qcnum = random.randint(0, settings["num_unknowns"] + settings["num_knowns"] -1)
guid=str(uuid.uuid4())
sql="SELECT add_hit(%s, %s, %s, %s, %s, %s, %s);"
cur2.execute(sql,("", guid, hittype_id, lang_id, 0, 0, 0))
hit_id = cur2.fetchone()[0]
if(not(hit_id in sentcounts)):
sentcounts.append(hit_id)
logging.info("Batch added")
sents = []
sentids = []
for item in batchiter:
doc_id = item[4]
candidates = controls.pull_candidates(doc_id.split('_')[0])
idsql = 'SELECT sentence from esl_sentences where doc_id=%s;'
cur2.execute(idsql, (doc_id,))
sents.append(cur2.fetchone()[0])
sentids.append(doc_id)
b = controls.best_control(sents, candidates, dfs)
print b
cid = controls.insert_into_db("CONTROL "+b, cur2)
conn.commit()
#outfile.write(b+'\n')
for s in enumerate(sentids):
print s
i = s[0]
if(i == qcnum):
sql="INSERT INTO esl_hits_data (hit_id, esl_sentence_id, language_id, sentence_num) VALUES (%s,%s,%s,%s);"
cur2.execute(sql,(hit_id, cid, lang_id, i))
#group senteces into groups of 5 for HITs
sents = []
sentids = []
docids = []
idsforhit = [0]*(settings["num_unknowns"] + settings["num_knowns"])
for item in batchiter:
doc_id = item[4]
docids.append(doc_id.split('_')[0])
idsql = 'SELECT sentence from esl_sentences where doc_id=%s;'
cur2.execute(idsql, (doc_id,))
sents.append(cur2.fetchone()[0])
sentids.append(doc_id)
if(args.reload):
doc = list(docids)[0]
print "Running control query for doc", doc_id
candidates = controls.pull_candidates(doc)
#choose the control sentence that fits best with the 5 real sentences
b = controls.best_control(sents, candidates, dfs)
bb = b[0][0]
print bb
logfile.write(bb+'\n')
newb = generrors.randerr(bb)
cid = controls.insert_into_db(hit_id, newb, bb, cur2, qcnum)
else:
bb = cachesents.pop()
newb = generrors.randerr(bb)
cid = controls.insert_into_db(hit_id, newb, bb, cur2, qcnum)
if(cid == -1):
print "Error inserting control sentence to DB"
break;
