def test():
juice = Juice(
key_name="test",
image="http://pics.dmm.co.jp/mono/movie/1dandy044/1dandy044pl.jpg",
download="http://www.jandown.com/link.php?ref=CUWjjPJP0q",
)
juice.put()
def harvest():
squeezed=memcache.get('Squeezed::lemons')
if squeezed is None:
squeezed=Squeezed.get_by_key_name('squeezed')
if squeezed is None:
fresh=(baseurl+lemon['url'] for lemon in lemons())
else:
fresh=(baseurl+lemon['url'] for lemon in lemons() if lemon['url'] not in squeezed.lemons)
else:
fresh=(baseurl+lemon['url'] for lemon in lemons() if lemon['url'] not in squeezed.lemons)
bucket=[]
for lemon in fresh:
logging.info('squeezing '+lemon)
juices = scrapemark.scrape("""
<span class='tpc_title'></span>
{*
<img src='{{ [juices].image }}' border=0>
<a href='{{ [juices].download }}' target=_blank></a>
*}
""", url=lemon)['juices']
logging.info(juices)
for juice in juices:
try:
juice=Juice(key_name=lemon, image=juice['image'], download=juice['download'])
juice.put()
except BadValueError:
logging.info(juice)
bucket.append(lemon)
if squeezed is None:
squeezed=Squeezed(key_name='squeezed', lemons=bucket)
else:
squeezed.lemons.extend(bucket)
squeezed.put()
memcache.set('Squeezed::lemons', squeezed)
