feat(Flag Dependency): Evaluate flags that depend on other flags (#267)

Co-authored-by: linear-code[bot] <222613912+linear-code[bot]@users.noreply.github.com>
Co-authored-by: Evandro Myller <22429+emyller@users.noreply.github.com>
Co-authored-by: Evandro Myller <evandro.myller@flagsmith.com>
Co-authored-by: Claude Opus 5.5 (1M context) <noreply@anthropic.com>

Author: 3 people

.github/workflows/pytest.yml

@@ -20,6 +20,7 @@ jobs:
         uses: actions/checkout@v6
         with:
           fetch-depth: 0
+          submodules: recursive
 
       - name: Set up Python ${{ matrix.python-version }}
         uses: actions/setup-python@v5

.github/workflows/sdk-spec.yml

@@ -11,7 +11,7 @@ permissions:
 jobs:
   sdk-spec:
     # Keep the ref in `uses:` and `spec-ref` in sync.
-    uses: Flagsmith/sdk-spec/.github/workflows/run-spec.yml@v0.6.0
+    uses: Flagsmith/sdk-spec/.github/workflows/run-spec.yml@v0.7.0
     with:
-      spec-ref: v0.6.0
+      spec-ref: v0.7.0
       language: python

.gitmodules

@@ -1,3 +1,6 @@
+[submodule "tests/engine_tests/engine-test-data"]
+	path = tests/engine_tests/engine-test-data
+	url = https://github.com/Flagsmith/engine-test-data.git
 [submodule "tests/data/sdk-benchmarks"]
 	path = tests/data/sdk-benchmarks
 	url = https://github.com/Flagsmith/sdk-benchmarks.git

flagsmith/flagsmith.py

@@ -519,6 +519,7 @@ def _get_identity_flags_from_document(
         return Flags.from_evaluation_context(
             context=context,
             overrides_index=overrides_index,
+            flag_dependent_features=local_evaluation.flag_dependent_features,
             analytics_processor=self._analytics_processor,
             default_flag_handler=self.default_flag_handler,
         )

flagsmith/local_evaluation.py

@@ -5,7 +5,7 @@
 from functools import cached_property
 
 from flag_engine import engine
-from flag_engine.context.types import SegmentContext
+from flag_engine.context.types import SegmentContext, SegmentRule
 
 from flagsmith.api.types import EnvironmentModel
 from flagsmith.mappers import (
@@ -17,6 +17,7 @@
     Flag,
     SegmentOverridesIndex,
     build_segment_overrides_index,
+    find_flag_dependent_features,
 )
 from flagsmith.types import (
     FeatureMetadata,
@@ -25,6 +26,13 @@
     TraitMapping,
 )
 
+IDENTITY_FREE_PROPERTY_PREFIXES = (
+    "$.environment.",
+    "$.environment[",
+    "$.flags.",
+    "$.flags[",
+)
+
 
 class LocalEvaluation:
     """Per-environment-document state for local evaluation."""
@@ -40,6 +48,9 @@ def __init__(
         self.overrides_index: SegmentOverridesIndex = build_segment_overrides_index(
             context
         )
+        self.flag_dependent_features = find_flag_dependent_features(
+            self.overrides_index
+        )
         self._identity_overrides = {
             identifier: (segment_context, self._index_overrides(segment_context))
             for identifier, segment_context in (
@@ -63,11 +74,20 @@ def from_environment_document(
 
     @cached_property
     def environment_flags(self) -> typing.Dict[str, Flag]:
-        """Environment flags, which depend on the environment document only."""
-        context_without_segments = self.context.copy()
-        context_without_segments.pop("segments", None)
+        """Environment flags, which depend on the environment document only.
+
+        Evaluated without an identity, so segments reading traits or identity
+        context are left out.
+        """
+        environment_context = self.context.copy()
+        if segments := environment_context.get("segments"):
+            environment_context["segments"] = {
+                segment_key: segment_context
+                for segment_key, segment_context in segments.items()
+                if _is_identity_free(segment_context["rules"])
+            }
         evaluation_result = engine.get_evaluation_result(
-            context=context_without_segments,
+            context=environment_context,
         )
         return {
             feature_name: Flag.from_evaluation_result(flag_result)
@@ -109,3 +129,14 @@ def _index_overrides(
             for override in segment_context.get("overrides") or ()
             if (feature_name := override["name"])
         }
+
+
+def _is_identity_free(rules: typing.Iterable[SegmentRule]) -> bool:
+    return all(
+        all(
+            condition["property"].startswith(IDENTITY_FREE_PROPERTY_PREFIXES)
+            for condition in rule.get("conditions") or ()
+        )
+        and _is_identity_free(rule.get("rules") or ())
+        for rule in rules
+    )

flagsmith/models.py

@@ -4,7 +4,7 @@
 from dataclasses import dataclass, field
 
 from flag_engine import engine
-from flag_engine.context.types import SegmentContext
+from flag_engine.context.types import SegmentContext, SegmentRule
 
 from flagsmith.analytics import AnalyticsProcessor
 from flagsmith.exceptions import FlagsmithFeatureDoesNotExistError
@@ -23,6 +23,8 @@
 ]
 SegmentOverridesIndex = typing.MutableMapping[FeatureName, SegmentContexts]
 
