def __init__(self, settings):
self.settings = settings
self.is_stopped = Event()
self.pid = os.getpid()
# channels
self.url_queue = Queue()
self.content_queue = Queue()
self.orch_queue = Queue()
thread_count = self.settings.thread_count
# create sub processes
self.content_processor = ContentProcessor(
"content_processor", self.content_queue, self.orch_queue, 1,
self.settings)
self.content_process = Process(target=self.content_processor.run)
self.url_processor = UrlProcessor(
"url_processor", self.url_queue, self.orch_queue, thread_count,
self.settings)
self.url_process = Process(target=self.url_processor.run)
class Orchestrator(object):
"""Orchestrator
Orchestrator creates and starts the control flow. It builds a multiprocess
framework, in which there are three object, the orchestrator, the url
processor, and the content processor. The url processor and the content
processor run in a subprocess, receive jobs from the orchestrator, execute
the jobs, send back the results to the orchestrator. They communicate with
queues.
Attributes:
settings: a Settings instance for the orchestrator
url_queue: queue, the channel for the orchestrator sending jobs
(materials) to the url processor
content_queue: queue, the channel for the orchestrator sending jobs
(materials) to the content processor
orch_queue: queue, the channel for the orchestrator receiving the
executing result from the url/content processor
"""
def __init__(self, settings):
self.settings = settings
self.is_stopped = Event()
self.pid = os.getpid()
# channels
self.url_queue = Queue()
self.content_queue = Queue()
self.orch_queue = Queue()
thread_count = self.settings.thread_count
# create sub processes
self.content_processor = ContentProcessor(
"content_processor", self.content_queue, self.orch_queue, 1,
self.settings)
self.content_process = Process(target=self.content_processor.run)
self.url_processor = UrlProcessor(
"url_processor", self.url_queue, self.orch_queue, thread_count,
self.settings)
self.url_process = Process(target=self.url_processor.run)
def _feed_seed_urls(self):
"""Read urls from url_list_file and put them into url_queue
"""
# read seed urls from file
url_list_file = self.settings.url_list_file
with file(url_list_file) as urls_file:
urls = [line.strip() for line in urls_file.readlines()]
# put the urls to url_queue
for url in urls:
material = UrlMaterial(url=url, depth=0)
self.url_queue.put(material)
logging.debug("feed seed url %s", material.url)
def _loop(self):
"""Main loop of this instance
The loop read products from `orch_queue`. For every product, if it is
1) a UrlProduct instance, then build a content matertial and send
it to content_queue
2) a ContentProduct instance, then (a) check depth, (b) build url
materials if depth < max_depth, (c) send each of the materials
to url_queue
The loop would also break if could not get a product with 10 seconds
"""
while not self.is_stopped.is_set():
try:
product = self.orch_queue.get(timeout=5)
except Empty:
logging.info("orchestrator loop broken for timeout")
self.stop()
break
except:
raise
# dispatching
if isinstance(product, UrlProduct):
material = ContentMaterial(url=product.url,
depth=product.depth,
filepath=product.filepath,
encoding=product.encoding)
self.content_queue.put(material)
elif isinstance(product, ContentProduct):
for url in product.urls:
material = UrlMaterial(url=url, depth=product.depth + 1)
self.url_queue.put(material)
def start(self):
"""Start the orchestrator
"""
# feed seed urls
self._feed_seed_urls()
# start processors and loopping
self.content_process.start()
self.url_process.start()
logging.info("orchestrator loop")
self._loop()
self.stop()
def stop(self):
"""Stop the orchestrator
"""
# close processors and wait
if not self.is_stopped.is_set() and self.pid == os.getpid():
self.is_stopped.set()
# logging.info("@@@@ %d %d", self.pid, os.getpid())
# stop processors
self.url_processor.stop()
self.content_processor.stop()
# wait processes exiting
self.content_process.join(timeout=3)
self.url_process.join(timeout=3)
logging.info("orchestrator exit")
