def validate_link(url,whitelist=False):
if url:
url=sanitize_url(url)
if url:
if whitelist and domain(url) not in DOMAIN_WHITELIST:
print "Domain %s not in whitelist." % domain(url)
return False
try:
lbu = LinksByUrl._byID(LinksByUrl._key_from_url(url))
except tdb_cassandra.NotFound:
return url
link_id36s = lbu._values()
links = Link._byID36(link_id36s, data=True, return_dict=False)
links = [l for l in links if not l._deleted]
if len(links)==0:
return url
print "Link %s exists..." % url
return False
def port_cassaurls(after_id=None, estimate=15231317):
from r2.models import Link, LinksByUrl
from r2.lib.db import tdb_cassandra
from r2.lib.db.operators import desc
from r2.lib.db.tdb_cassandra import CL
from r2.lib.utils import fetch_things2, in_chunks, progress
q = Link._query(Link.c._spam == (True, False), sort=desc("_date"), data=True)
if after_id:
q._after(Link._byID(after_id, data=True))
q = fetch_things2(q, chunk_size=500)
q = progress(q, estimate=estimate)
q = (l for l in q if getattr(l, "url", "self") != "self" and not getattr(l, "is_self", False))
chunks = in_chunks(q, 500)
for chunk in chunks:
with LinksByUrl._cf.batch(write_consistency_level=CL.ONE) as b:
for l in chunk:
k = LinksByUrl._key_from_url(l.url)
if k:
b.insert(k, {l._id36: l._id36})
def port_cassaurls(after_id=None, estimate=15231317):
from r2.models import Link, LinksByUrl
from r2.lib.db import tdb_cassandra
from r2.lib.db.operators import desc
from r2.lib.db.tdb_cassandra import CL
from r2.lib.utils import fetch_things2, in_chunks, progress
q = Link._query(Link.c._spam == (True, False),
sort=desc('_date'),
data=True)
if after_id:
q._after(Link._byID(after_id, data=True))
q = fetch_things2(q, chunk_size=500)
q = progress(q, estimate=estimate)
q = (l for l in q if getattr(l, 'url', 'self') != 'self'
and not getattr(l, 'is_self', False))
chunks = in_chunks(q, 500)
for chunk in chunks:
with LinksByUrl._cf.batch(write_consistency_level=CL.ONE) as b:
for l in chunk:
k = LinksByUrl._key_from_url(l.url)
if k:
b.insert(k, {l._id36: l._id36})