+FLAGS_JSONPATH_PREFIX = "$.flags"
+
 
 def build_segment_overrides_index(
     context: SDKEvaluationContext,
@@ -44,6 +46,35 @@ def build_segment_overrides_index(
     return index
 
 
+def find_flag_dependent_features(
+    overrides_index: SegmentOverridesIndex,
+) -> typing.FrozenSet[FeatureName]:
+    """Features overridden by a segment with a condition on another flag.
+
+    Such a condition may read any flag in the environment, transitively,
+    so these features can only be resolved against the full context.
+    """
+    return frozenset(
+        feature_name
+        for feature_name, segment_contexts in overrides_index.items()
+        if any(
+            _rules_read_flags(segment_context["rules"])
+            for segment_context in segment_contexts.values()
+        )
+    )
+
+
+def _rules_read_flags(rules: typing.Iterable[SegmentRule]) -> bool:
+    return any(
+        any(
+            condition["property"].startswith(FLAGS_JSONPATH_PREFIX)
+            for condition in rule.get("conditions") or ()
+        )
+        or _rules_read_flags(rule.get("rules") or ())
+        for rule in rules
+    )
+
+
 @dataclass(frozen=True)
 class ExperimentMetadata:
     """The running experiment a flag was evaluated under; identity evaluations only."""
@@ -130,6 +161,7 @@ class Flags:
     # paths (`from_evaluation_result` / `from_api_flags`).
     _context: typing.Optional[SDKEvaluationContext] = None
     _overrides_index: typing.Optional[SegmentOverridesIndex] = None
+    _flag_dependent_features: typing.AbstractSet[FeatureName] = frozenset()
     _fully_materialised: bool = False
 
     @classmethod
@@ -154,6 +186,7 @@ def from_evaluation_context(
         cls,
         context: SDKEvaluationContext,
         overrides_index: SegmentOverridesIndex,
+        flag_dependent_features: typing.AbstractSet[FeatureName],
         analytics_processor: typing.Optional[AnalyticsProcessor],
         default_flag_handler: typing.Optional[typing.Callable[[str], DefaultFlag]],
     ) -> Flags:
@@ -162,14 +195,17 @@ def from_evaluation_context(
         No engine work is done here — flags are resolved on first access
         via :meth:`_resolve_flag`. Reusing the same `overrides_index`
         across calls amortises its construction cost (it's rebuilt only
-        when the environment doc refreshes, not per identity).
+        when the environment doc refreshes, not per identity). The same
+        goes for `flag_dependent_features`, see
+        :func:`find_flag_dependent_features`.
         """
         return cls(
             flags={},
             default_flag_handler=default_flag_handler,
             _analytics_processor=analytics_processor,
             _context=context,
             _overrides_index=overrides_index,
+            _flag_dependent_features=flag_dependent_features,
         )
 
     @classmethod
@@ -201,16 +237,19 @@ def all_flags(self) -> typing.List[Flag]:
 
         :return: list of Flag objects.
         """
-        if self._context is not None and not self._fully_materialised:
-            result = engine.get_evaluation_result(self._context)
-            for feature_name, flag_result in result["flags"].items():
-                if feature_name not in self.flags:
-                    self.flags[feature_name] = Flag.from_evaluation_result(
-                        flag_result,
-                    )
-            self._fully_materialised = True
+        self._materialise()
         return list(self.flags.values())
 
+    def _materialise(self) -> None:
+        """Resolve every feature in the lazy context, once."""
+        if self._context is None or self._fully_materialised:
+            return
+        result = engine.get_evaluation_result(self._context)
+        for feature_name, flag_result in result["flags"].items():
+            if feature_name not in self.flags:
+                self.flags[feature_name] = Flag.from_evaluation_result(flag_result)
+        self._fully_materialised = True
+
     def is_feature_enabled(self, feature_name: str) -> bool:
         """
         Check whether a given feature is enabled.
@@ -276,13 +315,20 @@ def _resolve_flag(self, feature_name: str) -> Flag:
         that could override it, looked up in O(1) via the precomputed
         reverse index — so the engine's full pipeline runs against an
         input small enough to evaluate in ~1 µs.
+
+        Features overridden on other flags' values need the full context
+        instead, so they are resolved along with every other feature.
         """
         context = self._context
         overrides_index = self._overrides_index
         # `get_flag` / `all_flags` gate this call behind the same
         # non-None checks; assert here so type checkers can narrow.
         assert context is not None and overrides_index is not None
 
+        if feature_name in self._flag_dependent_features:
+            self._materialise()
+            return self.flags[feature_name]
+
         trimmed: SDKEvaluationContext = {
             **context,
             "features": {feature_name: context["features"][feature_name]},

sdk-spec/steps.py

@@ -39,6 +39,17 @@ def create_client_with_events(
     )
 
 
+# Create the SDK client with local evaluation
+def create_client_with_local_evaluation(
+    environment_key: str, api_url: str
+) -> Flagsmith:
+    return Flagsmith(
+        environment_key=environment_key,
+        api_url=api_url,
+        enable_local_evaluation=True,
+    )
+
+
 # When the application gets the environment flags
 def get_environment_flags(flagsmith: Flagsmith) -> list[SpecFlag]:
     return as_spec_flags(flagsmith.get_environment_flags())