feat: read file and validate opossum file structure

Author: Hellgartner

src/opossum_lib/cli.py

@@ -80,8 +80,8 @@ def convert_after_valid_input(
         the_spdx_file = spdx[0]
         return convert_spdx_to_opossum_information(the_spdx_file)
     else:
-        opossum_file = opossum_files[0]
-        return read_opossum_file(opossum_file)
+        opossum_input_file = opossum_files[0]
+        return read_opossum_file(opossum_input_file)
 
 
 if __name__ == "__main__":

src/opossum_lib/opossum/read_opossum_file.py

@@ -2,7 +2,9 @@
 #
 # SPDX-License-Identifier: Apache-2.0
 import logging
+import sys
 from datetime import datetime
+from zipfile import ZipFile
 
 from opossum_lib.opossum.opossum_file import (
     Metadata,
@@ -14,6 +16,29 @@
 
 def read_opossum_file(filename: str) -> OpossumInformation:
     logging.info(f"Converting opossum to opossum {filename}")
+
+    try:
+        with (
+            ZipFile(filename, "r") as input_zip_file,
+        ):
+            if "input.json" not in input_zip_file.namelist():
+                logging.error(
+                    f"Opossum file {filename} is corrupt"
+                    f" and does not contain 'input.json'"
+                )
+                sys.exit(1)
+            if "output.json" in input_zip_file.namelist():
+                logging.error(
+                    f"Opossum file {filename} also contains"
+                    f" 'output.json' which cannot be processed"
+                )
+                sys.exit(1)
+
+
+    except Exception as e:
+        # handle the exception
+        print(f"Error reading file {filename}: {e}")
+
     dummy_metadata = Metadata(
         projectId="test id",
         fileCreationDate=datetime.now().isoformat(),

src/opossum_lib/spdx/convert_to_opossum.py

@@ -49,6 +49,7 @@
 
 
 def convert_spdx_to_opossum_information(filename: str) -> OpossumInformation:
+    logging.info(f"Converting {filename} to opossum information.")
     try:
         document: SpdxDocument = parse_file(filename)
 

tests/data/opossum_input.opossum

@@ -0,0 +1,35 @@
+# SPDX-FileCopyrightText: TNG Technology Consulting GmbH <https://www.tngtech.com>
+#
+# SPDX-License-Identifier: Apache-2.0
+from pathlib import Path
+
+import pytest
+from _pytest.logging import LogCaptureFixture
+
+from opossum_lib.opossum.read_opossum_file import read_opossum_file
+
+
+def test_read_opossum_file_corrupted_file_exits_1(caplog: LogCaptureFixture) -> None:
+    input_path = (
+        Path(__file__).resolve().parent.parent
+        / "data"
+        / "opossum_input_corrupt.opossum"
+    )
+
+    with pytest.raises(SystemExit) as system_exit:
+        read_opossum_file(str(input_path))
+    assert system_exit.value.code == 1
+    assert "is corrupt and does not contain 'input.json'" in caplog.messages[0]
+
+
+def test_read_opossum_file_with_result_json_exits_1(caplog: LogCaptureFixture) -> None:
+    input_path = (
+        Path(__file__).resolve().parent.parent
+        / "data"
+        / "opossum_input_with_result.opossum"
+    )
+
+    with pytest.raises(SystemExit) as system_exit:
+        read_opossum_file(str(input_path))
+    assert system_exit.value.code == 1
+    assert "also contains 'output.json' which cannot be processed" in caplog.messages[0]