The parsing and namespace cleaning isn't terribly expensive, but it does
add up. It was adding an extra 5sec when processing 100,000 basic
message stanzas.
Based on profiling, using around 35 stream handlers quarters the number
of basic message stanzas that can be processed in a second, in
comparison to only using the bare minimum of four handlers.
To help, we can drop handlers for stream features once the session
has started. So that we can re-enable these handlers when a stream
must restart, the 'stream_start' event has been added which fires
whenever a stream header is received.
The 'stream_start' event is a more generic replacement for the
existing start_stream_handler() method.
This allows applications to filter out sensitive information, such
as passwords, so that it won't appear in the logs.
It does mean that the debug logs will not show the actual received
data, and there will be no indication of tampering, unless the
filter author explicitly logs and notes that a change was made.
A filter accepts and returns a stanza, but potentially modified.
To prevent sending/receiving a stanza, a filter may return None.
Incoming:
self.add_filter('in', in_filter)
Outgoing:
self.add_filter('out', out_filter)
Filters are applied in the order thay are added. However, you may
add an order parameter, which is the place in the list to insert the
filter:
self.add_filter('in', in_filter, order=0)
May set self.disconnect_wait=True so that all disconnect
calls wait for the send queue to empty, unless explicitly
overridden with wait=False.
The session_end now fires before closing the socket so
that final stanzas may be sent, such as unavailable presences
for components.
Calling reconnect() simultaneously from multiple threads (like when
using XEP-0199 keepalive) could break because the connection state
can transition and break the state expectations in one of the
reconnect() calls.
Note that using % in a string will _always_ perform the sting substitutions, because the strings are constructed before the function is called. So log.debug('%s' % expensiveoperation()) will take about the same CPU time whether or not the logging level is DEBUG or INFO. if you use , no substitutions are performed unless the string is actually logged
Note that using % in a string will _always_ perform the sting substitutions, because the strings are constructed before the function is called. So log.debug('%s' % expensiveoperation()) will take about the same CPU time whether or not the logging level is DEBUG or INFO. if you use , no substitutions are performed unless the string is actually logged
The processing loop was continuing to call __read_xml after </stream>
was received, which caused SyntaxErrors (can't find starting element).
This should fix issue #102
