ccache_hash.c

harvest是一个下载html网页得机器人

	int theHKey;
	char errorReply[SERV_REPLY_LENGTH];

	/* Start by looking at the oldest connection. */
	locSockRec = TimeOutListOld;

	/* Search the list for the oldest inactive connection. */
	while ((locSockRec != NULL) && (locSockRec->sockStateStor != INACTIVE))
		locSockRec = locSockRec->listPrev;

	/* If we didn't find one, all are active. */
	if (locSockRec == NULL)
		return FALSE;

	/* Strip it from the expiration list. */
	if (locSockRec->listNext == NULL)
		TimeOutListOld = locSockRec->listPrev;
	else
		locSockRec->listNext->listPrev = locSockRec->listPrev;

	if (locSockRec->listPrev == NULL)
		TimeOutListNew = locSockRec->listNext;
	else
		locSockRec->listPrev->listNext = locSockRec->listNext;


	/* Strip it from the active hash. */
	theHKey = GetHKey(locSockRec->theHostIP);

	if (locSockRec->hashPrev == NULL) {
		SockHash[theHKey] = locSockRec->hashNext;
		if (SockHash[theHKey] != NULL)
			SockHash[theHKey]->hashPrev = NULL;
	} else
		locSockRec->hashPrev->hashNext = locSockRec->hashNext;

	if (locSockRec->hashNext != NULL)
		locSockRec->hashNext->hashPrev = locSockRec->hashPrev;

	/* Disconnect the socket. */
	CancelTimer(locSockRec->timerid);	/* Cancel the timer. */

	Disconnect(locSockRec->theSocket, errorReply);

	free_url(locSockRec->socketInfo);
	free((char *) locSockRec);

	return TRUE;
}

/**************************************************************************
**  function: TouchDate()
**
**  parameters: A record to update.
**
**  postconditions:  Last access time field is updated and this connection
**  is moved to the front of the expiration list.
***************************************************************************/
static void TouchDate(SockCntlRec * theRec)
{

	/* If it's already first don't need to move it. */
	if (theRec == TimeOutListNew)
		return;

	/* Move it to the front of the list. */
	theRec->listPrev->listNext = theRec->listNext;
	if (theRec->listNext != NULL)
		theRec->listNext->listPrev = theRec->listPrev;

	if (theRec == TimeOutListOld)
		TimeOutListOld = theRec->listPrev;

	theRec->listPrev = NULL;
	theRec->listNext = TimeOutListNew;

	TimeOutListNew->listPrev = theRec;
	TimeOutListNew = theRec;

	CancelTimer(theRec->timerid);
	theRec->timerid = SetTimer(timeOut, ExpireSocket);
}

/**********************************************************************
** function: SockRetrieve()
**
** parameters: An allocated URL and the way the returned data is to 
** be buffered.
**
** postconditions: A new SockCntlRec will be returned, which will either
** represent a new one or a preopened connection.
***********************************************************************/
static SockCntlRec *SockRetrieve(URL * theURL, BufferStatus sockRetrType)
{
	unsigned long theIPnum;
	unsigned int theHKey;
	SockCntlRec *newSock;
	URL *newURL;

	/* If invalid data or not initialized, just return a NULL. */
	if (theURL == NULL)
		return (NULL);

	if (!initialized)
		SockInit(NULL);

	/*Get the IP of the host we're looking for. */
	if ((theIPnum = gethostinhex(theURL->host)) == -1)
		return (NULL);

	DoError(NO_ERROR, NULL);

	/* Find the hash bucket to look in. */
	theHKey = GetHKey(theIPnum);

	/* Try to find a connection in that bucket. */
	newSock = SearchList(SockHash[theHKey], theURL);

	if (newSock != (NULL)) {	/* Move to front of list and update time. */
		TouchDate(newSock);
		newSock->sockStateStor = sockRetrType;
		return (newSock);
	}
	/* Need to open a new conection. */
	/* If we are already at the maxConnections, try and disconnect one. */
	if (currConnections == maxConnections) {
		if (!DisconnectOne())
			return (NULL);
	}
	/* Get a new connnection. */
	newURL = dup_url(theURL);	/* save a copy */
	newSock = GetConnection(newURL, theIPnum);

	if (newSock == NULL)
		return (NULL);

	/* Set it up. */
	newSock->sockStateStor = sockRetrType;

	newSock->hashPrev = NULL;
	newSock->hashNext = SockHash[theHKey];

	if (SockHash[theHKey] != NULL)
		SockHash[theHKey]->hashPrev = newSock;

	SockHash[theHKey] = newSock;

	newSock->listPrev = NULL;
	newSock->listNext = TimeOutListNew;

	if (TimeOutListNew != NULL)
		TimeOutListNew->listPrev = newSock;
	else
		TimeOutListOld = newSock;

	TimeOutListNew = newSock;

	return (newSock);
}

/*************************************************************************
** function: SockGetData()
**
** parameters: Need a URL struct parsed by liburl and return string for
**	fetched filename  dcm 6/28/94
**
** postcondition:  Data is returned as requested if possible.
**************************************************************************/
DataReturn *SockGetData(URL * myURL, BufferStatus theStatus, char *filename)
{

	SockCntlRec *mySocket;
	DataReturn *myDataReturn;
	int errNum;
	char errorReply[SERV_REPLY_LENGTH];

	if (!initialized)
		SockInit(NULL);

	/* If not an FTP url, then we can't ftp anything. */
	if (myURL->type != URL_FTP) {
		DoError(BAD_URL_TYPE, NULL);
		return NULL;
	}
	/* Get a socket. */

