How to redirect tensorflow import errors from `stderr` to `stdout`

Question

Question

When you import tensorflow, it prints to stderr - bug report.

Currently this is confusing my monitoring system and logging these informative messages as errors.

I would like to redirect these messages from stderr to stdout.

In theory, this should work:

def redirect_tensorflow_logs():
    print("Before - Outside redirect block", file=sys.stderr)
    with redirect_stderr(sys.stdout):
        print("Before -  Inside redirect block", file=sys.stderr)
        import tensorflow as tf
        print("After - Inside redirect block", file=sys.stderr)
    print("After - Outside redirect block", file=sys.stderr)

Unfortunately, it is not. This is the output I am getting:

Output - stderr:

Before - Outside redirect block
2024-01-10 14:34:44.164579: I tensorflow/core/platform/cpu_feature_guard.cc:182] This TensorFlow binary is optimized to use available CPU instructions in performance-critical operations.
To enable the following instructions: AVX2 FMA, in other operations, rebuild TensorFlow with the appropriate compiler flags.
After - Outside redirect block

Output - stdout:

Before -  Inside redirect block
After - Inside redirect block

Is there a way to redirect these messages from stderr to stdout?

Alternate solutions

I know that set os.environ['TF_CPP_MIN_LOG_LEVEL'] = '3' (https://stackoverflow.com/a/42121886/251589) to remove these messages entirely. I would prefer to not do this because at some point, I will want to address these and hiding them will make this harder for future devs to understand what is going on.

I could also redirect ALL output from my program from stderr to stdout via something like this: python -m myprogram 2>&1. This would stop my monitoring system from raising errors but may cause me to miss some items in the future that are real errors.