def process_message(msg):
from r2.lib.db.queries import (
add_queries,
add_to_commentstree_q,
get_comments,
)
vote_data = json.loads(msg.body)
hook = hooks.get_hook('vote.validate_vote_data')
if hook.call_until_return(msg=msg, vote_data=vote_data) is False:
# Corrupt records in the queue. Ignore them.
print "Ignoring invalid vote by %s on %s %s" % (
vote_data.get('user_id', '<unknown>'),
vote_data.get('thing_fullname', '<unknown>'), vote_data)
return
timer = g.stats.get_timer("comment_vote_processor")
timer.start()
user = Account._byID(vote_data.pop("user_id"))
comment = Comment._by_fullname(vote_data.pop("thing_fullname"))
print "Processing vote by %s on %s %s" % (user, comment, vote_data)
try:
vote = Vote(
user,
comment,
direction=vote_data["direction"],
date=datetime.utcfromtimestamp(vote_data["date"]),
data=vote_data["data"],
event_data=vote_data.get("event_data"),
)
except TypeError as e:
# a vote on an invalid type got in the queue, just skip it
g.log.exception("Invalid type: %r", e.message)
return
vote.commit()
timer.intermediate("create_vote_object")
vote_valid = vote.is_automatic_initial_vote or vote.effects.affects_score
comment_valid = not (comment._spam or comment._deleted)
if vote_valid and comment_valid:
author = Account._byID(comment.author_id)
add_queries(
queries=[get_comments(author, sort, 'all') for sort in SORTS],
insert_items=comment,
)
timer.intermediate("author_queries")
# update the score periodically when a comment has many votes
update_threshold = g.live_config['comment_vote_update_threshold']
update_period = g.live_config['comment_vote_update_period']
num_votes = comment.num_votes
if num_votes <= update_threshold or num_votes % update_period == 0:
add_to_commentstree_q(comment)
timer.stop()
timer.flush()
def get_comment_scores(link, sort, comment_ids, timer):
"""Retrieve cached sort values for all comments on a post.
Arguments:
* link_id -- id of the Link containing the comments.
* sort -- a string indicating the attribute on the comments to use for
generating sort values.
Returns a dictionary from cid to a numeric sort value.
"""
from r2.lib.db import queries
from r2.models import CommentScoresByLink
if not comment_ids:
# no comments means no scores
return {}
if sort == "_date":
# comment ids are monotonically increasing, so we can use them as a
# substitute for creation date
scores_by_id = {comment_id: comment_id for comment_id in comment_ids}
else:
scores_by_id36 = CommentScoresByLink.get_scores(link, sort)
# we store these id36ed, but there are still bits of the code that
# want to deal in integer IDs
scores_by_id = {
int(id36, 36): score
for id36, score in scores_by_id36.iteritems()
}
scores_needed = set(comment_ids) - set(scores_by_id.keys())
if scores_needed:
g.stats.simple_event('comment_tree_bad_sorter')
missing_comments = Comment._byID(scores_needed,
data=True,
return_dict=False)
# queue the missing comments to be added to the comments tree, which
# will trigger adding their scores
for comment in missing_comments:
queries.add_to_commentstree_q(comment)
if sort == "_qa":
scores_by_missing_id36 = _get_qa_comment_scores(
link, missing_comments)
scores_by_missing = {
int(id36, 36): score
for id36, score in scores_by_missing_id36.iteritems()
}
else:
scores_by_missing = {
comment._id: getattr(comment, sort)
for comment in missing_comments
}
scores_by_id.update(scores_by_missing)
timer.intermediate('sort')
return scores_by_id
def process_message(msg):
from r2.lib.comment_tree import write_comment_scores
from r2.lib.db.queries import (
add_queries,
add_to_commentstree_q,
get_comments,
)
from r2.models.builder import get_active_sort_orders_for_link
vote_data = json.loads(msg.body)
hook = hooks.get_hook('vote.validate_vote_data')
if hook.call_until_return(msg=msg, vote_data=vote_data) is False:
