app.py

# infra/hf_spaces/reranker/app.py
# Serves cross-encoder/ms-marco-MiniLM-L-6-v2 reranking over HTTP.
# Model is loaded from /app/model_cache (baked into the Docker image at build time).

from contextlib import asynccontextmanager
from typing import Annotated

from fastapi import FastAPI
from pydantic import BaseModel, Field
from sentence_transformers import CrossEncoder

# Pipeline retrieves top-20, reranks to top-5. Cap at 64 to guard against abuse.
_MAX_TEXTS = 64
_MAX_QUERY_LEN = 512



class RerankRequest(BaseModel):
    query: str = Field(..., max_length=_MAX_QUERY_LEN)
    # Remove max_length on individual strings — token truncation handles long text safely
    texts: list[str] = Field(..., max_length=_MAX_TEXTS)
    top_k: int = Field(5, ge=1, le=20)


class RerankResponse(BaseModel):
    # Indices into the input texts list, sorted by descending relevance.
    indices: list[int]
    scores: list[float]


@asynccontextmanager
async def lifespan(app: FastAPI):
    app.state.model = CrossEncoder(
        "cross-encoder/ms-marco-MiniLM-L-6-v2",
        cache_folder="/app/model_cache",
    )
    yield
    app.state.model = None


app = FastAPI(
    title="PersonaBot Reranker",
    lifespan=lifespan,
    docs_url=None,
    redoc_url=None,
    openapi_url=None,
)


@app.get("/health")
async def health() -> dict[str, str]:
    if app.state.model is None:
        return {"status": "loading"}
    return {"status": "ok"}


@app.post("/rerank", response_model=RerankResponse)
async def rerank(request: RerankRequest) -> RerankResponse:
    if not request.texts:
        return RerankResponse(indices=[], scores=[])

    pairs = [(request.query, text) for text in request.texts]
    raw_scores: list[float] = [float(s) for s in app.state.model.predict(pairs)]

    # Sort by score descending, return top_k
    ranked = sorted(enumerate(raw_scores), key=lambda x: x[1], reverse=True)
    ranked = ranked[: request.top_k]

    return RerankResponse(
        indices=[i for i, _ in ranked],
        scores=[s for _, s in ranked],
    )