Merge pull request #203 from RouxRC/pr-fix-timeout

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

diff --git a/twitter/stream.py b/twitter/stream.py
index 96e92d739d229580c8876036b8f3768f59a91957..6ecc49ed760a54591477274c0c1517556900d04e 100644 (file)
--- a/twitter/stream.py
+++ b/twitter/stream.py
@@ -12,33 +12,38 @@ import sys, select, time
 
 from .api import TwitterCall, wrap_response, TwitterHTTPError
 
-def recv_chunk(sock):  # -> bytearray:
+python27_3 = sys.version_info >= (2, 7)
+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.
+    header = sock.recv(8)  # Scan for an up to 16MiB chunk size (0xffffff).
+    crlf = header.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.
-
+        size = int(header[:crlf], 16)  # Decode the chunk size. Rarely exceeds 8KiB.
+        chunk = bytearray(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)
+        if size <= 3:  # E.g. an HTTP chunk with just a keep-alive delimiter or end of stream (0).
+            chunk[:size] = header[start:start + size]
+        # There are several edge cases (size == [4-6]) as the chunk size exceeds the length
+        # of the initial read of 8 bytes. With Twitter, these do not, in practice, occur. The
+        # shortest JSON message starts with '{"limit":{'. Hence, it exceeds in size the edge cases
+        # and eliminates the need to address them.
+        else:  # There is more to read in the chunk.
+            end = len(header) - start
+            chunk[:end] = header[start:]
+            if python27_3:  # When possible, use less memory by reading directly into the buffer.
+                buffer = memoryview(chunk)[end:]  # Create a view into the bytearray to hold the rest of the chunk.
+                sock.recv_into(buffer)
+            else:  # less efficient for python2.6 compatibility
+                chunk[end:] = sock.recv(max(0, size - 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()
-
 
 class TwitterJSONIter(object):
 
@@ -54,7 +59,7 @@ class TwitterJSONIter(object):
         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)
         sock.setblocking(self.block and not self.timeout)
-        buf = u''
+        buf = ''
         json_decoder = json.JSONDecoder()
         timer = time.time()
         while True:
@@ -63,25 +68,22 @@ class TwitterJSONIter(object):
                 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
             try:
                 buf = buf.lstrip()  # Remove any keep-alive delimiters to detect hangups.
-                if self.timeout:
+                if self.timeout and not buf:  # This is a non-blocking read.
                     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:
+                    if not ready_to_read[0] and time.time() - timer > self.timeout:
                         yield {'timeout': True}
-                else:
-                    buf += recv_chunk(sock).decode('utf-8')
-                if not buf and self.block:
+                        continue
+                timer = time.time()
+                buf += recv_chunk(sock).decode('utf-8')
+                if not buf:
                     yield {'hangup': True}
+                    break
             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