Snippets

Filipp Ozinov Async HTTP Fetcher

Created by Filipp Ozinov last modified
import time
import thread
import httplib
import socket
import select
import os
import Queue
from StringIO import StringIO
import random


# ------------------------------------------------------------------------------------
class ERROR:
	CONNECTION_FAILED = -100
	CONNECTION_DROPED = -101
	MAX_QUEUE_SIZE = -110

# ------------------------------------------------------------------------------------
class SETTINGS:
	# After INACTIVE_TIMEOUT after last received data
	# connection will be removed from pool.
	INACTIVE_TIMEOUT = 3 * 60

	# Timeout for connecting to host. Host connections are blocking - do not use too big value.
	CONNECTION_TIMEOUT = 5

	# Number of retries if failure.
	DEFAULT_ATTEMPTS = 3

	# Send error response if too many requests incoming.
	MAX_QUEUE_SIZE = 500

	
#   Implementation
# ====================================================================================

# ------------------------------------------------------------------------------------
class _StrSocket():
	def __init__(self, response_str):
		self._file = StringIO(response_str)
	def makefile(self, *args, **kwargs):
		return self._file

# ------------------------------------------------------------------------------------
class _Connection:

	# ------------------------------------------------------------------------------------
	def __init__(self, host):
		self.host = host
		self.conn = httplib.HTTPConnection(host, timeout=SETTINGS.CONNECTION_TIMEOUT)
		self.conn.connect()
		self.conn.sock.setblocking(0)
		self.clear()
		self.lastResponse = time.time()

	# ------------------------------------------------------------------------------------
	def clear(self):
		self.__buffer = ''
		self.request = None

	# ------------------------------------------------------------------------------------
	def fileno(self):
		return self.conn.sock.fileno()

	# ------------------------------------------------------------------------------------
	def sendRequest(self, httpRequest):
		self.request = httpRequest
		self.conn._HTTPConnection__state = httplib._CS_IDLE
		self.conn.request(httpRequest.method, httpRequest.query, httpRequest.data)

	# ------------------------------------------------------------------------------------
	def processResponse(self):
		resp = self.conn.sock.recv(1024)
		if len(resp) > 0:
			self.lastResponse = time.time()
		self.__buffer += resp
		source = _StrSocket(self.__buffer)
		response = httplib.HTTPResponse(source)
		response.begin()
		status = response.status
		data = response.read()
		return (status, data, self.request)

	# ------------------------------------------------------------------------------------
	def isActive(self):
		return time.time() < self.lastResponse + SETTINGS.INACTIVE_TIMEOUT

# ------------------------------------------------------------------------------------
class _ConnectionManager:
	def __init__(self):
		self.__connections = {} # host -> list of connections

	# ------------------------------------------------------------------------------------
	def getConnection(self, host):
		hostConnections = self.__connections.get(host, None)
		if hostConnections is None:
			return _Connection(host)
		n = random.randint(0, len(hostConnections) - 1)
		conn = hostConnections[n]
		del hostConnections[n]
		if len(hostConnections) == 0:
			del self.__connections[host]
		return conn

	# ------------------------------------------------------------------------------------
	def returnConnection(self, connection):
		connection.clear()
		self.__connections.setdefault(connection.host, []).append(connection)

	# ------------------------------------------------------------------------------------
	def removeOldConnections(self):
		newConnections = {}
		for host, connections in self.__connections.iteritems():
			goodConnections = []
			for conn in connections:
				if conn.isActive():
					goodConnections.append(conn)
			if len(goodConnections) > 0:
				newConnections[host] = goodConnections
		self.__connections = newConnections


# ------------------------------------------------------------------------------------
class _AsyncFetcher:

	# ------------------------------------------------------------------------------------
	def __init__(self):
		self.__epoll = select.epoll()
		self.__requestQueue = Queue.Queue()
		self.__responseQueue = Queue.Queue()
		self.__connections = {} # fileno => http connection
		self.__connectionManager = _ConnectionManager()
		self.pipeToThread = os.pipe()
		self.pipeToMain = os.pipe()
		thread.start_new_thread(self.__workerThread, ())
		thread.start_new_thread(self.__mainThread, ())

