Introduction

This module extendedlogging provides a layer on top of autologging which extends standard python logging.

The main purpose is to improve usability: make it simpler to configure tracing to file in combination with console logging.

Some extra features:

• tracing lines are by default never split over multiple lines - newlines in strings are logged as '\n'
• tracing timestamps are in microsecond resolution instead of milliseconds
• tracing lines have a configurable maximum size
• exceptions are logged in trace file automatically as logging.ERROR

Demo

Tracing

Given the following code:

# demo.py

import extendedlogging

# define a little example class and instrument all its functions with autologging
@extendedlogging.traced
class myclass():
    def __init__(self, *args, **kwargs):
        self.f()
        extendedlogging.debug('__init__ done')
    def f(self):
        extendedlogging.info('hi!')
        pass
    def g(self):
        return 3

# configure tracing
extendedlogging.configure(tracing=True)

# run some code to produce logging events
c = myclass('some_argument')
c.g()
extendedlogging.warning('done')

With the given code, the logging to stdout looks something like:

INFO   :f:hi!
WARNING:<module>:done

With the given code, the tracing to file looks something like:

2022-04-09 14:06:00,855471:TRACE:demo.py,6:root.__init__:CALL *('some_argument',) **{}
2022-04-09 14:06:00,855649:TRACE:demo.py,9:root.f:CALL *() **{}
2022-04-09 14:06:00,855793:INFO:demo.py,10:root.f:hi!
2022-04-09 14:06:00,855901:TRACE:demo.py,9:root.f:RETURN None
2022-04-09 14:06:00,856035:DEBUG:demo.py,8:root.__init__:__init__ done
2022-04-09 14:06:00,856065:TRACE:demo.py,6:root.__init__:RETURN None
2022-04-09 14:06:00,856093:TRACE:demo.py,12:root.g:CALL *() **{}
2022-04-09 14:06:00,856115:TRACE:demo.py,12:root.g:RETURN 3
2022-04-09 14:06:00,856161:WARNING:demo.py,21:root.<module>:done

For more demo's, see the code snippets in the folders demos and tests.

When your program is multithreaded or using multiprocessing, then the flags tread_names resp. process_names need to be enabled explicitly in the configuration, to prevent garbage/inconsistent tracing.

Viewer

A Trace-and-Timing viewer ttviewer is also included, which can visualize log files. Consider the fibonacci demo program demo_fib.py.

Multiprocessing/multithreading demo:

Testing, dependencies

• to install dependencies, run:
  • pip install -r REQUIREMENTS.pip
  • (this assumes pip is pip3 - it should also work for python2 though)
• to run all tests locally, run: python tests/test_extendedlogging.py
• the ttviewer tooling requires catapult and a browser (default google-chrome, also firefox seems to work)