test: unit and integration test pdf check thrown error code and message

Author: mpolomdeepsense

sample-docs/failing-encrypted.pdf

@@ -1155,3 +1155,45 @@ def test_include_slide_notes(monkeypatch, test_default, include_slide_notes, tes
         assert "Here are important notes" == df["text"][0]
     else:
         assert "Here are important notes" != df["text"][0]
+
+
+@pytest.mark.parametrize(
+    ("pdf_name", "expected_error_message"),
+    [
+        ("failing-encrypted.pdf", "File is encrypted. Please decrypt it with password."),
+        (
+            "failing-invalid.pdf",
+            "File does not appear to be a valid PDF. Error: Stream has ended unexpectedly",
+        ),
+        (
+            "failing-missing-root.pdf",
+            "File does not appear to be a valid PDF. Error: Cannot find Root object in pdf",
+        ),
+        (
+            "failing-missing-pages.pdf",
+            "File does not appear to be a valid PDF. Error: Invalid object in /Pages",
+        ),
+    ],
+)
+@pytest.mark.parametrize(
+    "strategy",
+    [
+        "auto",
+        "fast",
+        "hi_res",
+        "ocr_only",
+    ],
+)
+def test_failing_pdfs_return_422(pdf_name: str, expected_error_message: str, strategy: str):
+    client = TestClient(app)
+    test_file = Path(__file__).parent.parent.parent / "sample-docs" / pdf_name
+
+    with open(test_file, "rb") as f:
+        response = client.post(
+            MAIN_API_ROUTE,
+            files=[("files", (str(test_file), f))],
+            data={"strategy": strategy},
+        )
+
+    assert response.status_code == 422
+    assert expected_error_message == str(response.json()["detail"])

sample-docs/failing-invalid.pdf

@@ -0,0 +1,55 @@
+from __future__ import annotations
+
+import io
+from pathlib import Path
+
+import pytest
+from fastapi import HTTPException
+from pypdf import PdfReader
+
+from prepline_general.api.general import _check_pdf
+
+TEST_ASSETS_DIR = Path(__file__).parent.parent.parent / "sample-docs"
+
+
+def _open_pdf(pdf_path: str) -> io.BytesIO:
+    with open(pdf_path, "rb") as f:
+        pdf_content = f.read()
+    return io.BytesIO(pdf_content)
+
+
+def test_check_pdf_with_valid_pdf():
+    pdf_path = str(TEST_ASSETS_DIR / "list-item-example.pdf")
+    pdf = _open_pdf(pdf_path)
+
+    result = _check_pdf(pdf)
+    assert isinstance(result, PdfReader)
+
+
+@pytest.mark.parametrize(
+    ("pdf_name", "expected_error_message"),
+    [
+        ("failing-encrypted.pdf", "File is encrypted. Please decrypt it with password."),
+        (
+            "failing-invalid.pdf",
+            "File does not appear to be a valid PDF. Error: Stream has ended unexpectedly",
+        ),
+        (
+            "failing-missing-root.pdf",
+            "File does not appear to be a valid PDF. Error: Cannot find Root object in pdf",
+        ),
+        (
+            "failing-missing-pages.pdf",
+            "File does not appear to be a valid PDF. Error: Invalid object in /Pages",
+        ),
+    ],
+)
+def test_check_pdf_with_invalid_pdf(pdf_name: str, expected_error_message: str):
+    pdf_path = str(TEST_ASSETS_DIR / pdf_name)
+    pdf = _open_pdf(pdf_path)
+
+    with pytest.raises(HTTPException) as exc_info:
+        _check_pdf(pdf)
+
+    assert exc_info.value.status_code == 422
+    assert expected_error_message == str(exc_info.value.detail)