	# ------------------------------------------------------------------------------------
	def fetch(self, request):
		if self.__responseQueue.qsize() > SETTINGS.MAX_QUEUE_SIZE or\
				self.__responseQueue.qsize() > SETTINGS.MAX_QUEUE_SIZE:
			request.callback(ERROR.MAX_QUEUE_SIZE, '', request)
			return
		self.__requestQueue.put(request)
		os.write(self.pipeToThread[1], '\n')

	# ------------------------------------------------------------------------------------
	def __mainThread(self):
		pipeIn = self.pipeToMain[0]
		while True:
			os.read(pipeIn, 1)
			while not self.__responseQueue.empty():
				status, data, request = self.__responseQueue.get()
				request.callback(status, data, request)

	# ------------------------------------------------------------------------------------
	def __workerThread(self):
		pipeIn = self.pipeToThread[0]
		self.__epoll.register(pipeIn, select.EPOLLIN)
		last10SecondsTime = time.time()
		while True:
			try:
				events = self.__epoll.poll(1)
				for fileno, event in events:
					if fileno == pipeIn:
						os.read(pipeIn, 1)
						while not self.__requestQueue.empty():
							request = self.__requestQueue.get()
							self.__processRequest(request)
					if fileno in self.__connections:
						self.__processIncoming(fileno)
				if time.time() > last10SecondsTime + 10.0:
					last10SecondsTime = time.time()
					self.__connectionManager.removeOldConnections()
			except Exception as e:
				print e


	# ------------------------------------------------------------------------------------
	def __publishResponse(self, status, data, request):
		self.__responseQueue.put((status, data, request))
		os.write(self.pipeToMain[1], '\n')

	# ------------------------------------------------------------------------------------
	def __processRequest(self, request):
		try:
			conn = self.__connectionManager.getConnection(request.host)
		except (socket.gaierror, socket.timeout, socket.error):
			if request.attempts <= 0:
				self.__publishResponse(ERROR.CONNECTION_FAILED, '', request)
			else:
				request.attempts -= 1
				self.__processRequest(request)
			return
		fno = conn.fileno()
		self.__connections[fno] = conn
		self.__epoll.register(fno, select.EPOLLIN)
		conn.sendRequest(request)

	# ------------------------------------------------------------------------------------
	def __processIncoming(self, fileno):
		conn = self.__connections[fileno]
		try:
			res = conn.processResponse()
		except (httplib.IncompleteRead, socket.error):
			return
		except httplib.BadStatusLine:
			self.__epoll.unregister(fileno)
			del self.__connections[fileno]
			if conn.request.attempts <= 0:
				self.__publishResponse(ERROR.CONNECTION_DROPED, '', conn.request)
			else:
				conn.request.attempts -= 1
				self.__processRequest(conn.request)
			return

		self.__epoll.unregister(fileno)
		status, data, callback = res
		self.__publishResponse(status, data, callback)
		self.__connectionManager.returnConnection(conn)
		del self.__connections[fileno]

# Interface
# ====================================================================================

# ------------------------------------------------------------------------------------
class HttpRequest:
	def __init__(self, host, method, query, data, callback):
		self.host = host
		self.method = method
		self.query = query
		self.data = data
		self.callback = callback
		self.attempts = SETTINGS.DEFAULT_ATTEMPTS

# ------------------------------------------------------------------------------------
_g_asyncFetcher = _AsyncFetcher()

# ------------------------------------------------------------------------------------
def fetch(request):
	_g_asyncFetcher.fetch(request)



#   Usage sample
# ====================================================================================

# ------------------------------------------------------------------------------------
def sampleCallback(status, data, request):
	print 'fetched:', status, len(data)
	print data

# ------------------------------------------------------------------------------------
if __name__ == '__main__':
	while True:
		fetch(HttpRequest('google.com:80', 'GET', '/', None, sampleCallback))
		time.sleep(4.0)

Comments (0)

HTTPS SSH

You can clone a snippet to your computer for local editing. Learn more.