Skip to main content

A dual vector foil (二向箔) that squashes any Python objects into your console.

Project description


Build Status PyPI version

A dual vector foil(二向箔) that squashes any Python objects into your console


Simply speaking, dvf (dual vector foil) is a recursive pretty printer for any objects in Python. It allows you to inspect Python object in a simple and comprehensive way. Checkout the following example:


Example on Flask app

Flask app is very complex Python object, and dvf can use paging (less) to wrap the output. If the GIF doesn't show immediately, be patient, it's about 10 MB large (click to zoom in):


If your eyes are sharp enough, you'll find a warning at the end of the gif. That's because dvf tries to access some attributes of Flask that are only valid in a request context. The warning is quite common for complex objects.


pip install dvf

The project is still under development, so any report on bugs is highly appreciated. The development is under Python 3.7 and Python 3.6 is also tested. The package provides no Python <= 3.5 support.


Why not dir or __dict__

There is already an amazing inspection package pdir, which emphasize on the usage of modules and objects, while dvf is aiming at data and internal structure of objects. As a result, dvf will by default omit any object attributes that have type of function, module or class, and will try its best to expand any iterable to see what really lies in .

Safety concern

As you might have gaused, it's not wise to use dvf on untrusted object because dvf will have to evoke some methods of the object to evaluate attributes. Is this a foundamental flaw of dvf? I think not. Because if an object is really malicious, it can delete your system when it's imported, why wait untill dvf to check it?

Deal with loops

The biggest problem of dvf is loops in objects. The following class has a pointer points to himself. A simple recursion implementation of dvf will result in an infinite loop.

class Foo:

    def __init__(self):
        self.another_me = self

To solve this economically, dvf records every object it has visited and omit them next time it meet the object. That's why sometimes a complete view of certain objects is not possible.

Another troublesome case is object creation during attribute access. A typical example is NumPy array, which has an attribute of T that returns the transpose of the array, which has another T that returns another new array. So there is also an infinite loop. To solve this dvf should be very cautious toward data descriptors. Some result gained from descriptors will not be expanded.

Todo list

  • User-custom searching and filtering
  • Docs on output format
  • Tests
  • Use prompt_toolkit to build an application that can handle wide output (horizontally scollable)

Project details

Download files

Download the file for your platform. If you're not sure which to choose, learn more about installing packages.

Source Distribution

dvf-0.1.2.tar.gz (7.9 kB view hashes)

Uploaded source

Built Distribution

dvf-0.1.2-py3-none-any.whl (9.6 kB view hashes)

Uploaded py3

Supported by

AWS AWS Cloud computing and Security Sponsor Datadog Datadog Monitoring Fastly Fastly CDN Google Google Download Analytics Microsoft Microsoft PSF Sponsor Pingdom Pingdom Monitoring Sentry Sentry Error logging StatusPage StatusPage Status page