def _load_term_to_db(fileName):
p = Pinyin()
count = 0
with open(fileName) as fp:
for line in fp:
print count
count += 1
if(count < 3225):
continue
try:
line = line.decode('gbk')
except Exception, err:
print line, err
continue
terms = line.split()
for term in terms:
if len(term) <= 1:
continue
pinyin = p.get_pinyin(term, "")
cond = {"_id": term}
if db.get_term(cond):
to = {"$inc": {"accFrequence": 1}}
db.update_term(cond, to)
else:
data = {"_id": term,
"pinyin": pinyin,
"accFrequence": 1,
"queryFrequence": 0}
db.insert_term(data)
def _load_term_to_db(fileName):
p = Pinyin()
count = 0
with open(fileName) as fp:
for line in fp:
print count
count += 1
if (count < 3225):
continue
try:
line = line.decode('gbk')
except Exception, err:
print line, err
continue
terms = line.split()
for term in terms:
if len(term) <= 1:
continue
pinyin = p.get_pinyin(term, "")
cond = {"_id": term}
if db.get_term(cond):
to = {"$inc": {"accFrequence": 1}}
db.update_term(cond, to)
else:
data = {
"_id": term,
"pinyin": pinyin,
"accFrequence": 1,
"queryFrequence": 0
}
db.insert_term(data)
def update_query(query):
if not isinstance(query, unicode):
query = query.decode("utf-8")
pinyin = p.get_pinyin(query, "")
cond = {"_id": query}
res = db.get_term(cond)
if res:
to = {"$inc": {"queryFrequence": 1}}
db.update_term(cond, to)
if res[u"queryFrequence"] == 2:
pinyinTree.insert_pinyin(pinyin)
else:
data = {"_id": query, "pinyin": pinyin, "accFrequence": 0, "queryFrequence": 1}
db.insert_term(data)
related = _get_related(query)
return related
def update_query(query):
if not isinstance(query, unicode):
query = query.decode('utf-8')
pinyin = p.get_pinyin(query, "")
cond = {"_id": query}
res = db.get_term(cond)
if res:
to = {"$inc": {"queryFrequence": 1}}
db.update_term(cond, to)
if res[u'queryFrequence'] == 2:
pinyinTree.insert_pinyin(pinyin)
else:
data = {
"_id": query,
"pinyin": pinyin,
"accFrequence": 0,
"queryFrequence": 1
}
db.insert_term(data)
related = _get_related(query)
return related