fix: Add support for custom span filtering

Signed-off-by: Cagri Yonca <cagri@ibm.com>

Author: CagriYonca

src/instana/options.py

@@ -144,6 +144,16 @@ def _add_instana_agent_span_filter(self) -> None:
                     }
                 ],
             },
+            {
+                "name": "filter-internal-sdk-spans-by-url",
+                "attributes": [
+                    {
+                        "key": "sdk.custom.tags.http.url",
+                        "values": ["com.instana"],
+                        "match_type": "contains",
+                    }
+                ],
+            },
         ])
 
     def _apply_env_stack_trace_config(self) -> None:

src/instana/util/span_utils.py

@@ -1,7 +1,7 @@
 # (c) Copyright IBM Corp. 2025
 
 
-from typing import Any, List
+from typing import Any, Dict, List, Optional
 
 from instana.util.config import SPAN_TYPE_TO_CATEGORY
 
@@ -36,8 +36,15 @@ def matches_rule(rule_attributes: List[Any], span_attributes: List[Any]) -> bool
                 rule_matched = True
 
         else:
+            span_value = None
             if key in span_attributes:
                 span_value = span_attributes[key]
+            elif "." in key:
+                # Support dot-notation paths for nested attributes
+                # e.g. "sdk.custom.tags.http.host" -> span["sdk.custom"]["tags"]["http.host"]
+                span_value = resolve_nested_key(span_attributes, key.split("."))
+
+            if span_value is not None:
                 for rule_value in target_values:
                     if match_key_filter(span_value, rule_value, match_type):
                         rule_matched = True
@@ -49,6 +56,36 @@ def matches_rule(rule_attributes: List[Any], span_attributes: List[Any]) -> bool
     return True
 
 
+def resolve_nested_key(data: Dict[str, Any], key_parts: List[str]) -> Optional[Any]:
+    """Resolve a dotted key path against a potentially nested dict.
+
+    Tries all possible prefix lengths so that keys which themselves contain
+    dots (e.g. ``sdk.custom`` or ``http.host``) are handled correctly.
+
+    Example::
+
+        # span_attributes = {"sdk.custom": {"tags": {"http.host": "example.com"}}}
+        resolve_nested_key(span_attributes, ["sdk", "custom", "tags", "http", "host"])
+        # -> "example.com"
+    """
+    if not key_parts or not isinstance(data, dict):
+        return None
+
+    # Try the longest prefix first so that keys with embedded dots are matched
+    # before shorter splits (e.g. prefer "sdk.custom" over "sdk").
+    for i in range(len(key_parts), 0, -1):
+        candidate = ".".join(key_parts[:i])
+        if candidate in data:
+            remaining = key_parts[i:]
+            if not remaining:
+                return data[candidate]
+            result = resolve_nested_key(data[candidate], remaining)
+            if result is not None:
+                return result
+
+    return None
+
+
 def match_key_filter(span_value: str, rule_value: str, match_type: str) -> bool:
     """Check if the first value matches the second value based on the match type."""
     # Guard against None values

tests/test_options.py

@@ -39,6 +39,16 @@
             }
         ],
     },
+    {
+        "name": "filter-internal-sdk-spans-by-url",
+        "attributes": [
+            {
+                "key": "sdk.custom.tags.http.url",
+                "values": ["com.instana"],
+                "match_type": "contains",
+            }
+        ],
+    },
 ]
 
 
@@ -184,6 +194,16 @@ def test_base_options_with_env_vars(self) -> None:
                         }
                     ],
                 },
+                {
+                    "name": "filter-internal-sdk-spans-by-url",
+                    "attributes": [
+                        {
+                            "key": "sdk.custom.tags.http.url",
+                            "values": ["com.instana"],
+                            "match_type": "contains",
+                        }
+                    ],
+                },
             ],
         }
 
@@ -296,6 +316,16 @@ def test_base_options_with_endpoint_file(self) -> None:
                         }
                     ],
                 },
+                {
+                    "name": "filter-internal-sdk-spans-by-url",
+                    "attributes": [
+                        {
+                            "key": "sdk.custom.tags.http.url",
+                            "values": ["com.instana"],
+                            "match_type": "contains",
+                        }
+                    ],
+                },
             ],
         }
         del self.base_options
@@ -377,6 +407,16 @@ def test_set_trace_configurations_by_env_variable(self) -> None:
                         }
                     ],
                 },
+                {
+                    "name": "filter-internal-sdk-spans-by-url",
+                    "attributes": [
+                        {
+                            "key": "sdk.custom.tags.http.url",
+                            "values": ["com.instana"],
+                            "match_type": "contains",
+                        }
+                    ],
+                },
             ],
         }
         assert not self.base_options.kafka_trace_correlation
@@ -512,6 +552,16 @@ def test_set_trace_configurations_by_in_code_configuration(self) -> None:
                         }
                     ],
                 },
+                {
+                    "name": "filter-internal-sdk-spans-by-url",
+                    "attributes": [
+                        {
+                            "key": "sdk.custom.tags.http.url",
+                            "values": ["com.instana"],
+                            "match_type": "contains",
+                        }
+                    ],
+                },
             ],
         }
 
