remdb 0.3.118__py3-none-any.whl → 0.3.141__py3-none-any.whl

rem/services/session/compression.py

@@ -170,12 +170,16 @@ class SessionMessageStore:
         entity_key = truncate_key(f"session-{session_id}-msg-{message_index}")
 
         # Create Message entity for assistant response
+        # Use pre-generated id from message dict if available (for frontend feedback)
         msg = Message(
+            id=message.get("id"),  # Use pre-generated ID if provided
             content=message.get("content", ""),
             message_type=message.get("role", "assistant"),
             session_id=session_id,
             tenant_id=self.user_id,  # Set tenant_id to user_id (application scoped to user)
             user_id=user_id or self.user_id,
+            trace_id=message.get("trace_id"),
+            span_id=message.get("span_id"),
             metadata={
                 "message_index": message_index,
                 "entity_key": entity_key,  # Store entity key for LOOKUP
@@ -284,11 +288,14 @@ class SessionMessageStore:
                 # Short assistant messages, user messages, and system messages stored as-is
                 # Store ALL messages in database for full audit trail
                 msg = Message(
+                    id=message.get("id"),  # Use pre-generated ID if provided
                     content=content,
                     message_type=message.get("role", "user"),
                     session_id=session_id,
                     tenant_id=self.user_id,  # Set tenant_id to user_id (application scoped to user)
                     user_id=user_id or self.user_id,
+                    trace_id=message.get("trace_id"),
+                    span_id=message.get("span_id"),
                     metadata={
                         "message_index": idx,
                         "timestamp": message.get("timestamp"),

rem/settings.py

@@ -33,14 +33,15 @@ Example .env file:
     AUTH__OIDC_CLIENT_ID=your-client-id
     AUTH__SESSION_SECRET=your-secret-key
 
-    # OpenTelemetry (disabled by default)
+    # OpenTelemetry (disabled by default - enable via env var when collector available)
+    # Standard OTLP collector ports: 4317 (gRPC), 4318 (HTTP)
     OTEL__ENABLED=false
     OTEL__SERVICE_NAME=rem-api
-    OTEL__COLLECTOR_ENDPOINT=http://localhost:4318
-    OTEL__PROTOCOL=http
+    OTEL__COLLECTOR_ENDPOINT=http://localhost:4317
+    OTEL__PROTOCOL=grpc
 
-    # Arize Phoenix (disabled by default)
-    PHOENIX__ENABLED=false
+    # Arize Phoenix (enabled by default - can be disabled via env var)
+    PHOENIX__ENABLED=true
     PHOENIX__COLLECTOR_ENDPOINT=http://localhost:6006/v1/traces
     PHOENIX__PROJECT_NAME=rem
 
@@ -241,6 +242,11 @@ class OTELSettings(BaseSettings):
         description="Export timeout in milliseconds",
     )
 
+    insecure: bool = Field(
+        default=True,
+        description="Use insecure (non-TLS) gRPC connection (default: True for local dev)",
+    )
+
 
 class PhoenixSettings(BaseSettings):
     """
@@ -267,8 +273,8 @@ class PhoenixSettings(BaseSettings):
     )
 
     enabled: bool = Field(
-        default=False,
-        description="Enable Phoenix integration (disabled by default for local dev)",
+        default=True,
+        description="Enable Phoenix integration (enabled by default)",
     )
 
     base_url: str = Field(
@@ -686,6 +692,91 @@ class S3Settings(BaseSettings):
     )
 
