backend/app/api/actions.py

"""On-demand actions and export endpoints.

These let the frontend trigger individual functions (re-extract with a chosen
schema, re-summarize) and download results — independent of the chat agent.
"""
from __future__ import annotations

from fastapi import APIRouter, HTTPException, Query
from fastapi.responses import Response

from app.core.logging import get_logger
from app.schemas.documents import Classification, ExtractionResult
from app.schemas.templates import all_template_names
from app.services import (
    anomaly,
    classification,
    export,
    extraction,
    storage,
    summary,
    vectorstore,
)

log = get_logger(__name__)
router = APIRouter(prefix="/api/documents", tags=["actions"])


@router.post("/{doc_id}/extract", response_model=ExtractionResult)
async def re_extract(doc_id: str, schema: str | None = Query(None),
                     provider: str | None = Query(None)):
    detail = storage.get(doc_id)
    if not detail:
        raise HTTPException(404, "Document not found")
    chunks = await vectorstore.query(doc_id, "key fields", k=40)
    doc_type = detail.classification.doc_type if detail.classification else None
    # `schema` is an OPTIONAL user-chosen framing; "auto" (or unset) keeps the
    # extractor fully adaptive.
    hint = schema if (schema and schema.lower() != "auto") else None
    result = await extraction.extract(detail.markdown or "", doc_type, chunks, provider,
                                      schema_hint=hint)
    detail.extraction = result
    detail.anomalies = await anomaly.detect(detail.markdown or "", result, doc_type, provider)
    storage.save(detail)
    return result


@router.post("/{doc_id}/classify", response_model=Classification)
async def re_classify(doc_id: str, provider: str | None = Query(None)):
    detail = storage.get(doc_id)
    if not detail:
        raise HTTPException(404, "Document not found")
    c = await classification.classify(detail.markdown or "", provider)
    detail.classification = c
    storage.save(detail)
    return c


@router.post("/{doc_id}/summarize")
async def re_summarize(doc_id: str, provider: str | None = Query(None)):
    detail = storage.get(doc_id)
    if not detail:
        raise HTTPException(404, "Document not found")
    s = await summary.summarize(detail.markdown or "", provider)
    detail.summary = s
    storage.save(detail)
    return {"summary": s}


@router.get("/{doc_id}/export")
async def export_document(doc_id: str, format: str = Query("json")):
    detail = storage.get(doc_id)
    if not detail:
        raise HTTPException(404, "Document not found")
    fmt = format.lower()
    if fmt not in export.EXPORTERS:
        raise HTTPException(400, f"Unsupported format. Use one of {list(export.EXPORTERS)}")
    fn, media, ext = export.EXPORTERS[fmt]
    data = fn(detail)
    safe = (detail.filename or doc_id).rsplit(".", 1)[0]
    return Response(
        content=data, media_type=media,
        headers={"Content-Disposition": f'attachment; filename="{safe}.{ext}"'},
    )


@router.get("/-/schemas")
async def list_schemas():
    return {"schemas": all_template_names()}