HayStackResponse

HayStack RAG API response model.

Since Haystack does not provide a standard REST API format, the API response of your custom-built Haystack RAG server should conform to the following format.

** For Normal **

`{
"code": 0,
"documents": [
{
"id": "0fab3c1c6c433368",
"content": "Test Content",
"blob": null,
"meta": {
"file_path": "test.txt"
},
"score": 1.2303546667099,
"embedding": [
0.01008153147995472,
-0.04555170238018036,
-0.024434546008706093
],
"sparse_embedding": null,
"error": null
}
]
}
`

You can build your Haystack RAG server like this:

` @app.post("/retrieve")
async def retriever(req: Request):
result = retriever.run(
query=req.query,
top_k=req.top_k,
)
documents = result["documents"]

return {
"code": 0,
"documents": documents,
"error": None
}
`

** For SentenceWindowRetriever **

` {
"code": 0,
"context_windows": [
"Test Content 0 "
],
"context_documents": [
{
"id": "9e4cd66e14377767fdc6b9",
"content": "Test Content 1",
"meta": {
"source": "usr_90.txt",
"source_id": "1528904be8e57",
"page_number": 1,
"split_id": 21,
"split_idx_start": 11199,
"_split_overlap": [
{
"doc_id": "55ee242643b2439c8",
"range": [433, 570]
},
{
"doc_id": "cdc785a04643cbd",
"range": [0, 146]
}
]
}
}
]
}
`

You can build your Haystack RAG server like this:

` @app.post("/retrieve")
async def run_retriever(req: RetrieverRequest):
retrieved = bm25.run(
query=req.query,
top_k=req.top_k
)
retrieved_documents = retrieved["documents"]

sentence_window = SentenceWindowRetriever(document_store=doc_store)

result = sentence_window.run(
retrieved_documents=retrieved_documents,
window_size=req.window_size
)

return {
"code": 0,
"context_windows": result["context_windows"],
"context_documents": result["context_documents"],
"error": None
}
`