Add checksums into UI JSON, and verify them

sacro/views.py

@@ -1,4 +1,5 @@
 import getpass
+import hashlib
 import html
 import json
 import logging
@@ -44,6 +45,10 @@ def __post_init__(self):
 
         self.version = self.raw_metadata["version"]
         self.update(self.raw_metadata["results"])
+        self.annotate()
+
+    def annotate(self):
+        """Add various useful annotations to the JSON data"""
 
         # add urls to JSON data
         for output, metadata in self.items():
@@ -57,6 +62,25 @@ def __post_init__(self):
                     "contents",
                 )
 
+        # add and check checksum data
+        checksums_dir = self.path.parent / "checksums"
+        for output, metadata in self.items():
+            for filedata in metadata["files"]:
+                filedata["checksum_valid"] = None
+                filedata["checksum"] = None
+
+                path = checksums_dir / (filedata["name"] + ".txt")
+                if not path.exists():
+                    continue
+
+                filedata["checksum"] = path.read_text(encoding="utf8")
+                actual_file = self.get_file_path(output, filedata["name"])
+                if not actual_file.exists():
+                    continue
+
+                checksum = hashlib.sha256(actual_file.read_bytes()).hexdigest()
+                filedata["checksum_valid"] = checksum == filedata["checksum"]
+
     def get_file_path(self, output, filename):
         """Return absolute path to output file"""
         if filename not in {

tests/test_views.py

@@ -22,6 +22,28 @@ def test_outputs(tmp_path):
     return views.Outputs(tmp_path / TEST_PATH.name)
 
 
+def test_outputs_annotation(test_outputs):
+    assert test_outputs.version == "0.4.0"
+    for metadata in test_outputs.values():
+        for filedata in metadata["files"]:
+            assert filedata["checksum"] is not None
+            assert filedata["checksum_valid"] is True
+            assert filedata["url"].startswith("/contents/?path=")
+
+
+def test_outputs_annotation_checksum_failed(test_outputs):
+    first_output = list(test_outputs)[0]
+    first_file = test_outputs[first_output]["files"][0]["name"]
+    checksum = test_outputs.path.parent / "checksums" / (first_file + ".txt")
+    checksum.write_text("bad checksum")
+
+    # re-annotate
+    test_outputs.annotate()
+
+    assert test_outputs[first_output]["files"][0]["checksum"] == "bad checksum"
+    assert test_outputs[first_output]["files"][0]["checksum_valid"] is False
+
+
 def test_index(test_outputs):
     request = RequestFactory().get(path="/", data={"path": str(test_outputs.path)})