def update_least_recently_updated(number_to_update, myredis, mydao):
(tiids_to_update, docs) = get_least_recently_updated_tiids_in_db(number_to_update, mydao)
update_docs_with_updater_timestamp(docs, mydao)
QUEUE_DELAY_IN_SECONDS = 0.25
mixpanel.track("Trigger:Update", {"Number Items":len(tiids_to_update), "Update Type":"Scheduled Least Recently"})
item.start_item_update(tiids_to_update, myredis, mydao, sleep_in_seconds=QUEUE_DELAY_IN_SECONDS)
return tiids_to_update
def update_github():
from totalimpact import item, tiredis
myredis = tiredis.from_url(os.getenv("REDISTOGO_URL"), db=0)
view_name = "queues/by_alias"
view_rows = db.view(view_name, include_docs=False)
row_count = 0
page_size = 500
start_key = ["url", "https://github.0000000"]
end_key = ["url", "https://github.zzzzzzzz"]
from couch_paginator import CouchPaginator
page = CouchPaginator(db, view_name, page_size, include_docs=False, start_key=start_key, end_key=end_key)
while page:
for row in page:
tiid = row.id
item.start_item_update([tiid], myredis, db, sleep_in_seconds=0.05)
row_count += 1
print "."
logger.info("%i. getting new page, last id was %s" % (row_count, row.id))
if page.has_next:
page = CouchPaginator(db, view_name, page_size, start_key=page.next, end_key=end_key, include_docs=True)
else:
page = None
print "number items = ", row_count
def update_github():
from totalimpact import item, tiredis
myredis = tiredis.from_url(os.getenv("REDISTOGO_URL"), db=0)
view_name = "queues/by_alias"
view_rows = db.view(view_name, include_docs=False)
row_count = 0
page_size = 500
start_key = ["url", "https://github.0000000"]
end_key = ["url", "https://github.zzzzzzzz"]
from couch_paginator import CouchPaginator
page = CouchPaginator(db, view_name, page_size, include_docs=False, start_key=start_key, end_key=end_key)
while page:
for row in page:
tiid = row.id
item.start_item_update([tiid], myredis, db, sleep_in_seconds=0.05)
row_count += 1
print "."
logger.info(u"%i. getting new page, last id was %s" %(row_count, row.id))
if page.has_next:
page = CouchPaginator(db, view_name, page_size, start_key=page.next, end_key=end_key, include_docs=True)
else:
page = None
print "number items = ", row_count
def refresh_from_tiids(tiids, myredis):
for tiid in tiids:
try:
item_obj = item_module.Item.from_tiid(tiid)
item = item_obj.as_old_doc()
item_module.start_item_update(tiid, item["aliases"], myredis)
except AttributeError:
logger.debug(u"couldn't find tiid {tiid} so not refreshing its metrics".format(
tiid=tiid))
return tiids
def products_refresh_post(tiids_string):
tiids = tiids_string.split(",")
for tiid in tiids:
try:
item_obj = item_module.Item.from_tiid(tiid)
item = item_obj.as_old_doc()
item_module.start_item_update(tiid, item["aliases"], myredis)
except AttributeError:
logger.debug(u"couldn't find tiid {tiid} so not refreshing its metrics".format(
tiid=tiid))
resp = make_response("true", 200)
return resp
def products_refresh_post(tiids_string):
tiids = tiids_string.split(",")
for tiid in tiids:
try:
item_obj = item_module.Item.from_tiid(tiid)
item = item_obj.as_old_doc()
item_module.start_item_update(tiid, item["aliases"], myredis)
except AttributeError:
logger.debug(
u"couldn't find tiid {tiid} so not refreshing its metrics".
format(tiid=tiid))
resp = make_response("true", 200)
return resp
def update_by_tiids(all_tiids, number_to_update, myredis):
tiids_to_update = all_tiids[0:min(number_to_update, len(all_tiids))]
now = datetime.datetime.utcnow().isoformat()
print "updating {number_to_update} of them now".format(number_to_update=number_to_update)
QUEUE_DELAY_IN_SECONDS = 0.25
for tiid in tiids_to_update:
item_obj = item_module.Item.query.get(tiid) # can use this method because don't need metrics
item_doc = item_obj.as_old_doc()
item_module.start_item_update(tiid, item_doc["aliases"], myredis)
item_obj.last_update_run = now
db.session.add(item_obj)
time.sleep(QUEUE_DELAY_IN_SECONDS)
db.session.commit()
return tiids_to_update
def refresh_from_tiids(tiids, analytics_credentials, priority, myredis):
item_objects = item_module.Item.query.filter(item_module.Item.tiid.in_(tiids)).all()
dicts_to_refresh = []
for item_obj in item_objects:
try:
tiid = item_obj.tiid
item_obj.set_last_refresh_start()
db.session.add(item_obj)
alias_dict = item_module.alias_dict_from_tuples(item_obj.alias_tuples)
dicts_to_refresh += [{"tiid":tiid, "aliases_dict": alias_dict}]
except AttributeError:
logger.debug(u"couldn't find tiid {tiid} so not refreshing its metrics".format(
tiid=tiid))
db.session.commit()
item_module.start_item_update(dicts_to_refresh, priority, myredis)
return tiids
def collection_update(cid=""):
# first, get the tiids in this collection:
try:
collection = mydao.get(cid)
tiids = collection["alias_tiids"].values()
except Exception:
logger.exception("couldn't get tiids in POST collection '{cid}'".format(
cid=cid
))
abort(404, "couldn't get tiids for this collection...maybe doesn't exist?")
