http.c

elinks下lynx是最重要的二个文本浏览器, 在linux下非常实用, lynx比elinks早的多, 目前好像停止开发, 这是lynx源代码

	BStrCat0(command, crlf);	/* Blank line means "end" of headers */

	BStrCat(command, anAnchor->post_data);
    }
    else
	BStrCat0(command, crlf);	/* Blank line means "end" of headers */

    if (TRACE) {
	CTRACE((tfp, "Writing:\n"));
	trace_bstring(command);
#ifdef USE_SSL
	CTRACE((tfp, "%s",
	       (anAnchor->post_data && !do_connect ? crlf : "")));
#else
	CTRACE((tfp, "%s",
	       (anAnchor->post_data ? crlf : "")));
#endif /* USE_SSL */
	CTRACE((tfp, "----------------------------------\n"));
    }

  _HTProgress (gettext("Sending HTTP request."));

#ifdef    NOT_ASCII  /* S/390 -- gil -- 0548 */
  {   char *p;

      for ( p = BStrData(command); p < BStrData(command) + BStrLen(command); p++ )
	  *p = TOASCII(*p);
  }
#endif /* NOT_ASCII */
  status = HTTP_NETWRITE(s, BStrData(command), BStrLen(command), handle);
  BStrFree(command);
  FREE(linebuf);
  if (status <= 0) {
      if (status == 0) {
	  CTRACE((tfp, "HTTP: Got status 0 in initial write\n"));
	  /* Do nothing. */
      } else if ((SOCKET_ERRNO == ENOTCONN ||
		  SOCKET_ERRNO == ECONNRESET ||
		  SOCKET_ERRNO == EPIPE) &&
		 !already_retrying &&
		 /* Don't retry if we're posting. */ !do_post) {
	    /*
	    **	Arrrrgh, HTTP 0/1 compatibility problem, maybe.
	    */
	    CTRACE((tfp, "HTTP: BONZO ON WRITE Trying again with HTTP0 request.\n"));
	    _HTProgress (RETRYING_AS_HTTP0);
	    HTTP_NETCLOSE(s, handle);
	    extensions = NO;
	    already_retrying = TRUE;
	    goto try_again;
      } else {
	  CTRACE((tfp, "HTTP: Hit unexpected network WRITE error; aborting connection.\n"));
	  HTTP_NETCLOSE(s, handle);
	  status = -1;
	  HTAlert(gettext("Unexpected network write error; connection aborted."));
	  goto done;
      }
  }

  CTRACE((tfp, "HTTP: WRITE delivered OK\n"));
  _HTProgress (gettext("HTTP request sent; waiting for response."));

  /*	Read the first line of the response
  **	-----------------------------------
  */
  {
    /* Get numeric status etc */
    BOOL end_of_file = NO;
    int buffer_length = INIT_LINE_SIZE;

    line_buffer = typecallocn(char, buffer_length);
    if (line_buffer == NULL)
	outofmem(__FILE__, "HTLoadHTTP");

    HTReadProgress (bytes_already_read = 0, 0);
    do {/* Loop to read in the first line */
	/*
	**  Extend line buffer if necessary for those crazy WAIS URLs ;-)
	*/
	if (buffer_length - length < LINE_EXTEND_THRESH) {
	    buffer_length = buffer_length + buffer_length;
	    line_buffer =
	      (char *)realloc(line_buffer, (buffer_length * sizeof(char)));
	    if (line_buffer == NULL)
		outofmem(__FILE__, "HTLoadHTTP");
	}
	CTRACE((tfp, "HTTP: Trying to read %d\n", buffer_length - length - 1));
	status = HTTP_NETREAD(s, line_buffer + length,
			      buffer_length - length - 1, handle);
	CTRACE((tfp, "HTTP: Read %d\n", status));
	if (status <= 0) {
	    /*
	     *	Retry if we get nothing back too.
	     *	Bomb out if we get nothing twice.
	     */
	    if (status == HT_INTERRUPTED) {
		CTRACE((tfp, "HTTP: Interrupted initial read.\n"));
		_HTProgress (CONNECTION_INTERRUPTED);
		HTTP_NETCLOSE(s, handle);
		status = HT_NO_DATA;
		goto clean_up;
	    } else if  (status < 0 &&
			(SOCKET_ERRNO == ENOTCONN ||
#ifdef _WINDOWS	/* 1997/11/09 (Sun) 16:59:58 */
			 SOCKET_ERRNO == ETIMEDOUT ||
#endif
			 SOCKET_ERRNO == ECONNRESET ||
			 SOCKET_ERRNO == EPIPE) &&
			!already_retrying && !do_post) {
		/*
		**  Arrrrgh, HTTP 0/1 compability problem, maybe.
		*/
		CTRACE((tfp, "HTTP: BONZO Trying again with HTTP0 request.\n"));
		HTTP_NETCLOSE(s, handle);
		FREE(line_buffer);
		FREE(line_kept_clean);

		extensions = NO;
		already_retrying = TRUE;
		_HTProgress (RETRYING_AS_HTTP0);
		goto try_again;
	    } else {
		CTRACE((tfp, "HTTP: Hit unexpected network read error; aborting connection; status %d.\n",
			   status));
		HTAlert(gettext("Unexpected network read error; connection aborted."));
		HTTP_NETCLOSE(s, handle);
		status = -1;
		goto clean_up;
	    }
	}

