Refactor uses of `print()` to `tf.print()` that cause side-effect errors

[khatchad]

For functions that fail refactoring preconditions due to side-effects caused by calls to print() (IO operations), refactor each such call to tf.print(). Then, the function in question would not fail refactoring preconditions due to calls to print() (it may fail for other reasons).

Suggested by @ansariahmad.

Regression

The side-effect for print() is found here:

Hybridize-Functions-Refactoring/edu.cuny.hunter.hybridize.core/src/edu/cuny/hunter/hybridize/core/wala/ml/PythonModRefWithBuiltinFunctions.java

Lines 58 to 62 in 444af9e

	if (typeReference.equals(PRINT_FUNCTION_TYPE_REFERENCE)) {
	// found a call to the built-in print function, which has side effects.
	// add a pointer to a fake global variable representing a modification to the output stream.
	this.result.add(GLOBAL_OUTPUT_STREAM_POINTER_KEY);
	}

However, I don't think this is where any changes would occur. Instead, we'd probably examine all of the found side-effects to see if they are due to print(). If they are, then we could try to proceed with the refactoring of print() to tf.print(). If there are other problems with the function, it probably doesn't make sense to do this transformation since the function fails the preconditions anyway.

Note that there may be differences in behavior between Python's print() and TensorFlow's tf.print(). So, we would probably want a developer option here to use tf.print() in place of print().

Note also that this should only apply to eager to graph execution transformations. For graph execution to eager, we still want to consider print() as side-effecting because replacing it with tf.print() would most likely violate semantics preservation.

[ansariahmad]

Hello Professor,

Would you please assign this issue to me? I will submit a report about it. I will check differences between print() and tf.print, when tf.print fails and what data types it does not support.

[khatchad]

Thanks @ansariahmad. If you're submitting a fix, please know that the build is currently broken right now. I need to upgrade some versions to get it working. Thanks.