 
+class DataLakeSettings(BaseSettings):
+    """
+    Data lake settings for experiment and dataset storage.
+
+    Data Lake Convention:
+        The data lake provides a standardized structure for storing datasets,
+        experiments, and calibration data in S3. Users bring their own bucket
+        and the version is pinned by default to v0 in the path.
+
+    S3 Path Structure:
+        s3://{bucket}/{version}/datasets/
+        ├── raw/                        # Raw source data + transformers
+        │   └── {dataset_name}/         # e.g., cns_drugs, codes, care
+        ├── tables/                     # Database table data (JSONL)
+        │   ├── resources/              # → resources table
+        │   │   ├── drugs/{category}/   # Psychotropic drugs
+        │   │   ├── care/stages/        # Treatment stages
+        │   │   └── crisis/             # Crisis resources
+        │   └── codes/                  # → codes table
+        │       ├── icd10/{category}/   # ICD-10 codes
+        │       └── cpt/                # CPT codes
+        └── calibration/                # Agent calibration
+            ├── experiments/            # Experiment configs + results
+            │   └── {agent}/{task}/     # e.g., siggy/risk-assessment
+            └── datasets/               # Shared evaluation datasets
+
+    Experiment Storage:
+        - Local: experiments/{agent}/{task}/experiment.yaml
+        - S3: s3://{bucket}/{version}/datasets/calibration/experiments/{agent}/{task}/
+
+    Environment variables:
+        DATA_LAKE__BUCKET_NAME - S3 bucket for data lake (required)
+        DATA_LAKE__VERSION - Path version prefix (default: v0)
+        DATA_LAKE__DATASETS_PREFIX - Datasets directory (default: datasets)
+        DATA_LAKE__EXPERIMENTS_PREFIX - Experiments subdirectory (default: experiments)
+    """
+
+    model_config = SettingsConfigDict(
+        env_prefix="DATA_LAKE__",
+        env_file=".env",
+        env_file_encoding="utf-8",
+        extra="ignore",
+    )
+
+    bucket_name: str | None = Field(
+        default=None,
+        description="S3 bucket for data lake storage (user-provided)",
+    )
+
+    version: str = Field(
+        default="v0",
+        description="API version for data lake paths",
+    )
+
+    datasets_prefix: str = Field(
+        default="datasets",
+        description="Root directory for datasets in the bucket",
+    )
+
+    experiments_prefix: str = Field(
+        default="experiments",
+        description="Subdirectory within calibration for experiments",
+    )
+
+    def get_base_uri(self) -> str | None:
+        """Get the base S3 URI for the data lake."""
+        if not self.bucket_name:
+            return None
+        return f"s3://{self.bucket_name}/{self.version}/{self.datasets_prefix}"
+
+    def get_experiment_uri(self, agent: str, task: str = "general") -> str | None:
+        """Get the S3 URI for an experiment."""
+        base = self.get_base_uri()
+        if not base:
+            return None
+        return f"{base}/calibration/{self.experiments_prefix}/{agent}/{task}"
+
+    def get_tables_uri(self, table: str = "resources") -> str | None:
+        """Get the S3 URI for a table directory."""
+        base = self.get_base_uri()
+        if not base:
+            return None
+        return f"{base}/tables/{table}"
+
+
 class ChunkingSettings(BaseSettings):
     """
     Document chunking settings for semantic text splitting.
@@ -1051,10 +1142,26 @@ class ModelsSettings(BaseSettings):
 
     @property
     def module_list(self) -> list[str]:
-        """Get modules as a list, filtering empty strings."""
-        if not self.import_modules:
-            return []
-        return [m.strip() for m in self.import_modules.split(";") if m.strip()]
+        """
+        Get modules as a list, filtering empty strings.
+
+        Auto-detects ./models folder if it exists and is importable.
+        """
+        modules = []
+        if self.import_modules:
+            modules = [m.strip() for m in self.import_modules.split(";") if m.strip()]
+
+        # Auto-detect ./models if it exists and is a Python package (convention over configuration)
+        from pathlib import Path
+
+        models_path = Path("./models")
+        if models_path.exists() and models_path.is_dir():
+            # Check if it's a Python package (has __init__.py)
+            if (models_path / "__init__.py").exists():
+                if "models" not in modules:
+                    modules.insert(0, "models")
+
+        return modules
 
 
 class SchemaSettings(BaseSettings):
@@ -1240,6 +1347,110 @@ class GitSettings(BaseSettings):
     )
 