# Corrupt records in the queue. Ignore them.
print "Ignoring invalid vote by %s on %s %s" % (
vote_data.get('user_id', '<unknown>'),
vote_data.get('thing_fullname', '<unknown>'), vote_data)
return
timer = g.stats.get_timer("comment_vote_processor")
timer.start()
user = Account._byID(vote_data.pop("user_id"))
comment = Comment._by_fullname(vote_data.pop("thing_fullname"))
print "Processing vote by %s on %s %s" % (user, comment, vote_data)
try:
vote = Vote(
user,
comment,
direction=vote_data["direction"],
date=datetime.utcfromtimestamp(vote_data["date"]),
data=vote_data["data"],
event_data=vote_data.get("event_data"),
)
except TypeError as e:
# a vote on an invalid type got in the queue, just skip it
g.log.exception("Invalid type: %r", e.message)
return
vote.commit()
timer.intermediate("create_vote_object")
vote_invalid = (not vote.effects.affects_score
and not vote.is_automatic_initial_vote)
comment_invalid = comment._spam or comment._deleted
if vote_invalid or comment_invalid:
timer.stop()
timer.flush()
return
author = Account._byID(comment.author_id)
add_queries(
queries=[get_comments(author, sort, 'all') for sort in SORTS],
insert_items=comment,
)
timer.intermediate("author_queries")
update_threshold = g.live_config['comment_vote_update_threshold']
update_period = g.live_config['comment_vote_update_period']
skip_score_update = (comment.num_votes > update_threshold
and comment.num_votes % update_period != 0)
# skip updating scores if this was the automatic initial vote. those
# updates will be handled by new_comment. Also only update scores
# periodically once a comment has many votes.
if not vote.is_automatic_initial_vote and not skip_score_update:
# check whether this link is using precomputed sorts, if it is
# we'll need to push an update to commentstree_q
link = Link._byID(comment.link_id)
if get_active_sort_orders_for_link(link):
# send this comment to commentstree_q where we will update
# CommentScoresByLink, CommentTree (noop), and CommentOrderer
add_to_commentstree_q(comment)
else:
# the link isn't using precomputed sorts, so just update the
# scores
write_comment_scores(link, [comment])
timer.intermediate("update_scores")
timer.stop()
timer.flush()
def get_comment_scores(link, sort, comment_ids, timer):
"""Retrieve cached sort values for all comments on a post.
Arguments:
* link_id -- id of the Link containing the comments.
* sort -- a string indicating the attribute on the comments to use for
generating sort values.
Returns a dictionary from cid to a numeric sort value.
"""
from r2.lib.db import queries
from r2.models import CommentScoresByLink
if not comment_ids:
# no comments means no scores
return {}
if sort == "_date":
# comment ids are monotonically increasing, so we can use them as a
# substitute for creation date
scores_by_id = {comment_id: comment_id for comment_id in comment_ids}
else:
scores_by_id36 = CommentScoresByLink.get_scores(link, sort)
# we store these id36ed, but there are still bits of the code that
# want to deal in integer IDs
scores_by_id = {
int(id36, 36): score
for id36, score in scores_by_id36.iteritems()
}
scores_needed = set(comment_ids) - set(scores_by_id.keys())
if scores_needed:
g.stats.simple_event('comment_tree_bad_sorter')
missing_comments = Comment._byID(
scores_needed, data=True, return_dict=False)
# queue the missing comments to be added to the comments tree, which
# will trigger adding their scores
for comment in missing_comments:
queries.add_to_commentstree_q(comment)
if sort == "_qa":
scores_by_missing_id36 = _get_qa_comment_scores(
link, missing_comments)
scores_by_missing = {
int(id36, 36): score
for id36, score in scores_by_missing_id36.iteritems()
}
else:
scores_by_missing = {
comment._id: getattr(comment, sort)
for comment in missing_comments
}
scores_by_id.update(scores_by_missing)
timer.intermediate('sort')
return scores_by_id
def process_message(msg):
from r2.lib.db.queries import (
add_queries,
add_to_commentstree_q,
get_comments,
)
vote_data = json.loads(msg.body)
hook = hooks.get_hook('vote.validate_vote_data')
if hook.call_until_return(msg=msg, vote_data=vote_data) is False:
