Contributing partial work towards pytest branch coverage

[jaltmayerpizzorno]

Hi Ned, I started working on this test to ensure that Coverage.py would provide coverage within pytest modules, but ran out of time before my trip tomorrow. I am also including a wrap_pytest function that I think should be sufficient to restore AST-based instrumentation on pytest modules despite assertion rewriting... but I couldn't get the tests to work, so never integrated it.

[ferdnyc]

I'm trying to understand the purpose of this code.

I get why contextlib.redirect_stdout is useful for capturing the output of other code that directly print()s data we'd prefer to send somewhere else. Its docs give the example of capturing a help() call in the REPL. Smart use of redirection, very clever.

But what's the point of using it just to surround an explicit print()? print already has support for redirecting its output somewhere other than sys.stdout.

What's the advantage of doing this, over...

with open("/tmp/foo.out", "a") as f:
    print(f"sysmon adding {arc=} (return)", file=f)

or even...

with open("/tmp/foo.out", "a") as f:
    f.write(f"sysmon adding {arc=} (return)\n")

?

[nedbat]

This code is temporary debugging code. It's written to be quick to insert, and then will be removed later. I have a snippet in my editor to add this:

import contextlib
with open("/tmp/foo.out", "a") as f:
    with contextlib.redirect_stdout(f):
        print(f"") 

This makes it easy for me to add print statements without having to remember file=f, or that f.write only accepts strings.