From ac93571dbccf1e220c58cddbae3d98d99d1f0517 Mon Sep 17 00:00:00 2001
From: Jakub Onderka <jakub.onderka@gmail.com>
Date: Sun, 25 Sep 2022 17:06:00 +0200
Subject: [PATCH] fix: Better validation for taxonomy files

---
 validate_all.py | 18 ++++++++++++++++++
 validate_all.sh |  7 +------
 2 files changed, 19 insertions(+), 6 deletions(-)
 create mode 100644 validate_all.py

diff --git a/validate_all.py b/validate_all.py
new file mode 100644
index 0000000..8d8359b
--- /dev/null
+++ b/validate_all.py
@@ -0,0 +1,18 @@
+import sys
+import glob
+import json
+from jsonschema import validate
+
+schema = json.load(open("schema.json", "r"))
+
+for taxonomy_file in glob.glob('./*/machinetag.json'):
+    print("Checking {}".format(taxonomy_file))
+    taxonomy = json.load(open(taxonomy_file, "r"))
+    validate(instance=taxonomy, schema=schema)
+
+    if "values" in taxonomy:
+        predicates = [predicate["value"] for predicate in taxonomy["predicates"]]
+        for value in taxonomy["values"]:
+            if value["predicate"] not in predicates:
+                print("ERROR: Predicate `{}` is missing".format(value["predicate"]))
+                sys.exit(1)
\ No newline at end of file
diff --git a/validate_all.sh b/validate_all.sh
index 0f4a1f0..025ccdf 100755
--- a/validate_all.sh
+++ b/validate_all.sh
@@ -20,11 +20,6 @@ if ! [ $((directories-2)) -eq $manifest_entries ]; then
     exit 1
 fi
 
-for dir in */machinetag.json
-do
-  echo -n "${dir}: "
-  jsonschema -i ${dir} schema.json
-  echo ''
-done
+python3 validate_all.py
 
 jsonschema -i mapping/mapping.json schema_mapping.json