Add support for composite samplers in declarative config (open-telemetry#5201)

* opentelemetry-sdk: add declarative config support for experimental rule based sampler

Assisted-by: Cursor

* Make match similar to java

Assisted-by: Cursor

* Add changelog

* Revert "Make match similar to java"

This reverts commit 0c14136.

---------

Co-authored-by: Lukas Hering <40302054+herin049@users.noreply.github.com>

Author: xrmx

.changelog/5201.added

@@ -0,0 +1 @@
+`opentelemetry-sdk`: Add `composite/development` samplers support to declarative file configuration

opentelemetry-sdk/src/opentelemetry/sdk/_configuration/_tracer_provider.py

@@ -12,6 +12,15 @@
     load_entry_point,
 )
 from opentelemetry.sdk._configuration._exceptions import ConfigurationError
+from opentelemetry.sdk._configuration.models import (
+    ExperimentalComposableRuleBasedSampler as RuleBasedSamplerConfig,
+)
+from opentelemetry.sdk._configuration.models import (
+    ExperimentalComposableRuleBasedSamplerRule as RuleBasedSamplerRuleConfig,
+)
+from opentelemetry.sdk._configuration.models import (
+    ExperimentalComposableSampler as ComposableSamplerConfig,
+)
 from opentelemetry.sdk._configuration.models import (
     OtlpGrpcExporter as OtlpGrpcExporterConfig,
 )
@@ -46,6 +55,25 @@
     SpanLimits,
     TracerProvider,
 )
+from opentelemetry.sdk.trace._sampling_experimental import (
+    ComposableSampler,
+    composable_always_off,
+    composable_always_on,
+    composable_parent_threshold,
+    composable_rule_based,
+    composable_traceid_ratio_based,
+    composite_sampler,
+)
+from opentelemetry.sdk.trace._sampling_experimental._rule_based import (
+    AllPredicate,
+    AlwaysMatchPredicate,
+    AttributePatternsPredicate,
+    AttributeValuesPredicate,
+    ParentPredicate,
+    PredicateT,
+    RulesT,
+    SpanKindPredicate,
+)
 from opentelemetry.sdk.trace.export import (
     BatchSpanProcessor,
     ConsoleSpanExporter,
@@ -59,6 +87,7 @@
     Sampler,
     TraceIdRatioBased,
 )
+from opentelemetry.trace import SpanKind as TraceSpanKind
 
 _logger = logging.getLogger(__name__)
 
@@ -185,6 +214,88 @@ def _create_span_processor(
     )
 
 
+def _create_experimental_composable_sampler(
+    config: ComposableSamplerConfig,
+) -> ComposableSampler:
+    """Create an experimental composable sampler from config"""
+    if config.always_on is not None:
+        return composable_always_on()
+    if config.always_off is not None:
+        return composable_always_off()
+    if config.parent_threshold is not None:
+        return composable_parent_threshold(
+            _create_experimental_composable_sampler(
+                config.parent_threshold.root
+            )
+        )
+    if config.probability is not None:
+        ratio = config.probability.ratio
+        return composable_traceid_ratio_based(
+            ratio if ratio is not None else 1.0
+        )
+    if config.rule_based is not None:
+        return composable_rule_based(
+            _create_rule_based_sampler_rules(config.rule_based)
+        )
+    raise ConfigurationError(
+        f"Unknown or unsupported experimental composable sampler type in config: {config!r}. "
+        "Supported types: always_on, always_off, parent_threshold, probability, rule_based."
+    )
+
+
+def _create_rule_based_sampler_rules(
+    config: RuleBasedSamplerConfig,
+) -> RulesT:
+    if config.rules is None:
+        return []
+    return [
+        (
+            _create_rule_based_sampler_rule_predicate(rule),
+            _create_experimental_composable_sampler(rule.sampler),
+        )
+        for rule in config.rules
+    ]
+
+
+def _create_rule_based_sampler_rule_predicate(
+    config: RuleBasedSamplerRuleConfig,
+) -> PredicateT:
+    predicates: list[PredicateT] = []
+    if config.attribute_values is not None:
+        predicates.append(
+            AttributeValuesPredicate(
+                config.attribute_values.key,
+                config.attribute_values.values,
+            )
+        )
+    if config.attribute_patterns is not None:
+        predicates.append(
+            AttributePatternsPredicate(
+                config.attribute_patterns.key,
+                config.attribute_patterns.included,
+                config.attribute_patterns.excluded,
+            )
+        )
+    if config.span_kinds is not None:
+        predicates.append(
+            SpanKindPredicate(
+                [
+                    TraceSpanKind[span_kind.value.upper()]
+                    for span_kind in config.span_kinds
+                ]
+            )
+        )
+    if config.parent is not None:
+        predicates.append(
+            ParentPredicate([parent.value for parent in config.parent])
+        )
+    if not predicates:
+        return AlwaysMatchPredicate()
+    if len(predicates) == 1:
+        return predicates[0]
+    return AllPredicate(predicates)
+
+
 def _create_sampler(config: SamplerConfig) -> Sampler:
     """Create a sampler from config.
 
