htwais.c

www工具包. 这是W3C官方支持的www支撑库. 其中提供通用目的的客户端的WebAPI: complete HTTP/1.1 (with caching, pipelining, PUT, POS

/*        *z++ = *p++ - '0';
	q = strchr(p , '=');
	if (!q) return 0;
*/
	s = strchr(q, ';');	/* (Check only) */
	if (!s) return 0;	/* Bad! No ';';	*/
        sor = z;		/* Remember where the size field was */
	z++;			/* Skip record size for now	*/
	for(p=q+1; *p!=';' ; ) {
	   if (*p == HEX_ESCAPE) {
	        char c;
		unsigned int b;
		p++;
	        c = *p++;
		b =   HTAsciiHexToChar(c);
		c = *p++;
		if (!c) break;	/* Odd number of chars! */
		*z++ = (b<<4) + HTAsciiHexToChar(c);
	    } else {
	        *z++ = *p++;	/* Record */
	    }
	}
	*sor = (z-sor-1);	/* Fill in size -- not counting size itself */
	p++;			/* After semicolon: start of next record */
    }
#endif /* OLD_CODE */

#ifdef HTDEBUG
    if (PROT_TRACE) {
	char *p;
	HTTRACE(PROT_TRACE, "WAIS........ id (%d bytes) is " _ (int)docid->size);
	for(p=docid->bytes; p<docid->bytes+docid->size; p++) {
	    if ((*p >= ' ') && (*p<= '~')) /* Assume ASCII! */
		HTTRACE(PROT_TRACE, "%c" _ *p);
	    else
		HTTRACE(PROT_TRACE, "<%x>" _ (unsigned)*p);
	}
	HTTRACE(PROT_TRACE, "\n");
    }
#endif /* HTDEBUG */

    return docid;		/* Ok */
    
} /* WAIS_from_WWW */


/*	Send a plain text record to the client		output_text_record()
**	--------------------------------------
*/

PRIVATE void output_text_record (
    HTStream *			target,
    WAISDocumentText *		record,
    boolean			quote_string_quotes,
    boolean                    binary)
{
  long count;
  if (binary) {
    (*target->isa->put_block)(target,
			      record->DocumentText->bytes,
			      record->DocumentText->size);
    return;
  }

  for(count = 0; count < record->DocumentText->size; count++){
    long ch = (unsigned char)record->DocumentText->bytes[count];
    if (ch == 27) {	/* What is this in for? Tim */

	    /* then we have an escape code */
	    /* if the next letter is '(' or ')', then ignore two letters */
	    if('(' == record->DocumentText->bytes[count + 1] ||
		')' == record->DocumentText->bytes[count + 1])
	    count += 1;             /* it is a term marker */
	    else count += 4;		/* it is a paragraph marker */
    } else if (ch == '\n' || ch == '\r') {
	    PUTC('\n');
    } else if ((ch=='\t') || isprint(ch)){
	    PUTC(ch);
    } 
  }
} /* output text record */



/*	Format A Search response for the client		display_search_response
**	---------------------------------------
** modified from tracy shen's version in wutil.c
** displays either a text record or a set of headlines.
*/
void display_search_response (HTStructured *		target,
			      SearchResponseAPDU *	response,
			      char *			database,
			      char *	 		keywords)
{
  WAISSearchResponse  *info;
  long i, k;
  char line[LINE_BUFFER_SIZE];	/* For building strings to display */
  
  BOOL archie =  strstr(database, "archie")!=0;	/* Specical handling */

  
  HTTRACE(PROT_TRACE, "WAIS........ Displaying search response\n");
  sprintf(line,
	  "Index %s contains the following %d item%s relevant to '%s'.\n",
	  database,
	  (int)(response->NumberOfRecordsReturned),
	  response->NumberOfRecordsReturned ==1 ? "" : "s",
	  keywords);
  PUTS(line);
  PUTS("The first figure for each entry is its relative score, ");
  PUTS("the second the number of lines in the item.");
  START(HTML_MENU);

  if ( response->DatabaseDiagnosticRecords != 0 ) {
    info = (WAISSearchResponse *)response->DatabaseDiagnosticRecords;
    i =0; 

    if (info->Diagnostics != NULL)
      HTshowDiags((HTStream*)target, info->Diagnostics);

    if ( info->DocHeaders != 0 ) {
      for (k=0; info->DocHeaders[k] != 0; k++ ) {
	WAISDocumentHeader* head = info->DocHeaders[k];
	char * headline = trim_junk(head->Headline);
	any * docid = head->DocumentID;
	char * docname;			/* printable version of docid */
	i++;

/*	Make a printable string out of the document id.
*/
	HTTRACE(PROT_TRACE, "HTWAIS:  %2ld: Score: %4ld, lines:%4ld '%s'\n" _ i _ 
		    (long int)(info->DocHeaders[k]->Score) _ 
		    (long int)(info->DocHeaders[k]->Lines) _ 
		    headline);

