gh-104240: return code unit metadata from codegen

[iritkatriel]

With this we can do:

from _testinternalcapi import compiler_codegen, optimize_cfg, assemble_code_object
import ast
import opcode
import textwrap
import types

a = ast.parse("(x+y)/2", mode='eval')
filename = "myfile.py"
eval_mode = 1  # 0=exec, 1=eval, 2=single

insts, metadata = compiler_codegen(a, filename, 0, eval_mode)

consts = [v[1] for v in sorted([(v, k) for k, v in metadata['consts'].items()])]
insts = optimize_cfg(insts, consts)

from test.test_compiler_assemble import IsolatedAssembleTests
IsolatedAssembleTests().complete_metadata(metadata)

co = assemble_code_object(filename, insts, metadata)

for x,y in [(3, 4), (-100, 200), (10, 18)]:
    f = types.FunctionType(co, {'x': x, 'y': y})
    print(f'avg of {x} and {y} is {f()}')

• Issue: combine compilation steps in a pipeline #104240

[bedevere-bot]

🤖 New build scheduled with the buildbot fleet by @iritkatriel for commit e3c83dd 🤖

If you want to schedule another build, you need to add the 🔨 test-with-refleak-buildbots label again.

[carljm]

Fix the MSVC warning, but otherwise LGTM!

[carljm]

Looks like MSVC wants an explicit cast here.

[carljm]

Oh, I had the files tab open for a while and didn't notice the buildbots coming back unhappy. Guess that should be fixed, too -- if it's actually related to this PR? Seems surprising to me for this PR to cause those tests to fail.

[iritkatriel]

Oh, I had the files tab open for a while and didn't notice the buildbots coming back unhappy. Guess that should be fixed, too -- if it's actually related to this PR? Seems surprising to me for this PR to cause those tests to fail.

These buildbots seem broken for a while actually.

[AlexWaygood]

Oh, I had the files tab open for a while and didn't notice the buildbots coming back unhappy. Guess that should be fixed, too -- if it's actually related to this PR? Seems surprising to me for this PR to cause those tests to fail.

These buildbots seem broken for a while actually.

Looks like I'm the guilty culprit. I can repro the refleaks locally with 1b19bd1 but not with the previous commit.

[AlexWaygood]

Oh, I had the files tab open for a while and didn't notice the buildbots coming back unhappy. Guess that should be fixed, too -- if it's actually related to this PR? Seems surprising to me for this PR to cause those tests to fail.

These buildbots seem broken for a while actually.

Looks like I'm the guilty culprit. I can repro the refleaks locally with 1b19bd1 but not with the previous commit.

Should be fixed by 9196da4.