Quick IPython tutorial

IPython can be used as an improved replacement for the Python prompt, and for that you don’t really need to read any more of this manual. But in this section we’ll try to summarize a few tips on how to make the most effective use of it for everyday Python development, highlighting things you might miss in the rest of the manual (which is getting long). We’ll give references to parts in the manual which provide more detail when appropriate.

The following article by Jeremy Jones provides an introductory tutorial about IPython: http://www.onlamp.com/pub/a/python/2005/01/27/ipython.html


Tab completion

TAB-completion, especially for attributes, is a convenient way to explore the structure of any object you’re dealing with. Simply type object_name.<TAB> and a list of the object’s attributes will be printed (see the readline section for more). Tab completion also works on file and directory names, which combined with IPython’s alias system allows you to do from within IPython many of the things you normally would need the system shell for.

Explore your objects

Typing object_name? will print all sorts of details about any object, including docstrings, function definition lines (for call arguments) and constructor details for classes. The magic commands %pdoc, %pdef, %psource and %pfile will respectively print the docstring, function definition line, full source code and the complete file for any object (when they can be found). If automagic is on (it is by default), you don’t need to type the ‘%’ explicitly. See this section for more.

The %run magic command

The %run magic command allows you to run any python script and load all of its data directly into the interactive namespace. Since the file is re-read from disk each time, changes you make to it are reflected immediately (in contrast to the behavior of import). I rarely use import for code I am testing, relying on %run instead. See this section for more on this and other magic commands, or type the name of any magic command and ? to get details on it. See also this section for a recursive reload command. %run also has special flags for timing the execution of your scripts (-t) and for executing them under the control of either Python’s pdb debugger (-d) or profiler (-p). With all of these, %run can be used as the main tool for efficient interactive development of code which you write in your editor of choice.

Debug a Python script

Use the Python debugger, pdb. The %pdb command allows you to toggle on and off the automatic invocation of an IPython-enhanced pdb debugger (with coloring, tab completion and more) at any uncaught exception. The advantage of this is that pdb starts inside the function where the exception occurred, with all data still available. You can print variables, see code, execute statements and even walk up and down the call stack to track down the true source of the problem (which often is many layers in the stack above where the exception gets triggered). Running programs with %run and pdb active can be an efficient to develop and debug code, in many cases eliminating the need for print statements or external debugging tools. I often simply put a 1/0 in a place where I want to take a look so that pdb gets called, quickly view whatever variables I need to or test various pieces of code and then remove the 1/0. Note also that ‘%run -d’ activates pdb and automatically sets initial breakpoints for you to step through your code, watch variables, etc. The output caching section has more details.

Use the output cache

All output results are automatically stored in a global dictionary named Out and variables named _1, _2, etc. alias them. For example, the result of input line 4 is available either as Out[4] or as _4. Additionally, three variables named _, __ and ___ are always kept updated with the for the last three results. This allows you to recall any previous result and further use it for new calculations. See the output caching section for more.

Suppress output

Put a ‘;’ at the end of a line to suppress the printing of output. This is useful when doing calculations which generate long output you are not interested in seeing. The _* variables and the Out[] list do get updated with the contents of the output, even if it is not printed. You can thus still access the generated results this way for further processing.

Input cache

A similar system exists for caching input. All input is stored in a global list called In , so you can re-execute lines 22 through 28 plus line 34 by typing ‘exec In[22:29]+In[34]’ (using Python slicing notation). If you need to execute the same set of lines often, you can assign them to a macro with the %macro function. See here for more.

Use your input history

The %hist command can show you all previous input, without line numbers if desired (option -n) so you can directly copy and paste code either back in IPython or in a text editor. You can also save all your history by turning on logging via %logstart; these logs can later be either reloaded as IPython sessions or used as code for your programs.

Define your own system aliases

Even though IPython gives you access to your system shell via the ! prefix, it is convenient to have aliases to the system commands you use most often. This allows you to work seamlessly from inside IPython with the same commands you are used to in your system shell. IPython comes with some pre-defined aliases and a complete system for changing directories, both via a stack (see %pushd, %popd and %dhist) and via direct %cd. The latter keeps a history of visited directories and allows you to go to any previously visited one.

Call system shell commands

Use Python to manipulate the results of system commands. The ‘!!’ special syntax, and the %sc and %sx magic commands allow you to capture system output into Python variables.

Use Python variables when calling the shell

Expand python variables when calling the shell (either via ‘!’ and ‘!!’ or via aliases) by prepending a $ in front of them. You can also expand complete python expressions. See our shell section for more details.

Use profiles

Use profiles to maintain different configurations (modules to load, function definitions, option settings) for particular tasks. You can then have customized versions of IPython for specific purposes. This section has more details.

Embed IPython in your programs

A few lines of code are enough to load a complete IPython inside your own programs, giving you the ability to work with your data interactively after automatic processing has been completed. See here for more.

Use the Python profiler

When dealing with performance issues, the %run command with a -p option allows you to run complete programs under the control of the Python profiler. The %prun command does a similar job for single Python expressions (like function calls).

Use IPython to present interactive demos

Use the IPython.demo.Demo class to load any Python script as an interactive demo. With a minimal amount of simple markup, you can control the execution of the script, stopping as needed. See here for more.

