VertexAISearchRetriever return Japanese documents as Unicode

[Freezaa9]

HI,

We are currently benchmarking VertexAISearchRetriever against VectorSearchVectorStore

And we are working with document in japanese.

With VectorSearchVectorStore retriever we can retrieve the japanese document without problem,
but when we switch to VertexAISearchRetriever the page content in japanese is in unicode. While in the Vertex ai Search console in GCP we can visualize the document in normal Japanese.

The data store is created with one structuted jsonl file from GCS.

Code example:

query = "how does dior sauvage smell like ?"

retriever = VertexAISearchRetriever(    
    project_id=PROJECT_ID,
    location_id=LOCATION_ID,

    data_store_id=DATA_STORE_ID,
    max_documents=10,
    engine_data_type=1,
)

result = retriever.invoke(query)
for doc in result:
    print(doc)
    print(type(doc))
    print(doc.page_content)

What we retrieve:

page_content='{"content": "\u8a73\u7d30: \n\u30bd\u30f4\u30a1\u30fc\u30b8\u30e5 \u30d1\u30eb\u30d5\u30a1\u30f3\u306f\u3001\u3044\u3063\u305d\u3046\u6fc3\u539a\u306b\u8868\u73fe\u3055"}' metadata={'id': '5e73bbc4b8dc180d0836ab21d1f9b0ac', 'name': 'projects/XXXXXX/locations/XXXX/collections/default_collection/dataStores/XXXX/branches/0/documents/5e73bbc4b8dc180d0836ab21d1f9b0ac'}
<class 'langchain_core.documents.base.Document'>

Thanks for your help