def started(self):
"""
Return True if the server is started, False otherwise.
"""
client = None
try:
client = TimblClient(port=self.port)
except ConnectionError:
del client
return False
return True
def __init__(self, host, port, settings, **kwargs):
"""
Initializes a MBClassifier, sets up a L{TimblServer}
and connects to this server via an instance of L{TimblClient}.
Args:
- host (str): Host specifies the server address (localhost by default)
- port (int): Port specifies the server tcp communicating port.
- settings (dict): the settings used by Timbl (see :mod:`config`)
"""
self.server = TimblServer(
host=host,
port=port,
features=settings,
classifier=self.__class__.__name__,
)
self.server.run()
self.client = TimblClient(host=host, port=port)
# register the kill method to make sure the timblserver is killed
# at exiting the interactive shell
atexit.register(self.kill)
def __init__(self, host, port, settings, **kwargs):
"""
Initializes a MBClassifier, sets up a L{TimblServer}
and connects to this server via an instance of L{TimblClient}.
Args:
- host (str): Host specifies the server address (localhost by default)
- port (int): Port specifies the server tcp communicating port.
- settings (dict): the settings used by Timbl (see :mod:`config`)
"""
self.server = TimblServer(host=host, port=port, features=settings,
classifier=self.__class__.__name__,)
self.server.run()
self.client = TimblClient(host=host, port=port)
# register the kill method to make sure the timblserver is killed
# at exiting the interactive shell
atexit.register(self.kill)
class MBClassifier(object):
"""
Abstract Class representing a Memory-Based Classifier. The classifier
sets up a :class:`mbmp.server.TimblServer` instance and connects to this
server via a :class:`mbmp.client.TimblClient`.
"""
def __init__(self, host, port, settings, **kwargs):
"""
Initializes a MBClassifier, sets up a L{TimblServer}
and connects to this server via an instance of L{TimblClient}.
Args:
- host (str): Host specifies the server address (localhost by default)
- port (int): Port specifies the server tcp communicating port.
- settings (dict): the settings used by Timbl (see :mod:`config`)
"""
self.server = TimblServer(host=host, port=port, features=settings,
classifier=self.__class__.__name__,)
self.server.run()
self.client = TimblClient(host=host, port=port)
# register the kill method to make sure the timblserver is killed
# at exiting the interactive shell
atexit.register(self.kill)
def __repr__(self):
return '<%s connected to %s>' % (
self.__class__.__name__, self.client.name)
def __enter__(self):
return self
def __exit__(self, type, value, traceback):
"""
Within a 'with' statement, always kill the server and close the
connection to the server in case of an error.
"""
self.kill()
def kill(self):
"""
Close the connection between the client and the server and
kill the server if not in use by another process.
"""
self.client.close()
if server_in_use(self.server):
sys.stderr.write(
'Connection closed. Cannot kill timbl server; server in '
'use by other instance or by someone else.\n')
else:
if self.server._process is not None:
self.server.kill()
def _classify(self, iterable, size=5):
"""
Helper function to query the server.
Args:
- iterable (iterable): the item to be classified
- size (int): the window size used be the classifier
Returns:
List of predictions per element of iterable
"""
return [self.client.query(elt)
for elt in format_testitem(iterable, size)]
def classify(self, word):
"""
Return a list of outcome classes for each element of WORD.
Args:
- word (str): a string representing a word.
Returns:
List of predicted outcomes per letter of word.
"""
raise NotImplementedError('MBClassifier is an abstract interface')
def pprint_parse(self, parse):
"""
Return a pretty print of the classification.
Args:
- parse: the parse returned by :func:`mbmp.MBClassifier.classify`
Returns:
A string representation of parse
"""
raise NotImplementedError('MBClassifier is an abstract interface')
def trees(self, parse):
"""
Return a tree-structure of the classification using the NLTK-Tree
Class.
Args:
- parse: the parse returned by :func:`mbmp.MBClassifier.classify`
Yields:
a generator with all tree structures obtained.
"""
raise NotImplementedError('MBClassifier is an abstract interface')
class MBClassifier(object):
"""
Abstract Class representing a Memory-Based Classifier. The classifier
sets up a :class:`mbmp.server.TimblServer` instance and connects to this
server via a :class:`mbmp.client.TimblClient`.
"""
def __init__(self, host, port, settings, **kwargs):
"""
Initializes a MBClassifier, sets up a L{TimblServer}
and connects to this server via an instance of L{TimblClient}.
Args:
- host (str): Host specifies the server address (localhost by default)
- port (int): Port specifies the server tcp communicating port.
- settings (dict): the settings used by Timbl (see :mod:`config`)
"""
self.server = TimblServer(
host=host,
port=port,
features=settings,
classifier=self.__class__.__name__,
)
self.server.run()
self.client = TimblClient(host=host, port=port)
# register the kill method to make sure the timblserver is killed
# at exiting the interactive shell
atexit.register(self.kill)
def __repr__(self):
return '<%s connected to %s>' % (self.__class__.__name__,
self.client.name)
def __enter__(self):
return self
def __exit__(self, type, value, traceback):
"""
Within a 'with' statement, always kill the server and close the
connection to the server in case of an error.
"""
self.kill()
def kill(self):
"""
Close the connection between the client and the server and
kill the server if not in use by another process.
"""
self.client.close()
if server_in_use(self.server):
sys.stderr.write(
'Connection closed. Cannot kill timbl server; server in '
'use by other instance or by someone else.\n')
else:
if self.server._process is not None:
self.server.kill()
def _classify(self, iterable, size=5):
"""
Helper function to query the server.
Args:
- iterable (iterable): the item to be classified
- size (int): the window size used be the classifier
Returns:
List of predictions per element of iterable
"""
return [
self.client.query(elt) for elt in format_testitem(iterable, size)
]
def classify(self, word):
"""
Return a list of outcome classes for each element of WORD.
Args:
- word (str): a string representing a word.
Returns:
List of predicted outcomes per letter of word.
"""
raise NotImplementedError('MBClassifier is an abstract interface')
def pprint_parse(self, parse):
"""
Return a pretty print of the classification.
Args:
- parse: the parse returned by :func:`mbmp.MBClassifier.classify`
Returns:
A string representation of parse
"""
raise NotImplementedError('MBClassifier is an abstract interface')
def trees(self, parse):
"""
Return a tree-structure of the classification using the NLTK-Tree
Class.
Args:
- parse: the parse returned by :func:`mbmp.MBClassifier.classify`
Yields:
a generator with all tree structures obtained.
"""
raise NotImplementedError('MBClassifier is an abstract interface')
