def get_image(image_url, actually_store=True):
"""
This is the job that gets queued when a tweet needs to be analyzed
"""
redis_queue = utils.get_rq_redis_conn()
start = time.time() #Lets store some timing info
image = retrieve_image(image_url) #Go get that image
if image is not None: #As long as we have a vaid image and its not None (aka Null)
image = process_image(image) #Do our image processings
if actually_store: #If our configu file says to store the image for reals
key = store_to_vipr(image) #store to vipr
store_to_redis(key) #and keep track of it in redis
end = time.time() # and record how long it took
if random.randint(1, 10) < 5: #about 50% of the time we should
redis_queue.lpush("stats:execution-times", end - start) #send in an update on execution time
redis_queue.incr("stats:tweets-processed") #and also record that we processed another tweet.
def watch_stream():
twitter_creds = utils.twitter_creds()
redis_queue = utils.get_rq_redis_conn()
hashtag = redis_queue.get("hashtag")
q = utils.get_rq()
twitter_api = TwitterAPI(
consumer_key=twitter_creds['consumer_key'].encode('ascii','ignore'),
consumer_secret=twitter_creds['consumer_secret'].encode('ascii','ignore'),
access_token_key=twitter_creds['access_token'].encode('ascii','ignore'),
access_token_secret=twitter_creds['token_secret'].encode('ascii','ignore')
) #setup the twitter streaming connectors.
watcher_logger.info("Waiting for tweets...")
while True:
try:
for tweet in twitter_api.request('statuses/filter', {'track': hashtag}).get_iterator(): #for each one of thise
if hashtag != redis_queue.get("hashtag"):
watcher_logger.info("Hashtag changed from {}, breaking loop to restart with new hashtag".format(hashtag))
hashtag = redis_queue.get("hashtag")
break
#watcher_logger.debug("Tweet Received: {}".format(hashtag)) #Log it
redis_queue.incr("stats:tweets") #Let Redis know we got another one.
watcher_logger.debug("received tweet with tag {}".format(hashtag))
try:
if tweet['entities']['media'][0]['type'] == 'photo': #Look for the photo. If its not there, will throw a KeyError, caught below
if 'retweeted' not in tweet:
watcher_logger.info("Tweet was a RT - ignoring")
continue
watcher_logger.info("Dispatching tweet ({}) with URL {}".format(hashtag,tweet['entities']['media'][0]['media_url'])) # log it
q.enqueue(
get_image,
tweet['entities']['media'][0]['media_url'],
ttl=60,
result_ttl=60,
timeout=60
) #add a job to the queue, calling get_image() with the image URL and a timeout of 60s
except KeyError as e:
watcher_logger.debug("Caught a key error for tweet, expected behavior, so ignoring: {}".format(e.message))
except Exception as e:
watcher_logger.critical("UNEXPECTED EXCEPTION: {}".format(e))
except httplib.IncompleteRead as e:
watcher_logger.warn("HTTP Exception {}".format(e))
except ProtocolError as e:
watcher_logger.warn("Protocol Exception {}".format(e))
from rq.decorators import job #funtion decoration
import dweepy #see dweet.io
import logging, logging.config
import utils
import datetime
logging.config.dictConfig(utils.get_log_dict())
logger = logging.getLogger('vascodagama.dashboard')
redis_images = utils.get_images_redis_conn()
r = utils.get_rq_redis_conn()
q = utils.get_rq()
#Setup our redis and RQ connections. see twitter_watch for more details.
configstuff = utils.configstuff()
@job("dashboard", connection=r, timeout=10, result_ttl=10)
def send_update(metric, value): #method for sending updates about metrics as needed.
logger.debug("Sending update for {}: {}".format(metric, value))
dweepy.dweet_for(configstuff['dweet_thing'], {metric: value})
def update_dashboard(): # the primary function.
logger.info("updating")
import json #json functions
logging.config.dictConfig(utils.get_log_dict())
worker_logger = logging.getLogger("vascodagama.worker")
watcher_logger = logging.getLogger("vascodagama.watcher")
logger = logging.getLogger('vascodagama.images')
#setup flask
app = Flask(__name__)
#connect to redis
redis_images = utils.get_images_redis_conn()
#Setup a connection that will be used by RQ (each redis connection instance only talks to 1 DB)
redis_queue = utils.get_rq_redis_conn()
#gets a list of random URLS from refis.
def get_random_urls(count=100):
pipe_keys = redis_images.pipeline() #setup 2 batches
pipe_urls = redis_images.pipeline()
keys = []
for i in range(0, count): # get 'count' random keys
pipe_keys.randomkey()
for key in pipe_keys.execute(): #for each one of those random keys
pipe_urls.hget(key, "url") #get the URL property.
urls = pipe_urls.execute() #the list of URLs is the result.
return list(set(urls)) #return it.
import json #json functions
logging.config.dictConfig(utils.get_log_dict())
worker_logger = logging.getLogger("vascodagama.worker")
watcher_logger = logging.getLogger("vascodagama.watcher")
logger = logging.getLogger('vascodagama.images')
#setup flask
app = Flask(__name__)
#connect to redis
redis_images = utils.get_images_redis_conn()
#Setup a connection that will be used by RQ (each redis connection instance only talks to 1 DB)
redis_queue = utils.get_rq_redis_conn()
#gets a list of random URLS from refis.
def get_random_urls(count=100):
pipe_keys = redis_images.pipeline() #setup 2 batches
pipe_urls = redis_images.pipeline()
keys = []
for i in range(0,count): # get 'count' random keys
pipe_keys.randomkey()
for key in pipe_keys.execute(): #for each one of those random keys
pipe_urls.hget(key,"url") #get the URL property.
urls = pipe_urls.execute() #the list of URLs is the result.
return list(set(urls)) #return it.
from rq.decorators import job #funtion decoration
import dweepy #see dweet.io
import logging, logging.config
import utils
import datetime
logging.config.dictConfig(utils.get_log_dict())
logger = logging.getLogger('vascodagama.dashboard')
redis_images = utils.get_images_redis_conn()
r = utils.get_rq_redis_conn()
q = utils.get_rq()
#Setup our redis and RQ connections. see twitter_watch for more details.
configstuff = utils.configstuff()
@job("dashboard", connection=r, timeout=10, result_ttl=10)
def send_update(metric,
value): #method for sending updates about metrics as needed.
logger.debug("Sending update for {}: {}".format(metric, value))
dweepy.dweet_for(configstuff['dweet_thing'], {metric: value})
def update_dashboard(): # the primary function.
logger.info("updating")
#For each one of the metrics, collected the data and issue a job to actually send that out.
get_queue_len()
# logging.debug("{}: {}".format("tweets in queue",tweets_in_queue))
