Using a debugger and curses at the same time? - python

I'm calling python -m pdb myapp.py, when an exception fires, and I'd normally be thrown back to the pdb interpreter to investigate the problem. However this exception is being thrown after I've called through curses.wrapper() and entered curses mode, rendering the pdb interpreter useless. How can I work around this?

James` answer is a good and I've upvoted it but I'd also consider trying to split the logic and presentation layers of my program. Keep the curses part a thin layer on top of a library and write a simple driver that invokes the correct routines to recreate the error. Then you can dive in and do what's necessary.
Another way I can think of is to create a function called debug or something that throws you back into the regular screen and invokes pdb. Then stick it just before the code that raises the exception and run your program. Something like
def debug(stdscr):
curses.nocbreak()
stdscr.keypad(0)
curses.echo()
curses.endwin()
import pdb; pdb.set_trace()
Apparently, this is similar to what is done with the curses.wrapper function. It's mentioned briefly at http://www.amk.ca/python/howto/curses/.

Not being familiar with Python, this may not be exactly what you want. But apparently, winpdb can attach to a script - just like gdb can to a running process (IIUC).
http://winpdb.org/docs/launch-time/
Don't be mislead by the name, it is platform independent.

use pyclewn
you can use pyclewn with vim.
or use pdb-clone,the core of pyclewn
its good ,its like gdb ,can remote debug

Related

how to halt python program after pdb.set_trace()

When debugging scripts in Python (2.7, running on Linux) I occasionally inject pdb.set_trace() (note that I'm actually using ipdb), e.g.:
import ipdb as pdb
try:
do_something()
# I'd like to look at some local variables before running do_something_dangerous()
pdb.set_trace()
except:
pass
do_something_dangerous()
I typically run my script from the shell, e.g.
python my_script.py
Sometimes during my debugging session I realize that I don't want to run do_something_dangerous(). What's the easiest way to halt program execution so that do_something_dangerous() is not run and I can quit back to the shell?
As I understand it pressing ctrl-d (or issuing the debugger's quit command) will simply exit ipdb and the program will continue running (in my example above). Pressing ctrl-c seems to raise a KeyboardInterrupt but I've never understood the context in which it was raised.
I'm hoping for something like ctrl-q to simply take down the entire process, but I haven't been able to find anything.
I understand that my example is highly contrived, but my question is about how to abort execution from pdb when the code being debugged is set up to catch exceptions. It's not about how to restructure the above code so it works!
I found that ctrl-z to suspend the python/ipdb process, followed by 'kill %1' to terminate the process works well and is reasonably quick for me to type (with a bash alias k='kill %1'). I'm not sure if there's anything cleaner/simpler though.
From the module docs:
q(uit)
Quit from the debugger. The program being executed is aborted.
Specifically, this will cause the next debugger function that gets called to raise a BdbQuit exception.

Why python debugger always get this timeout waiting for response on 113 when using Pycharm?

Bigger image
Especially I run code perhaps running a little long time(10 mins roughly), and hit the break point.
The python debugger always show me this kind of error "timeout waiting for response on 113"
I circle them in red in screencut.
And I use Pycharm as my python IDE, is it just issue for Pycharm IDE? Or Python debugger issue?
And if Pycharm is not recommended, can anyone give me better IDE which be able to debug efficiently.
I had a similar thing happen to me a few months ago, it turned out I had a really slow operation within a __repr__() for a variable I had on the stack. When PyCharm hits a breakpoint it grabs all of the variables in the current scope and calls __repr__ on them. Here's an amusement that demonstrates this issue:
import time
class Foo(object):
def __repr__(self):
time.sleep(100)
return "look at me"
if __name__ == '__main__':
a = Foo()
print "set your breakpoint here"
PyCharm will also call __getattribute__('__class__'). If you have a __getattribute__ that's misbehaving that could trip you up as well.
This may not be what's happening to you but perhaps worth considering.
As you are on Windows, for debugging such & most things I use the good old PythonWin IDE:
This IDE + Debugger runs in the same process as the debugged stuff!
This way, being in direct touch with real objects, like pdb in simple interactive shell, but having a usable GUI, is a big advantage most of the time. And this way there are no issues of transferring vast objects via repr/pickle or so between processes, no delays, no issues of timeouts etc.
If a step takes a long time, PythonWin will also simply wait and not respond before ... (unless one issues a break signal/KeyboardInterrupt via the PythonWin system tray icon).
And the interactive shell of PythonWin is also fully usable during the debugging - with namespace inside the current frame.
It's an old question but reply can be helpful.
Delete the .idea folder from the project root dir. It will clean up the Pycharm's database and the debugger will stop timing out. It works for me on Windows.

