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

 */

#ifdef USE_SSL
use_tunnel:
  /*
  ** If this is an https document
  ** then do the SSL stuff here
  */
  if (did_connect || !strncmp(url, "https", 5)) {
      SSL_handle = handle = HTGetSSLHandle();
      SSL_set_fd(handle, s);
#if SSLEAY_VERSION_NUMBER >= 0x0900
      if (!try_tls)
	  handle->options|=SSL_OP_NO_TLSv1;
#endif /* SSLEAY_VERSION_NUMBER >= 0x0900 */
      HTSSLInitPRNG();
      status = SSL_connect(handle);

      if (status <= 0) {
#if SSLEAY_VERSION_NUMBER >= 0x0900
	  if (try_tls) {
	      _HTProgress(gettext("Retrying connection without TLS."));
	      try_tls = FALSE;
	      if (did_connect)
		  HTTP_NETCLOSE(s, handle);
	      goto try_again;
	  } else {
	      unsigned long SSLerror;
	      CTRACE((tfp,
"HTTP: Unable to complete SSL handshake for '%s', SSL_connect=%d, SSL error stack dump follows\n",
				url, status));
	      SSL_load_error_strings();
	      while((SSLerror=ERR_get_error())!=0) {
		  CTRACE((tfp,"HTTP: SSL: %s\n",ERR_error_string(SSLerror,NULL)));
	      }
	      HTAlert("Unable to make secure connection to remote host.");
	      if (did_connect)
		  HTTP_NETCLOSE(s, handle);
	      status = HT_NOT_LOADED;
	      goto done;
	  }
#else
	  unsigned long SSLerror;
	  CTRACE((tfp,
"HTTP: Unable to complete SSL handshake for '%s', SSL_connect=%d, SSL error stack dump follows\n",
				url, status));
	  SSL_load_error_strings();
	  while((SSLerror=ERR_get_error())!=0) {
	      CTRACE((tfp,"HTTP: SSL: %s\n",ERR_error_string(SSLerror,NULL)));
	  }
	  HTAlert("Unable to make secure connection to remote host.");
	  if (did_connect)
	      HTTP_NETCLOSE(s, handle);
	  status = HT_NOT_LOADED;
	  goto done;
#endif /* SSLEAY_VERSION_NUMBER >= 0x0900 */
      }

      X509_NAME_oneline(X509_get_subject_name(SSL_get_peer_certificate(handle)),
		        ssl_dn, sizeof(ssl_dn));
      if ((cert_host = strstr(ssl_dn, "/CN=")) == NULL) {
	  HTSprintf0(&msg,
		     gettext("SSL error:Can't find common name in certificate-Continue?"));
	   if (! HTForcedPrompt(ssl_noprompt, msg, YES)) {
	      status = HT_NOT_LOADED;
	      FREE(msg);
	      goto done;
	  }
      } else {
	  cert_host += 4;
	  if ((p = strchr(cert_host, '/')) != NULL)
	      *p = '\0';
	  if ((p = strchr(cert_host, ':')) != NULL)
	      *p = '\0';
	  ssl_host = HTParse(url, "", PARSE_HOST);
	  if ((p = strchr(ssl_host, ':')) != NULL)
	      *p = '\0';
	  if (strcasecomp(ssl_host, cert_host)) {
	      HTSprintf0(&msg,
			 gettext("SSL error:host(%s)!=cert(%s)-Continue?"),
			 ssl_host,
			 cert_host);
	      if (! HTForcedPrompt(ssl_noprompt, msg, YES)) {
		  status = HT_NOT_LOADED;
		  FREE(msg);
		  goto done;
	      }
	  }
      }

      HTSprintf0(&msg,
		 gettext("Secure %d-bit %s (%s) HTTP connection"),
		 SSL_get_cipher_bits(handle, NULL),
		 SSL_get_cipher_version(handle),
		 SSL_get_cipher(handle));
      _HTProgress(msg);
      FREE(msg);
  }
#endif /* USE_SSL */

