def run(self):
"""
Drops all indices
"""
global_coll = solr.global_collection()
for doc in solr.get_all_docs_by_query(global_coll, '*:*', fields='id'):
solr.collection_for_wiki(doc['id']).drop()
global_coll.drop()
solr.all_pages_collection().drop()
solr.all_topics_collection().drop()
solr.all_user_pages_collection().drop()
solr.wiki_user_collection().drop()
solr.user_collection().drop()
def get_row(self):
"""
Returns the row from the DB as a dict
:return: row data
:rtype: dict
"""
for row in solr.get_all_docs_by_query(solr.all_pages_collection(),
'id:%s_%s' % (str(self.page_id), str(self.wiki_id))):
return row
def get_row(self):
"""
Returns the row from the DB as a dict
:return: row data
:rtype: dict
"""
for row in solr.get_all_docs_by_query(solr.all_pages_collection(),
'id:%s' % self.doc_id,
fields=','.join(self.fields)):
return row
def run(self):
"""
Drops all indices
"""
global_coll = solr.global_collection()
print 'global'
global_coll.optimize()
for doc in solr.get_all_docs_by_query(global_coll, '*:*', fields='id'):
print doc['id']
solr.collection_for_wiki(doc['id']).optimize()
print 'all pages'
solr.all_pages_collection().optimize()
print 'all topics'
solr.all_topics_collection().optimize()
print 'all user pages'
solr.all_user_pages_collection().optimize()
print 'wiki user'
solr.wiki_user_collection().optimize()
print 'user'
solr.user_collection().optimize()
def analyze_pages_globally():
print "Analyzing All Pages..."
page_collection = solr.all_pages_collection()
authorities = []
for page_doc in solr.get_all_docs_by_query(page_collection, '*:*'):
authorities.append(page_doc['authority_f'])
page_scaler = MinMaxScaler(authorities)
docs = []
counter = 0
for page_doc in solr.get_all_docs_by_query(page_collection, '*:*'):
docs.append({'id': page_doc['id'], 'scaled_authority_f': {'set': page_scaler.scale(page_doc['authority_f'])}})
counter += 1
if counter % 500:
page_collection.add(docs)
docs = []
page_collection.commit()
def get_pages(self, limit=10, offset=None, **kwargs):
"""
Gets most authoritative pages for a topic using Authority DB and Wikia API data
:param limit: Number of results we want
:type limit: int
:param offset: offset
:type offset: int
:return: a list of objects reflecting page results
:rtype: list
"""
collection = solr.all_pages_collection()
return solr.get_docs_by_query_with_limit(collection,
self.topic,
limit=limit,
offset=offset,
boost='scaled_authority_f',
fields=','.join(PageModel.fields),
**sans_q(kwargs))
def get_pages(self, limit=10, offset=None, for_api=False):
"""
Gets most authoritative pages for a topic using Authority DB and Wikia API data
:param limit: Number of results we want
:type limit: int
:param offset: offset
:type offset: int
:param for_api: if it's for the api, we add less
:type for_api: bool
:return: a list of objects reflecting page results
:rtype: list
"""
collection = solr.all_pages_collection()
return solr.get_docs_by_query_with_limit(collection,
self.topic,
limit=limit,
offset=offset,
boost='scaled_authority_f')
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()
def add_with_metadata(wiki_data, docs):
"""
For a grouping of docs, gets metadata from SearchController, and generates user documents.
Then commits. Let's hear it for asynchronous request handling.
:param wiki_data: a dict representing the data we've retrieved from the Wikia API
:type wiki_data: dict
:param docs: a list of docs ready to be uploaded to solr
:type docs: list
:return: a list of tuples including user ids and accompanying user name
:rtype: list
"""
docs = filter(lambda x: x, docs)
params = {
u'controller': u'WikiaSearchIndexerController',
u'method': u'get',
u'service': u'All',
u'ids': u'|'.join([doc['id'].split('_').pop() for doc in docs if doc]) # doc can be none here LOL
}
r = requests.get(u"%swikia.php" % wiki_data['url'], params=params)
response = r.json()
contents = response.get('contents', [])
author_pages = []
pa = PageAuthorityService()
user_dict = {}
for doc in docs:
for search_doc in contents:
if 'id' not in search_doc:
continue
if doc['id'] == search_doc['id']:
doc.update(dict(
attr_title=search_doc['title_en'],
title_s=search_doc['title_en'],
url_s=search_doc['url'],
hub_s=search_doc['hub']
))
users_txt = []
user_ids_is = []
total_contribs_f = 0.0
pa_response = pa.get(doc['id'])
if pa_response['status'] != 200:
continue
wiki_id, page_id = doc['id'].split('_')
for contrib in pa_response[doc['id']]:
user_dict[(contrib['userid'], contrib['user'])] = 1
users_txt.append(contrib['user'])
user_ids_is.append(contrib['userid'])
total_contribs_f += contrib['contribs']
author_pages.append({
'id': '%s_%s' % (doc['id'], contrib['userid']),
'doc_id_s': {'set': doc['id']},
'wiki_id_i': wiki_id,
'page_id_i': page_id,
'user_id_i': '%s' % contrib['userid'],
'type_s': {'set': 'PageUser'},
'name_txt_en': {'set': contrib['user']},
'name_s': {'set': contrib['user']},
'contribs_f': {'set': contrib['contribs']},
'attr_entities': {'set': doc['attr_entities']['set']},
'doc_authority_f': {'set': doc['authority_f']['set']},
'user_page_authority_f': {'set': contrib['contribs'] * doc['authority_f']['set']},
'hub_s': doc['hub_s']
})
doc['attr_users'] = {'set': users_txt}
doc['user_ids_is'] = {'set': user_ids_is}
doc['total_contribs_f'] = {'set': total_contribs_f}
update_docs = list(docs) + list(author_pages)
solr.collection_for_wiki(wiki_data['id']).add(update_docs)
solr.all_pages_collection().add(docs)
solr.all_user_pages_collection().add(author_pages)
return user_dict.keys()