Python Debugging Notes

The most basic way for debugging python would be to print variable information during the code execution, but we all know that this has some implications that make the rest of development a more sadistic process, as it introduces code that will have to be removed before the code goes to products.

Debugging live code can be improved by using breakpoints, calling the debugger within the code brings up a shell where you can use the debugger commands. Some of the tools I use are (you can install them with pip, package names are in parenthesis):

The pdb and ipdb work are fairly similar, but the ipdb has some great features that come with ipython, such as autocomplete and highlights.

The most different here is the PuDB, that features some kind of a full UI for debugging, see the image bellow.

Now I wrote the following code the I need to step-by-step debug, so let's use each tool to check the values for i.

 1 #!/bin/python
 3 def my_loop():
 4     for i in range(20):
 5         import pdb; pdb.set_trace()
 6         print i
 8 def main():
 9     my_loop()
11 if __name__ == "__main__":
12     main()


To create a breakpoint using pdb just past the following code:

import pdb; pdb.set_trace()

That puts us into the shell:

> .../tmp/
-> print i
(Pdb) _

Now if we enter ? and hit enter we can see the help for the documented commands.

(Pdb) ?

Documented commands (type help <topic>):
EOF    bt         cont      enable  jump  pp       run      unt
a      c          continue  exit    l     q        s        until
alias  cl         d         h       list  quit     step     up
args   clear      debug     help    n     r        tbreak   w
b      commands   disable   ignore  next  restart  u        whatis
break  condition  down      j       p     return   unalias  where

Miscellaneous help topics:
exec  pdb

Undocumented commands:
retval  rv

One command that can save you when you are lost is the list command. It basically prints the source code, for your current location, with a few lines before and after.

(Pdb) list
  1     #!/bin/python
  3     def my_loop():
  4         for i in range(20):
  5             import pdb; pdb.set_trace()
  6  ->         print i
  8     def main():
  9         my_loop()
 11     if __name__ == "__main__":

Basic usage commands:

  • c or continue to go to the next break point (if there is no other break point it will finish execution);
  • n or next will go to next line
  • s or step will make a "step into" kind of thing
  • u or up will "step up"

You can also call your script as argument for pdb, it will start it in the debugger where you can use all commands from the beginning of your script.


ipdb is very similar to pdb but has all the cool features from the ipython shell. From the start you will see, at least, better colors. It also has the same commands as the pdb, so this is just a matter of test, or project limitations. You can always enter help in the command line to get a more detailed list of the "documented" commands.


Let's start our script from command line with pudb by calling pudb

You can see that it opens a really nice and busy view of the code we are in. You can call help by typing "?". The hotkeys n, s, c and u have the same meaning, so it feels like home.

One cool thing you can do with this that is easier in PuDB is that by hitting b you can set a break point where the cursor is. You can also call <ctrl>+p to edit your preferences. Even set the default python shell to ipython, if you want.

To call it, as a breakpoint, from within your code, you can paste the line bellow where you want it to pause execution and bring up the interface.

from pudb import set_trace; set_trace()

Celery Remote Debugging

For the lack of a better place to add this note, and as this is also debugging. The documentation is at

To remote debug celery tasks you basically open give rdb a port, then you can connect with telnet.

from celery.contrib import rdb
rdb = rdb.Rdb(port=6899, port_search_limit=100, port_skew=0)

rdb.set_trace() # set breakpoint

comments powered by Disqus
Fork me on GitHub