#ifdef    NOT_ASCII  /* S/390 -- gil -- 0564 */
	{   char *p;

	    for ( p = line_buffer + length; p < line_buffer + length + status; p++ )
		*p = FROMASCII(*p);
	}
#endif /* NOT_ASCII */

	bytes_already_read += status;
	HTReadProgress (bytes_already_read, 0);

#ifdef UCX  /* UCX returns -1 on EOF */
	if (status == 0 || status == -1)
#else
	if (status == 0)
#endif
	{
	    end_of_file = YES;
	    break;
	}
	line_buffer[length+status] = 0;

	if (line_buffer) {
	    FREE(line_kept_clean);
	    line_kept_clean = (char *)malloc(buffer_length * sizeof(char));
	    if (line_kept_clean == NULL)
		outofmem(__FILE__, "HTLoadHTTP");
	    memcpy(line_kept_clean, line_buffer, buffer_length);
	    real_length_of_line = length + status;
	}

	eol = strchr(line_buffer + length, LF);
	/* Do we *really* want to do this? */
	if (eol && eol != line_buffer && *(eol-1) == CR)
	    *(eol-1) = ' ';

	length = length + status;

	/* Do we really want to do *this*? */
	if (eol)
	    *eol = 0;		/* Terminate the line */
    }
    /*	All we need is the first line of the response.	If it's a HTTP/1.0
    **	response, then the first line will be absurdly short and therefore
    **	we can safely gate the number of bytes read through this code
    **	(as opposed to below) to ~1000.
    **
    **	Well, let's try 100.
    */
    while (!eol && !end_of_file && bytes_already_read < 100);
  } /* Scope of loop variables */

  /* save total length, in case we decide later to show it all - kw */
  rawlength = length;

  /*	We now have a terminated unfolded line.  Parse it.
  **	--------------------------------------------------
  */
  CTRACE((tfp, "HTTP: Rx: %s\n", line_buffer));

  /*
  **  Kludge to work with old buggy servers and the VMS Help gateway.
  **  They can't handle the third word, so we try again without it.
  */
  if (extensions &&	  /* Old buggy server or Help gateway? */
      (0==strncmp(line_buffer,"<TITLE>Bad File Request</TITLE>",31) ||
       0==strncmp(line_buffer,"Address should begin with",25) ||
       0==strncmp(line_buffer,"<TITLE>Help ",12) ||
       0==strcmp(line_buffer,
		 "Document address invalid or access not authorised"))) {
      FREE(line_buffer);
      FREE(line_kept_clean);
      extensions = NO;
      already_retrying = TRUE;
      CTRACE((tfp, "HTTP: close socket %d to retry with HTTP0\n", s));
      HTTP_NETCLOSE(s, handle);
      /* print a progress message */
      _HTProgress (RETRYING_AS_HTTP0);
      goto try_again;
  }


  {
    int fields;
    char server_version[VERSION_LENGTH+1];

    server_version[0] = 0;

    fields = sscanf(line_buffer, "%20s %d",
		    server_version,
		    &server_status);

    CTRACE((tfp, "HTTP: Scanned %d fields from line_buffer\n", fields));

    if (http_error_file) {     /* Make the status code externally available */
	FILE *error_file;
#ifdef SERVER_STATUS_ONLY
	error_file = fopen(http_error_file, TXT_W);
	if (error_file) {		/* Managed to open the file */
	    fprintf(error_file, "error=%d\n", server_status);
	    fclose(error_file);
	}
#else
	error_file = fopen(http_error_file, TXT_A);
	if (error_file) {		/* Managed to open the file */
	    fprintf(error_file, "   URL=%s (%s)\n", url, METHOD);
	    fprintf(error_file, "STATUS=%s\n", line_buffer);
	    fclose(error_file);
	}
#endif /* SERVER_STATUS_ONLY */
    }

    /*
    **	Rule out a non-HTTP/1.n reply as best we can.
    */
    if (fields < 2 || !server_version[0] || server_version[0] != 'H' ||
	server_version[1] != 'T' || server_version[2] != 'T' ||
	server_version[3] != 'P' || server_version[4] != '/' ||
	server_version[6] != '.') {
	/*
	 *  Ugh!  An HTTP0 reply,
	 */
	HTAtom * encoding;

