I have a code similar to this:
try:
if x:
statement1
statement2
statement3
elif y:
statement4
statement5
statement6
I've done the following before:
try:
doing = "statement1"
statement1
doing = "statement2"
statement2
doing = "statement3"
statement3
doing = "statement4"
statement4
except:
print "exception occurred doing ", doing
The advantage over printing checkpoints is there's no log output unless there actually is an exception.
You should run your program in a debugger, such as pdb. This will allow you to run your code normally, and then examine the environment when something unexpected like this occurs.
Given a script named 'main.py', run it like this:
python -m pdb main.py
Then, when your program starts, it will start in the debugger. Type c
to continue until the next breakpoint (or crash). Then, you can examine the environment by doing things like print spam.eggs
. You can also set breakpoints by doing pdb.set_trace()
(I commonly do import pdb; pdb.set_trace()
).
Additionally, what do you mean that it is "okay" for 'statement 3' to raise the exception? Are you expecting the exception? If so, it might be better to write a try/except block around this statement, so that the program can continue.
Building on JJ above..
The advantage of using system errors over statements is they record more specific information which will aid debugging later (believe me I get a lot)
eg. I record them to a text file, so after my programs have automatically run overnight on the server, I can retrieve any issues, and have enough information to quicken the repair!
More Info... Traceback & Sys
import traceback
import sys
try:
print 1/0
except Exception as e:
print '1', e.__doc__
print '2', sys.exc_info()
print '3', sys.exc_info()[0]
print '4', sys.exc_info()[1]
print '5', sys.exc_info()[2], 'Sorry I mean line...',traceback.tb_lineno(sys.exc_info()[2])
ex_type, ex, tb = sys.exc_info()
print '6', traceback.print_tb(tb)
Yields
> 1 Second argument to a division or modulo operation was zero.
> 2 (<type 'exceptions.ZeroDivisionError'>, ZeroDivisionError('integer division
> or modulo by zero',), <traceback object at 0x022DCF30>)
> 3 <type 'exceptions.ZeroDivisionError'>
> 4 integer division or modulo by zero
> 5 <traceback object at 0x022DCF30> Sorry I mean line... 5
> 6 File "Z:\Programming\Python 2.7\Error.py", line 5, in <module>
> print 1/0
None
>>>
You should wrap the statements you care about more tightly. Extracting the line number from the traceback is going to be involved and fragile.
Edit your source code, so that you remove one line at a time, until the error disappears, and that should point you closer to the problem.
what about this:
try:
if x:
print 'before statement 1'
statement1
print 'before statement 2' #ecc. ecc.
statement2
statement3
elif y:
statement4
statement5
statement6
else:
raise
except:
statement7
this is the straightforward workaround but I suggest to use a debugger
or even better, use the sys module :D
try:
if x:
print 'before statement 1'
statement1
print 'before statement 2' #ecc. ecc.
statement2
statement3
elif y:
statement4
statement5
statement6
else:
raise
except:
print sys.exc_traceback.tb_lineno
#this is the line number, but there are also other infos