	Freeze();
	mySocket = SockRetrieve(myURL, theStatus);
	Thaw();

	if (!mySocket)
		return NULL;

	if (GetError() != noErr) {
		mySocket->sockStateStor = INACTIVE;
		return NULL;
	}
	myDataReturn = (DataReturn *) malloc(sizeof(DataReturn));

	/* Set flags for FTP layer. */
	if (theStatus == MEMORY_ONLY) {
		myDataReturn->inMemory = TRUE;
		myDataReturn->useTempFile = FALSE;
	} else {
		myDataReturn->inMemory = FALSE;

		if (theStatus == TEMP)
			myDataReturn->useTempFile = TRUE;
		else if (filename == NULL) {
			myDataReturn->useTempFile = FALSE;
			StripFileName(myDataReturn->fileName,
			    myURL->pathname);
		} else {
			myDataReturn->useTempFile = FALSE;
			strcpy(myDataReturn->fileName, filename);
		}
	}

	/* Try to retrieve the requested file. */
	errNum = Retrieve(myURL->pathname + 1, mySocket->theSocket, NULL, FALSE,
	    TRUE, errorReply, myDataReturn);

	mySocket->sockStateStor = INACTIVE;

	if (errNum < 0) {
		if (errNum == SERV_REPLY_ERROR)
			DoError(errNum, errorReply);
		else
			DoError(errNum, NULL);

		DestroyDataReturn(myDataReturn);

		return (NULL);
	}
	return (myDataReturn);
}


/******************************************************************************
** function name: ExpireSocket()
**
** preconditions: A timer must have already been set for this socket.
**                The whichTimer should identify the socket to be expired.
**
** postconditions:  The socket is expired and the connection is closed.
**
******************************************************************************/
int ExpireSocket(int whichTimer)
{
	SockCntlRec *locSockRec;
	int theHKey;
	char errorReply[SERV_REPLY_LENGTH];

	if (!initialized)
		return 1;


	locSockRec = TimeOutListOld;

	/* Try to find the correct socket. */
	while ((locSockRec != NULL) && (locSockRec->timerid != whichTimer))
		locSockRec = locSockRec->listPrev;

	if (locSockRec == NULL)
		return 1;

	/* If it's still being used reset the timer. */
	if (locSockRec->sockStateStor != INACTIVE) {
		locSockRec->timerid = SetTimer(timeOut, ExpireSocket);
		return 1;
	}
	/* Strip it from the expiration list. */
	if (locSockRec->listNext == NULL)
		TimeOutListOld = locSockRec->listPrev;
	else
		locSockRec->listNext->listPrev = locSockRec->listPrev;

	if (locSockRec->listPrev == NULL)
		TimeOutListNew = locSockRec->listNext;
	else
		locSockRec->listPrev->listNext = locSockRec->listNext;

	/* Strip it from the active hash. */
	theHKey = GetHKey(locSockRec->theHostIP);

	if (locSockRec->hashPrev == NULL) {
		SockHash[theHKey] = locSockRec->hashNext;
		if (SockHash[theHKey] != NULL)
			SockHash[theHKey]->hashPrev = NULL;
	} else
		locSockRec->hashPrev->hashNext = locSockRec->hashNext;

	if (locSockRec->hashNext != NULL)
		locSockRec->hashNext->hashPrev = locSockRec->hashPrev;

	/* Disconnect the socket. */
	Disconnect(locSockRec->theSocket, errorReply);

	free_url(locSockRec->socketInfo);
	free((char *) locSockRec);
	return 1;
}


/******************************************************************************
** function name: ShutDownCache()
**
** preconditions: The cache must have previously been initialized.
**
** postconditions:  Timer's are all cleared, everything is disconnected.
**
******************************************************************************/
void ShutDownCache()
{

	if (!initialized)
		return;

	/* Freeze Timers and disconnect all sockets. */
	Freeze();
	while (DisconnectOne());

	initialized = FALSE;

	free((char *) SockHash);

	/* Reset signals. */
	signal(SIGALRM, NULL);
	signal(SIGUSR1, NULL);

	return;
}


/******************************************************************************
** function name: DestroyDataReturn()
**
** preconditions: Need an allocated DataReturn.
**
** postconditions:  Memory is freed.
**
******************************************************************************/

void DestroyDataReturn(theDataReturn)
     DataReturn *theDataReturn;
{
	free(theDataReturn->buffer);
	free(theDataReturn);
	theDataReturn = NULL;
}


/******************************************************************************
** function name: DumpHash()
**
** preconditions: SIGUSR1 must be set up to call this function.
**
** postconditions: On a SIGUSR the hash tables internals are printed.
**
******************************************************************************/

static void DumpHash()
{
	int i;
	SockCntlRec *theSocket;

	/*
	 * If we're not initialized for some reason, print out an 
	 * appropriate message.
	 */
	if (!initialized) {
		printf("Sorry, the hash table has not been initialized.\n");
		printf("No data to output.\n");
		return;
	}
	/* Go through the slots to find all connections. */
	for (i = 0; i < HASH_SLOTS; i++) {
		/* If we find one, print out its stuff. */
		if (SockHash[i] != NULL) {
			printf("Hash Slot %d:\n", i);
			theSocket = SockHash[i];
			/* Print out ALL nodes in the linked list. */
			while (theSocket != NULL) {
				printf("\t%s\n", theSocket->socketInfo->url);
				theSocket = theSocket->hashNext;
			}
		}
	}
}