@@ -779,6 +829,16 @@ def test_tracing_filter_environment_variables(self) -> None:
                         }
                     ],
                 },
+                {
+                    "name": "filter-internal-sdk-spans-by-url",
+                    "attributes": [
+                        {
+                            "key": "sdk.custom.tags.http.url",
+                            "values": ["com.instana"],
+                            "match_type": "contains",
+                        }
+                    ],
+                },
             ],
         }
 

tests/util/test_span_utils.py

@@ -1,6 +1,13 @@
 # (c) Copyright IBM Corp. 2025
 
-from instana.util.span_utils import matches_rule, match_key_filter, get_span_kind
+from collections import defaultdict
+
+from instana.util.span_utils import (
+    get_span_kind,
+    match_key_filter,
+    matches_rule,
+    resolve_nested_key,
+)
 
 
 class TestSpanUtils:
@@ -144,3 +151,111 @@ def test_matches_rule_with_none_attribute_value(self) -> None:
             {"key": "http.method", "values": ["GET"], "match_type": "strict"}
         ]
         assert matches_rule(rule_method, span_attrs)
+
+    def test_resolve_nested_key_embedded_dot_keys(self) -> None:
+        """Resolves sdk.custom.tags.http.host through a defaultdict structure —
+        the exact layout produced by real SDK spans."""
+        sdk_custom = defaultdict(dict)
+        sdk_custom["tags"] = defaultdict(str)
+        sdk_custom["tags"]["http.host"] = "agent.com.instana.io"
+
+        assert (
+            resolve_nested_key(
+                {"sdk.custom": sdk_custom}, ["sdk", "custom", "tags", "http", "host"]
+            )
+            == "agent.com.instana.io"
+        )
+
+    def test_resolve_nested_key_returns_none_when_missing(self) -> None:
+        """Returns None when the dotted path does not exist in the data."""
+        assert (
+            resolve_nested_key(
+                {"sdk.custom": {"tags": {}}}, ["sdk", "custom", "tags", "http", "host"]
+            )
+            is None
+        )
+
+    def test_resolve_nested_key_with_empty_key_parts(self) -> None:
+        """Returns None when key_parts is an empty list."""
+        data = {"sdk.custom": {"tags": {"http.host": "example.com"}}}
+        assert resolve_nested_key(data, []) is None
+
+    def test_resolve_nested_key_with_non_dict_data(self) -> None:
+        """Returns None when data is not a dictionary."""
+        # Test with string
+        assert resolve_nested_key("not a dict", ["key"]) is None
+
+        # Test with list
+        assert resolve_nested_key(["not", "a", "dict"], ["key"]) is None
+
+        # Test with None
+        assert resolve_nested_key(None, ["key"]) is None
+
+        # Test with integer
+        assert resolve_nested_key(42, ["key"]) is None
+
+    def test_matches_rule_sdk_span_host_match(self) -> None:
+        """SDK span whose sdk.custom.tags.http.host contains 'com.instana' should be filtered."""
+        sdk_custom = defaultdict(dict)
+        sdk_custom["tags"] = {"http.host": "agent.com.instana.io"}
+        span_attrs = {
+            "type": "sdk",
+            "kind": 3,
+            "sdk.name": "my-span",
+            "sdk.custom": sdk_custom,
+        }
+
+        rule = [
+            {
+                "key": "sdk.custom.tags.http.host",
+                "values": ["com.instana"],
+                "match_type": "contains",
+            }
+        ]
+        assert matches_rule(rule, span_attrs)
+
+    def test_matches_rule_sdk_span_host_no_match(self) -> None:
+        """SDK span with an unrelated host should NOT be filtered."""
+        sdk_custom = defaultdict(dict)
+        sdk_custom["tags"] = {"http.host": "myapp.example.com"}
+        span_attrs = {
+            "type": "sdk",
+            "kind": 3,
+            "sdk.name": "my-span",
+            "sdk.custom": sdk_custom,
+        }
+
+        rule = [
+            {
+                "key": "sdk.custom.tags.http.host",
+                "values": ["com.instana"],
+                "match_type": "contains",
+            }
+        ]
+        assert not matches_rule(rule, span_attrs)
+
+    def test_matches_rule_sdk_span_url_match(self) -> None:
+        """SDK span whose sdk.custom.tags.http.url contains 'com.instana' should be filtered.
+
+        Covers the span shape:
+          data.sdk.custom.tags.http.url = 'http://localhost:42699/com.instana.plugin.python.89262'
+        """
+        sdk_custom = defaultdict(dict)
+        sdk_custom["tags"] = {
+            "http.url": "http://localhost:42699/com.instana.plugin.python.89262"
+        }
+        span_attrs = {
+            "type": "sdk",
+            "kind": 3,
+            "sdk.name": "HEAD",
+            "sdk.custom": sdk_custom,
+        }
+
+        rule = [
+            {
+                "key": "sdk.custom.tags.http.url",
+                "values": ["com.instana"],
+                "match_type": "contains",
+            }
+        ]
+        assert matches_rule(rule, span_attrs)