Move variables out of the iterator class and into the generator function. This slight...

[z_archive/twitter.git] / twitter / stream.py

diff --git a/twitter/stream.py b/twitter/stream.py
index b11a89363ff508de6d65d2f3d8eb260af0efc6cb..fa332cf18148bba01af8d71970c0dccc6bd46afb 100644 (file)
--- a/twitter/stream.py
+++ b/twitter/stream.py
@@ -12,15 +12,37 @@ import sys, select, time
 
 from .api import TwitterCall, wrap_response, TwitterHTTPError
 
+def recv_chunk(sock):
+    buf = sock.recv(10)  # Scan for an up to a 4GiB chunk size (0xffffffff).
+    if buf:
+        crlf = buf.find(b'\r\n')  # Find the HTTP chunk size.
+        if crlf > 0:
+            remaining = int(buf[:crlf], 16)  # Decode the chunk size.
+            chunk = bytearray(remaining)  # Create the chunk buffer.
+
+            start = crlf + 2  # Add in the length of the header's CRLF pair.
+            end = len(buf) - start
+
+            chunk[:end] = buf[start:]
+            chunk[end:] = sock.recv(remaining - end)
+
+            sock.recv(2)  # Read the trailing CRLF pair. Throw it away.
+
+            return chunk
+    return b''
+
+##  recv_chunk()
+
+
 class TwitterJSONIter(object):
 
     def __init__(self, handle, uri, arg_data, block=True, timeout=None):
         self.decoder = json.JSONDecoder()
         self.handle = handle
-        self.buf = b""
+        self.uri = uri
+        self.arg_data = arg_data
         self.block = block
         self.timeout = timeout
-        self.timer = time.time()
 
 
     def __iter__(self):
@@ -31,39 +53,41 @@ class TwitterJSONIter(object):
         sock.setsockopt(socket.SOL_SOCKET, socket.SO_KEEPALIVE, 1)
         if not self.block or self.timeout:
             sock.setblocking(False)
+        buf = b''
+        timer = time.time()
         while True:
             try:
-                utf8_buf = self.buf.decode('utf8').lstrip()
+                utf8_buf = buf.decode('utf-8').lstrip()
                 res, ptr = self.decoder.raw_decode(utf8_buf)
-                self.buf = utf8_buf[ptr:].encode('utf8')
+                buf = utf8_buf[ptr:].encode('utf-8')
                 yield wrap_response(res, self.handle.headers)
-                self.timer = time.time()
+                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)
             try:
                 if self.timeout:
                     ready_to_read = select.select([sock], [], [], self.timeout)
                     if ready_to_read[0]:
-                        self.buf += sock.recv(1024)
-                        if time.time() - self.timer > self.timeout:
+                        buf += recv_chunk(sock)
+                        if time.time() - timer > self.timeout:
                             yield {"timeout":True}
                     else:
                         yield {"timeout":True}
                 else:
-                    self.buf += sock.recv(1024)
+                    buf += recv_chunk(sock)
             except SSLError as e:
                 if (not self.block or self.timeout) and (e.errno == 2):
                     # Apparently this means there was nothing in the socket buf
                     pass
                 else:
                     raise
+            except urllib_error.HTTPError as e:
+                raise TwitterHTTPError(e, self.uri, 'json', self.arg_data)
 
 def handle_stream_response(req, uri, arg_data, block, timeout=None):
     handle = urllib_request.urlopen(req,)