def main(file_path):
#
# Instantiation and start all the actors
# - No Args constructor to separate instantiation and initialization
#
storage_manager = DataStorageManager()
word_freq_manager = WordFrequencyManager()
word_indexer = WordsIndexer()
word_index_controller = WordIndexController()
#
# Initialization and wiring (ref of an actor passed to another one)
#
send_message(storage_manager, ['init', file_path])
send_message(word_freq_manager, ['init', FREQUENCY_THRESHOLD])
send_message(word_indexer, ['init', PAGE_SIZE])
# Passing references to other actors is not a violation. BUT DIRECTLY CALLING METHODS OF THE ACTORS IS!
send_message(word_index_controller,
['init', storage_manager, word_freq_manager, word_indexer])
# Start the execution
send_message(word_index_controller, ['run'])
# At this point we need to "block" main thread and wait for the all the actors to finish
[
t.join() for t in [
storage_manager, word_freq_manager, word_indexer,
word_index_controller
]
]
def _get_pages(self, words):
self.words = words
# self.count = len(words)
# Tell word_indexer to get you back the pages corresponding to each word
[
send_message(self.word_indexer, ['get_pages_for_word', self, word])
for word in words
]
# Force and "EOF" here? or die ?
send_message(self.word_indexer, ['eof', self])
def _get_words(self, recipient):
# Filter
_filtered_word_freqs = {}
for word in self._word_freqs.keys():
if self._word_freqs[word] <= self._frequency_threshold:
_filtered_word_freqs[word] = self._word_freqs[word]
self._word_freqs = _filtered_word_freqs
# Sort and send. TODO Really necessary?
# Assume the medium will deliver words in order
words = [
entry[0] for entry in sorted(self._word_freqs.items(),
key=operator.itemgetter(0),
reverse=False)
]
send_message(recipient, ['words', words])
def _process_lines(self, lines):
for line_number, line in enumerate(lines):
for w in line.split():
send_message(self.word_freq_manager, ['increment_count', w])
send_message(self.word_indexer, ['index_word', w, line_number])
# Question: At this point we assume that word_freq_manager processed already all the words
# but how one can be sure about that?
send_message(self.word_freq_manager, ['get_words', self])
def _get_lines(self, recipient):
""" Returns the list of lines in storage """
send_message(recipient, ['lines', self._lines])
def _run(self):
send_message(self.storage_manager, ['lines', self])
def _done(self):
for recipient in [
self.storage_manager, self.word_freq_manager,
self.word_indexer, self
]:
send_message(recipient, ['die'])
def _eof(self, recipient):
send_message(recipient, ['eof'])
def _get_pages_for_word(self, recipient, word):
if word in self._words_and_pages.keys():
send_message(recipient,
['pages_for_word', word, self._words_and_pages[word]])