def analyze_wikis_globally():
print "Analyzing Wikis..."
wiki_collection = solr.existing_collection(solr.global_collection())
wiki_docs = [doc for doc in solr.get_all_docs_by_query(wiki_collection, '*:*')]
scaler = MinMaxScaler([doc['total_authority_f'] for doc in wiki_docs])
new_docs = []
for doc in wiki_docs:
new_docs.append({'id': doc['id'], 'scaled_authority_f': {'set': scaler.scale(doc['total_authority_f'])}})
if len(new_docs) > 10:
try:
wiki_collection.add(new_docs)
except ReadTimeout:
sleep(5)
try:
wiki_collection.add(new_docs)
except ReadTimeout:
pass
new_docs = []
try:
wiki_collection.add(new_docs)
except ReadTimeout:
sleep(5)
try:
wiki_collection.add(new_docs)
except ReadTimeout:
pass
wiki_collection.commit()
def analyze_users_globally():
print "Analyzing Users..."
user_collection = solr.existing_collection(solr.user_collection())
wiki_user_collection = solr.wiki_user_collection()
id_to_docs = dict()
for user_doc in solr.get_all_docs_by_query(wiki_user_collection, '*:*'):
# these are gonna be wiki-id_user-id
doc_id = user_doc['id'].split('_').pop()
if doc_id not in id_to_docs:
id_to_docs[doc_id] = dict(id=doc_id,
attr_entities={'set': []},
name_s={'set': user_doc['name_s']},
name_txt_en={'set': user_doc['name_txt_en']},
wikis_is={'set': []},
attr_wikis={'set': []},
authorities_fs={'set': []},
total_authority_f={'set': 0},
scaled_authority_f={'set': 0})
try:
map(id_to_docs[doc_id]['attr_entities']['set'].append, user_doc['attr_entities'])
id_to_docs[doc_id]['wikis_is']['set'].append(user_doc['wiki_id_i'])
id_to_docs[doc_id]['attr_wikis']['set'].append(user_doc['wiki_name_txt'])
id_to_docs[doc_id]['authorities_fs']['set'].append(user_doc['total_page_authority_f'])
except KeyError:
pass # zero f***s
id_to_total_authorities = dict([(uid, sum(doc['authorities_fs']['set'])) for uid, doc in id_to_docs.items()])
user_scaler = MinMaxScaler(id_to_total_authorities.values())
for uid, total_authority in id_to_total_authorities.items():
id_to_docs[uid]['total_authority_f']['set'] = total_authority
id_to_docs[uid]['scaled_authority_f']['set'] = user_scaler.scale(total_authority)
user_collection.add(id_to_docs.values())
user_collection.commit()
def ingest_data(wiki_id):
"""
Create Solr documents for a given wiki ID
:param wiki_id: the ID of the wiki (int or str)
:type wiki_id: int
:return:
"""
# make sure all pages and all user pages exists
solr.existing_collection(solr.all_pages_collection())
solr.existing_collection(solr.all_user_pages_collection())
resp = requests.get(u'http://www.wikia.com/api/v1/Wikis/Details', params={u'ids': wiki_id})
items = resp.json()['items']
if wiki_id not in items:
print u"Wiki doesn't exist?"
