[feat] mem0 instrumentation: change to use memory handler (#89)

* feat:mem0 instrumentation: change to use memory handler

Author: sunym1991

.github/workflows/loongsuite_test_0.yml

@@ -374,44 +374,6 @@ jobs:
       - name: Run tests
         run: tox -c tox-loongsuite.ini -e py313-test-loongsuite-instrumentation-dashscope-latest -- -ra
 
-  py39-test-loongsuite-instrumentation-mem0-oldest_ubuntu-latest:
-    name: LoongSuite loongsuite-instrumentation-mem0-oldest 3.9 Ubuntu
-    runs-on: ubuntu-latest
-    timeout-minutes: 30
-    steps:
-      - name: Checkout repo @ SHA - ${{ github.sha }}
-        uses: actions/checkout@v4
-
-      - name: Set up Python 3.9
-        uses: actions/setup-python@v5
-        with:
-          python-version: "3.9"
-
-      - name: Install tox
-        run: pip install tox-uv
-
-      - name: Run tests
-        run: tox -c tox-loongsuite.ini -e py39-test-loongsuite-instrumentation-mem0-oldest -- -ra
-
-  py39-test-loongsuite-instrumentation-mem0-latest_ubuntu-latest:
-    name: LoongSuite loongsuite-instrumentation-mem0-latest 3.9 Ubuntu
-    runs-on: ubuntu-latest
-    timeout-minutes: 30
-    steps:
-      - name: Checkout repo @ SHA - ${{ github.sha }}
-        uses: actions/checkout@v4
-
-      - name: Set up Python 3.9
-        uses: actions/setup-python@v5
-        with:
-          python-version: "3.9"
-
-      - name: Install tox
-        run: pip install tox-uv
-
-      - name: Run tests
-        run: tox -c tox-loongsuite.ini -e py39-test-loongsuite-instrumentation-mem0-latest -- -ra
-
   py310-test-loongsuite-instrumentation-mem0-oldest_ubuntu-latest:
     name: LoongSuite loongsuite-instrumentation-mem0-oldest 3.10 Ubuntu
     runs-on: ubuntu-latest

.pre-commit-config.yaml

@@ -18,4 +18,4 @@ repos:
     hooks:
       - id: rstcheck
         additional_dependencies: ['rstcheck[sphinx]']
-        args: ["--report-level", "warning"]
+        args: ["--report-level", "warning"]

CHANGELOG-loongsuite.md

@@ -13,6 +13,9 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0
 
 ### Fixed
 
+- `loongsuite-instrumentation-mem0`: use memory handler
+  ([#89](https://github.com/alibaba/loongsuite-python-agent/pull/89))
+
 - Add `from __future__ import annotations` to fix Python 3.9 compatibility for union type syntax (`X | Y`)
   ([#80](https://github.com/alibaba/loongsuite-python-agent/pull/80))
 

instrumentation-loongsuite/loongsuite-instrumentation-mem0/README.md

@@ -45,8 +45,15 @@ export OTEL_EXPORTER_OTLP_TRACES_ENDPOINT=<trace_endpoint>
 export OTEL_EXPORTER_OTLP_PROTOCOL=http/protobuf
 export OTEL_SERVICE_NAME=mem0-demo
 
+# Enable GenAI experimental semantic conventions (required for GenAI content/event features)
+export OTEL_SEMCONV_STABILITY_OPT_IN=gen_ai_latest_experimental
+
 # (Optional) Capture message content – may contain sensitive data
-export OTEL_INSTRUMENTATION_GENAI_CAPTURE_MESSAGE_CONTENT=true
+# Must be one of: NO_CONTENT | SPAN_ONLY | EVENT_ONLY | SPAN_AND_EVENT
+export OTEL_INSTRUMENTATION_GENAI_CAPTURE_MESSAGE_CONTENT=SPAN_ONLY
+
+# (Optional) Emit GenAI events (LogRecord). Requires a LoggerProvider exporter in your app.
+export OTEL_INSTRUMENTATION_GENAI_EMIT_EVENT=true
 ```
 
 ### Option 1: Using opentelemetry-instrument
@@ -69,7 +76,9 @@ If everything is working, you should see spans for:
 You can also start your application with `loongsuite-instrument` to forward data to LoongSuite/Jaeger:
 
