def create_account_handler(self):
'''
Note that the verification goes on in model/form.py.
'''
# create the new user; post to db via sqlalchemy
new_user = model.User()
new_user.username = request.params['username']
new_user.fullname = " ".join(
[request.params['first_name'], request.params['last_name']])
new_user.experience = request.params['experience']
new_user._set_password(request.params['password'])
new_user.email = request.params['email']
# These are for citation settings,
# initialized to True to make everything in the citation visible by default.
new_user.show_journal = True
new_user.show_authors = True
new_user.show_keywords = True
Session.add(new_user)
Session.commit()
# send out an email
greeting_message = """
Hi, %s.\n
Thanks for signing up at abstrackr (%s). You
should be able to sign up now with username %s (only you know your password).
This is just a welcome email to say hello, and that we've got your email.
Should you ever need to reset your password, we'll send you instructions
to this email. In the meantime, happy screening!
-- The Brown EPC.
""" % (new_user.fullname, url('/', qualified=True), new_user.username)
try:
self.send_email_to_user(new_user, "welcome to abstrackr",
greeting_message)
except:
# this almost certainly means we're on our Windows dev box :)
pass
###
# log this user in programmatically (issue #28)
rememberer = request.environ['repoze.who.plugins']['cookie']
identity = {'repoze.who.userid': new_user.username}
response.headerlist = response.headerlist + \
rememberer.remember(request.environ, identity)
rememberer.remember(request.environ, identity)
# if they were originally trying to join a review prior to
# registering, then join them now. (issue #8).
if 'then_join' in request.params and request.params['then_join'] != '':
redirect(
url(controller="review",
action="join",
review_code=request.params['then_join']))
else:
redirect(url(controller="account", action="login"))
def _clear_this_user_locks(self, all_assignments):
priorities_locked_by_this_user = self._get_all_priorities_locked_by_this_user(all_assignments)
for p in priorities_locked_by_this_user:
p.is_out = 0
p.locked_by = None
Session.add(p)
Session.commit()
def _clear_this_user_locks(self, all_assignments):
priorities_locked_by_this_user = self._get_all_priorities_locked_by_this_user(all_assignments)
for p in priorities_locked_by_this_user:
p.is_out = 0
p.locked_by = None
Session.add(p)
Session.commit()
def create_account_handler(self):
'''
Note that the verification goes on in model/form.py.
'''
# create the new user; post to db via sqlalchemy
new_user = model.User()
new_user.username = request.params['username']
new_user.fullname = " ".join([request.params['first_name'], request.params['last_name']])
new_user.experience = request.params['experience']
new_user._set_password(request.params['password'])
new_user.email = request.params['email']
# These are for citation settings,
# initialized to True to make everything in the citation visible by default.
new_user.show_journal = True
new_user.show_authors = True
new_user.show_keywords = True
Session.add(new_user)
Session.commit()
# send out an email
greeting_message = """
Hi, %s.\n
Thanks for signing up at abstrackr (%s). You
should be able to sign up now with username %s (only you know your password).
This is just a welcome email to say hello, and that we've got your email.
Should you ever need to reset your password, we'll send you instructions
to this email. In the meantime, happy screening!
-- The Brown EPC.
""" % (new_user.fullname, url('/', qualified=True), new_user.username)
try:
self.send_email_to_user(new_user, "welcome to abstrackr", greeting_message)
except:
# this almost certainly means we're on our Windows dev box :)
pass
###
# log this user in programmatically (issue #28)
rememberer = request.environ['repoze.who.plugins']['cookie']
identity = {'repoze.who.userid': new_user.username}
response.headerlist = response.headerlist + \
rememberer.remember(request.environ, identity)
rememberer.remember(request.environ, identity)
# if they were originally trying to join a review prior to
# registering, then join them now. (issue #8).
if 'then_join' in request.params and request.params['then_join'] != '':
redirect(url(controller="review", action="join", review_code=request.params['then_join']))
else:
redirect(url(controller="account", action="login"))
def gen_token_to_reset_pwd(self, user):
# generate a random token for the user to reset their password; stick
# it in the database
make_token = lambda N: ''.join(random.choice(string.ascii_uppercase + string.digits) for x in range(N))
reset_pwd_q = Session.query(model.ResetPassword)
existing_tokens = [entry.token for entry in reset_pwd_q.all()]
token_length=10
cur_token = make_token(token_length)
while cur_token in existing_tokens:
cur_token = make_code(token_length)
reset = model.ResetPassword()
reset.token = cur_token
reset.user_email = user.email
Session.add(reset)
Session.commit()
return cur_token
def gen_token_to_reset_pwd(self, user):
# generate a random token for the user to reset their password; stick
# it in the database
make_token = lambda N: ''.join(random.choice(string.ascii_uppercase + string.digits) for x in range(N))
reset_pwd_q = Session.query(model.ResetPassword)
existing_tokens = [entry.token for entry in reset_pwd_q.all()]
token_length=10
cur_token = make_token(token_length)
while cur_token in existing_tokens:
cur_token = make_code(token_length)
reset = model.ResetPassword()
reset.token = cur_token
reset.user_email = user.email
Session.add(reset)
Session.commit()
return cur_token
def _create_reviews(p_id, iter_size, which_iter):
lock_file_path = join(dirname(abspath(__file__)), '_delete_lock.lck')