@@ -200,14 +311,21 @@ def _create_sampler(config: SamplerConfig) -> Sampler:
     if config.trace_id_ratio_based is not None:
         ratio = config.trace_id_ratio_based.ratio
         return TraceIdRatioBased(ratio if ratio is not None else 1.0)
+    if config.composite_development is not None:
+        return composite_sampler(
+            _create_experimental_composable_sampler(
+                config.composite_development
+            )
+        )
     if config.parent_based is not None:
         return _create_parent_based_sampler(config.parent_based)
     if config.additional_properties:
         name = next(iter(config.additional_properties))
         return load_entry_point("opentelemetry_sampler", name)()
     raise ConfigurationError(
         f"Unsupported sampler type in config: {config!r}. "
-        "Supported types: always_on, always_off, trace_id_ratio_based, parent_based."
+        "Supported types: always_on, always_off, composite_development, "
+        "trace_id_ratio_based, parent_based."
     )
 
 

opentelemetry-sdk/src/opentelemetry/sdk/trace/_sampling_experimental/_rule_based.py

@@ -3,11 +3,18 @@
 
 from __future__ import annotations
 
+import logging
 from collections.abc import Sequence
+from fnmatch import fnmatchcase
 from typing import Protocol
 
 from opentelemetry.context import Context
-from opentelemetry.trace import Link, SpanKind, TraceState
+from opentelemetry.trace import (
+    Link,
+    SpanKind,
+    TraceState,
+    get_current_span,
+)
 from opentelemetry.util.types import AnyValue, Attributes
 
 from ._composable import ComposableSampler, SamplingIntent
@@ -32,6 +39,9 @@ class AttributePredicate:
     """An exact match of an attribute value"""
 
     def __init__(self, key: str, value: AnyValue):
+        logging.warning(
+            "This is deprecated, use AttributeValuesPredicate instead"
+        )
         self.key = key
         self.value = value
 
@@ -52,6 +62,172 @@ def __str__(self):
         return f"{self.key}={self.value}"
 
