Support further Python Interop for callables in Estimation and QIR Compilation #2091
Conversation
…mpilation This follows up on #2054 by adding the enhanced support for Q# callables exposed into Python and Python argument arrays in resource estimation and QIR compilation.
swernli
requested review from
billti,
ivanbasov,
idavis and
minestarks
as code owners
![github-advanced-security[bot]](https://avatars.githubusercontent.com/in/57789?s=60&v=4){kind=small}
| qsharp.eval("operation Program() : Result { use q = Qubit(); return MResetZ(q); }") | ||
| operation = qsharp.compile("Program()") | ||
| qir = str(operation) | ||
| assert "define void @ENTRYPOINT__main()" in qir | ||
| assert '"required_num_qubits"="1" "required_num_results"="1"' in qir | ||
|
|
||
|
|
||
| def test_compile_qir_str_from_python_callable() -> None: | ||
| qsharp.init(target_profile=qsharp.TargetProfile.Base) | ||
| qsharp.eval("operation Program() : Result { use q = Qubit(); return MResetZ(q); }") | ||
| operation = qsharp.compile(qsharp.code.Program) | ||
| qir = str(operation) | ||
| assert "define void @ENTRYPOINT__main()" in qir | ||
| assert '"required_num_qubits"="1" "required_num_results"="1"' in qir | ||
|
|
||
|
|
||
| def test_compile_qir_str_from_python_callable_with_single_arg() -> None: | ||
| qsharp.init(target_profile=qsharp.TargetProfile.Base) | ||
| qsharp.eval( | ||
| "operation Program(nQubits : Int) : Result[] { use qs = Qubit[nQubits]; MResetEachZ(qs) }" | ||
| ) |
Check notice
Code scanning / devskim
If untrusted data (data from HTTP requests, user submitted files, etc.) is included in an eval statement it can allow an attacker to inject their own code. Note test
| qsharp.eval( | ||
| "operation Program(nQubits : Int[]) : Result[] { use qs = Qubit[nQubits[1]]; MResetEachZ(qs) }" | ||
| ) |
Check notice
Code scanning / devskim
If untrusted data (data from HTTP requests, user submitted files, etc.) is included in an eval statement it can allow an attacker to inject their own code. Note test
| qsharp.eval( | ||
| """ | ||
| operation Test() : Unit { | ||
| use qs = Qubit[10]; | ||
| for q in qs { | ||
| T(q); | ||
| M(q); | ||
| } | ||
| } | ||
| """ | ||
| ) |
Check notice
Code scanning / devskim
If untrusted data (data from HTTP requests, user submitted files, etc.) is included in an eval statement it can allow an attacker to inject their own code. Note test
| qsharp.eval( | ||
| """ | ||
| operation Test(nQubits : Int) : Unit { | ||
| use qs = Qubit[nQubits]; | ||
| for q in qs { | ||
| T(q); | ||
| M(q); | ||
| } | ||
| } | ||
| """ | ||
| ) |
Check notice
Code scanning / devskim
If untrusted data (data from HTTP requests, user submitted files, etc.) is included in an eval statement it can allow an attacker to inject their own code. Note test
| pub fn fir_to_qir( | ||
| fir_store: &qsc_fir::fir::PackageStore, | ||
| capabilities: TargetCapabilityFlags, | ||
| compute_properties: Option<PackageStoreComputeProperties>, | ||
| entry: &ProgramEntry, | ||
| ) -> Result<String, qsc_partial_eval::Error> { | ||
| let mut program = get_rir_from_compilation(fir_store, compute_properties, entry, capabilities)?; | ||
| let compute_properties = compute_properties; |
There was a problem hiding this comment.
Whoops, this looks like it got left behind from an earlier experiment.
There was a problem hiding this comment.
What about the third sibling, .run() ? I think it would make sense to allow that to take a callable, as well.
And of course, there's .circuit(), which can already take a callable, but uses a different pattern... which is unfortunate, because this new pattern is better. I wouldn't be opposed to updating .circuit() to allow the new pattern as well, but I understand if you don't want to touch it in this PR.
| program, | ||
| entry, | ||
| entry: Some(entry), | ||
| } | ||
| } | ||
|
|
||
| fn new_from_package_id( | ||
| package_store: &'a PackageStore, | ||
| compute_properties: &'a PackageStoreComputeProperties, | ||
| package_id: PackageId, | ||
| capabilities: TargetCapabilityFlags, | ||
| ) -> Self { | ||
| // Create the entry-point callable. | ||
| let mut resource_manager = ResourceManager::default(); |
There was a problem hiding this comment.
Note: refactor new_* methods to use shared private fn.
| @@ -639,6 +639,43 @@ impl Interpreter { | |||
| }) | |||
| } | |||
|
|
|||
| pub fn qirgen_from_callable( | |||
This follows up on #2054 by adding the enhanced support for Q# callables exposed into Python and Python argument arrays in resource estimation and QIR compilation.