Update app/data_indexing.py
Browse files- app/data_indexing.py +3 -15
app/data_indexing.py
CHANGED
|
@@ -91,31 +91,19 @@ class DataIndexer:
|
|
| 91 |
except Exception as e:
|
| 92 |
print(e)
|
| 93 |
|
| 94 |
-
def search(
|
| 95 |
-
|
| 96 |
-
print('text query:', text_query)
|
| 97 |
-
|
| 98 |
-
filter = None
|
| 99 |
-
if hybrid_search and self.source_index:
|
| 100 |
-
source_docs = self.source_index.similarity_search(text_query, 50)
|
| 101 |
-
print("source_docs", source_docs)
|
| 102 |
-
filter = {"source": {"$in":[doc.page_content for doc in source_docs]}}
|
| 103 |
-
|
| 104 |
-
# vector = self.embedding_client.feature_extraction(text_query)
|
| 105 |
vector = self.embedding_client.embed_query(text_query)
|
| 106 |
result = self.index.query(
|
| 107 |
vector=vector,
|
| 108 |
top_k=top_k,
|
| 109 |
include_metadata=True,
|
| 110 |
-
filter=filter
|
| 111 |
)
|
| 112 |
|
| 113 |
docs = []
|
| 114 |
for res in result["matches"]:
|
| 115 |
metadata = res["metadata"]
|
| 116 |
-
|
| 117 |
-
|
| 118 |
-
docs.append(text)
|
| 119 |
return docs
|
| 120 |
|
| 121 |
|
|
|
|
| 91 |
except Exception as e:
|
| 92 |
print(e)
|
| 93 |
|
| 94 |
+
def search(text_query, top_k=5):
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 95 |
vector = self.embedding_client.embed_query(text_query)
|
| 96 |
result = self.index.query(
|
| 97 |
vector=vector,
|
| 98 |
top_k=top_k,
|
| 99 |
include_metadata=True,
|
|
|
|
| 100 |
)
|
| 101 |
|
| 102 |
docs = []
|
| 103 |
for res in result["matches"]:
|
| 104 |
metadata = res["metadata"]
|
| 105 |
+
text = metadata.pop('text')
|
| 106 |
+
docs.append(text)
|
|
|
|
| 107 |
return docs
|
| 108 |
|
| 109 |
|