 
+class AlwaysMatchPredicate:
+    def __call__(
+        self,
+        parent_ctx: Context | None,
+        name: str,
+        span_kind: SpanKind | None,
+        attributes: Attributes,
+        links: Sequence[Link] | None,
+        trace_state: TraceState | None,
+    ) -> bool:
+        return True
+
+    def __str__(self) -> str:
+        return "AlwaysMatch"
+
+
+class AllPredicate:
+    def __init__(self, predicates: Sequence[PredicateT]):
+        self._predicates = tuple(predicates)
+
+    def __call__(
+        self,
+        parent_ctx: Context | None,
+        name: str,
+        span_kind: SpanKind | None,
+        attributes: Attributes,
+        links: Sequence[Link] | None,
+        trace_state: TraceState | None,
+    ) -> bool:
+        return all(
+            predicate(
+                parent_ctx,
+                name,
+                span_kind,
+                attributes,
+                links,
+                trace_state,
+            )
+            for predicate in self._predicates
+        )
+
+    def __str__(self) -> str:
+        return " && ".join(str(predicate) for predicate in self._predicates)
+
+
+class AttributeValuesPredicate:
+    def __init__(self, key: str, values: Sequence[str]):
+        self._key = key
+        self._values = frozenset(values)
+
+    def __call__(
+        self,
+        parent_ctx: Context | None,
+        name: str,
+        span_kind: SpanKind | None,
+        attributes: Attributes,
+        links: Sequence[Link] | None,
+        trace_state: TraceState | None,
+    ) -> bool:
+        if not attributes or self._key not in attributes:
+            return False
+        return any(
+            str(value) in self._values
+            for value in _attribute_values(attributes[self._key])
+        )
+
+    def __str__(self) -> str:
+        values = ",".join(sorted(self._values))
+        return f"{self._key} in [{values}]"
+
+
+class AttributePatternsPredicate:
+    def __init__(
+        self,
+        key: str,
+        included: Sequence[str] | None = None,
+        excluded: Sequence[str] | None = None,
+    ):
+        self._key = key
+        self._included = tuple(included or ())
+        self._excluded = tuple(excluded or ())
+
+    def __call__(
+        self,
+        parent_ctx: Context | None,
+        name: str,
+        span_kind: SpanKind | None,
+        attributes: Attributes,
+        links: Sequence[Link] | None,
+        trace_state: TraceState | None,
+    ) -> bool:
+        if not attributes or self._key not in attributes:
+            return False
+        return any(
+            self._matches_value(str(value))
+            for value in _attribute_values(attributes[self._key])
+        )
+
+    def _matches_value(self, value: str) -> bool:
+        included = not self._included or any(
+            fnmatchcase(value, pattern) for pattern in self._included
+        )
+        excluded = any(
+            fnmatchcase(value, pattern) for pattern in self._excluded
+        )
+        return included and not excluded
+
+    def __str__(self) -> str:
+        return f"{self._key} matches"
+
+
+class SpanKindPredicate:
+    def __init__(self, span_kinds: Sequence[SpanKind]):
+        self._span_kinds = frozenset(span_kinds)
+
+    def __call__(
+        self,
+        parent_ctx: Context | None,
+        name: str,
+        span_kind: SpanKind | None,
+        attributes: Attributes,
+        links: Sequence[Link] | None,
+        trace_state: TraceState | None,
+    ) -> bool:
+        return span_kind in self._span_kinds
+
+    def __str__(self) -> str:
+        kinds = ",".join(kind.name.lower() for kind in self._span_kinds)
+        return f"span_kind in [{kinds}]"
+
+
+class ParentPredicate:
+    def __init__(self, parents: Sequence[str]):
+        self._parents = frozenset(parents)
+
+    def __call__(
+        self,
+        parent_ctx: Context | None,
+        name: str,
+        span_kind: SpanKind | None,
+        attributes: Attributes,
+        links: Sequence[Link] | None,
+        trace_state: TraceState | None,
+    ) -> bool:
+        parent_span_context = get_current_span(parent_ctx).get_span_context()
+        if not parent_span_context.is_valid:
+            parent = "none"
+        elif parent_span_context.is_remote:
+            parent = "remote"
+        else:
+            parent = "local"
+        return parent in self._parents
+
+    def __str__(self) -> str:
+        parents = ",".join(self._parents)
+        return f"parent in [{parents}]"
+
+
+def _attribute_values(value):
+    if isinstance(value, Sequence) and not isinstance(
+        value, (str, bytes, bytearray)
+    ):
+        return value
+    return (value,)
+
+
 RulesT = Sequence[tuple[PredicateT, ComposableSampler]]
 
 _non_sampling_intent = SamplingIntent(