|
|
|
|
from cppy.cp_util import *
|
|
|
|
|
'''
|
|
|
|
|
注册回调的一个变体
|
|
|
|
|
提供一个中心消息管理器,统一管理消息的订阅和通知
|
|
|
|
|
这是一个示例性质的原型,具体分布式环境下需要调整
|
|
|
|
|
'''
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
#################################################
|
|
|
|
|
# The event management
|
|
|
|
|
#################################################
|
|
|
|
|
class EventManager:
|
|
|
|
|
def __init__(self):
|
|
|
|
|
self._subscriptions = {}
|
|
|
|
|
|
|
|
|
|
def subscribe(self, event_type, handler):
|
|
|
|
|
self._subscriptions.setdefault(event_type, []).append(handler)
|
|
|
|
|
|
|
|
|
|
def publish(self, event):
|
|
|
|
|
event_type = event[0]
|
|
|
|
|
for h in self._subscriptions.get(event_type, []):
|
|
|
|
|
h(event)
|
|
|
|
|
|
|
|
|
|
#################################################
|
|
|
|
|
# The application entities
|
|
|
|
|
#################################################
|
|
|
|
|
class DataStorage:
|
|
|
|
|
""" Models the contents of the file """
|
|
|
|
|
def __init__(self, event_manager):
|
|
|
|
|
self._event_manager = event_manager
|
|
|
|
|
self._event_manager.subscribe('load', self.load)
|
|
|
|
|
self._event_manager.subscribe('start', self.produce_words)
|
|
|
|
|
|
|
|
|
|
def load(self, event):
|
|
|
|
|
self._data = extract_file_words( event[1] )
|
|
|
|
|
|
|
|
|
|
def produce_words(self, event):
|
|
|
|
|
for w in self._data:
|
|
|
|
|
self._event_manager.publish(('word', w))
|
|
|
|
|
self._event_manager.publish(('eof', None))
|
|
|
|
|
|
|
|
|
|
class StopWordFilter:
|
|
|
|
|
""" Models the stop word filter """
|
|
|
|
|
def __init__(self, event_manager):
|
|
|
|
|
self._stop_words = []
|
|
|
|
|
self._event_manager = event_manager
|
|
|
|
|
self._event_manager.subscribe('load', self.load)
|
|
|
|
|
self._event_manager.subscribe('word', self.is_stop_word)
|
|
|
|
|
|
|
|
|
|
def load(self, event):
|
|
|
|
|
self._stop_words = get_stopwords()
|
|
|
|
|
|
|
|
|
|
def is_stop_word(self, event):
|
|
|
|
|
word = event[1]
|
|
|
|
|
if word not in self._stop_words:
|
|
|
|
|
self._event_manager.publish(('valid_word', word))
|
|
|
|
|
|
|
|
|
|
class WordFrequencyCounter:
|
|
|
|
|
""" Keeps the word frequency data """
|
|
|
|
|
def __init__(self, event_manager):
|
|
|
|
|
self._word_freqs = {}
|
|
|
|
|
self._event_manager = event_manager
|
|
|
|
|
self._event_manager.subscribe('valid_word', self.increment_count)
|
|
|
|
|
self._event_manager.subscribe('print', self.print_freqs)
|
|
|
|
|
|
|
|
|
|
def increment_count(self, event):
|
|
|
|
|
word = event[1]
|
|
|
|
|
self._word_freqs[word] = self._word_freqs.get(word, 0) + 1
|
|
|
|
|
|
|
|
|
|
def print_freqs(self, event):
|
|
|
|
|
print_word_freqs ( sort_dict (self._word_freqs) )
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
class WordFrequencyApplication:
|
|
|
|
|
def __init__(self, event_manager):
|
|
|
|
|
self._event_manager = event_manager
|
|
|
|
|
self._event_manager.subscribe('run', self.run)
|
|
|
|
|
self._event_manager.subscribe('eof', self.stop)
|
|
|
|
|
|
|
|
|
|
def run(self, event):
|
|
|
|
|
path_to_file = event[1]
|
|
|
|
|
self._event_manager.publish(('load', path_to_file))
|
|
|
|
|
self._event_manager.publish(('start', None))
|
|
|
|
|
|
|
|
|
|
def stop(self, event):
|
|
|
|
|
self._event_manager.publish(('print', None))
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
if __name__ == "__main__":
|
|
|
|
|
em = EventManager()
|
|
|
|
|
DataStorage(em), StopWordFilter(em), WordFrequencyCounter(em)
|
|
|
|
|
WordFrequencyApplication(em)
|
|
|
|
|
em.publish(('run', testfilepath ))
|