class HaikuFindingThread(threading.Thread):
def __init__(self, results_queue):
self.results_queue = results_queue
self.terminate_request_flag = False
self.haiku_finder = HaikuFinder()
self.untex_thread_class = UntexWrapper()
self.last_parsed_id = get_last_parsed_id()
threading.Thread.__init__(self)
def terminate_request(self):
self.terminate_request_flag = True
def terminate(self):
if not no_dictionary_update:
logging.info("Saving dictionary")
self.haiku_finder.save_dict()
logging.info("Saving last_parsed_id")
set_last_parsed_id(self.last_parsed_id)
def run(self):
for (article_id, raw_tex) in rssparse(arxiv_feed_url, self.last_parsed_id):
if self.terminate_request_flag:
critical("Worker thread caught terminate_request_flag, terminating")
self.terminate()
return
self.last_parsed_id = article_id
logging.info("Attempting raw tex from article_id: "+ str(article_id))
if not raw_tex:
error("Got empty raw_tex from article_id: " + str(article_id))
continue
haiku_list = []
try:
raw_text = self.untex_thread_class.run_untex(raw_tex)
haiku_list = self.haiku_finder.find_haiku_in_text(raw_text)
except RuntimeError as e:
print_with_warning(str(e))
if len(haiku_list)==0:
logging.info("Found no Haiku in article_id: " + str(article_id))
else:
for haiku in haiku_list:
self.results_queue.put(haiku + " (" + str(article_id) + ") #arXivHaiku")
logging.info("Found haiku in article_id :" + str(article_id) + " : " + haiku)
print_with_ok("Got to the end of the list :(")
logging.info("Worker thread finished")
self.terminate()
def __init__(self, results_queue): self.results_queue = results_queue self.terminate_request_flag = False self.haiku_finder = HaikuFinder() self.untex_thread_class = UntexWrapper() self.last_parsed_id = get_last_parsed_id() threading.Thread.__init__(self)
