logzai-otlp 1.0.6

Lightweight OpenTelemetry logging client for LogzAI

logzai-otlp

Lightweight OpenTelemetry logging client for LogzAI.

• Transport: OTLP over HTTP or gRPC
• Structured logs: pass arbitrary keyword args as attributes
• Graceful shutdown: buffers flushed on process exit

Installation

pip install logzai-otlp

Quick start (HTTP OTLP)

from logzai_otlp import logzai

logzai.init(
    ingest_token="your-ingest-token",
    ingest_endpoint="https://ingest.logzai.com",  # Base OTLP endpoint
    service_name="orders-api",
    environment="prod",
    protocol="http",  # default
)

logzai.info("User logged in", user_id="123", method="oauth")
logzai.error("Payment failed", order_id="42", reason="card_declined")

# optional – logs are flushed automatically at process exit
logzai.shutdown()

Using standard logging

This client installs an OpenTelemetry handler on the logzai logger. You can use the standard library logging API if you prefer:

import logging
from logzai_otlp import logzai

logzai.init(ingest_token="your-token", ingest_endpoint="https://ingest.logzai.com")

logger = logging.getLogger("logzai")
logger.info("Inventory updated", extra={"sku": "A-42", "qty": 3})

API

logzai.init(
    ingest_token: str,
    ingest_endpoint: str = "http://ingest.logzai.com",
    min_level: int = logging.DEBUG,
    *,
    service_name: str = "app",
    service_namespace: str = "default",
    environment: str = "prod",
    protocol: str = "http",  # "http" | "grpc"
    mirror_to_console: bool = False,
) -> None

• ingest_token: Your LogzAI ingest token; sent as x-ingest-token header
• ingest_endpoint: Base OTLP collector endpoint (paths /logs and /traces are appended automatically)
  • HTTP example: https://ingest.logzai.com
  • gRPC example: ingest.logzai.com:4317
• min_level: Minimum logging level (default: logging.DEBUG)
• service_name: OpenTelemetry service.name resource attribute
• service_namespace: OpenTelemetry service.namespace resource attribute
• environment: OpenTelemetry deployment.environment resource attribute
• protocol: http (default) or grpc
• mirror_to_console: Also log to console/stdout (default: False)

Convenience logging helpers:

logzai.debug(message: str, exc_info: bool = False, **attrs) -> None
logzai.info(message: str, exc_info: bool = False, **attrs) -> None
logzai.warning(message: str, exc_info: bool = False, **attrs) -> None
logzai.warn(message: str, exc_info: bool = False, **attrs) -> None  # alias of warning
logzai.error(message: str, exc_info: bool = True, **attrs) -> None
logzai.critical(message: str, exc_info: bool = True, **attrs) -> None
logzai.exception(message: str, **attrs) -> None  # logs with full exception info
logzai.shutdown() -> None  # flush and shutdown provider

All extra keyword arguments are sent as structured log attributes. The exc_info parameter controls whether exception information is automatically captured when available.

Tracing

LogzAI now includes distributed tracing capabilities alongside logging:

from logzai_otlp import logzai

logzai.init(ingest_token="your-token", ingest_endpoint="https://ingest.logzai.com")

# Create spans manually
span = logzai.start_span("database_query")
span.set_attribute("table", "users")
span.end()

# Use context manager for automatic span lifecycle
with logzai.span("api_request") as span:
    logzai.info("Processing request", request_id="123")
    span.set_attribute("method", "GET")
    span.set_attribute("path", "/api/users")
    # span ends automatically, even on exceptions

Tracing methods:

logzai.start_span(name: str, **kwargs) -> Span
logzai.span(name: str, **kwargs)  # context manager
logzai.set_span_attribute(span: Span, key: str, value: Any) -> None

Plugins

LogzAI supports a plugin system that enables framework integrations and custom functionality. Plugins receive the LogzAI instance and can extend logging/tracing behavior.

Built-in Plugins

LogzAI includes built-in plugins for popular frameworks:

FastAPI Plugin

Automatically logs HTTP requests/responses and creates spans for distributed tracing:

from fastapi import FastAPI
from logzai_otlp import logzai, fastapi_plugin

app = FastAPI()

logzai.init(
    ingest_token="your-token",
    ingest_endpoint="https://ingest.logzai.com",
    service_name="my-api"
)

# Enable FastAPI instrumentation
logzai.plugin('fastapi', fastapi_plugin, {
    "app": app,  # Required: your FastAPI app
    "log_request_body": True,  # Optional: log request bodies
    "slow_request_threshold_ms": 500  # Optional: warn on slow requests
})

@app.post("/login")
async def login(username: str):
    # All logs here are automatically in the POST /login span
    logzai.info("User logging in", username=username)
    return {"status": "ok"}

# Each request creates a span with:
# - Route: POST /login
# - Status code, duration, client IP
# - All logs during request handling

PydanticAI Plugin

Automatically logs all agent usage and messages with distributed tracing:

from logzai_otlp import logzai, pydantic_ai_plugin
from pydantic_ai import Agent

# Initialize LogzAI
logzai.init(
    ingest_token="your-token",
    ingest_endpoint="https://ingest.logzai.com",
    service_name="my-ai-app"
)

# Enable PydanticAI instrumentation
logzai.plugin('pydantic-ai', pydantic_ai_plugin, {
    "include_messages": True  # Optional: include full message history
})

# Use PydanticAI normally - all calls are automatically logged and traced
agent = Agent('openai:gpt-4', instructions="You are helpful")
result = await agent.run("Hello!")

# Creates spans for:
# - Agent execution time (pydantic_ai.agent.run)
# - Span attributes: agent name, model, provider, token usage
#
# Logs will include:
# - Token usage (input/output/total)
# - Model and provider info
# - Full message history (if enabled)
# - User prompts and responses

Creating Custom Plugins

Plugin functions receive the LogzAI instance and optional config, and return an optional cleanup function:

from typing import Optional
from logzai_otlp import logzai

def my_plugin(instance, config: Optional[dict] = None):
    """Plugin receives LogzAI instance and optional config."""

    # Add custom functionality
    def custom_log(message, **kwargs):
        instance.info(f"[CUSTOM] {message}", **kwargs)

    instance.custom_log = custom_log

    # Optional: return cleanup function (sync or async)
    def cleanup():
        if hasattr(instance, 'custom_log'):
            delattr(instance, 'custom_log')

    return cleanup

# Initialize and register
logzai.init(ingest_token="token", ingest_endpoint="https://ingest.logzai.com")
logzai.plugin("my-plugin", my_plugin, {"enabled": True})

# Use plugin
logzai.custom_log("Hello from plugin!")

# Cleanup (optional - happens automatically on shutdown)
logzai.unregister_plugin("my-plugin")

Plugin API

# Register a plugin
logzai.plugin(
    name: str,                    # Unique identifier
    plugin_func: LogzAIPlugin[T], # Plugin function
    config: Optional[T] = None    # Optional configuration
) -> None

# Unregister a plugin
logzai.unregister_plugin(name: str) -> bool

# Plugins are automatically cleaned up on shutdown
logzai.shutdown()

Plugin Lifecycle:

• Plugins execute immediately upon registration
• Can return optional cleanup function (sync or async)
• Cleaned up in reverse order (LIFO) during shutdown
• Duplicate registration replaces existing plugin with warning

Requirements

• Python >= 3.9
• opentelemetry-sdk>=1.27.0
• opentelemetry-exporter-otlp>=1.27.0

License

MIT – see LICENSE for details.