	CTRACE((tfp, "--- Talking HTTP0.\n"));

	format_in = HTFileFormat(url, &encoding, NULL);
	/*
	**  Treat all plain text as HTML.
	**  This sucks but its the only solution without
	**  without looking at content.
	*/
	if (!strncmp(HTAtom_name(format_in), "text/plain",10)) {
	    CTRACE((tfp, "HTTP: format_in being changed to text/HTML\n"));
	    format_in = WWW_HTML;
	}
	if (!IsUnityEnc(encoding)) {
	    /*
	    **	Change the format to that for "www/compressed".
	    */
	    CTRACE((tfp, "HTTP: format_in is '%s',\n", HTAtom_name(format_in)));
	    StrAllocCopy(anAnchor->content_type, HTAtom_name(format_in));
	    StrAllocCopy(anAnchor->content_encoding, HTAtom_name(encoding));
	    format_in = HTAtom_for("www/compressed");
	    CTRACE((tfp, "        Treating as '%s' with encoding '%s'\n",
			"www/compressed", HTAtom_name(encoding)));
	}

	start_of_data = line_kept_clean;
    } else {
	/*
	**  Set up to decode full HTTP/1.n response. - FM
	*/
	format_in = HTAtom_for("www/mime");
	CTRACE((tfp, "--- Talking HTTP1.\n"));

	/*
	**  We set start_of_data to "" when !eol here because there
	**  will be a put_block done below; we do *not* use the value
	**  of start_of_data (as a pointer) in the computation of
	**  length (or anything else) when !eol.  Otherwise, set the
	**  value of length to what we have beyond eol (i.e., beyond
	**  the status line). - FM
	*/
	start_of_data = eol ? eol + 1 : "";
	length = eol ? length - (start_of_data - line_buffer) : 0;

	/*
	**  Trim trailing spaces in line_buffer so that we can use
	**  it in messages which include the status line. - FM
	*/
	while (line_buffer[strlen(line_buffer)-1] == ' ')
	       line_buffer[strlen(line_buffer)-1] = '\0';

	/*
	**  Take appropriate actions based on the status. - FM
	*/
	switch (server_status/100) {
	  case 1:
	    /*
	    **	HTTP/1.1 Informational statuses.
	    **	100 Continue.
	    **	101 Switching Protocols.
	    **	> 101 is unknown.
	    **	We should never get these, and they have only
	    **	the status line and possibly other headers,
	    **	so we'll deal with them by showing the full
	    **	header to the user as text/plain. - FM
	    */
	    HTAlert(gettext("Got unexpected Informational Status."));
	    do_head = TRUE;
	    break;

	  case 2:
	    /*
	    **	Good: Got MIME object! (Successful) - FM
	    */
	    if (do_head) {
		/*
		 *  If HEAD was requested, show headers (and possibly
		 *  bogus body) for all 2xx status codes as text/plain - KW
		 */
		HTProgress(line_buffer);
		break;
	    }
	    switch (server_status) {
	      case 204:
		/*
		 *  No Content.
		 */
		HTAlert(line_buffer);
		HTTP_NETCLOSE(s, handle);
		HTNoDataOK = 1;
		status = HT_NO_DATA;
		goto clean_up;

	      case 205:
		/*
		 *  Reset Content.  The server has fulfilled the
		 *  request but nothing is returned and we should
		 *  reset any form content.  We'll instruct the
		 *  user to do that, and restore the current
		 *  document. - FM
		 */
		HTAlert(gettext("Request fulfilled.  Reset Content."));
		HTTP_NETCLOSE(s, handle);
		status = HT_NO_DATA;
		goto clean_up;

	      case 206:
		/*
		 *  Partial Content.  We didn't send a Range
		 *  so something went wrong somewhere.	Show
		 *  the status message and restore the current
		 *  document. - FM
		 */
		HTAlert(line_buffer);
		HTTP_NETCLOSE(s, handle);
		status = HT_NO_DATA;
		goto clean_up;