mixpanel.track("Trigger:Update",
{"Number Items":len(tiids), "Report Id":cid, "Update Type":"webapp"},
request)
item_module.start_item_update(tiids, myredis, mydao)
resp = make_response("true", 200)
resp.mimetype = "application/json"
return resp
def refresh_from_tiids(tiids, analytics_credentials, priority, myredis):
item_objects = item_module.Item.query.filter(
item_module.Item.tiid.in_(tiids)).all()
dicts_to_refresh = []
for item_obj in item_objects:
try:
tiid = item_obj.tiid
item_obj.set_last_refresh_start()
db.session.add(item_obj)
alias_dict = item_module.alias_dict_from_tuples(
item_obj.alias_tuples)
dicts_to_refresh += [{"tiid": tiid, "aliases_dict": alias_dict}]
except AttributeError:
logger.debug(
u"couldn't find tiid {tiid} so not refreshing its metrics".
format(tiid=tiid))
db.session.commit()
item_module.start_item_update(dicts_to_refresh, priority, myredis)
return tiids
def collection_metrics_refresh(cid=""):
# first, get the tiids in this collection:
try:
coll_doc = collection.get_collection_doc(cid)
tiids = coll_doc["alias_tiids"].values()
except (TypeError, AttributeError):
logger.exception(
u"couldn't get tiids in POST collection '{cid}'".format(cid=cid))
abort_custom(500, "Error doing collection_update")
for tiid in tiids:
try:
item_obj = item_module.Item.from_tiid(tiid)
item = item_obj.as_old_doc()
item_module.start_item_update(tiid, item["aliases"], myredis)
except AttributeError:
logger.debug(
u"couldn't find tiid {tiid} in {cid} so not refreshing its metrics"
.format(cid=cid, tiid=tiid))
resp = make_response("true", 200)
return resp
def collection_metrics_refresh(cid=""):
# first, get the tiids in this collection:
try:
coll_doc = collection.get_collection_doc(cid)
tiids = coll_doc["alias_tiids"].values()
except (TypeError, AttributeError):
logger.exception(u"couldn't get tiids in POST collection '{cid}'".format(
cid=cid
))
abort_custom(500, "Error doing collection_update")
for tiid in tiids:
try:
item_obj = item_module.Item.from_tiid(tiid)
item = item_obj.as_old_doc()
item_module.start_item_update(tiid, item["aliases"], myredis)
except AttributeError:
logger.debug(u"couldn't find tiid {tiid} in {cid} so not refreshing its metrics".format(
cid=cid, tiid=tiid))
resp = make_response("true", 200)
return resp
def update_active_publisher_items(number_to_update, myredis, mydao):
all_tiids = []
all_docs = []
for publisher in active_publishers:
for journal_dict in active_publishers[publisher]["journals"]:
min_year = 2011 #only update 2012-2013 right now
(tiids_from_doi_prefix, docs_from_doi_prefix) = get_least_recently_updated_items_from_doi_prefix(min_year, journal_dict["doi_prefix"], myredis, mydao)
logger.info("doi prefix {prefix} has {num_tiids} items published since {min_year} last updated more than 24 hours ago".format(
num_tiids=len(tiids_from_doi_prefix), prefix=journal_dict["doi_prefix"], min_year=min_year))
all_tiids += tiids_from_doi_prefix
all_docs += docs_from_doi_prefix
print "recent items for active publishers that were last updated more than a day ago, n=", len(all_tiids)
tiids_to_update = all_tiids[0:min(number_to_update, len(all_tiids))]
docs_to_update = all_docs[0:min(number_to_update, len(all_docs))]
response = update_docs_with_updater_timestamp(docs_to_update, mydao)
print "updating {number_to_update} of them now".format(number_to_update=number_to_update)
QUEUE_DELAY_IN_SECONDS = 0.25
mixpanel.track("Trigger:Update", {"Number Items":len(tiids_to_update), "Update Type":"Scheduled Registered"})
item.start_item_update(tiids_to_update, myredis, mydao, sleep_in_seconds=QUEUE_DELAY_IN_SECONDS)
return tiids_to_update
