Brandon Invergo

It seems that many, if not most people nowadays use IPython instead of the default Python interpreter. I’ve used it briefly in the past but, while I found a few features to be useful, such as the persistent command history and tab-completion, I found it to be otherwise quite bloated with features that I would never use (and which, furthermore, employ a strange and non-Python meta-syntax).

It appears to be an under-appreciated fact that you can run a script automatically when you launch the Python interpreter. I certainly did not know about it for some time. Using this feature, I’ve found that I can implement a few basic features in the standard Python interpreter that greatly improve its usage (for me), without going full hog with other features. The script is available on Gitlab.

Many of its capabilities should come as no surprise to people who are familiar with Python: tab-completion, persistent history, etc. Others are non-standard and required a bit of hacking to get right: syntax-highlighted completion results, auto-indentation, and printable functions.

To use this script, just set the PYTHONSTARTUP environment variable to point to it, e.g. if the file is in $HOME/.config/python/ put this in your .bashrc:

export PYTHONSTARTUP=$HOME/.config/python/python-startup.py

The script is compatible with both Python 2 and Python 3.

I highly recommend reading the code and modifying it to suit your tastes. Of course, if you hack up some cool new feature, I’d love to hear about it [1].

Features

Pre-loaded modules

This comes as a bit of a side-effect: in order to implement some of the functions in the start-up script, some modules had to be imported. The benefit of this is that you won’t need to import some modules in your interactive sessions, particularly os.path and sys.

Also, for Python 2, the print function is imported from __future__. Additionally, the pprint function is automatically imported. pprint pretty-prints objects, making it much easier to view large lists or dicts, for example.

Persistent history

Your command history is saved between sessions, so you can recall commands that you previously entered. Note that your Python 2 and Python 3 histories are stored separately (though this is easy enough to change if you don’t want that). By default, the command histories are saved in $HOME/.local/share/python.

Colorized, history-aware prompts

Rather than the traditional >>> (standard) and ... (continuation) prompts, this script provides prompts that show the current line in the command history file. They are also colorized: the standard prompt is green while the continuation prompt is yellow. These colors are easily customized in the script.

Before:

>>> def foo(bar):
...     print(bar)

After (you’ll have to imagine the colors in your mind):

[26]> def foo(bar):
(26)>     print(bar)

Highlighted tab-completion of commands

When you are entering commands, you can hit the <Tab> key to complete a command.

That’s not terribly new and surprising. However, this script also highlights the completion results according to their syntactic function: modules in yellow, functions in blue, etc. It’s surprisingly useful.

Indentation keyboard shortcuts

Since indentation is so important in Python but since we are occupying <Tab> with completion, I added some handy shortcuts to handle indentation for you. Ctrl-j will indent four spaces while Ctrl-u will unindent four spaces. These commands behave the way you would expect: you can use them with the cursor residing anywhere in the line, even in the middle of text (as opposed to having to move to the beginning of the line and entering the spaces manually) [2].

Context-specific auto-indentation

To save you some keystrokes, your commands will be automatically indented as necessary. For example, if you enter an if command, then the next line will automatically indent itself one level inwards. If you then unindent a level, all lines after that will be at the new indentation level. Note that the indentation will happen as soon as you start typing, not as soon as the prompt appears like you might expect [3].

Printable functions

This one’s a bit non-standard, so it bears some explaining. I found that sometimes in my interactive sessions, I had forgotten the specific details of a function definition. However, going back through the command history would only show it one line at a time, and in reverse at that. So, the start-up script provides a function decorator, saved_function that allows you to recall and print the definition of a function that was entered at the command-line [4]. Simply pass it the line in the command history where the function definition will begin (i.e. the next line after the one where the decorator is being entered).

For example:

[27]> @saved_function(28)
(27)> def foo(bar):
(27)>     print(bar)
[30]> # ...other stuff...
# later... more work...hey, what does function foo do again?
[48]> print(foo)
def foo(bar):
    print(bar)

Footnotes: