar = cookie_jar_new ();
      if (opt.cookies_input && !cookies_loaded_p)
	{
	  cookie_jar_load (wget_cookie_jar, opt.cookies_input);
	  cookies_loaded_p = 1;
	}
    }

  *newloc = NULL;

  /* Warn on (likely bogus) wildcard usage in HTTP.  */
  if (opt.ftp_glob && has_wildcards_p (u->path))
    logputs (LOG_VERBOSE, _("Warning: wildcards not supported in HTTP.\n"));

  xzero (hstat);

  /* Determine the local filename.  */
  if (local_file && *local_file)
    hstat.local_file = local_file;
  else if (local_file && !opt.output_document)
    {
      *local_file = url_file_name (u);
      hstat.local_file = local_file;
    }
  else
    {
      dummy = url_file_name (u);
      hstat.local_file = &dummy;
      /* be honest about where we will save the file */
      if (local_file && opt.output_document)
        *local_file = HYPHENP (opt.output_document) ? NULL : xstrdup (opt.output_document);
    }

  if (!opt.output_document)
    locf = *hstat.local_file;
  else
    locf = opt.output_document;

  hstat.referer = referer;

  filename_len = strlen (*hstat.local_file);
  filename_plus_orig_suffix = alloca (filename_len + sizeof (".orig"));

  if (opt.noclobber && file_exists_p (*hstat.local_file))
    {
      /* If opt.noclobber is turned on and file already exists, do not
	 retrieve the file */
      logprintf (LOG_VERBOSE, _("\
File `%s' already there; not retrieving.\n\n"), *hstat.local_file);
      /* If the file is there, we suppose it's retrieved OK.  */
      *dt |= RETROKF;

      /* #### Bogusness alert.  */
      /* If its suffix is "html" or "htm" or similar, assume text/html.  */
      if (has_html_suffix_p (*hstat.local_file))
	*dt |= TEXTHTML;

      xfree_null (dummy);
      return RETROK;
    }

  use_ts = 0;
  if (opt.timestamping)
    {
      int local_dot_orig_file_exists = 0;

      if (opt.backup_converted)
	/* If -K is specified, we'll act on the assumption that it was specified
	   last time these files were downloaded as well, and instead of just
	   comparing local file X against server file X, we'll compare local
	   file X.orig (if extant, else X) against server file X.  If -K
	   _wasn't_ specified last time, or the server contains files called
	   *.orig, -N will be back to not operating correctly with -k. */
	{
	  /* Would a single s[n]printf() call be faster?  --dan

	     Definitely not.  sprintf() is horribly slow.  It's a
	     different question whether the difference between the two
	     affects a program.  Usually I'd say "no", but at one
	     point I profiled Wget, and found that a measurable and
	     non-negligible amount of time was lost calling sprintf()
	     in url.c.  Replacing sprintf with inline calls to
	     strcpy() and number_to_string() made a difference.
	     --hniksic */
	  memcpy (filename_plus_orig_suffix, *hstat.local_file, filename_len);
	  memcpy (filename_plus_orig_suffix + filename_len,
		  ".orig", sizeof (".orig"));

	  /* Try to stat() the .orig file. */
	  if (stat (filename_plus_orig_suffix, &st) == 0)
	    {
	      local_dot_orig_file_exists = 1;
	      local_filename = filename_plus_orig_suffix;
	    }
	}      

      if (!local_dot_orig_file_exists)
	/* Couldn't stat() <file>.orig, so try to stat() <file>. */
	if (stat (*hstat.local_file, &st) == 0)
	  local_filename = *hstat.local_file;

      if (local_filename != NULL)
	/* There was a local file, so we'll check later to see if the version
	   the server has is the same version we already have, allowing us to
	   skip a download. */
	{
	  use_ts = 1;
	  tml = st.st_mtime;
#ifdef WINDOWS
	  /* Modification time granularity is 2 seconds for Windows, so
	     increase local time by 1 second for later comparison. */
	  tml++;
#endif
	  local_size = st.st_size;
	  got_head = 0;
	}
    }
  /* Reset the counter.  */
  count = 0;
  *dt = 0;
  /* THE loop */
  do
    {
      /* Increment the pass counter.  */
      ++count;
      sleep_between_retrievals (count);
      /* Get the current time string.  */
      tms = time_str (NULL);
      /* Print fetch message, if opt.verbose.  */
      if (opt.verbose)
	{
	  char *hurl = url_string (u, 1);
	  char tmp[256];
	  strcpy (tmp, "        ");
	  if (count > 1)
	    sprintf (tmp, _("(try:%2d)"), count);
	  logprintf (LOG_VERBOSE, "--%s--  %s\n  %s => `%s'\n",
		     tms, hurl, tmp, locf);
#ifdef WINDOWS
	  ws_changetitle (hurl);
#endif
	  xfree (hurl);
	}