return
api_data = items[wiki_id]
wiki_data = {
'id': api_data['id'],
'wam_f': {'set': api_data['wam_score']},
'title_s': {'set': api_data['title']},
'attr_title': {'set': api_data['title']},
'attr_desc': {'set': api_data['desc']}
}
for key in api_data['stats'].keys():
wiki_data['%s_i' % key] = {'set': api_data['stats'][key]}
wiki_api_data = requests.get(u'%swikia.php' % (api_data[u'url']),
params={u'method': u'getForWiki',
u'service': u'CrossWikiCore',
u'controller': u'WikiaSearchIndexerController'}).json()[u'contents']
wiki_data[u'hub_s'] = wiki_api_data[u'hub_s']
# easier
api_data[u'hub_s'] = wiki_api_data[u'hub_s']
collection = solr.existing_collection(solr.collection_for_wiki(wiki_id))
use_caching(is_read_only=True, shouldnt_compute=True)
wpe = WikiPageToEntitiesService().get_value(wiki_id)
if not wpe:
print u"NO WIKI PAGE TO ENTITIES SERVICE FOR", wiki_id
return False
documents = []
grouped_futures = []
pages_to_authority = WikiAuthorityService().get_value(str(wiki_data['id']))
for counter, (doc_id, entity_data) in enumerate(wpe.items()):
documents.append({
'id': doc_id,
'attr_entities': {'set': list(set(entity_data.get(u'redirects', {}).values()
+ entity_data.get(u'titles')))},
'type_s': {'set': 'Page'},
'authority_f': {'set': pages_to_authority.get(doc_id, 0)},
'hub_s': wiki_api_data['hub_s']
})
if counter != 0 and counter % 1500 == 0:
grouped_futures.append(
group(add_with_metadata.s(api_data, grouping) for grouping in iter_grouper(15, documents))()
)
documents = []
grouped_futures.append(
group(add_with_metadata.s(api_data, grouping) for grouping in iter_grouper(15, documents))()
)
# block on completion of all grouped futures
completed = 0
total = 0
while len(filter(lambda x: not x.ready(), grouped_futures)) > 1:
new_completed = 0
new_total = 0
for future in grouped_futures:
new_completed += future.completed_count()
new_total += len(future.results)
if completed != new_completed or total != new_total:
completed = new_completed
total = new_total
print "Grouped Tasks: (%d/%d)" % (completed, total)
sleep(2)
all_user_tuples = []
for future in grouped_futures:
result = get_with_backoff(future, [])
map(all_user_tuples.extend, result)
all_user_tuples = list(set(all_user_tuples))
if not all_user_tuples:
print "Empty user tuples, bailing"
return
# assign the unique user ids to the first variable, and the unique usernames to the second
all_user_ids, all_users = zip(*all_user_tuples)
collection.commit()
solr.all_pages_collection().commit()
solr.all_user_pages_collection().commit()
wiki_data['attr_entities'] = {'set': []}
for count, entities in WikiEntitiesService().get_value(str(wiki_id)).items():
for entity in entities:
map(wiki_data['attr_entities']['set'].append, [entity] * int(count)) # goddamnit count isn't int
wiki_data['user_ids_is'] = {'set': all_user_ids}
wiki_data['attr_users'] = {'set': all_users}
wiki_data['total_authority_f'] = {'set': sum(pages_to_authority.values())}
wiki_data['authorities_fs'] = {'set': pages_to_authority.values()}
wiki_collection = solr.existing_collection(solr.global_collection())
wiki_collection.add([wiki_data])
wiki_collection.commit()
print "Committed wiki data"
print "Retrieving user docs..."
futures = group(build_wiki_user_doc.s(api_data, user_tuple) for user_tuple in all_user_tuples)()
future_result_len = len(futures.results)
while not futures.ready():
print "Progress: (%d/%d)" % (futures.completed_count(), future_result_len)
sleep(2)
user_docs = get_with_backoff(futures, [])
if not user_docs:
print "User docs was empty. Possibly connection problems."
return
authority_scaler = MinMaxScaler([doc['total_page_authority_f']['set'] for doc in user_docs])
contribs_scaler = MinMaxScaler([doc['total_contribs_f']['set'] for doc in user_docs])
for doc in user_docs:
scaled_authority = authority_scaler.scale(doc['total_page_authority_f']['set'])
scaled_contribs = contribs_scaler.scale(doc['total_contribs_f']['set'])
doc['scaled_authority_f'] = {'set': scaled_authority}
doc['scaled_contribs_f'] = {'set': scaled_contribs}
doc['scaled_contribs_authority_f'] = {'set': scaled_authority * scaled_contribs}
wiki_user_collection = solr.existing_collection(solr.wiki_user_collection())
wiki_user_collection.add(user_docs)
wiki_user_collection.commit()
print "Analyzing topics"
futures = group(get_wiki_topic_doc.s(wiki_data['id'], topic)
for topic in list(set(wiki_data['attr_entities']['set'])))()
future_result_len = len(futures.results)
counter = 0
while not futures.ready():
if counter % 5 == 0:
print "Progress: (%d/%d)" % (futures.completed_count(), future_result_len)
sleep(2)
counter += 1
topic_docs = get_with_backoff(futures, [])
if not topic_docs:
print "No topics, probably a connection error"
return
collection.add(topic_docs)
collection.commit()
topic_collection = solr.existing_collection(solr.all_topics_collection())
topic_collection.add(topic_docs)
topic_collection.commit()