  /*	Ask that node for the document,
  **	omitting the host name & anchor
  */
  {
    char * p1 = (HTParse(url, "", PARSE_PATH|PARSE_PUNCTUATION));

#ifdef USE_SSL
    if (do_connect) {
	METHOD = "CONNECT";
	BStrCopy0(command, "CONNECT ");
    } else
#endif /* USE_SSL */
    if (do_post) {
	METHOD = "POST";
	BStrCopy0(command, "POST ");
    } else if (do_head) {
	METHOD = "HEAD";
	BStrCopy0(command, "HEAD ");
    } else {
	METHOD = "GET";
	BStrCopy0(command, "GET ");
    }

    /*
    **	If we are using a proxy gateway don't copy in the first slash
    **	of say: /gopher://a;lkdjfl;ajdf;lkj/;aldk/adflj
    **	so that just gopher://.... is sent.
    */
#ifdef USE_SSL
    if (using_proxy && !did_connect) {
	if (do_connect)
	    BStrCat0(command, connect_host);
	else
	    BStrCat0(command, p1+1);
    }
#else
    if (using_proxy)
	BStrCat0(command, p1+1);
#endif /* USE_SSL */
    else
	BStrCat0(command, p1);
    FREE(p1);
  }
  if (extensions) {
      BStrCat0(command, " ");
      BStrCat0(command, HTTP_VERSION);
  }

  BStrCat0(command, crlf);	/* CR LF, as in rfc 977 */

  if (extensions) {
      int n, i;
      char * host = NULL;

      if ((host = HTParse(anAnchor->address, "", PARSE_HOST)) != NULL) {
	  strip_userid(host);
	  HTBprintf(&command, "Host: %s%c%c", host, CR,LF);
	  FREE(host);
      }

      if (!HTPresentations)
	  HTFormatInit();
      n = HTList_count(HTPresentations);

      first_Accept = TRUE;
      len = 0;
      for (i = 0; i < n; i++) {
	  HTPresentation *pres =
			(HTPresentation *)HTList_objectAt(HTPresentations, i);
	  if (pres->get_accept) {
	      if (pres->quality < 1.0) {
		  if (pres->maxbytes > 0) {
		      sprintf(temp, ";q=%4.3f;mxb=%ld",
				    pres->quality, pres->maxbytes);
		  } else {
		      sprintf(temp, ";q=%4.3f", pres->quality);
		  }
	      } else if (pres->maxbytes > 0) {
		  sprintf(temp, ";mxb=%ld", pres->maxbytes);
	      } else {
		  temp[0] = '\0';
	      }
	      HTSprintf0(&linebuf, "%s%s%s",
			    (first_Accept ?
			       "Accept: " : ", "),
			    HTAtom_name(pres->rep),
			    temp);
	      len += strlen(linebuf);
	      if (len > 252 && !first_Accept) {
		  BStrCat0(command, crlf);
		  HTSprintf0(&linebuf, "Accept: %s%s",
				HTAtom_name(pres->rep),
				temp);
		  len = strlen(linebuf);
	      }
	      BStrCat0(command, linebuf);
	      first_Accept = FALSE;
	  }
      }
      HTBprintf(&command, "%s*/*;q=0.01%c%c",
		    (first_Accept ?
		       "Accept: " : ", "), CR, LF);
      first_Accept = FALSE;
      len = 0;

      /*
       * FIXME:  suppressing the "Accept-Encoding" in this case is done to work
       * around limitations of the presentation logic used for the command-line
       * "-base" option.  The remote site may transmit the document gzip'd, but
       * the ensuing logic in HTSaveToFile() would see the mime-type as gzip
       * rather than text/html, and not prepend the base URL.  This is less
       * efficient than accepting the compressed data and uncompressing it,
       * adding the base URL but is simpler than augmenting the dump's
       * presentation logic -TD
       */
      if (LYPrependBaseToSource && dump_output_immediately) {
	  CTRACE((tfp, "omit Accept-Encoding to work-around interaction with -source\n"));
      } else {
	  char *list = 0;
#if defined(USE_ZLIB) || defined(GZIP_PATH)
	  StrAllocCopy(list, "gzip");
#endif
#if defined(USE_ZLIB) || defined(COMPRESS_PATH)
	  if (list != 0)
	      StrAllocCat(list, ", ");
	  StrAllocCat(list, "compress");
#endif
	  if (list != 0) {
	      HTBprintf(&command, "Accept-Encoding: %s%c%c", list, CR, LF);
	      free(list);
	  }
      }