 ```bash
-export OTEL_INSTRUMENTATION_GENAI_CAPTURE_MESSAGE_CONTENT=true
+export OTEL_SEMCONV_STABILITY_OPT_IN=gen_ai_latest_experimental
+export OTEL_INSTRUMENTATION_GENAI_CAPTURE_MESSAGE_CONTENT=SPAN_ONLY
+export OTEL_INSTRUMENTATION_GENAI_EMIT_EVENT=true
 
 loongsuite-instrument \
     --traces_exporter console \
@@ -94,7 +103,9 @@ You can control the Mem0 instrumentation using environment variables.
 |-----------------------------------------------------------|---------|-----------------------------------------------------------------------------|
 | `OTEL_INSTRUMENTATION_MEM0_ENABLED`                       | `true`  | Enable or disable the Mem0 instrumentation entirely.                       |
 | `OTEL_INSTRUMENTATION_MEM0_INNER_ENABLED`                 | `false` | Enable internal phases (Vector Store, Graph Store, Rerank).              |
-| `OTEL_INSTRUMENTATION_GENAI_CAPTURE_MESSAGE_CONTENT`      | `false` | Capture input/output message content (may contain PII or sensitive data).  |
+| `OTEL_SEMCONV_STABILITY_OPT_IN`                           | *(empty)* | Set to `gen_ai_latest_experimental` to enable GenAI experimental semantics (required for content/event). |
+| `OTEL_INSTRUMENTATION_GENAI_CAPTURE_MESSAGE_CONTENT`      | `NO_CONTENT` | Content capturing mode: `NO_CONTENT`, `SPAN_ONLY`, `EVENT_ONLY`, `SPAN_AND_EVENT` (may contain PII/sensitive data). |
+| `OTEL_INSTRUMENTATION_GENAI_EMIT_EVENT`                   | `false` | Emit GenAI events (`LogRecord`). Requires configuring a `LoggerProvider`. |
 
 ### Configuration Examples
 
@@ -103,7 +114,11 @@ You can control the Mem0 instrumentation using environment variables.
 export OTEL_INSTRUMENTATION_MEM0_INNER_ENABLED=true
 
