pdf/gateway/app/routers/documents/crud.py

import re
import httpx
from urllib.parse import urlparse
from fastapi import APIRouter, File, HTTPException, UploadFile, status
from pydantic import BaseModel

from app.schemas.document import (
    DocumentInfoResponse,
    PageInfoResponse,
    PermissionsResponse,
)
from app.services import engine
from app.services.store import document_store

router = APIRouter(prefix="/documents", tags=["documents"])


class RemoteImportRequest(BaseModel):
    stream_url: str
    upload_url: str
    auth_token: str | None = None
    resource_id: str | None = None


@router.post("/import-remote", response_model=DocumentInfoResponse, status_code=status.HTTP_201_CREATED)
async def import_remote_document(req: RemoteImportRequest) -> DocumentInfoResponse:
    if not engine.is_available():
        raise HTTPException(
            status_code=status.HTTP_501_NOT_IMPLEMENTED,
            detail="Engine bridge (bindings/python) not yet available.",
        )

    stream_parsed = urlparse(req.stream_url)
    upload_parsed = urlparse(req.upload_url)
    if stream_parsed.scheme not in ("http", "https") or upload_parsed.scheme not in ("http", "https"):
        raise HTTPException(status_code=400, detail="Invalid URL scheme. Only http/https are allowed.")

    headers = {}
    if req.auth_token:
        headers["Authorization"] = f"Bearer {req.auth_token}"

    MAX_SIZE = 50 * 1024 * 1024  # 50 MB limit
    async with httpx.AsyncClient(timeout=60.0, follow_redirects=True) as client:
        try:
            async with client.stream("GET", req.stream_url, headers=headers) as res:
                if res.status_code != 200:
                    await res.aread()
                    raise HTTPException(
                        status_code=res.status_code,
                        detail=f"Failed to stream remote document: {res.text[:200]}",
                    )
                
                content_length = res.headers.get("Content-Length")
                if content_length and int(content_length) > MAX_SIZE:
                    raise HTTPException(status_code=400, detail="Document exceeds maximum allowed size (50MB).")
                
                bytes_data_arr = bytearray()
                async for chunk in res.aiter_bytes():
                    bytes_data_arr.extend(chunk)
                    if len(bytes_data_arr) > MAX_SIZE:
                        raise HTTPException(status_code=400, detail="Document exceeds maximum allowed size (50MB).")
                
                bytes_data = bytes(bytes_data_arr)
        except Exception as err:
            if isinstance(err, HTTPException):
                raise err
            raise HTTPException(
                status_code=status.HTTP_400_BAD_REQUEST,
                detail=f"Failed to connect to remote stream_url: {err!s}",
            )

        cd = res.headers.get("Content-Disposition", "")
        match = re.search(r'filename="?([^";]+)"?', cd)
        filename = match.group(1) if match else f"remote_document_{req.resource_id or 'file'}.pdf"

    try:
        pdfengine = engine.require()
        doc = pdfengine.PdfDocument.load_from_memory(bytes_data, "")
        info = document_store.add_document(filename, bytes_data, doc)
        info["remote_context"] = {
            "stream_url": req.stream_url,
            "upload_url": req.upload_url,
            "auth_token": req.auth_token,
            "resource_id": req.resource_id,
        }
        return make_document_response(info)
    except Exception as e:
        raise HTTPException(
            status_code=status.HTTP_400_BAD_REQUEST,
            detail=f"Failed to load remote PDF into engine: {e!s}",
        )


def make_document_response(d: dict) -> DocumentInfoResponse:
    pages_list = []
    if "doc_instance" in d:
        doc = d["doc_instance"]
        for i in range(doc.page_count):
            try:
                page = doc.get_page(i)
                pages_list.append(PageInfoResponse(index=i, width=page.width, height=page.height))
            except Exception:
                pass
    perms = d.get("permissions")
    return DocumentInfoResponse(
        id=d["id"],
        filename=d["filename"],
        sizeBytes=d["sizeBytes"],
        totalPages=d["totalPages"],
        pageWidth=d.get("pageWidth", 612.0),
        pageHeight=d.get("pageHeight", 792.0),
        uploadedAt=d["uploadedAt"],
        status=d["status"],
        pages=pages_list,
        permissions=PermissionsResponse(**perms) if perms else PermissionsResponse(),
    )


@router.post("", response_model=DocumentInfoResponse, status_code=status.HTTP_201_CREATED)
async def upload_document(file: UploadFile = File(...), password: str = "") -> DocumentInfoResponse:
    if not engine.is_available():
        raise HTTPException(
            status_code=status.HTTP_501_NOT_IMPLEMENTED,
            detail="Engine bridge (bindings/python) not yet available.",
        )

    bytes_data = await file.read()
    try:
        pdfengine = engine.require()
        doc = pdfengine.PdfDocument.load_from_memory(bytes_data, password)
        info = document_store.add_document(file.filename, bytes_data, doc)
        return make_document_response(info)
    except ValueError as e:
        detail = str(e)
        if "Password required" in detail:
            raise HTTPException(
                status_code=status.HTTP_401_UNAUTHORIZED, detail="Password required"
            )
        elif "Invalid password" in detail:
            raise HTTPException(status_code=status.HTTP_401_UNAUTHORIZED, detail="Invalid password")
        else:
            raise HTTPException(status_code=status.HTTP_400_BAD_REQUEST, detail=detail)
    except Exception as e:
        raise HTTPException(
            status_code=status.HTTP_400_BAD_REQUEST, detail=f"Failed to load PDF: {e!s}"
        )


@router.get("", response_model=list[DocumentInfoResponse])
def list_documents() -> list[DocumentInfoResponse]:
    if not engine.is_available():
        raise HTTPException(
            status_code=status.HTTP_501_NOT_IMPLEMENTED,
            detail="Engine bridge (bindings/python) not yet available.",
        )

    docs = document_store.list_documents()
    return [make_document_response(d) for d in docs]


@router.get("/{document_id}", response_model=DocumentInfoResponse)
def get_document(document_id: str) -> DocumentInfoResponse:
    if not engine.is_available():
        raise HTTPException(
            status_code=status.HTTP_501_NOT_IMPLEMENTED,
            detail="Engine bridge (bindings/python) not yet available.",
        )

    d = document_store.get_document(document_id)
    if not d:
        raise HTTPException(status_code=status.HTTP_404_NOT_FOUND, detail="Document not found")

    return make_document_response(d)


@router.delete("/{document_id}")
def delete_document(document_id: str):
    if not engine.is_available():
        raise HTTPException(
            status_code=status.HTTP_501_NOT_IMPLEMENTED,
            detail="Engine bridge (bindings/python) not yet available.",
        )

    deleted = document_store.delete_document(document_id)
    if not deleted:
        raise HTTPException(status_code=status.HTTP_404_NOT_FOUND, detail="Document not found")

    return {"success": True}