if not isfile(lock_file_path):
Session.query(
model.Citation).filter(model.Citation.project_id != p_id).delete()
Session.query(
model.Label).filter(model.Label.project_id != p_id).delete()
Session.commit()
open(lock_file_path, 'w+').close()
u_id = 2629
k_init = 400
c_count = len(
Session.query(model.Citation).filter_by(project_id=p_id).all())
k_inc = 100
for itercount in range(iter_size * which_iter,
iter_size * which_iter + iter_size):
### THIS is the code for one run of the experiment
## labeled citation counter
labeled_citation_counter = 0
labels = Session.query(model.Label).filter_by(project_id=p_id).all()
user = Session.query(model.User).filter_by(id=u_id).first()
citations = Session.query(
model.Citation).filter_by(project_id=p_id).all()
print len(citations)
c_count = len(citations)
r_sample = defaultdict(list)
sample_indexes = sample(range(c_count), k_init)
C_r = []
for ii in sample_indexes:
C_r.append(citations[ii])
for cc in C_r:
for ll in Session.query(model.Label).filter_by(
project_id=p_id).filter_by(study_id=cc.id).all():
r_sample[ll.study_id].append(ll)
new_review = model.Project()
new_review.leaders.append(user)
new_review.initial_round_size = 0
new_review.tag_privacy = True
Session.add(new_review)
Session.flush()
state_dict = defaultdict(int)
citation_dict = {}
for c in citations:
citation = model.Citation()
citation.project_id = new_review.id
citation.title = c.title
citation.abstract = c.abstract
citation.keywords = c.keywords
citation.refman = c.refman
model.Session.add(citation)
Session.flush()
citation_dict[citation.id] = c.id
if c.id in r_sample:
labeled_citation_counter += 1
state_dict[citation.id] = 1
for t in r_sample[c.id]:
label = model.Label()
label.project_id = new_review.id
label.study_id = citation.id
label.label = t.label
model.Session.add(label)
print new_review.id
Session.commit()
## i is a counter for the current increment
i = 0
while True:
## we want to change the increment size if there are a certain number of citations is labeled
#if labeled_citation_counter > 15000:
# k_inc = 2000
#elif labeled_citation_counter > 5000:
# k_inc = 1000
#else:
# k_inc = 500
r_sample = defaultdict(list)
print "EXPERIMENT NO: " + str(itercount)
make_predictions(new_review.id)
######################## here's where I record the results
preds_for_review = Session.query(model.Prediction).filter(
model.Prediction.project_id == new_review.id).all()
path_to_preds_out = os.path.join(
"_exports",
"predictions_%d_%d_of_%d.csv" % (p_id, i, itercount))
with open(path_to_preds_out, 'w+') as fout:
csv_out = csv.writer(fout)
preds_file_headers = [
"citation_id", "refman", "title",
"predicted p of being relevant",
"'hard' screening prediction*", "state"
]
csv_out.writerow(preds_file_headers)
sorted_preds = sorted(preds_for_review,
key=lambda x: x.predicted_probability,
reverse=True)
for pred in sorted_preds:
citation = Session.query(model.Citation).filter(
model.Citation.id == pred.study_id).first()
#citation = self._get_citation_from_id(pred.study_id)
citation_title = citation.title.encode('ascii', 'ignore')
row_str = [
citation.id, citation.refman, citation_title,
pred.predicted_probability, pred.prediction,
state_dict[citation.id]
]
csv_out.writerow(row_str)
######################### ---------------------------
i += 1
if labeled_citation_counter >= c_count:
break
P_a = []
for pa in Session.query(model.Prediction).filter_by(
project_id=new_review.id).order_by(
model.Prediction.predicted_probability.desc()).all():
if state_dict[pa.study_id] == 0:
P_a.append(pa)
if len(P_a) == k_inc:
break
if len(P_a) == 0:
print "~~~NO PREDS!!!"
ccc = [
label
for label in Session.query(model.Citation.id).filter_by(
project_id=new_review.id).filter(
~model.Citation.labels.any()).limit(k_inc)
]
print len(ccc)
for cc in ccc:
labeled_citation_counter += 1
state_dict[cc.id] = 1
for ll in Session.query(model.Label).filter_by(
study_id=citation_dict[cc.id]).all():
label = model.Label()
label.project_id = new_review.id
label.study_id = cc.id
label.label = ll.label
model.Session.add(label)
else:
for pp in P_a:
labeled_citation_counter += 1
state_dict[pp.study_id] = 2
for ll in Session.query(
model.Label).filter_by(project_id=p_id).filter_by(
study_id=citation_dict[pp.study_id]).all():
label = model.Label()
label.project_id = new_review.id
label.study_id = pp.study_id
label.label = ll.label
model.Session.add(label)
Session.commit()
print len(
Session.query(
model.Label).filter_by(project_id=new_review.id).all())
return
def _set_assignment_done_status(self, all_assignments):
for a in all_assignments:
b_assignment_done = controller_globals._check_assignment_done(a)
a.done = b_assignment_done
Session.add(a)
Session.commit()
found = 0
not_found = 0
with open(FILE_PATH, 'r') as f:
reader = csv.DictReader(f, delimiter='\t')
for row in reader:
#print(row['id'], row['title'], row['abstract'])
citation = citations_q.filter_by(title=row['title'], project_id=PROJECT_ID).first()
if not citation:
print('could not find title matching with %s' % row['title'])
not_found += 1
else:
found += 1
citation.refman = row['id']
Session.add(citation)
Session.commit()
print("found %s" % found)
print("not found %s" % not_found)
##############################################################################################
### This was for getting PMID's from the titles
#citations_q = Session.query(model.Citation)
#citations = citations_q.filter(model.Citation.pmid == 0).all()
#
#print(len(citations))
#
#for c in citations:
# id = c.id
def _set_assignment_done_status(self, all_assignments):
for a in all_assignments:
b_assignment_done = controller_globals._check_assignment_done(a)
a.done = b_assignment_done
Session.add(a)
Session.commit()
