def cache_fill_loop():
global sources
while True: # fill cache up to min_cache_imgs
if Crawler.info()["images"] < min_cache_imgs_before_refill:
while Crawler.info()["images"] < min_cache_imgs:
random.choice(sources).crawl()
# sleep for non-invasive threading ;)
time.sleep(1.337)
def cache_status():
info = Crawler.info()
msg = "images cached: %d (%d bytes) - already crawled: %d (%d bytes)" % \
(info["images"], info["images_size"], info["blacklist"], info["blacklist_size"])
logger.info(msg)
for crawler in sources:
for site in sources[crawler]:
key = crawler + "_" + site
if key in info["images_per_site"]:
factor = 1
if crawler in factors and site in factors[crawler]:
factor = factors[crawler][site]
count = info["images_per_site"][key]
bar = "|"
for i in range(0, count / 5):
if i < min_cache_imgs_before_refill / 5:
bar += "#"
else:
bar += "*"
sitestats = ("%15s - %-15s with factor %4.1f: %2d Images " + bar) % (crawler, site, factor, count)
logger.info(sitestats)
msg += "\r\n" + sitestats
return msg
def cache_fill_loop():
global sources
while True: # fill cache up to min_cache_imgs per site
info = Crawler.info()
for crawler in sources:
for site in sources[crawler]:
key = crawler + "_" + site
if key not in info["images_per_site"] or info["images_per_site"][key] < min_cache_imgs_before_refill:
try:
sources[crawler][site].crawl()
info = Crawler.info()
except Exception as e:
logger.error("Error in crawler %s - %s: %s" % (crawler, site, e))
break
# sleep for non-invasive threading ;)
time.sleep(1.337)
def cache_status():
info = Crawler.info()
msg = "images cached: %d (%d bytes) - already crawled: %d (%d bytes)" %\
(info["images"], info["images_size"], info["blacklist"], info["blacklist_size"])
logger.info(msg)
return msg
