trace execution path of python interpreter

Question

The following works nice to see what the interpreter does:

python -m trace --ignore-dir=$HOME/lib64:$HOME/lib:/usr -t path-to/script.py

But there are too many lines. I would like to hide the lines which happen during importing a file.

Example: I am not interested in lines like this:

saved_filter.py(9): class SavedFilter(models.Model):
saved_filter.py(10):     name = models.TextField(max_length=256)
saved_filter.py(11):     user = models.ForeignKey('auth.User', null=True, blank=True)

I could not find a solution in the docs: https://docs.python.org/2/library/trace.html

Update If there is a different way to get the result, for example a different python package (not trace), this would be a good solution, too.

Update 2 The tracing should be non-interactive.

Update 3

I tried the solution provided by Martin v. Löwis. It works in some cases, but not all.

file foo.py

import bar

def main():
    f=bar.Foo()
    f.my_func()

if __name__=='__main__':
    main()

file bar.py

class Foo(object):
    def my_func(self):
        def inner():
            print('#in inner')
            return 'inner'
        print('#in my_func()')
        inner()
        return 1

If I call foo.py, the wanted result looks similar to this:

foo.py: f=bar.Foo() foo.py: f.my_func() bar.py: print('#in my_func()') bar.py: inner() bar.py: print('#in inner') bar.py: return 'inner' bar.py: return 1

Result of trace2.py

> python tmp/trace2-orig.py --trace tmp/foo.py 
 --- modulename: foo, funcname: <module>
 --- modulename: bar, funcname: <module>
 --- modulename: bar, funcname: Foo
bar.py(1): class Foo(object):               <======= Import lines
bar.py(2):     def my_func(self):
 --- modulename: foo, funcname: main
foo.py(4):     f=bar.Foo()
foo.py(5):     f.my_func()
 --- modulename: bar, funcname: my_func
bar.py(3):         def inner():
bar.py(6):         print('#in my_func()')
#in my_func()
bar.py(7):         inner()
 --- modulename: bar, funcname: inner
bar.py(4):             print('#in inner')
#in inner
bar.py(5):             return 'inner'
bar.py(8):         return 1
 --- modulename: trace, funcname: _unsettrace
trace.py(80):         sys.settrace(None)

Unfortunately there is still class Foo(object) which is something executed during import.

I guess the detection of code loading and executing does not cover all cases.

Try to collect info by Trace object method call inside your script to collect info about your target function and print given CoverageResults aftert hat. — Andrew Svetlov
– Andrew Svetlov, Commented Dec 9, 2014 at 20:43

Martin v. Löwis · Accepted Answer · 2014-12-13 10:44:05Z

1

If you create a script trace2.py as

import trace

OrigTrace = trace.Trace
class Trace2(trace.Trace):
    def localtrace_trace(self, frame, why, arg):
        if why == "line" and frame.f_code.co_name == '<module>':
            return self.localtrace
        return OrigTrace.localtrace_trace(self, frame, why, arg)

trace.Trace=Trace2
trace.main()

and run python -m trace2 -t script.py you will not see trace output from lines that are on the module level.

answered Dec 13, 2014 at 10:44

Martin v. Löwis

128k20 gold badges205 silver badges238 bronze badges

Sign up to request clarification or add additional context in comments.

1 Comment

guettli Over a year ago

Thank you very much. I tried your solution. It does not catch all cases of code loading yet. See updated question.

M.javid · Accepted Answer · 2014-12-08 21:29:53Z

0

you can use pdb library, by this library you can make breakpoint and/or inject python statements to the break point line by interactive console UI and doing more other actions. ;)

answered Dec 8, 2014 at 21:29

M.javid

6,6344 gold badges44 silver badges58 bronze badges

1 Comment

guettli Over a year ago

I updated the question: The tracing should be non-interactive. AFAIK pdb is made for interactive debugging. Please tell me, if I am missing something.

lunixbochs · Accepted Answer · 2014-12-09 08:59:16Z

0

If you just want to hide code paths during import, just pipe it through a filter like this one:

import re
import sys

cur = None
skip_re = re.compile(r'^(?P<filename>.+)?\((\d*)\):\s*import')
for line in sys.stdin:
    if cur and not line.startswith(cur):
        continue
    cur = None
    match = skip_re.match(line)
    if match:
        cur = match.group('filename')
    sys.stdout.write(line)

answered Dec 9, 2014 at 8:59

lunixbochs

22.5k2 gold badges41 silver badges48 bronze badges

Collectives™ on Stack Overflow

trace execution path of python interpreter

3 Answers 3

1 Comment

1 Comment

Comments

Your Answer

Linked

Hot Network Questions

Collectives™ on Stack Overflow

3 Answers 3

1 Comment

1 Comment

Comments

Your Answer

Sign up or log in

Post as a guest

Linked

Related