	      default:
		/*
		 *  200 OK.
		 *  201 Created.
		 *  202 Accepted.
		 *  203 Non-Authoritative Information.
		 *  > 206 is unknown.
		 *  All should return something to display.
		 */
#if defined(USE_SSL) && !defined(DISABLE_NEWS)
		if (do_connect) {
		    CTRACE((tfp, "HTTP: Proxy tunnel to '%s' established.\n",
				connect_host));
		    do_connect = FALSE;
		    url = connect_url;
		    FREE(line_buffer);
		    FREE(line_kept_clean);
		    if (!strncmp(connect_url, "snews", 5)) {
			CTRACE((tfp,
			"      Will attempt handshake and snews connection.\n"));
			status = HTNewsProxyConnect(s, url, anAnchor,
						    format_out, sink);
			goto done;
		    }
		    did_connect = TRUE;
		    already_retrying = TRUE;
		    eol = 0;
		    bytes_already_read = 0;
		    had_header = NO;
		    length = 0;
		    doing_redirect = FALSE;
		    permanent_redirection = FALSE;
		    target = NULL;
		    CTRACE((tfp,
			"      Will attempt handshake and resubmit headers.\n"));
		    goto use_tunnel;
		}
#endif /* USE_SSL */
		HTProgress(line_buffer);
	    } /* case 2 switch */
	    break;

	  case 3:
	    /*
	    **	Various forms of Redirection. - FM
	    **	300 Multiple Choices.
	    **	301 Moved Permanently.
	    **	302 Found (temporary; we can, and do, use GET).
	    **	303 See Other (temporary; always use GET).
	    **	304 Not Modified.
	    **	305 Use Proxy.
	    **	306 Set Proxy.
	    **	307 Temporary Redirect with method retained.
	    **	> 308 is unknown.
	    */
	    if (no_url_redirection || do_head || keep_mime_headers) {
		/*
		 *  If any of these flags are set, we do not redirect,
		 *  but instead show what was returned to the user as
		 *  text/plain. - FM
		 */
		HTProgress(line_buffer);
		break;
	    }

	    if (server_status == 300) { /* Multiple Choices */
		/*
		 *  For client driven content negotiation.  The server
		 *  should be sending some way for the user-agent to
		 *  make a selection, so we'll show the user whatever
		 *  the server returns.  There might be a Location:
		 *  header with the server's preference present, but
		 *  the choice should be up to the user, someday based
		 *  on an Alternates: header, and a body always should
		 *  be present with descriptions and links for the
		 *  choices (i.e., we use the latter, for now). - FM
		 */
		HTAlert(line_buffer);
		if (traversal) {
		    HTTP_NETCLOSE(s, handle);
		    status = -1;
		    goto clean_up;
		}
		if (!dump_output_immediately &&
		    format_out == HTAtom_for("www/download")) {
		    /*
		     *	Convert a download request to
		     *	a presentation request for
		     *	interactive users. - FM
		     */
		    format_out = WWW_PRESENT;
		}
		break;
	    }

	    if (server_status == 304) { /* Not Modified */
		/*
		 *  We didn't send an "If-Modified-Since" header,
		 *  so this status is inappropriate.  We'll deal
		 *  with it by showing the full header to the user
		 *  as text/plain. - FM
		 */
		HTAlert(gettext("Got unexpected 304 Not Modified status."));
		do_head = TRUE;
		break;
	    }

	    if (server_status == 305 ||
		server_status == 306 ||
		server_status > 307) {
		/*
		 *  Show user the content, if any, for 305, 306,
		 *  or unknown status. - FM
		 */
		HTAlert(line_buffer);
		if (traversal) {
		    HTTP_NETCLOSE(s, handle);
		    status = -1;
		    goto clean_up;
		}
		if (!dump_output_immediately &&
		    format_out == HTAtom_for("www/download")) {
		    /*
		     *	Convert a download request to
		     *	a presentation request for
		     *	interactive users. - FM
		     */
		    format_out = WWW_PRESENT;
		}
		break;
	    }

	    /*
	     *	We do not load the file, but read the headers for
	     *	the "Location:", check out that redirecting_url
	     *	and if it's acceptible (e.g., not a telnet URL
	     *	when we have that disabled), initiate a new fetch.
	     *	If that's another redirecting_url, we'll repeat the
	     *	checks, and fetch initiations if acceptible, until
	     *	we reach the actual URL, or the redirection limit
	     *	set in HTAccess.c is exceeded.	If the status was 301
	     *	indicating that the relocation is permanent, we set
	     *	the permanent_redirection flag to make it permanent
	     *	for the current anchor tree (i.e., will persist until
	     *	the tree is freed or the client exits).  If the
	     *	redirection would include POST content, we seek
	     *	confirmation from an interactive user, with option to
	     *	use 303 for 301 (but not for 307), and otherwise refuse
	     *	the redirection.  We also don't allow permanent
	     *	redirection if we keep POST content.  If we don't find
	     *	the Location header or it's value is zero-length, we
	     *	display whatever the server returned, and the user
	     *	should RELOAD that to try again, or make a selection
	     *	from it if it contains links, or Left-Arrow to the
	     *	previous document. - FM
	     */
	    {
	      if ((dump_output_immediately || traversal) &&
		  do_post &&
		  server_status != 303 &&
		  server_status != 302 &&
		  server_status != 301) {
		  /*