def user_search(user, days_ago=None, nosmoothing=False):
from jv3.study.content_analysis import activity_logs_for_user
global search_cache
global search_query_cache
alogs = wuw.reduceRepeatLogsValues(
activity_logs_for_user(user, None, days_ago))
searches = []
queries = []
last_time = 0
for al_i in range(len(alogs)):
al = alogs[al_i]
if al["action"] == 'search':
try:
query = JSONDecoder().decode(al["search"])
except:
continue
if type(query) == dict:
key = None
if 'search' in query: key = 'search'
if 'query' in query: key = 'query'
if key is not None:
# no empty searches pls
if len(query[key].strip()) > 0 and nosmoothing or long(
al['when']) - long(last_time) > (
10 * 1000): # 10 second smoothing
queries.append(query[key])
al['query'] = query[key]
al['hits'] = query.get('hits', [])
searches.append(al)
last_time = al['when']
elif al["action"] == 'clear-search' and (
nosmoothing or long(al['when']) - long(last_time) >
(10 * 1000)):
al['query'] = ''
searches.append(al)
last_time = al["when"]
search_cache[user.id] = searches
search_query_cache[user.id] = queries
return searches, nltk.FreqDist(queries)
def user_search(user,days_ago=None,nosmoothing=False):
from jv3.study.content_analysis import activity_logs_for_user
global search_cache
global search_query_cache
alogs = wuw.reduceRepeatLogsValues(activity_logs_for_user(user,None,days_ago))
searches = []
queries = []
last_time = 0
for al_i in range(len(alogs)):
al = alogs[al_i]
if al["action"] == 'search':
try:
query = JSONDecoder().decode(al["search"])
except:
continue
if type(query) == dict:
key = None
if 'search' in query: key = 'search'
if 'query' in query: key = 'query'
if key is not None:
# no empty searches pls
if len(query[key].strip()) > 0 and nosmoothing or long(al['when'])-long(last_time) > (10*1000): # 10 second smoothing
queries.append(query[key])
al['query'] = query[key]
al['hits'] = query.get('hits',[])
searches.append(al)
last_time = al['when']
elif al["action"] == 'clear-search' and (nosmoothing or long(al['when'])-long(last_time) > (10*1000)):
al['query'] = ''
searches.append(al)
last_time = al["when"]
search_cache[user.id] = searches
search_query_cache[user.id] = queries
return searches,nltk.FreqDist(queries)
def note_ss(note,filter_top=False):
from jv3.study.content_analysis import activity_logs_for_user
from jv3.study.ca_load import jid2nidforuser
global __sigscroll_startend_cache_flat
SSCF = __sigscroll_startend_cache_flat
def compute_duration(note):
def dur(send):
if type(send) == tuple: return send[1]-send[0]
return send
xd = SSCF.get(note["id"],[])
if len(xd) > 1:
return reduce(lambda x,y: dur(x)+dur(y),xd)
elif len(xd) == 1:
return dur(xd[0])
return -1
if note["id"] in SSCF :
return {'sigscroll_counts': len(SSCF.get(note["id"],[])), 'sigscroll_duration': compute_duration(note) }
## populate for this uer
alogs = activity_logs_for_user(note["owner_id"],None)
# if len(alogs) == 0:
# ## means we have no activitylogs for that user
# from jv3.study.content_analysis import _notes_to_features
# SSCF.update( [ (n["id"],[]) for n in [_notes_to_values(x) for x in Note.objects.filter(owner=n["owner_id"])] ] )
# return
debug__all_nids = []
next_is_top = True
toplist_jids = [] # things to block
alogs.sort(key=lambda x: x["when"])
print "activity logs", len(alogs)
for al_i in range(len(alogs)):
print al_i
al = alogs[al_i]
if al["action"] == 'sidebar-open':
next_is_top = True
continue
if not al["action"] == "significant-scroll":
continue
if al["search"] is None:
print "skipping"
continue
al["search"] = json.loads(al["search"])
if next_is_top:
toplist_jids = [long(nv["id"]) for nv in al["search"]["note_visibilities"]]
##print "TOPLIST :: %s " % repr(toplist_jids)
next_is_top = False
for nv in al["search"]["note_visibilities"]:
try :
jid = int(nv["id"]) ## this returns the _jid_ not id!
