def usersearch(query):
global user_searcher_
analyzer = WhitespaceAnalyzer(Version.LUCENE_CURRENT)
if query is None:
return {}
#parse query using lucene parser and get docs
p_query = QueryParser(Version.LUCENE_CURRENT, "user", analyzer).parse(query)
print p_query
scoreDocs = user_searcher_.search(p_query, 50).scoreDocs
print "%s total matching documents." % len(scoreDocs)
docs = []
for scoreDoc in scoreDocs:
doc = user_searcher_.doc(scoreDoc.doc)
user = doc.get("user")
profile = UserProfiles.get_user_profile_(user)
p = {}
if profile:
if profile[user]["location"] != None:
p["hl"] = profile[user]["location"]
if "status" in profile[user]:
p["sts"] = {}
p["sts"]["text"] = profile[user]["status"]["text"]
#print p["sts"]["text"]
p["sts"]["created_at"] = profile[user]["status"]["created_at"]
p["foc"] = profile[user]["followers_count"]
p["frc"] = profile[user]["friends_count"]
p["sc"] = profile[user]["statuses_count"]
p["name"] = profile[user]["name"]
p["url"] = profile[user]["url"]
p["des"] = profile[user]["description"]
docs.append({"user": doc.get("user"), "locations": doc.get("locs"),
"profile": p})
else:
docs.append({"user": doc.get("user"), "locations": doc.get("locs")})
return cjson.encode(docs)
def update_profile_information(users, docs, locations,
query_location, query_terms,
reverse_geocode=True):
unique_users = set()
f = open("no_profiles.txt", "a+")
no_profile_accounts = [x.strip() for x in f.readlines()]
profiles = UserProfiles.get_user_profile_info(users)
#profiles = OnlineUser.get_location_time_info(users)
for rd in docs[:]:
#foursquare removed as we can consider it as a spam account
#introduced this logic because sometimes the user write the wrong username while referring to a user
#we get the profile from twitter but need to make sure that the username is corrected
if rd["user"] not in profiles:
found = False
for i in profiles.keys():
#the above logic might introduce duplicates so we remove those
if rd["user"] == i.lower():
rd["user"] = i
found = True
if not found:
if rd["user"] not in no_profile_accounts:
f.write(rd["user"]+"\n")
#we could augment the tweets but leaving that for now in case the user already exists
#but not doing that thinking not many people should be making mistakes so we won't miss
#out on many endorsements.
docs.remove(rd)
continue
if rd["user"] == "foursquare" or rd["user"] in unique_users:
docs.remove(rd)
continue
unique_users.add(rd["user"])
"""
#Not dealing with last online time, because of the real time twitter call
#slowdown. Some profiles might have last online time but we don't want to
#make it part of ranking yet. Just propose that it can be done.
if 'last_online' in profiles[rd['user']]:
rd['details']['lot'] = float(profiles[rd['user']]['last_online'])
rd['details']['t'] = profiles[rd['user']]['status']['created_at']#last online time
else:
rd['details']['lot'] = float(0) #make it something else
"""
rd["details"]["h"] = 0
rd["profile"] = {"hl": ""}
if profiles[rd["user"]]["location"] != None:
rd["profile"]["hl"] = cjson.encode(profiles[rd["user"]]["location"])
loc = profiles[rd["user"]]["location"].lower()
for l in locations:
if l in loc:
rd["details"]["h"] = locations[l]["dwt"]
break
if rd["details"]["h"] == 0:
coords = None
if "location_coords" in profiles[rd["user"]]:
coords = profiles[rd["user"]]["location_coords"]
else:
#coords = LI.get(loc, True)
coords = LI.get(loc, False) #not invoking geocoder because of query limits
if coords[0] is not None and coords[1] is not None and query_location is not None:
wdl = get_weighted_distance_location(query_location[1]["lat"],
query_location[1]["lng"],
coords[0],
coords[1])
if wdl:
rd["details"]["h"] = wdl["dwt"]
if "status" in profiles[rd["user"]]:
rd["profile"]["sts"] = {}
if "geo" in profiles[rd["user"]]["status"] and profiles[rd["user"]]["status"]["geo"] != None:
geo_field = profiles[rd["user"]]["status"]["geo"]
if reverse_geocode:
try:
geo_results = Geocoder.reverse_geocode(float(geo_field["coordinates"][0]),
float(geo_field["coordinates"][1]))
if geo_results:
rd["profile"]["cl"] = str(geo_results[0])
rd["profile"]["sts"]["geo"] = str(geo_results[0])
except GeocoderError as e:
print str(e)
rd["profile"]["sts"]["text"] = json.dumps(profiles[rd["user"]]["status"]["text"])
rd["profile"]["sts"]["created_at"] = json.dumps(profiles[rd["user"]]["status"]["created_at"])
rd["profile"]["foc"] = profiles[rd["user"]]["followers_count"]
rd["profile"]["frc"] = profiles[rd["user"]]["friends_count"]
rd["profile"]["sc"] = profiles[rd["user"]]["statuses_count"]
rd["profile"]["name"] = profiles[rd["user"]]["name"]
rd["profile"]["url"] = json.dumps(profiles[rd["user"]]["url"])
rd["profile"]["des"] = json.dumps(profiles[rd["user"]]["description"])
rd["details"]["term_des_count"] = 0
for i in query_terms:
if i in rd["profile"]["des"].lower():
rd["details"]["term_des_count"] += 1
if not rd['tweets'] and rd["details"]["term_des_count"] == 0:
docs.remove(rd)
continue
if "profile_image_url" in profiles[rd["user"]]:
rd["profile"]["pic"] = json.dumps(profiles[rd["user"]]["profile_image_url"])
else:
url = UserProfiles.get_profile_image_url(rd["user"])
if url:
rd["profile"]["pic"] = url
f.close()
return docs