Finding a line of a C module called by a python script that segfaults

I have a caller.py which repeatedly calls routines from some_c_thing.so, which was created from some_c_thing.c. When I run it, it segfaults - is there a way for me to detect which line of c code is segfaulting?
This might work:
make sure the native library is compiled with debug symbols (-g switch for gcc).
Run python under gdb and let it crash:
gdb --args python caller.py
run # tell gdb to run the program
# script runs and crashes
bt # print backtrace, which should show the crashing line
If crash happens in the native library code, then this should reveal the line.
If native library code just corrupts something or violates some postconditions, and crash happens in Python interpreter's code, then this will not be helpful. In that case your options are code review, adding debug prints (first step would be to just log entry and exit of each C function to detect which is the last C function called before crash, then adding more fine-grained logging for variable values etc), and finally using debugger to see what happens by using the usual debugger techniques (breakpoints, stepping, watches...).
Take Python and the .so file(s) out of the equation. See what params are being passed, if any, and call the routines from a debugger capable of stepping through C code and binaries.
Here is a link to an article describing a simple C debugging process, in case you're not familiar with debugging C (command line interface). Here is another link on using NetBeans to debug C. Also using Eclipse...
This could help: gdb: break in shared library loaded by python (might also turn out to be a dupe)
segfault... Check if the number of variables or the types of variables you passed to that c function (in .so) are correct. If not aligned, usually it's a segfault.

How to use python's gevent.monkey.path_sys() with ipdb debugger?

I am using python's gevent library and do not want raw_input (or more specifically the event loop in cmd.Cmd) to block when awaiting user input. So as a result I use gevent.monkey.patch_sys() to ensure that my other greenlets may run when waiting for user input. Works great except that it seems to interact with readline.
For example, I no longer have history and auto-complete in ipython's ipdb debugger since the arrow keys no longer work. This can be seen with this simple snippet:
from gevent import monkey
monkey.patch_sys()
import ipdb; ipdb.set_trace()
# now hit arrow keys at the prompt
I get the following:
ipdb> ^[[A^[[A^[[A^[[A^[[A^[[A^[[A^[[A
*** SyntaxError: invalid syntax (<stdin>, line 1)
I have traced it to what I think could be an issue with python's readline as I know that ipython depends on it.
Also this appears to be a similar situation:
https://github.com/gevent/gevent/issues/6
but in my case I want to specifically use auto completion and history in the debugger.
I am running on OSX within iterm. Is this something specific to a console setting? Or is something in the patch fundamentally breaking readline?
Any ideas on how to resolve?

stopping execution of Python program on exception in ipython

I have a long running Python program that raises exception at some point. Is there some way to run this from ipython session and stop on the exception so I could examine the live data?
You may want ipython -i yourscript.py, which will execute your script in the interpreter environment. But this won't let you inspect the local environment where the exception happened, for example local variables within a function – you'll just be able to inspect globals. You probably want this instead:
In [1]: %run test.py
<exception occurs>
In [2]: %debug test.py
If you're not familiar with using PDB, check out some docs first.
Edit thanks to Thomas K
yes, depending on how you are setup. you can import your program and run it like any other module inside a try except block.
import yourprogram
try:
yourprogram.main_function(args)
except:
print "we blew up, investigate why"
If your program is not in a function you may need to put the try block around your import.
The problem with this approach is that the variables you are wanting to look at may be no longer in scope. I usually use print statements or log messages at various points to figure out what is not looking like I am expecting.

Categories