X-Git-Url: https://jfr.im/git/z_archive/twitter.git/blobdiff_plain/67ddbde4bccbb5ec89c6fc8529bf8e3215b8c389..443e409dca0f1808eb10692e14d1a666d0a03f66:/twitter/stream.py diff --git a/twitter/stream.py b/twitter/stream.py index 3848a2c..96e92d7 100644 --- a/twitter/stream.py +++ b/twitter/stream.py @@ -8,56 +8,102 @@ except ImportError: import json from ssl import SSLError import socket +import sys, select, time + +from .api import TwitterCall, wrap_response, TwitterHTTPError + +def recv_chunk(sock): # -> bytearray: + + buf = sock.recv(8) # Scan for an up to 16MiB chunk size (0xffffff). + crlf = buf.find(b'\r\n') # Find the HTTP chunk size. + + if crlf > 0: # If there is a length, then process it + + remaining = int(buf[:crlf], 16) # Decode the chunk size. + + start = crlf + 2 # Add in the length of the header's CRLF pair. + end = len(buf) - start + + chunk = bytearray(remaining) + + if end < remaining: + chunk[:end] = buf[start:] + chunk[end:] = sock.recv(remaining - end) + sock.recv(2) # Read the trailing CRLF pair. Throw it away. + else: # E.g. an HTTP chunk with just a keep-alive delimiter. + chunk[:remaining] = buf[start:start + remaining] + + return chunk + + return bytearray() + +## recv_chunk() -from .api import TwitterCall, wrap_response class TwitterJSONIter(object): - def __init__(self, handle, uri, arg_data, block=True): - self.decoder = json.JSONDecoder() + def __init__(self, handle, uri, arg_data, block=True, timeout=None): self.handle = handle - self.buf = b"" + self.uri = uri + self.arg_data = arg_data self.block = block + self.timeout = timeout + def __iter__(self): - sock = self.handle.fp._sock.fp._sock + sock = self.handle.fp.raw._sock if sys.version_info >= (3, 0) else self.handle.fp._sock.fp._sock sock.setsockopt(socket.SOL_SOCKET, socket.SO_KEEPALIVE, 1) - if not self.block: - sock.setblocking(False) + sock.setblocking(self.block and not self.timeout) + buf = u'' + json_decoder = json.JSONDecoder() + timer = time.time() while True: try: - utf8_buf = self.buf.decode('utf8').lstrip() - res, ptr = self.decoder.raw_decode(utf8_buf) - self.buf = utf8_buf[ptr:].encode('utf8') + buf = buf.lstrip() + res, ptr = json_decoder.raw_decode(buf) + buf = buf[ptr:] yield wrap_response(res, self.handle.headers) + timer = time.time() continue except ValueError as e: - if self.block: - pass - else: - yield None - except urllib_error.HTTPError as e: - raise TwitterHTTPError(e, uri, self.format, arg_data) - # this is a non-blocking read (ie, it will return if any data is available) + if self.block: pass + else: yield None try: - self.buf += sock.recv(1024) - except SSLError as e: - if (not self.block) and (e.errno == 2): - # Apparently this means there was nothing in the socket buf - pass + buf = buf.lstrip() # Remove any keep-alive delimiters to detect hangups. + if self.timeout: + ready_to_read = select.select([sock], [], [], self.timeout) + if ready_to_read[0]: + buf += recv_chunk(sock).decode('utf-8') # This is a non-blocking read. + if time.time() - timer > self.timeout: + yield {'timeout': True} + else: + yield {'timeout': True} else: - raise + buf += recv_chunk(sock).decode('utf-8') + if not buf and self.block: + yield {'hangup': True} + except SSLError as e: + # Error from a non-blocking read of an empty buffer. + if (not self.block or self.timeout) and (e.errno == 2): pass + else: raise + +def handle_stream_response(req, uri, arg_data, block, timeout=None): + try: + handle = urllib_request.urlopen(req,) + except urllib_error.HTTPError as e: + raise TwitterHTTPError(e, uri, 'json', arg_data) + return iter(TwitterJSONIter(handle, uri, arg_data, block, timeout=timeout)) -def handle_stream_response(req, uri, arg_data, block): - handle = urllib_request.urlopen(req,) - return iter(TwitterJSONIter(handle, uri, arg_data, block)) +class TwitterStreamCallWithTimeout(TwitterCall): + def _handle_response(self, req, uri, arg_data, _timeout=None): + return handle_stream_response(req, uri, arg_data, block=True, timeout=self.timeout) class TwitterStreamCall(TwitterCall): - def _handle_response(self, req, uri, arg_data): + def _handle_response(self, req, uri, arg_data, _timeout=None): return handle_stream_response(req, uri, arg_data, block=True) class TwitterStreamCallNonBlocking(TwitterCall): - def _handle_response(self, req, uri, arg_data): + def _handle_response(self, req, uri, arg_data, _timeout=None): return handle_stream_response(req, uri, arg_data, block=False) class TwitterStream(TwitterStreamCall): @@ -68,7 +114,7 @@ class TwitterStream(TwitterStreamCall): iterator that yields objects decoded from the stream. For example:: - twitter_stream = TwitterStream(auth=UserPassAuth('joe', 'joespassword')) + twitter_stream = TwitterStream(auth=OAuth(...)) iterator = twitter_stream.statuses.sample() for tweet in iterator: @@ -83,16 +129,19 @@ class TwitterStream(TwitterStreamCall): """ def __init__( self, domain="stream.twitter.com", secure=True, auth=None, - api_version='1', block=True): + api_version='1.1', block=True, timeout=None): uriparts = () uriparts += (str(api_version),) if block: - call_cls = TwitterStreamCall + if timeout: + call_cls = TwitterStreamCallWithTimeout + else: + call_cls = TwitterStreamCall else: call_cls = TwitterStreamCallNonBlocking TwitterStreamCall.__init__( self, auth=auth, format="json", domain=domain, callable_cls=call_cls, - secure=secure, uriparts=uriparts) + secure=secure, uriparts=uriparts, timeout=timeout, gzip=False)