	START(HTML_LI);
	sprintf(line, "%4ld  %4ld  ",
	    head->Score,
	    head->Lines);
	PUTS(line);

	if (archie) {
	    char * www_name = WWW_from_archie(headline);
	    if (www_name) {
		HTStartAnchor(target, NULL, www_name);
		PUTS(headline);
		
		END(HTML_A);
		HT_FREE(www_name);
	    } else {
		 PUTS(headline);
		 PUTS(" (bad file name)");
	    }
	} else { /* Not archie */
	    docname =  WWW_from_WAIS(docid);
	    if (docname) {
		char * dbname = HTEscape(database, URL_XPALPHAS);
                char types_array[1000]; /* bad */
                char *type_escaped;
                types_array[0] = 0;
                if (head->Types) {
                    int i;
                    for (i = 0; head->Types[i]; i++)
                      {
                        if (i)
                          strcat (types_array, ",");

                        type_escaped = HTEscape (head->Types[i], URL_XALPHAS);
                        strcat (types_array, type_escaped);
                        HT_FREE(type_escaped);
                      }
                    HTTRACE(PROT_TRACE, "WAIS........ Types_array `%s\'\n" _ 
			       types_array);
		} else {
                    strcat (types_array, "TEXT");
		}
		sprintf(line, "%s/%s/%d/%s",
                        dbname,
                        types_array,
                        (int)(head->DocumentLength),
                        docname);
#ifdef OLD_CODE
		sprintf(line, "%s/%s/%d/%s",		/* W3 address */
				    dbname,
		    head->Types ? head->Types[0] : "TEXT",
		    (int)(head->DocumentLength),
		    docname);
#endif /* OLD_CODE */
		HTStartAnchor(target, NULL, ( (head->Types) 
		      && (!strcmp(head->Types[0], "URL"))) ? 
			      headline : line); /* NT, Sep 93 */
		PUTS(headline);
		END(HTML_A);
		HT_FREE(dbname);
		HT_FREE(docname);
	    } else {
		 PUTS("(bad doc id)");
	    }
	  }
      } /* next document header */
    } /* if there were any document headers */
    
    if ( info->ShortHeaders != 0 ) {
      k =0;
      while (info->ShortHeaders[k] != 0 ) {
	i++;
	PUTS( "(Short Header record, can't display)");
      }
    }
    if ( info->LongHeaders != 0 ) {
      k =0;
      while (info->LongHeaders[k] != 0) {
	i++;
	PUTS( "\nLong Header record, can't display\n");
      }
    }
    if ( info->Text != 0 ) {
      k =0;
      while (info->Text[k] != 0) {
	i++;
	PUTS( "\nText record\n");
	output_text_record((HTStream*)target, info->Text[k++], false, false);
      }
    }
    if ( info->Headlines != 0 ) {
      k =0;
      while (info->Headlines[k] != 0) {
	i++;
	PUTS( "\nHeadline record, can't display\n");
	/* dsply_headline_record( info->Headlines[k++]); */
      }
    }
    if ( info->Codes != 0 ) {
      k =0;
      while (info->Codes[k] != 0) {
	i++;
	PUTS( "\nCode record, can't display\n");
	/* dsply_code_record( info->Codes[k++]); */
      }
    }
  }				/* Loop: display user info */
  END(HTML_MENU);
  PUTC('\n'); ;
}


/*	HTWAISCleanup
**	-----------
**      This function closes the connection and frees memory.
**      Returns YES on OK, else NO
*/
PRIVATE int HTWAISCleanup (HTRequest *req, int status)
{
 HTNet* net;
 HTStream* input;
 wais_info* theWAISinfo = NULL;


 if (req)
   {
    net = HTRequest_net(req);
    input = HTRequest_inputStream(req);

    /* Free stream with data TO network */
    if (HTRequest_isDestination(req))
      HTRequest_removeDestination(req);
     else
      if (input)
	{
	 if (status == HT_INTERRUPTED)
	   (*input->isa->abort)(input, NULL);
	  else
	   (*input->isa->_free)(input);

	 HTRequest_setInputStream(req, NULL);
	}

    if (net)
      {
       if ((theWAISinfo = (wais_info*)HTNet_context(net)))
	 {
	  theWAISinfo->state = HTWAIS_CLEANUP;

	  if (theWAISinfo->connection)
	    close_connection_to_server(theWAISinfo->connection);

	  if (theWAISinfo->wais_database)
	    HT_FREE(theWAISinfo->wais_database);

	  if (theWAISinfo->request_message)
	    s_free(theWAISinfo->request_message);

	  if (theWAISinfo->response_message)
	    s_free(theWAISinfo->response_message);