# Corrupt records in the queue. Ignore them.
print "Ignoring invalid vote by %s on %s %s" % (
vote_data.get('user_id', '<unknown>'),
vote_data.get('thing_fullname', '<unknown>'),
vote_data)
return
timer = g.stats.get_timer("comment_vote_processor")
timer.start()
user = Account._byID(vote_data.pop("user_id"))
comment = Comment._by_fullname(vote_data.pop("thing_fullname"))
print "Processing vote by %s on %s %s" % (user, comment, vote_data)
try:
vote = Vote(
user,
comment,
direction=vote_data["direction"],
date=datetime.utcfromtimestamp(vote_data["date"]),
data=vote_data["data"],
event_data=vote_data.get("event_data"),
)
except TypeError as e:
# a vote on an invalid type got in the queue, just skip it
g.log.exception("Invalid type: %r", e.message)
return
vote.commit()
timer.intermediate("create_vote_object")
vote_valid = vote.is_automatic_initial_vote or vote.effects.affects_score
comment_valid = not (comment._spam or comment._deleted)
if vote_valid and comment_valid:
author = Account._byID(comment.author_id)
add_queries(
queries=[get_comments(author, sort, 'all') for sort in SORTS],
insert_items=comment,
)
timer.intermediate("author_queries")
# update the score periodically when a comment has many votes
update_threshold = g.live_config['comment_vote_update_threshold']
update_period = g.live_config['comment_vote_update_period']
num_votes = comment.num_votes
if num_votes <= update_threshold or num_votes % update_period == 0:
add_to_commentstree_q(comment)
timer.stop()
timer.flush()
def process_message(msg):
from r2.lib.comment_tree import write_comment_scores
from r2.lib.db.queries import (
add_queries,
add_to_commentstree_q,
get_comments,
)
from r2.models.builder import get_active_sort_orders_for_link
vote_data = json.loads(msg.body)
hook = hooks.get_hook('vote.validate_vote_data')
if hook.call_until_return(msg=msg, vote_data=vote_data) is False:
# Corrupt records in the queue. Ignore them.
print "Ignoring invalid vote by %s on %s %s" % (
vote_data.get('user_id', '<unknown>'),
vote_data.get('thing_fullname', '<unknown>'),
vote_data)
return
timer = g.stats.get_timer("comment_vote_processor")
timer.start()
user = Account._byID(vote_data.pop("user_id"))
comment = Comment._by_fullname(vote_data.pop("thing_fullname"))
print "Processing vote by %s on %s %s" % (user, comment, vote_data)
try:
vote = Vote(
user,
comment,
direction=vote_data["direction"],
date=datetime.utcfromtimestamp(vote_data["date"]),
data=vote_data["data"],
event_data=vote_data.get("event_data"),
)
except TypeError as e:
# a vote on an invalid type got in the queue, just skip it
g.log.exception("Invalid type: %r", e.message)
return
vote.commit()
timer.intermediate("create_vote_object")
vote_invalid = (not vote.effects.affects_score and
not vote.is_automatic_initial_vote)
comment_invalid = comment._spam or comment._deleted
if vote_invalid or comment_invalid:
timer.stop()
timer.flush()
return
author = Account._byID(comment.author_id)
add_queries(
queries=[get_comments(author, sort, 'all') for sort in SORTS],
insert_items=comment,
)
timer.intermediate("author_queries")
update_threshold = g.live_config['comment_vote_update_threshold']
update_period = g.live_config['comment_vote_update_period']
skip_score_update = (comment.num_votes > update_threshold and
comment.num_votes % update_period != 0)
# skip updating scores if this was the automatic initial vote. those
# updates will be handled by new_comment. Also only update scores
# periodically once a comment has many votes.
if not vote.is_automatic_initial_vote and not skip_score_update:
# check whether this link is using precomputed sorts, if it is
# we'll need to push an update to commentstree_q
link = Link._byID(comment.link_id)
if get_active_sort_orders_for_link(link):
# send this comment to commentstree_q where we will update
# CommentScoresByLink, CommentTree (noop), and CommentOrderer
add_to_commentstree_q(comment)
else:
# the link isn't using precomputed sorts, so just update the
# scores
write_comment_scores(link, [comment])
timer.intermediate("update_scores")
timer.stop()
timer.flush()