      if (language && *language) {
	  HTBprintf(&command, "Accept-Language: %s%c%c", language, CR, LF);
      }

      if (pref_charset && *pref_charset) {
	  BStrCat0(command, "Accept-Charset: ");
	  StrAllocCopy(linebuf, pref_charset);
	  if (linebuf[strlen(linebuf)-1] == ',')
	      linebuf[strlen(linebuf)-1] = '\0';
	  LYLowerCase(linebuf);
	  if (strstr(linebuf, "iso-8859-1") == NULL)
	      StrAllocCat(linebuf, ", iso-8859-1;q=0.01");
	  if (strstr(linebuf, "us-ascii") == NULL)
	      StrAllocCat(linebuf, ", us-ascii;q=0.01");
	  BStrCat0(command, linebuf);
	  HTBprintf(&command, "%c%c", CR, LF);
      }

#if 0
      /*
      **  Promote 300 (Multiple Choices) replies, if supported,
      **  over 406 (Not Acceptable) replies. - FM
      **
      **  This used to be done in versions 2.7 and 2.8*, but violates
      **  the specs for transparent content negotiation and has the
      **  effect that servers supporting those specs will send 300
      **  (Multiple Choices) instead of a normal response (e.g. 200 OK),
      **  since they will assume that the client wants to make the
      **  choice.  It is not clear whether there are any servers or sites
      **  for which sending this header really improves anything.
      **
      **  If there ever is a need to send "Negotiate: trans" and really
      **  mean it, we should send "Negotiate: trans,trans" or similar,
      **  since that is semantically equivalent and some servers may
      **  ignore "Negotiate: trans" as a special case when it comes from
      **  Lynx (to work around the old faulty behavior). - kw
      **
      **  References:
      **  RFC 2295 (see also RFC 2296), and mail to lynx-dev and
      **  [e-mail] from Koen Holtman, Jan 1999.
      */
      if (!do_post) {
	  HTBprintf(&command, "Negotiate: trans%c%c", CR, LF);
      }
#endif /* 0 */

      /*
      **  When reloading give no-cache pragma to proxy server to make
      **  it refresh its cache. -- Ari L. <[e-mail]>
      **
      **  Also send it as a Cache-Control header for HTTP/1.1. - FM
      */
      if (reloading) {
	  HTBprintf(&command, "Pragma: no-cache%c%c", CR, LF);
	  HTBprintf(&command, "Cache-Control: no-cache%c%c", CR, LF);
      }

      if (LYUserAgent && *LYUserAgent) {
	  char *cp = LYSkipBlanks(LYUserAgent);
	  /* Won't send it at all if all blank - kw */
	  if (*cp != '\0')
	      HTBprintf(&command, "User-Agent: %.*s%c%c",
		      INIT_LINE_SIZE-15, LYUserAgent, CR, LF);
      } else {
	  HTBprintf(&command, "User-Agent: %s/%s  libwww-FM/%s%c%c",
		  HTAppName ? HTAppName : "unknown",
		  HTAppVersion ? HTAppVersion : "0.0",
		  HTLibraryVersion, CR, LF);
      }

      if (personal_mail_address && !LYNoFromHeader) {
	  HTBprintf(&command, "From: %s%c%c", personal_mail_address, CR,LF);
      }

      if (!(LYUserSpecifiedURL ||
	    LYNoRefererHeader || LYNoRefererForThis) &&
	  strcmp(HTLoadedDocumentURL(), "")) {
	  char *cp = LYRequestReferer;
	  if (!cp) cp = HTLoadedDocumentURL(); /* @@@ Try both? - kw */
	  BStrCat0(command, "Referer: ");
	  if (isLYNXIMGMAP(cp)) {
	      char *cp1 = trimPoundSelector(cp);
	      BStrCat0(command, cp + LEN_LYNXIMGMAP);
	      restorePoundSelector(cp1);
	  } else {
	      BStrCat0(command, cp);
	  }
	  HTBprintf(&command, "%c%c", CR, LF);
      }

