backend/app/api/documents.py

"""Document endpoints: upload, list, detail, page images, re-process, delete."""
from __future__ import annotations

import uuid
from datetime import datetime, timezone

from fastapi import APIRouter, BackgroundTasks, File, HTTPException, UploadFile
from fastapi.responses import FileResponse

from app.core.config import settings
from app.core.logging import get_logger
from app.schemas.documents import DocStatus, DocumentDetail, DocumentMeta
from app.services import storage, vectorstore
from app.services.pipeline import process_document

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

ALLOWED = {
    ".pdf", ".docx", ".doc", ".pptx", ".xlsx", ".html", ".md", ".txt",
    ".png", ".jpg", ".jpeg", ".tiff", ".bmp",
}


@router.post("", response_model=DocumentMeta)
async def upload(background: BackgroundTasks, file: UploadFile = File(...)):
    ext = "." + (file.filename or "").rsplit(".", 1)[-1].lower() if "." in (file.filename or "") else ""
    if ext not in ALLOWED:
        raise HTTPException(400, f"Unsupported file type '{ext}'. Allowed: {sorted(ALLOWED)}")

    doc_id = uuid.uuid4().hex[:12]
    dest = settings.uploads_path / f"{doc_id}{ext}"
    data = await file.read()
    dest.write_bytes(data)

    detail = DocumentDetail(
        id=doc_id,
        filename=file.filename or dest.name,
        content_type=file.content_type or "application/octet-stream",
        size_bytes=len(data),
        status=DocStatus.uploaded,
        created_at=datetime.now(timezone.utc).isoformat(),
    )
    storage.save(detail)

    # kick off async processing
    background.add_task(process_document, doc_id, dest)
    log.info("Uploaded %s (%d bytes) -> %s", file.filename, len(data), doc_id)
    return DocumentMeta.model_validate(detail.model_dump())


@router.get("", response_model=list[DocumentMeta])
async def list_documents():
    return storage.list_all()


@router.get("/{doc_id}", response_model=DocumentDetail)
async def get_document(doc_id: str):
    detail = storage.get(doc_id)
    if not detail:
        raise HTTPException(404, "Document not found")
    return detail


@router.get("/{doc_id}/pages/{page}.png")
async def get_page_image(doc_id: str, page: int):
    path = settings.renders_path / doc_id / f"page-{page}.png"
    if not path.exists():
        raise HTTPException(404, "Page image not found")
    return FileResponse(path, media_type="image/png")


@router.delete("/{doc_id}")
async def delete_document(doc_id: str):
    if not storage.get(doc_id):
        raise HTTPException(404, "Document not found")
    vectorstore.delete(doc_id)
    storage.delete(doc_id)
    return {"ok": True}