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)
if(len(candidates) > 0):
i = 0
b = controls.best_control(sents, candidates, dfs, nbest=5)
for bb in b:
bbuni = controls.touni(bb[0])
newb = generrors.randerr(bb[0])
cid = controls.insert_into_db(hit_id, newb, cur2)
if(not(cid == -1)):
outfile.write(bbuni+'\t')
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))
outfile.write('\n')
conn.commit()
else:
check = 0
cachedsents = codecs.open('controls.log.bk', encoding='utf-8', mode='r')
for hit in cachedsents.readlines():
guid=str(uuid.uuid4())
try:
sql="SELECT add_hit(%s, %s, %s, %s, %s, %s, %s);"
cur2.execute(sql,("", guid, hittype_id, lang_id, 0, 0, 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))
else:
idsql = 'SELECT id from esl_sentences where doc_id=%s;'
cur2.execute(idsql, (s[1],))
eslid = cur2.fetchone()[0]
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, eslid, lang_id, i))
sentnums.append(item[4])
#Insert 1 control sentence
ctrlsql = 'SELECT sentence, doc_id from esl_sentences where qc = 1 and doc=%s'
cur2.execute(ctrlsql, (doc,))
refs = cur2.fetchall()
controlnum = random.randint(0, len(refs)-1)
b = list(refs)[controlnum][0]
sentnum = list(refs)[controlnum][1]
timeout = 0
while(timeout < 20 and (sentnum in sentnums)):
controlnum = random.randint(0, len(refs)-1)
b = list(refs)[controlnum][0]
sentnum = list(refs)[controlnum][1]
timeout += 1
newb = generrors.randerr(b)
cid = controls.insert_into_db(hit_id, newb, b, cur2, qcnum)
if(cid == -1):
print "Error inserting control sentence to DB"
break;
else:
for i in range(0, settings['num_unknowns']+settings['num_knowns']):
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, n))
n += 1
else:
if(len(sents) > 0):
sent = sents[0]
sid = sent[0]
print sent
sql="INSERT INTO esl_hits_data(hit_id,esl_sentence_id,language_id,sentence_num)VALUES(%s,%s,%s,%s);"