      {
	char *abspath;
	char *docname;
	char *hostname;
	char *colon;
	int portnumber;
	char *auth, *cookie = NULL;
	BOOL secure = (BOOL) (strncmp(anAnchor->address, "https", 5) ?
							FALSE : TRUE);

	abspath = HTParse(arg, "", PARSE_PATH|PARSE_PUNCTUATION);
	docname = HTParse(arg, "", PARSE_PATH);
	hostname = HTParse(arg, "", PARSE_HOST);
	if (hostname &&
	    NULL != (colon = strchr(hostname, ':'))) {
	    *(colon++) = '\0';	/* Chop off port number */
	    portnumber = atoi(colon);
	} else if (!strncmp(arg, "https", 5)) {
	    portnumber = HTTPS_PORT;
	} else	{
	    portnumber = HTTP_PORT;
	}

	/*
	**  Add Authorization, Proxy-Authorization,
	**  and/or Cookie headers, if applicable.
	*/
	if (using_proxy) {
	    /*
	    **	If we are using a proxy, first determine if
	    **	we should include an Authorization header
	    **	and/or Cookie header for the ultimate target
	    **	of this request. - FM & AJL
	    */
	    char *host2 = NULL, *path2 = NULL;
	    int port2 = (strncmp(docname, "https", 5) ?
					   HTTP_PORT : HTTPS_PORT);
	    host2 = HTParse(docname, "", PARSE_HOST);
	    path2 = HTParse(docname, "", PARSE_PATH|PARSE_PUNCTUATION);
	    if (host2) {
		if ((colon = strchr(host2, ':')) != NULL) {
		    /* Use non-default port number */
		    *colon = '\0';
		    colon++;
		    port2 = atoi(colon);
		}
	    }
	    /*
	    **	This composeAuth() does file access, i.e., for
	    **	the ultimate target of the request. - AJL
	    */
	    auth_proxy = NO;
	    if ((auth = HTAA_composeAuth(host2, port2, path2,
					 auth_proxy)) != NULL &&
		*auth != '\0') {
		/*
		**  If auth is not NULL nor zero-length, it's
		**  an Authorization header to be included. - FM
		*/
		HTBprintf(&command, "%s%c%c", auth, CR, LF);
		CTRACE((tfp, "HTTP: Sending authorization: %s\n", auth));
	    } else if (auth && *auth == '\0') {
		/*
		**  If auth is a zero-length string, the user either
		**  cancelled or goofed at the username and password
		**  prompt. - FM
		*/
		if (!(traversal || dump_output_immediately) &&
			HTConfirm(CONFIRM_WO_PASSWORD)) {
		    show_401 = TRUE;
		} else {
		    if (traversal || dump_output_immediately)
			HTAlert(FAILED_NEED_PASSWD);
#ifdef USE_SSL
		    if (did_connect)
			HTTP_NETCLOSE(s, handle);
#endif /* USE_SSL */
		    BStrFree(command);
		    FREE(hostname);
		    FREE(docname);
		    FREE(abspath);
		    FREE(host2);
		    FREE(path2);
		    status = HT_NOT_LOADED;
		    goto done;
		}
	    } else {
		CTRACE((tfp, "HTTP: Not sending authorization (yet).\n"));
	    }
	    /*
	    **	Add 'Cookie:' header, if it's HTTP or HTTPS
	    **	document being proxied.
	    */
	    if (!strncmp(docname, "http", 4)) {
		cookie = LYAddCookieHeader(host2, path2, port2, secure);
	    }
	    FREE(host2);
	    FREE(path2);
	    /*
	    **	The next composeAuth() will be for the proxy. - AJL
	    */
	    auth_proxy = YES;
	} else {
	    /*
	    **	Add cookie for a non-proxied request. - FM
	    */
	    cookie = LYAddCookieHeader(hostname, abspath, portnumber, secure);
	    auth_proxy = NO;
	}
	/*
	**  If we do have a cookie set, add it to the request buffer. - FM
	*/
	if (cookie != NULL) {
	    if (*cookie != '$') {
		/*
		**  It's a historical cookie, so signal to the
		**  server that we support modern cookies. - FM
		*/
		BStrCat0(command, "Cookie2: $Version=\"1\"");
		BStrCat0(command, crlf);
		CTRACE((tfp, "HTTP: Sending Cookie2: $Version =\"1\"\n"));
	    }
	    if (*cookie != '\0') {
		/*
		**  It's not a zero-length string, so add the header.
		**  Note that any folding of long strings has been
		**  done already in LYCookie.c. - FM
		*/
		BStrCat0(command, "Cookie: ");
		BStrCat0(command, cookie);
		BStrCat0(command, crlf);
		CTRACE((tfp, "HTTP: Sending Cookie: %s\n", cookie));
	    }
	    FREE(cookie);
	}
	FREE(abspath);

