|
|
@@ -297,6 +297,13 @@ def tree_reposition(tree,sentence_index=None):
|
|
|
obj["wordOffset_end"] = wordOffset_end
|
|
|
wordOffset_begin = wordOffset_end
|
|
|
|
|
|
+ list_kv = obj.get("kv",[])
|
|
|
+ for _d in list_kv:
|
|
|
+ _d["position"]["key_begin_sentence"] = sentence_index
|
|
|
+ _d["position"]["key_end_sentence"] = sentence_index
|
|
|
+ _d["position"]["value_begin_sentence"] = sentence_index
|
|
|
+ _d["position"]["value_end_sentence"] = sentence_index
|
|
|
+
|
|
|
else:
|
|
|
sentence_index += 1
|
|
|
obj["sentence_index"] = sentence_index
|
|
|
@@ -1720,7 +1727,7 @@ if __name__ == '__main__':
|
|
|
_pd = Html2KVTree(html_content)
|
|
|
_pd.print_tree(_pd.tree,"-|")
|
|
|
|
|
|
- list_kv = _pd.extract_kv("资格条件")
|
|
|
+ list_kv = _pd.extract_kv("资质要求")
|
|
|
print(list_kv)
|
|
|
|
|
|
#获取预处理后的所有句子，该句子与kv值对应
|
luojiehua