      /* Default document type is empty.  However, if spider mode is
	 on or time-stamping is employed, HEAD_ONLY commands is
	 encoded within *dt.  */
      if (opt.spider || (use_ts && !got_head))
	*dt |= HEAD_ONLY;
      else
	*dt &= ~HEAD_ONLY;

      /* Decide whether or not to restart.  */
      if (opt.always_rest
	  && stat (locf, &st) == 0
	  && S_ISREG (st.st_mode))
	/* When -c is used, continue from on-disk size.  (Can't use
	   hstat.len even if count>1 because we don't want a failed
	   first attempt to clobber existing data.)  */
	hstat.restval = st.st_size;
      else if (count > 1)
	/* otherwise, continue where the previous try left off */
	hstat.restval = hstat.len;
      else
	hstat.restval = 0;

      /* Decide whether to send the no-cache directive.  We send it in
	 two cases:
	   a) we're using a proxy, and we're past our first retrieval.
	      Some proxies are notorious for caching incomplete data, so
	      we require a fresh get.
	   b) caching is explicitly inhibited. */
      if ((proxy && count > 1)	/* a */
	  || !opt.allow_cache	/* b */
	  )
	*dt |= SEND_NOCACHE;
      else
	*dt &= ~SEND_NOCACHE;

      /* Try fetching the document, or at least its head.  */
      err = gethttp (u, &hstat, dt, proxy);

      /* It's unfortunate that wget determines the local filename before finding
	 out the Content-Type of the file.  Barring a major restructuring of the
	 code, we need to re-set locf here, since gethttp() may have xrealloc()d
	 *hstat.local_file to tack on ".html". */
      if (!opt.output_document)
	locf = *hstat.local_file;

      /* Time?  */
      tms = time_str (NULL);
      /* Get the new location (with or without the redirection).  */
      if (hstat.newloc)
	*newloc = xstrdup (hstat.newloc);
      switch (err)
	{
	case HERR: case HEOF: case CONSOCKERR: case CONCLOSED:
	case CONERROR: case READERR: case WRITEFAILED:
	case RANGEERR: case FOPEN_EXCL_ERR:
	  /* Non-fatal errors continue executing the loop, which will
	     bring them to "while" statement at the end, to judge
	     whether the number of tries was exceeded.  */
	  free_hstat (&hstat);
	  printwhat (count, opt.ntry);
	  if (err == FOPEN_EXCL_ERR)
	    {
	      /* Re-determine the file name. */
	      if (local_file && *local_file)
		{
		  xfree (*local_file);
		  *local_file = url_file_name (u);
		  hstat.local_file = local_file;
		}
	      else
		{
		  xfree (dummy);
		  dummy = url_file_name (u);
		  hstat.local_file = &dummy;
		}
	      /* be honest about where we will save the file */
	      if (local_file && opt.output_document)
		*local_file = HYPHENP (opt.output_document) ? NULL : xstrdup (opt.output_document);
	      if (!opt.output_document)
		locf = *hstat.local_file;
	      else
		locf = opt.output_document;
	    }
	  continue;
	case HOSTERR: case CONIMPOSSIBLE: case PROXERR: case AUTHFAILED: 
	case SSLINITFAILED: case CONTNOTSUPPORTED:
	  /* Fatal errors just return from the function.  */
	  free_hstat (&hstat);
	  xfree_null (dummy);
	  return err;
	case FWRITEERR: case FOPENERR:
	  /* Another fatal error.  */
	  logputs (LOG_VERBOSE, "\n");
	  logprintf (LOG_NOTQUIET, _("Cannot write to `%s' (%s).\n"),
		     *hstat.local_file, strerror (errno));
	  free_hstat (&hstat);
	  xfree_null (dummy);
	  return err;
	case CONSSLERR:
	  /* Another fatal error.  */
	  logprintf (LOG_NOTQUIET, _("Unable to establish SSL connection.\n"));