	/*
	**  If we are using a proxy, auth_proxy should be YES, and
	**  we check here whether we want a Proxy-Authorization header
	**  for it.  If we are not using a proxy, auth_proxy should
	**  still be NO, and we check here for whether we want an
	**  Authorization header. - FM & AJL
	*/
	if ((auth = HTAA_composeAuth(hostname,
				     portnumber,
				     docname,
				     auth_proxy)) != NULL &&
	    *auth != '\0') {
	    /*
	    **	If auth is not NULL nor zero-length, it's
	    **	an Authorization or Proxy-Authorization
	    **	header to be included. - FM
	    */
	    HTBprintf(&command, "%s%c%c", auth, CR, LF);
	    CTRACE((tfp, (auth_proxy ?
			 "HTTP: Sending proxy authorization: %s\n" :
			 "HTTP: Sending authorization: %s\n"),
			auth));
	} else if (auth && *auth == '\0') {
	    /*
	    **	If auth is a zero-length string, the user either
	    **	cancelled or goofed at the username and password
	    **	prompt. - FM
	    */
	    if (!(traversal || dump_output_immediately) && HTConfirm(CONFIRM_WO_PASSWORD)) {
		if (auth_proxy == TRUE) {
		    show_407 = TRUE;
		} else {
		    show_401 = TRUE;
		}
	    } else {
		if (traversal || dump_output_immediately)
		    HTAlert(FAILED_NEED_PASSWD);
		BStrFree(command);
		FREE(hostname);
		FREE(docname);
		status = HT_NOT_LOADED;
		goto done;
	    }
	} else {
	    CTRACE((tfp, (auth_proxy ?
			 "HTTP: Not sending proxy authorization (yet).\n" :
			 "HTTP: Not sending authorization (yet).\n")));
	}
	FREE(hostname);
	FREE(docname);
      }
      auth_proxy = NO;
  }

    if (
#ifdef USE_SSL
	!do_connect &&
#endif /* USE_SSL */
	do_post) {
	CTRACE((tfp, "HTTP: Doing post, content-type '%s'\n",
		     anAnchor->post_content_type
		     ? anAnchor->post_content_type
		     : "lose"));
	HTBprintf(&command, "Content-type: %s%c%c",
		   anAnchor->post_content_type
		   ? anAnchor->post_content_type
		   : "lose",
		  CR, LF);

	HTBprintf(&command, "Content-length: %d%c%c",
		  !isBEmpty(anAnchor->post_data)
		   ? BStrLen(anAnchor->post_data)
		   : 0,
		  CR, LF);