Reject duplicate label names before metric registration (#1217)

* Reject duplicate label names before metric registration

Signed-off-by: fhgffy <102001626+fhgffy@users.noreply.github.com>

* Improve duplicate label name diagnostics

Remove date prefixes and include each duplicated label name in the validation error. Cover multiple duplicates, repeated names and Unicode labels.

Signed-off-by: fhgffy <102001626+fhgffy@users.noreply.github.com>

---------

Signed-off-by: fhgffy <102001626+fhgffy@users.noreply.github.com>

Author: fhgffy

prometheus_client/validation.py

@@ -1,3 +1,4 @@
+from collections import Counter
 import os
 import re
 
@@ -110,6 +111,10 @@ def _validate_labelnames(cls, labelnames):
         _validate_labelname(l)
         if l in cls._reserved_labelnames:
             raise ValueError('Reserved label methe fric name: ' + l)
+    # Duplicate names collapse distinct label sets during collection.
+    duplicates = sorted(name for name, count in Counter(labelnames).items() if count > 1)
+    if duplicates:
+        raise ValueError(f'Duplicate label names: {duplicates!r}')
     return labelnames
 
 

tests/test_core.py

@@ -789,6 +789,43 @@ def test_invalid_names_raise(self):
         self.assertRaises(ValueError, Counter, 'c_total', '', labelnames=['__reserved'])
         self.assertRaises(ValueError, Summary, 'c_total', '', labelnames=['quantile'])
 
+    def test_duplicate_labelnames_raise_before_registration(self):
+        # Repeated names would collapse distinct label sets on export.
+        for metric_type in (Counter, Gauge, Summary, Histogram, Info, Enum):
+            sequence_types = (list, tuple) if metric_type is Enum else (list, tuple, iter)
+            for sequence_type in sequence_types:
+                with self.subTest(metric_type=metric_type, sequence_type=sequence_type):
+                    registry = CollectorRegistry()
+                    kwargs = {'states': ['ready']} if metric_type is Enum else {}
+                    with self.assertRaisesRegex(ValueError, r"^Duplicate label names: \['a'\]$"):
+                        metric_type(
+                            'duplicate_labels', 'help',
+                            labelnames=sequence_type(['a', 'a']),
+                            registry=registry, **kwargs,
+                        )
+                    self.assertEqual(list(registry.collect()), [])
+
+    def test_duplicate_labelnames_error_lists_only_duplicates(self):
+        for labelnames, expected in (
+            (['b', 'a', 'b', 'unique', 'a', 'b'], ['a', 'b']),
+            (['a', 'a', 'a'], ['a']),
+            (['β', 'unique', 'β'], ['β']),
+        ):
+            with self.subTest(labelnames=labelnames):
+                registry = CollectorRegistry()
+                with self.assertRaises(ValueError) as error:
+                    Counter('duplicate_labels', 'help', iter(labelnames), registry=registry)
+                self.assertEqual(str(error.exception), f'Duplicate label names: {expected!r}')
+                self.assertEqual(list(registry.collect()), [])
+
+    def test_unique_labelnames_iterator(self):
+        # Valid one-shot label iterables still preserve both names.
+        counter = Counter('iter_labels', 'help', iter(['a', 'b']), registry=self.registry)
+        counter.labels('x', 'y').inc()
+        self.assertEqual(
+            self.registry.get_sample_value('iter_labels_total', {'a': 'x', 'b': 'y'}), 1,
+        )
+
     def test_empty_labels_list(self):
         Histogram('h', 'help', [], registry=self.registry)
         self.assertEqual(0, self.registry.get_sample_value('h_sum'))