Cosmetic/pretty printing of compiler output

[dowusu-antwi]

Fixes #144. (edit: Now with circuit drawings.) E.g.,

import qiskit_superstaq as qss
import qiskit
qc = qiskit.QuantumCircuit(1)
qc.h(0)
qc.draw()

provider = qss.SuperstaqProvider()
compile_result = provider.ibmq_compile(qc, target="ibmq_lagos_qpu")
print(compile_result.__repr_pretty__())
# CompilerOutput(    
#          ┌─────────┐┌─────────┐┌─────────┐
#     q_0: ┤ Rz(π/2) ├┤ Rx(π/2) ├┤ Rz(π/2) ├
#          └─────────┘└─────────┘└─────────┘,
#     {0: 0},
#     Schedule(
#         (0, ShiftPhase(-1.5707963268, DriveChannel(0))),
#         (0, ShiftPhase(-1.5707963268, ControlChannel(1))),
#         (
#             0,
#             Play(
#                 Drag(
#                     duration=160,
#                     sigma=40,
#                     beta=-1.053810766518306,
#                     amp=0.1987049754148708,
#                     angle=0.022620855474399307,
#                     name="X90p_d0",
#                 ),
#                 DriveChannel(0),
#                 name="X90p_d0",
#             ),
#         ),
#         (160, ShiftPhase(-1.5707963268, DriveChannel(0))),
#         (160, ShiftPhase(-1.5707963268, ControlChannel(1))),
#         name="sched25",
#     ),
#     None,
#     None,
#     None,
# )

[bharat-thotakura]

LGTM

[richrines1]

would it be possible/make sense to use the circuit diagrams instead of their reprs, assuming they're small enough?

[dowusu-antwi]

would it be possible/make sense to use the circuit diagrams instead of their reprs, assuming they're small enough?

Yes, by the way.

Working on handling an edge case: empty circuits give empty pulse schedules which black complains about.

[dowusu-antwi]

Done. (Also moved pretty printing to separate files per chat with Rich.)

[dowusu-antwi]

I bet preprocess_compiler_output and postprocess_compiler_output can be replaced by clever regex -- I would be glad to nuke as much of it as possible, but wanted to get something in since I've been dragging my feet.

edit: pretty_printing.py does feel like overkill -- I really would be glad to nuke it for a cleaner regex implementation, so I added profuse comments to make what I'm doing explicitly clear (to facilitate translation into regex, now or in the future)

[epelaaez]

lgtm (with micro nit)

[richrines1]

black needs to be moved from dev-requirements to requirements

[richrines1]

it would be nice to abbreviate things that are this long - i think one of the goals of adding a pretty repr is to make it so e.g. typing compiler_output in a console/notebook just prints something manageable and (somewhat) succinct

(i think textwrap.shorten does something like this?)

[richrines1]

to be picked up by ipython/jupyter/etc i think this needs to be

Suggested change

	def __repr_pretty__(self) -> str:
	def _repr_pretty_(self, p, cycle=False) -> None:
	...
	p.text(<string>)

(see https://ipython.readthedocs.io/en/stable/config/integrating.html#pretty-printing)

[richrines1]

is it possible/easy to avoid these spaces on black *blank* lines?

[dowusu-antwi]

Do you mean replace them with \t tab characters?

[richrines1]

oops sorry that was supposed to say "blank lines" - my thinking was just that there's no reason to include spaces in otherwise empty lines, e.g. this line could just be

Suggested change

	" \n"
	"\n"

[dowusu-antwi]

black needs to be moved from dev-requirements to requirements

Done.