	  HT_FREE(theWAISinfo->names);
	  HT_FREE(theWAISinfo->basetitle);

	  if (status < 0)
	    {
	     HTParentAnchor* anchor;
	     char* unescaped = NULL;
	     void* err_par;
	     unsigned int err_par_length;

	     if ((anchor = HTRequest_anchor(req)))   /* Be robust */
	       {
		char* arg;

		if ((arg = HTAnchor_physical(anchor)))
		  {
		   StrAllocCopy(unescaped, arg);
		   HTUnEscape(unescaped);
		  }
	       }

	     if (unescaped)
	       {
		err_par = (void*)unescaped;
		err_par_length = (unsigned int)(strlen(unescaped));
	       }
	      else
	       {
		err_par = (void*)"???";
		err_par_length = 3;
	       }

	     HTRequest_addError(req, ERR_FATAL, NO, HTERR_INTERNAL,
				err_par, err_par_length, "HTLoadWAIS");
	     if (unescaped) HT_FREE(unescaped);
	    }

	  /* Free and remove our own context structure for wais */
	  HT_FREE(theWAISinfo);
	  HTNet_setContext(net, NULL);
	 }
       /* End ``if (theWAISinfo)'' */

       /* Remove the request object */
       HTNet_delete(net, status);
      }

    return YES;
   }
  else
   return NO;
}




/*	Load Document from WAIS Server				HTLoadWAIS()
**	------------------------------
**
** On entry,
**	request		This is the request structure
** On exit,
**	returns		<0		Error has occured
**			HT_LOADED	OK
*/

#define MAX_KEYWORDS_LENGTH 4000
#define MAX_SERVER_LENGTH   1000
#define MAX_DATABASE_LENGTH 1000
#define MAX_SERVICE_LENGTH  1000

PRIVATE int HTWAISEvent(SOCKET soc, void * pVoid, HTEventType type)
{
 wais_info* theWAISinfo;  /* Specific protocol information */
 HTNet* net;              /* Generic protocol information */
 HTRequest* request;
 HTParentAnchor* anchor;
 const char * arg;
 HTStream* sink;
 HTFormat format_out;
#if 0
 static const char * error_header = "<h1>Access error</h1>\n<p>The following error occured in accesing a WAIS server:\n</p>\n";
#endif
 char* key;			  /* pointer to keywords in URL */
 long request_buffer_length;	/* how of the request is left */
 SearchResponseAPDU* retrieval_response = 0;
 char keywords[MAX_KEYWORDS_LENGTH + 1];
 char* server_name = NULL;
 char* service;
 char* docname = NULL;
 char* doctype = NULL;
 long document_length = -1;
 BOOL ok = NO;
 int status = HT_ERROR;

#if 0
 FILE* connection = 0;
 char* names;		/* Copy of arg to be hacked up */
 char* basetitle = NULL;
 char* wais_database = NULL;    /* name of current database */
 char*  www_database = NULL;    /* Same name escaped */
 char*  request_message = NULL; /* arbitrary message limit */
 char* response_message = NULL; /* arbitrary message limit */
#endif
    
#if 0
 extern FILE * connect_to_server();
#endif


 if ((theWAISinfo = (wais_info*)pVoid))   /* Be robust */
   {
    if ((net = theWAISinfo->net) == NULL) return HT_ERROR;
    if ((request = HTNet_request(net)) == NULL) return HT_ERROR;
    if ((anchor = HTRequest_anchor(request)))
      arg = HTAnchor_physical(anchor);
     else
      return HT_ERROR;

    sink = HTRequest_outputStream(request);
    format_out = HTRequest_outputFormat(request);
   }
  else
   return HT_ERROR;
    

 if (type == HTEvent_BEGIN)
   {
    theWAISinfo->state  = HTWAIS_BEGIN;
    theWAISinfo->result = HT_ERROR;
   }
  else
   if (type == HTEvent_CLOSE)
     {
     /* Interrupted */
      char interrupted[] = "request interruption";

      HTRequest_addError(request, ERR_FATAL, NO, HTERR_INTERRUPTED,
			 (void*)interrupted, (unsigned int)strlen(interrupted),
			 "HTLoadWAIS");
      HTWAISCleanup(request, HT_INTERRUPTED);
      return HT_OK;
     }
    else
     if (type == HTEvent_END)
       {
	HTWAISCleanup(request, (theWAISinfo ? theWAISinfo->result : HT_ERROR));
	return HT_OK;
       }


 /*	Decipher and check syntax of WWW address:
 **	----------------------------------------
 **
 **	First we remove the "wais:" if it was spcified.  920110
 */

 if (!acceptable_inited) init_acceptable();
 theWAISinfo->state = HTWAIS_PARSING_URL;
 HTTRACE(PROT_TRACE, "HTLoadWAIS.. Looking for \"%s\"\n" _ arg);

 theWAISinfo->names = HTParse(arg, "",
			      PARSE_HOST | PARSE_PATH | PARSE_PUNCTUATION);
 key = strchr(theWAISinfo->names, '?');

 if (key)
   {
    char* p;

    *key++ = 0;	/* Split off keywords */

    for (p=key; *p; p++) if (*p == '+') *p = ' ';
    HTUnEscape(key);
   }

 if (theWAISinfo->names[0] == '/')
   {
    server_name = theWAISinfo->names+1;