fix(celery): coerce non-string values in CeleryGetter.get() (#4360)

* fix(celery): coerce non-string values in CeleryGetter.get()

Celery's Context copies all message properties as instance attributes
via __dict__.update(), including non-string values like timelimit
(tuple of ints) and priority (int). The TextMapPropagator contract
requires string values from Getter.get(), but CeleryGetter returned
these non-string values as-is, causing TraceState.from_header() to
crash with TypeError when calling re.split() on an int.

Coerce all non-string values to strings before returning them.

See #4359

* Add CHANGELOG entry for CeleryGetter fix

* Apply ruff formatting

Author: DrMeers

CHANGELOG.md

@@ -51,6 +51,8 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0
 
 ### Fixed
 
+- `opentelemetry-instrumentation-celery`: Coerce non-string values to strings in `CeleryGetter.get()` to prevent `TypeError` in `TraceState.from_header()` when Celery request attributes contain ints
+  ([#4360](https://github.com/open-telemetry/opentelemetry-python-contrib/pull/4360))
 - `opentelemetry-docker-tests`: Replace deprecated `SpanAttributes` from `opentelemetry.semconv.trace` with `opentelemetry.semconv._incubating.attributes`
  ([#4339](https://github.com/open-telemetry/opentelemetry-python-contrib/pull/4339))
 - `opentelemetry-instrumentation-confluent-kafka`: Skip `recv` span creation when `poll()` returns no message or `consume()` returns an empty list, avoiding empty spans on idle polls

instrumentation/opentelemetry-instrumentation-celery/src/opentelemetry/instrumentation/celery/__init__.py

@@ -104,8 +104,18 @@ def get(self, carrier, key):
         value = getattr(carrier, key, None)
         if value is None:
             return None
-        if isinstance(value, str) or not isinstance(value, Iterable):
+        # Celery's Context copies all message properties as instance
+        # attributes, including non-string values like timelimit (tuple
+        # of ints).  The TextMapPropagator contract requires string
+        # values, so coerce anything that isn't already a string.
+        if isinstance(value, str):
             value = (value,)
+        elif isinstance(value, Iterable):
+            value = tuple(
+                str(v) if not isinstance(v, str) else v for v in value
+            )
+        else:
+            value = (str(value),)
         return value
 
     def keys(self, carrier):

instrumentation/opentelemetry-instrumentation-celery/tests/test_getter.py

@@ -36,7 +36,40 @@ def test_get_iter(self):
         getter = CeleryGetter()
         mock_obj.test = ["val"]
         val = getter.get(mock_obj, "test")
-        self.assertEqual(val, ["val"])
+        self.assertEqual(val, ("val",))
+
+    def test_get_int(self):
+        """Non-string scalar values should be coerced to strings.
+
+        Celery's Context stores some attributes as ints (e.g. priority).
+        The TextMapPropagator contract requires string values; passing
+        an int to re.split() in TraceState.from_header() causes a
+        TypeError.
+        """
+        mock_obj = mock.Mock()
+        getter = CeleryGetter()
+        mock_obj.test = 42
+        val = getter.get(mock_obj, "test")
+        self.assertEqual(val, ("42",))
+
+    def test_get_iter_with_non_string_elements(self):
+        """Iterable values containing non-strings should be coerced.
+
+        Celery's timelimit attribute is a tuple of ints, e.g. (300, 60).
+        """
+        mock_obj = mock.Mock()
+        getter = CeleryGetter()
+        mock_obj.test = (300, 60)
+        val = getter.get(mock_obj, "test")
+        self.assertEqual(val, ("300", "60"))
+
+    def test_get_iter_with_mixed_types(self):
+        """Iterables with a mix of strings and non-strings."""
+        mock_obj = mock.Mock()
+        getter = CeleryGetter()
+        mock_obj.test = ["val", 123]
+        val = getter.get(mock_obj, "test")
+        self.assertEqual(val, ("val", "123"))
 
     def test_keys(self):
         getter = CeleryGetter()