Imagine you have a test which works in one environment, and fails in a different environment.
If it is reproducible, then you can add this snippet, and compare the files created in /tmp/out*log.
You can modify skip_filename() and truncate_filename() to you needs.
This is copy+paste code, use it only temporarily.
The code uses sys.settrace() from the standard library
and PySnooper which you can install via pip install pysnooper.
You can copy+paste this snippet into the code which you want to trace.
In my case I wanted to see how my code behaves different under a differen Python version.
It will create a file /tmp/out-... with every line which the interpreter executed.
You can compare two created files with vimdiff on the shell, with the GUI meld or any
other diff-tool.
from pysnooper.tracer import get_path_and_source_from_frame
stream = io.open('/tmp/out-%s.log' % sys.version.split()[0], 'wt', encoding='utf8')
def tracer(frame, event, arg):
filename, source = get_path_and_source_from_frame(frame)
if skip_filename(filename):
return
filename = truncate_filename(filename)
source_line = source[frame.f_lineno - 1]
stream.write('{event}, {filename}:{line_number} {source_line}\n'.format(
event=event, filename=filename, source_line=source_line, line_number=frame.f_lineno))
return tracer
def truncate_filename(filename):
for remove in ['/home/me/foo/projects/']: # modify this to fit your liking
filename = filename.replace(remove, '')
return filename
def skip_filename(filename):
if filename.startswith('<'):
return True
if 'lib/python' in filename:
# Skip /usr/lib/ and lib/python/site-packages
# I only want to trace my code.
return True
sys.settrace(tracer)
I don't think it makes sense that I create a new tracing library for Python. There are already several.
See my Python Tips "Tracing"