From d81b3f62823b1961d4473fbd8fb4f5d17de0902b Mon Sep 17 00:00:00 2001
From: Simon Davy <simon.davy@thedatalab.org>
Date: Mon, 16 Oct 2023 17:13:57 +0100
Subject: [PATCH] Add very basic json validation.

---
 sacro/models.py      | 17 +++++++++++++++++
 tests/test_models.py | 24 ++++++++++++++++++++++++
 2 files changed, 41 insertions(+)

diff --git a/sacro/models.py b/sacro/models.py
index 87132bd..6384ae0 100644
--- a/sacro/models.py
+++ b/sacro/models.py
@@ -20,6 +20,9 @@ def load_from_path(path):
 class ACROOutputs(dict):
     """An ACRO json output file"""
 
+    class InvalidFile(Exception):
+        pass
+
     path: Path
     version: str = None
     config: dict = field(default_factory=dict)
@@ -30,6 +33,20 @@ def __post_init__(self):
         if config_path.exists():
             self.config = json.loads(config_path.read_text())
 
+        # super basic structural validation
+        try:
+            assert "version" in self.raw_metadata
+            assert "results" in self.raw_metadata
+            assert len(self.raw_metadata["results"]) > 0
+            for result in self.raw_metadata["results"].values():
+                assert "files" in result
+                assert len(result["files"]) > 0
+                for filedata in result["files"]:
+                    assert "name" in filedata
+
+        except AssertionError as exc:
+            raise self.InvalidFile(f"{self.path} is not a valid ACRO json file: {exc}")
+
         self.version = self.raw_metadata["version"]
         self.update(self.raw_metadata["results"])
         self.annotate()
diff --git a/tests/test_models.py b/tests/test_models.py
index f957a2c..1135fbd 100644
--- a/tests/test_models.py
+++ b/tests/test_models.py
@@ -1,3 +1,10 @@
+import json
+
+import pytest
+
+from sacro import models
+
+
 def test_outputs_annotation(test_outputs):
     assert test_outputs.version == "0.4.0"
     for metadata in test_outputs.values():
@@ -30,3 +37,20 @@ def test_outputs_annotation_checksum_failed(test_outputs):
 
     assert test_outputs[first_output]["files"][0]["checksum"] == "bad checksum"
     assert test_outputs[first_output]["files"][0]["checksum_valid"] is False
+
+
+@pytest.mark.parametrize(
+    "data",
+    [
+        {},
+        {"version": "1"},
+        {"version": "1", "results": {}},
+        {"version": "1", "results": {"name": {"files": []}}},
+        {"version": "1", "results": {"name": {"notfiles": "foo"}}},
+    ],
+)
+def test_validation(data, tmp_path):
+    bad_json = tmp_path / "bad.json"
+    bad_json.write_text(json.dumps(data))
+    with pytest.raises(models.ACROOutputs.InvalidFile):
+        models.ACROOutputs(bad_json)