|
|
@@ -166,9 +166,25 @@ class TrunksService:
|
|
|
result = []
|
|
|
for i, targetSentence in enumerate(targetSentences):
|
|
|
cleanedTarget = sentence_util.clean_text(targetSentence)
|
|
|
+ #cleanedTarget长度小于5的不进行匹配
|
|
|
+ if len(cleanedTarget)<5:
|
|
|
+ continue
|
|
|
if cleanedTarget in cleanedContent:
|
|
|
result.append(i)
|
|
|
|
|
|
+ # 补齐连续下标
|
|
|
+ if result:
|
|
|
+ result.sort()
|
|
|
+ filled_result = []
|
|
|
+ prev = result[0]
|
|
|
+ filled_result.append(prev)
|
|
|
+ for current in result[1:]:
|
|
|
+ if current - prev <= 2:
|
|
|
+ for i in range(prev + 1, current):
|
|
|
+ filled_result.append(i)
|
|
|
+ filled_result.append(current)
|
|
|
+ prev = current
|
|
|
+ return filled_result
|
|
|
return result
|
|
|
|
|
|
_cache = {}
|