 
+class DBListenerSettings(BaseSettings):
+    """
+    PostgreSQL LISTEN/NOTIFY database listener settings.
+
+    The DB Listener is a lightweight worker that subscribes to PostgreSQL
+    NOTIFY events and dispatches them to external systems (SQS, REST, custom).
+
+    Architecture:
+        - Single-replica deployment (to avoid duplicate processing)
+        - Dedicated connection for LISTEN (not from connection pool)
+        - Automatic reconnection with exponential backoff
+        - Graceful shutdown on SIGTERM
+
+    Use Cases:
+        - Sync data changes to external systems (Phoenix, webhooks)
+        - Trigger async jobs without polling
+        - Event-driven architectures with PostgreSQL as event source
+
+    Example PostgreSQL trigger:
+        CREATE OR REPLACE FUNCTION notify_feedback_insert()
+        RETURNS TRIGGER AS $$
+        BEGIN
+            PERFORM pg_notify('feedback_sync', json_build_object(
+                'id', NEW.id,
+                'table', 'feedbacks',
+                'action', 'insert'
+            )::text);
+            RETURN NEW;
+        END;
+        $$ LANGUAGE plpgsql;
+
+    Environment variables:
+        DB_LISTENER__ENABLED - Enable the listener worker (default: false)
+        DB_LISTENER__CHANNELS - Comma-separated PostgreSQL channels to listen on
+        DB_LISTENER__HANDLER_TYPE - Handler type: 'sqs', 'rest', or 'custom'
+        DB_LISTENER__SQS_QUEUE_URL - SQS queue URL (for handler_type=sqs)
+        DB_LISTENER__REST_ENDPOINT - REST endpoint URL (for handler_type=rest)
+        DB_LISTENER__RECONNECT_DELAY - Initial reconnect delay in seconds
+        DB_LISTENER__MAX_RECONNECT_DELAY - Maximum reconnect delay in seconds
+
+    References:
+        - PostgreSQL NOTIFY: https://www.postgresql.org/docs/current/sql-notify.html
+        - Brandur's Notifier: https://brandur.org/notifier
+    """
+
+    model_config = SettingsConfigDict(
+        env_prefix="DB_LISTENER__",
+        env_file=".env",
+        env_file_encoding="utf-8",
+        extra="ignore",
+    )
+
+    enabled: bool = Field(
+        default=False,
+        description="Enable the DB Listener worker (disabled by default)",
+    )
+
+    channels: str = Field(
+        default="",
+        description=(
+            "Comma-separated list of PostgreSQL channels to LISTEN on. "
+            "Example: 'feedback_sync,entity_update,user_events'"
+        ),
+    )
+
+    handler_type: str = Field(
+        default="rest",
+        description=(
+            "Handler type for dispatching notifications. Options: "
+            "'sqs' (publish to SQS), 'rest' (POST to endpoint), 'custom' (Python handlers)"
+        ),
+    )
+
+    sqs_queue_url: str = Field(
+        default="",
+        description="SQS queue URL for handler_type='sqs'",
+    )
+
+    rest_endpoint: str = Field(
+        default="http://localhost:8000/api/v1/internal/events",
+        description=(
+            "REST endpoint URL for handler_type='rest'. "
+            "Receives POST with {channel, payload, source} JSON body."
+        ),
+    )
+
+    reconnect_delay: float = Field(
+        default=1.0,
+        description="Initial delay (seconds) between reconnection attempts",
+    )
+
+    max_reconnect_delay: float = Field(
+        default=60.0,
+        description="Maximum delay (seconds) between reconnection attempts (exponential backoff cap)",
+    )
+
+    @property
+    def channel_list(self) -> list[str]:
+        """Get channels as a list, filtering empty strings."""
+        if not self.channels:
+            return []
+        return [c.strip() for c in self.channels.split(",") if c.strip()]
+
+
 class TestSettings(BaseSettings):
     """
     Test environment settings.
@@ -1347,18 +1558,30 @@ class Settings(BaseSettings):
     migration: MigrationSettings = Field(default_factory=MigrationSettings)
     storage: StorageSettings = Field(default_factory=StorageSettings)
     s3: S3Settings = Field(default_factory=S3Settings)
+    data_lake: DataLakeSettings = Field(default_factory=DataLakeSettings)
     git: GitSettings = Field(default_factory=GitSettings)
     sqs: SQSSettings = Field(default_factory=SQSSettings)
+    db_listener: DBListenerSettings = Field(default_factory=DBListenerSettings)
     chunking: ChunkingSettings = Field(default_factory=ChunkingSettings)
     content: ContentSettings = Field(default_factory=ContentSettings)
     schema_search: SchemaSettings = Field(default_factory=SchemaSettings)
     test: TestSettings = Field(default_factory=TestSettings)
 
 
+# Auto-load .env file from current directory if it exists
+# This happens BEFORE config file loading, so .env takes precedence
+from pathlib import Path
+from dotenv import load_dotenv
+
+_dotenv_path = Path(".env")
+if _dotenv_path.exists():
+    load_dotenv(_dotenv_path, override=False)  # Don't override existing env vars
+    logger.debug(f"Loaded environment from {_dotenv_path.resolve()}")
+
 # Load configuration from ~/.rem/config.yaml before initializing settings
 # This allows user configuration to be merged with environment variables
-# Set REM_SKIP_CONFIG_FILE=true to disable (useful for development with .env)
-if not os.getenv("REM_SKIP_CONFIG_FILE", "").lower() in ("true", "1", "yes"):
+# Set REM_SKIP_CONFIG=1 to disable (useful for development with .env)
+if not os.getenv("REM_SKIP_CONFIG", "").lower() in ("true", "1", "yes"):
     try:
         from rem.config import load_config, merge_config_to_env