flogging

Structured Logging Documentation

This module provides structured logging capabilities that allow you to output either human-readable logs or JSON-formatted structured logs. These features are especially useful for debugging in development and for generating easily parseable logs in production environments.

Overview

The logging setup supports two main modes:

1. Human-readable logs: Provides logs in a colored, properly formatted output for local development.
2. Structured logs: Outputs logs as JSON objects, with each log record represented as a single line. This mode is ideal for production environments where logs need to be processed by log aggregation systems.

Key Functions

• add_logging_args(): Adds logging configuration options to an argparse.ArgumentParser instance, making it easy to configure logging via command-line arguments.
• setup(): Directly configures logging by specifying the logging level and format (structured or human-readable).
• set_context(): Assigns a custom context to be logged with each message in structured logging mode.
• log_multipart(): Logs large messages by splitting them into chunks and compressing the data.

---

Usage

Basic Setup with setup()

To initialize logging in your application, call the setup() function. You can specify whether to enable structured logging or use the default human-readable format.

from my_logging_module import setup

# Initialize logging
setup(level="INFO", structured=False)  # Human-readable format

# Enable structured logging for production
setup(level="INFO", structured=True)

Parameters for setup()

• level: The logging level (e.g., "DEBUG", "INFO", "WARNING", "ERROR"). This can be a string or an integer.
• structured: A boolean that controls whether structured logging is enabled. Set to True for JSON logs.
• allow_trailing_dot: Prevents log messages from having a trailing dot unless explicitly allowed.
• level_from_msg: An optional function to dynamically change the logging level based on the content of the message.
• ensure_utf8_streams: Ensures that stdout and stderr use UTF-8 encoding.

Adding Logging Arguments with add_logging_args()

You can easily integrate logging configuration options into your command-line interface using add_logging_args(). This function automatically adds command-line flags for setting the logging level and format.

Command-Line Flags

• --log-level: Set the logging verbosity (e.g., "DEBUG", "INFO", "WARNING").
• --log-structured: Enable structured logging (outputs logs in JSON format).

Environment Variables

You can also set the logging level and format using environment variables:

• LOG_LEVEL: Set the logging level.
• LOG_STRUCTURED: Enable structured logging.

LOG_LEVEL=DEBUG LOG_STRUCTURED=1 python my_app.py

Structured Logging in Production

To enable structured logging (JSON logs), you can either set the --log-structured flag when running your application or configure it programmatically using setup():

python my_app.py --log-level DEBUG --log-structured

In structured logging mode, each log entry is a JSON object with the following fields:

• level: The log level (e.g., "info", "error").
• msg: The log message.
• source: The file and line number where the log occurred.
• time: The timestamp of the log event.
• thread: The thread ID in a shortened format.
• name: The logger name.

Example structured log output:

{
  "level": "info",
  "msg": "Application started",
  "source": "app.py:42",
  "time": "2023-09-23T14:22:35.000+00:00",
  "thread": "f47c",
  "name": "my_app"
}

Custom Context with set_context()

In structured logging mode, you can attach additional context to each log message by calling set_context(). This context is logged alongside the usual fields, allowing you to track custom metadata.

from my_logging_module import set_context

# Set custom context
set_context({"user_id": "12345", "transaction_id": "abcde"})

# The custom context will now appear in each structured log message

Handling Large Log Messages with log_multipart()

When logging large messages (e.g., serialized data or files), the log_multipart() function compresses and splits the message into smaller chunks to prevent issues with log size limits.

from my_logging_module import log_multipart

# Log a large message
log_multipart(logging.getLogger(), b"Large data to be logged")

This function will automatically split the message and log each chunk, ensuring the entire message is captured.

---

Customizing the Logging Format

Human-Readable Logs

By default, when not using structured logging, logs are output in a colored format, with color-coding based on the log level:

• DEBUG: Gray
• INFO: Cyan
• WARNING: Yellow
• ERROR/CRITICAL: Red

You can further customize the format by modifying the AwesomeFormatter class, which is used for formatting logs in human-readable mode. It also shortens thread IDs for easier readability.

Enforcing Logging Standards

To enforce standards in your logging messages, such as preventing trailing dots in log messages, the module provides the check_trailing_dot() decorator. This can be applied to logging functions to raise an error if a message ends with a dot:

from my_logging_module import check_trailing_dot

@check_trailing_dot
def log_message(record):
    # Your custom logging logic
    pass

---

Best Practices

• Use human-readable logs in development for easier debugging.
• Switch to structured logging in production to enable easier parsing and aggregation by log management tools.
• Set custom contexts to include additional metadata in your logs, such as user IDs or request IDs, to improve traceability in production.
• Use multipart logging to handle large log messages that might otherwise exceed log size limits.

---

Example

Here's a full example of how to use structured logging with command-line configuration:

import argparse
import logging
from flogging import add_logging_args, set_context, setup

# Initialize logging
setup(level="INFO", structured=False)  # Human-readable format
# Create argument parser
parser = argparse.ArgumentParser(description="My Application")
add_logging_args(parser)

# Parse arguments and setup logging
args = parser.parse_args()

# Set additional context for structured logging
set_context({"request_id": "123abc"})

# Start logging messages
logger = logging.getLogger("my_app")
logger.info("Application started")