tests/test_uploads_pipeline.py

"""End-to-end smoke test for the D-light upload pipeline.

Stays local: drives the FastAPI app via Starlette TestClient and exercises
the unit-level pipeline helpers. Skips the LLM `extract-text` body so we
don't need a live NVIDIA key — the offline fallback is what we want here.
"""

from __future__ import annotations

import io
from pathlib import Path

from fastapi.testclient import TestClient

from app.services.proj_doc_pipeline import (
    IngestArtifact,
    collect_text_window,
    extract_pages_text,
    stamp_index,
)

TEST_PDF = Path(__file__).parent / "test.pdf"


def test_pipeline_helpers_text_extraction():
    """Pure pipeline layer: native text + stamp heuristic on tests/test.pdf."""
    if not TEST_PDF.exists():
        # No fixture PDF — skip; the assertion below is meant to be smoke.
        return
    raw = TEST_PDF.read_bytes()
    artifact = extract_pages_text(raw)
    assert artifact.sha256, "sha256 must be present"
    assert artifact.page_count >= 1, "at least one page expected"
    assert isinstance(artifact.pages, list)
    # Stamp index is keyed on label incl. "(стр. N)" suffix.
    stamps = stamp_index(artifact.pages)
    assert isinstance(stamps, dict)
    # Window helper must return a non-empty string for any page.
    if artifact.pages:
        window = collect_text_window(
            artifact.pages, around=0, span=2
        )
        assert window and isinstance(window, str)


def test_pipeline_artifact_default_empty_text():
    """Direct call against the dataclass — no IO."""
    artifact = IngestArtifact(
        file_name="x.pdf",
        mime_type="application/pdf",
        sha256="0" * 64,
        page_count=0,
    )
    assert artifact.full_text(1) == ""
    assert artifact.has_native_text is False


def test_uploads_blueprint_endpoint_accepts_pdf(minimal_app_client: TestClient):
    """POST /uploads/blueprint → 200, page list non-empty (no Supabase)."""
    if not TEST_PDF.exists():
        return
    client = minimal_app_client
    files = {"file": ("test.pdf", TEST_PDF.read_bytes(), "application/pdf")}
    resp = client.post("/uploads/blueprint", files=files)
    assert resp.status_code == 200, resp.text
    body = resp.json()
    assert body["file_name"] == "test.pdf"
    assert body["page_count"] >= 1
    assert isinstance(body["pages"], list)
    assert body["stored_in_supabase"] is False  # no SB creds in tests

    status = client.get(
        f"/uploads/blueprint/{body['upload_id']}"
    )
    assert status.status_code in (200, 404), status.text
    # If Supabase isn't reachable we fall back to the on-disk cache → 200.
    if status.status_code == 200:
        s = status.json()
        assert s["upload_id"] == body["upload_id"]
        assert s["page_count"] == body["page_count"]


def test_uploads_blueprint_rejects_oversize(minimal_app_client: TestClient):
    client = minimal_app_client
    payload = b"x" * (51 * 1024 * 1024)  # 51 MB > MAX_BYTES=50MB
    files = {
        "file": ("huge.pdf", io.BytesIO(payload), "application/pdf")
    }
    resp = client.post("/uploads/blueprint", files=files)
    assert resp.status_code == 413
    assert "too large" in resp.json()["detail"]


def test_uploads_blueprint_rejects_unsupported_mime(minimal_app_client: TestClient):
    client = minimal_app_client
    files = {
        "file": ("odd.avi", b"abcd", "video/x-msvideo")
    }
    resp = client.post("/uploads/blueprint", files=files)
    assert resp.status_code == 415