 # Enable content capture (be careful with sensitive data)
-export OTEL_INSTRUMENTATION_GENAI_CAPTURE_MESSAGE_CONTENT=true
+export OTEL_SEMCONV_STABILITY_OPT_IN=gen_ai_latest_experimental
+export OTEL_INSTRUMENTATION_GENAI_CAPTURE_MESSAGE_CONTENT=SPAN_AND_EVENT
+
+# Enable event emission (requires LoggerProvider exporter)
+export OTEL_INSTRUMENTATION_GENAI_EMIT_EVENT=true
 ```
 
 ## Semantic Conventions Status

instrumentation-loongsuite/loongsuite-instrumentation-mem0/pyproject.toml

@@ -18,7 +18,6 @@ classifiers = [
     "License :: OSI Approved :: Apache Software License",
     "Programming Language :: Python",
     "Programming Language :: Python :: 3",
-    "Programming Language :: Python :: 3.9",
     "Programming Language :: Python :: 3.10",
     "Programming Language :: Python :: 3.11",
     "Programming Language :: Python :: 3.12",
@@ -29,6 +28,7 @@ dependencies = [
     "opentelemetry-api ~=1.37",
     "opentelemetry-instrumentation ~=0.58b0",
     "opentelemetry-semantic-conventions ~=0.58b0",
+    "opentelemetry-util-genai ~= 0.2b0",
 ]
 
 [project.optional-dependencies]

instrumentation-loongsuite/loongsuite-instrumentation-mem0/src/opentelemetry/instrumentation/mem0/__init__.py

@@ -26,6 +26,7 @@
 from opentelemetry.instrumentation.mem0.version import __version__
 from opentelemetry.instrumentation.utils import unwrap
 from opentelemetry.semconv.schemas import Schemas
+from opentelemetry.util.genai.extended_handler import ExtendedTelemetryHandler
 
 # Module-level logger
 logger = logging.getLogger(__name__)
@@ -121,6 +122,19 @@ def _instrument(self, **kwargs: Any) -> None:
         if not tracer_provider:
             tracer_provider = trace_api.get_tracer_provider()
 
+        meter_provider = kwargs.get("meter_provider")
+
+        # Optional: logger provider for GenAI events (util will no-op if not provided)
+        logger_provider = kwargs.get("logger_provider")
+
+        # Create util GenAI handler (strong dependency, no fallback).
+        # Avoid singleton here so tests (and multiple tracer providers) don't leak across runs.
+        telemetry_handler = ExtendedTelemetryHandler(
+            tracer_provider=tracer_provider,
+            meter_provider=meter_provider,
+            logger_provider=logger_provider,
+        )
+
         # Create tracer
         tracer = trace_api.get_tracer(
             "opentelemetry.instrumentation.mem0",
@@ -129,10 +143,9 @@ def _instrument(self, **kwargs: Any) -> None:
             schema_url=Schemas.V1_28_0.value,
         )
 
-        # Wrap ThreadPoolExecutor.submit to support context propagation
         # Execute instrumentation (traces only, metrics removed)
-        self._instrument_memory_operations(tracer)
-        self._instrument_memory_client_operations(tracer)
+        self._instrument_memory_operations(telemetry_handler)
+        self._instrument_memory_client_operations(telemetry_handler)
         # Sub-phases controlled by toggle, avoid binding wrapper when disabled to reduce overhead
         if mem0_config.is_internal_phases_enabled():
             self._instrument_vector_operations(tracer)
@@ -379,7 +392,7 @@ def _wrapper(wrapped, instance, args, kwargs):
 
         return _wrapper
 
-    def _instrument_memory_operations(self, tracer):
+    def _instrument_memory_operations(self, telemetry_handler):
         """Instrument Memory and AsyncMemory operations."""
         try:
             if (
@@ -393,7 +406,7 @@ def _instrument_memory_operations(self, tracer):
                 )
                 return
 
-            wrapper = MemoryOperationWrapper(tracer)
+            wrapper = MemoryOperationWrapper(telemetry_handler)
 
             # Instrument Memory (sync)
             for method in self._public_methods_of(
@@ -431,7 +444,7 @@ def _instrument_memory_operations(self, tracer):
         except Exception as e:
             logger.debug(f"Failed to instrument Memory operations: {e}")
 
-    def _instrument_memory_client_operations(self, tracer):
+    def _instrument_memory_client_operations(self, telemetry_handler):
         """Instrument MemoryClient and AsyncMemoryClient operations."""
         try:
             if (
@@ -445,7 +458,7 @@ def _instrument_memory_client_operations(self, tracer):
                 )
                 return
 
-            wrapper = MemoryOperationWrapper(tracer)
+            wrapper = MemoryOperationWrapper(telemetry_handler)
 
             # Instrument MemoryClient (sync)
             for method in self._public_methods_of(

instrumentation-loongsuite/loongsuite-instrumentation-mem0/src/opentelemetry/instrumentation/mem0/config.py

@@ -5,15 +5,6 @@
 from __future__ import annotations
 
 import os
-from dataclasses import dataclass
-from typing import Optional
-
-OTEL_INSTRUMENTATION_GENAI_CAPTURE_MESSAGE_CONTENT = (
-    "OTEL_INSTRUMENTATION_GENAI_CAPTURE_MESSAGE_CONTENT"
-)
-OTEL_INSTRUMENTATION_GENAI_CAPTURE_MESSAGE_CONTENT_MAX_LENGTH = (
-    "OTEL_INSTRUMENTATION_GENAI_CAPTURE_MESSAGE_CONTENT_MAX_LENGTH"
-)
 
 SLOW_REQUEST_THRESHOLD_SECONDS = 5.0
 
@@ -58,51 +49,6 @@ def first_present_bool(keys: list[str], default: bool) -> bool:
     return default
 
 
-@dataclass
-class GenAITelemetryOptions:
-    """GenAI telemetry configuration options."""
-
-    capture_message_content: Optional[bool] = None
-    capture_message_content_max_length: Optional[int] = None
-
-    def __post_init__(self):
-        if self.capture_message_content is None:
-            self.capture_message_content = (
-                os.getenv(
-                    OTEL_INSTRUMENTATION_GENAI_CAPTURE_MESSAGE_CONTENT, "false"
-                ).lower()
-                == "true"
-            )
-
-        if self.capture_message_content_max_length is None:
-            self.capture_message_content_max_length = int(
-                os.getenv(
-                    OTEL_INSTRUMENTATION_GENAI_CAPTURE_MESSAGE_CONTENT_MAX_LENGTH,
-                    "1048576",
-                )
-            )
-
-    def should_capture_content(self) -> bool:
-        """Check if content capture is enabled."""
-        return self.capture_message_content
-
-    def truncate_content(self, content: str) -> str:
-        """Truncate content to max length if needed."""
-        if not content:
-            return content
-        if len(content) > self.capture_message_content_max_length:
-            return (
-                content[: self.capture_message_content_max_length]
-                + "...[truncated]"
-            )
-        return content
-
-    @classmethod
-    def from_env(cls) -> "GenAITelemetryOptions":
-        """Create configuration from environment variables."""
-        return cls()
-
-
 class Mem0InstrumentationConfig:
     """Mem0 instrumentation configuration."""
 
@@ -124,16 +70,6 @@ def is_internal_phases_enabled() -> bool:
     )
 
 
-def should_capture_content() -> bool:
-    """Check if message content capture is enabled."""
-    return GenAITelemetryOptions.from_env().should_capture_content()
-
-
 def get_slow_threshold_seconds() -> float:
     """Get slow request threshold in seconds."""
     return SLOW_REQUEST_THRESHOLD_SECONDS
-
-
-def get_telemetry_options() -> GenAITelemetryOptions:
-    """Get GenAI telemetry configuration options."""
-    return GenAITelemetryOptions.from_env()