debug__all_nids.append(jid)
## omit nots that are at the top of the list
if filter_top and jid in toplist_jids:
print "filter top and jid in toplist continuing"
continue
nid = jid2nidforuser(al["owner"],jid) ## convert to NID (guaranteed unique)
if nv.has_key("exitTime") and nv.has_key("entryTime"):
ap = SSCF.get(nid,[])
if nv["entryTime"] == nv["exitTime"]:
print " case 1 ",jid
## this is to get around the bug in 0.4.5-7 which
## results in (start,start) for no-scroll open-close, and search/idle
ap.append( (nv["entryTime"],long(al["when"])) )
else:
print "case 2 "
ap.append( (nv["entryTime"],nv["exitTime"]) )
SSCF[nid] = ap
except:
print "noncritical warn %s " % repr(sys.exc_info())
pass
## filter all the newdudes
print SSCF
SSCF.update( dict([ (nid,adjacent_filtered(views)) for nid,views in SSCF.iteritems() ]) ) # if (nid in new_dudes) ] ) )
print "all debug__all_nids",len(set(debug__all_nids))
return {'sigscroll_counts': len(SSCF.get(note["id"],[])),'sigscroll_duration': compute_duration(note) }
def note_ss(note, filter_top=False):
from jv3.study.content_analysis import activity_logs_for_user
from jv3.study.ca_load import jid2nidforuser
global __sigscroll_startend_cache_flat
SSCF = __sigscroll_startend_cache_flat
def compute_duration(note):
def dur(send):
if type(send) == tuple: return send[1] - send[0]
return send
xd = SSCF.get(note["id"], [])
if len(xd) > 1:
return reduce(lambda x, y: dur(x) + dur(y), xd)
elif len(xd) == 1:
return dur(xd[0])
return -1
if note["id"] in SSCF:
return {
'sigscroll_counts': len(SSCF.get(note["id"], [])),
'sigscroll_duration': compute_duration(note)
}
## populate for this uer
alogs = activity_logs_for_user(note["owner_id"], None)
# if len(alogs) == 0:
# ## means we have no activitylogs for that user
# from jv3.study.content_analysis import _notes_to_features
# SSCF.update( [ (n["id"],[]) for n in [_notes_to_values(x) for x in Note.objects.filter(owner=n["owner_id"])] ] )
# return
debug__all_nids = []
next_is_top = True
toplist_jids = [] # things to block
alogs.sort(key=lambda x: x["when"])
print "activity logs", len(alogs)
for al_i in range(len(alogs)):
print al_i
al = alogs[al_i]
if al["action"] == 'sidebar-open':
next_is_top = True
continue
if not al["action"] == "significant-scroll":
continue
if al["search"] is None:
print "skipping"
continue
al["search"] = json.loads(al["search"])
if next_is_top:
toplist_jids = [
long(nv["id"]) for nv in al["search"]["note_visibilities"]
]
##print "TOPLIST :: %s " % repr(toplist_jids)
next_is_top = False
for nv in al["search"]["note_visibilities"]:
try:
jid = int(nv["id"]) ## this returns the _jid_ not id!
debug__all_nids.append(jid)
## omit nots that are at the top of the list
if filter_top and jid in toplist_jids:
print "filter top and jid in toplist continuing"
continue
nid = jid2nidforuser(
al["owner"], jid) ## convert to NID (guaranteed unique)
if nv.has_key("exitTime") and nv.has_key("entryTime"):
ap = SSCF.get(nid, [])
if nv["entryTime"] == nv["exitTime"]:
print " case 1 ", jid
## this is to get around the bug in 0.4.5-7 which
## results in (start,start) for no-scroll open-close, and search/idle
ap.append((nv["entryTime"], long(al["when"])))
else:
print "case 2 "
ap.append((nv["entryTime"], nv["exitTime"]))
SSCF[nid] = ap
except:
print "noncritical warn %s " % repr(sys.exc_info())
pass
## filter all the newdudes
print SSCF
SSCF.update(
dict([
(nid, adjacent_filtered(views)) for nid, views in SSCF.iteritems()
])) # if (nid in new_dudes) ] ) )
print "all debug__all_nids", len(set(debug__all_nids))
return {
'sigscroll_counts': len(SSCF.get(note["id"], [])),
'sigscroll_duration': compute_duration(note)
}
