🐛 trim bad characters from the start of k8s labels (#104)

Author: danielgafni

dagster_ray/kuberay/utils.py

@@ -8,8 +8,13 @@ def normalize_k8s_label_values(labels: dict[str, str]) -> dict[str, str]:
 
     cleanup_regex = re.compile(r"[^a-zA-Z0-9-_.]+")
 
+    banned_starting_characters = ["-", "_", "."]
+
     for key, value in labels.items():
-        # daniel~!@my.domain -> daniel-my-domain
-        labels[key] = cleanup_regex.sub("", value.replace("@", "-").replace(".", "-"))[:63]
+        # -daniel~!@my.domain -> daniel-my-domain
+        with_maybe_bad_start = cleanup_regex.sub("", value.replace("@", "-").replace(".", "-"))
+        while with_maybe_bad_start and with_maybe_bad_start[0] in banned_starting_characters:
+            with_maybe_bad_start = with_maybe_bad_start[1:]
+        labels[key] = with_maybe_bad_start[:63]
 
     return labels

tests/kuberay/test_utils.py

@@ -10,6 +10,8 @@ def test_normalize_k8s_label_values():
             "user-dirty": "daniel!`~@my.org",
             "alphanumeric": "abc123",
             "long": 64 * "a",
+            "badstart": "-foo",
+            "badstart_after_initial_replace": "@foo",
         }
     ) == {
         "foo": "bar",
@@ -18,4 +20,6 @@ def test_normalize_k8s_label_values():
         "user-dirty": "daniel-my-org",
         "alphanumeric": "abc123",
         "long": 63 * "a",
+        "badstart": "foo",
+        "badstart_after_initial_replace": "foo",
     }