Run doctests

Run your doctests from within IPython for development and debugging. The special %doctest_mode command toggles a mode where the prompt, output and exceptions display matches as closely as possible that of the default Python interpreter. In addition, this mode allows you to directly paste in code that contains leading ‘>>>’ prompts, even if they have extra leading whitespace (as is common in doctest files). This combined with the ‘%history -tn’ call to see your translated history (with these extra prompts removed and no line numbers) allows for an easy doctest workflow, where you can go from doctest to interactive execution to pasting into valid Python code as needed.

Source code handling tips

IPython is a line-oriented program, without full control of the terminal. Therefore, it doesn’t support true multiline editing. However, it has a number of useful tools to help you in dealing effectively with more complex editing.

The %edit command gives a reasonable approximation of multiline editing, by invoking your favorite editor on the spot. IPython will execute the code you type in there as if it were typed interactively. Type %edit? for the full details on the edit command.

If you have typed various commands during a session, which you’d like to reuse, IPython provides you with a number of tools. Start by using %hist to see your input history, so you can see the line numbers of all input. Let us say that you’d like to reuse lines 10 through 20, plus lines 24 and 28. All the commands below can operate on these with the syntax:

%command 10-20 24 28

where the command given can be:

  • %macro <macroname>: this stores the lines into a variable which, when called at the prompt, re-executes the input. Macros can be edited later using ‘%edit macroname’, and they can be stored persistently across sessions with ‘%store macroname’ (the storage system is per-profile). The combination of quick macros, persistent storage and editing, allows you to easily refine quick-and-dirty interactive input into permanent utilities, always available both in IPython and as files for general reuse.
  • %edit: this will open a text editor with those lines pre-loaded for further modification. It will then execute the resulting file’s contents as if you had typed it at the prompt.
  • %save <filename>: this saves the lines directly to a named file on disk.

While %macro saves input lines into memory for interactive re-execution, sometimes you’d like to save your input directly to a file. The %save magic does this: its input sytnax is the same as %macro, but it saves your input directly to a Python file. Note that the %logstart command also saves input, but it logs all input to disk (though you can temporarily suspend it and reactivate it with %logoff/%logon); %save allows you to select which lines of input you need to save.

Lightweight ‘version control’

When you call %edit with no arguments, IPython opens an empty editor with a temporary file, and it returns the contents of your editing session as a string variable. Thanks to IPython’s output caching mechanism, this is automatically stored:

In [1]: %edit

IPython will make a temporary file named: /tmp/ipython_edit_yR-HCN.py

Editing... done. Executing edited code...

hello - this is a temporary file

Out[1]: "print 'hello - this is a temporary file'\n"

Now, if you call ‘%edit -p’, IPython tries to open an editor with the same data as the last time you used %edit. So if you haven’t used %edit in the meantime, this same contents will reopen; however, it will be done in a new file. This means that if you make changes and you later want to find an old version, you can always retrieve it by using its output number, via ‘%edit _NN’, where NN is the number of the output prompt.

Continuing with the example above, this should illustrate this idea:

In [2]: edit -p

IPython will make a temporary file named: /tmp/ipython_edit_nA09Qk.py

Editing... done. Executing edited code...

hello - now I made some changes

Out[2]: "print 'hello - now I made some changes'\n"

In [3]: edit _1

IPython will make a temporary file named: /tmp/ipython_edit_gy6-zD.py

Editing... done. Executing edited code...

hello - this is a temporary file

IPython version control at work :)

Out[3]: "print 'hello - this is a temporary file'\nprint 'IPython version control at work :)'\n"

This section was written after a contribution by Alexander Belchenko on the IPython user list.

Effective logging

A very useful suggestion sent in by Robert Kern follows:

I recently happened on a nifty way to keep tidy per-project log files. I made a profile for my project (which is called “parkfield”):

include ipythonrc

# cancel earlier logfile invocation:

logfile ''

execute import time

execute __cmd = '/Users/kern/research/logfiles/parkfield-%s.log rotate'

execute __IP.magic_logstart(__cmd % time.strftime('%Y-%m-%d'))

I also added a shell alias for convenience:

alias parkfield="ipython -pylab -profile parkfield"

Now I have a nice little directory with everything I ever type in, organized by project and date.

Logging to a file

Here is an alternative logging solution that lets you record your sessions in a daily time-stamped log-files.

Add the following lines or make it a-like to your ipy_user_conf.py:

from time import strftime

def main():

        ldir = '/home/$YOUR_USERNAME_HERE/.ipython/'
        filename = os.path.join(ldir, strftime('%Y-%m-%d')+".py")
        notnew = os.path.exists(filename)
        ip.IP.logger.logstart(logfname=filename, logmode='append')
        log_write = ip.IP.logger.log_write
        if notnew:
            log_write("# =================================")
            log_write("#!/usr/bin/env python \n# %s.py \n"
                      "# IPython automatic logging file" %
        log_write("# %s \n# =================================" %
        print " Logging to "+filename

    except RuntimeError:
        print " Already logging to "+ip.IP.logger.logfname

Contribute your own: If you have your own favorite tip on using IPython efficiently for a certain task (especially things which can’t be done in the normal Python interpreter), don’t hesitate to send it!