starknet_os: aliases test

[nimrod-starkware]

No description provided.

[reviewable-StarkWare]

This change is 

[nimrod-starkware]

• starknet_os: add aliases test from non empty storage #7714
• starknet_os: aliases test #7680 👈 (View in Graphite)
• starknet_os: run tests directly #7675
• starknet_os: remove program input for python tests #7672
• main-v0.14.0

This stack of pull requests is managed by Graphite. Learn more about stacking.

[dorimedini-starkware]

Reviewed 5 of 5 files at r1, all commit messages.
Reviewable status: all files reviewed, 9 unresolved discussions (waiting on @amosStarkware)

---

crates/starknet_os/src/tests/aliases.rs line 56 at r1 (raw file):

#[case(Vec::from_iter((0..128).map(Felt::from)), (0..128).collect::<Vec<_>>(), HashMap::from_iter([(0.into(), 128.into())]))]
#[case(Vec::from_iter((0..129).map(Felt::from)), (0..129).collect::<Vec<_>>(), HashMap::from_iter([(0.into(), 129.into()), (128.into(), 128.into())]))]
#[case(vec![13.into(), 500.into(), 11.into(), 2000.into(), 2001.into(), 13.into(), 501.into(), 98.into(), 222.into(), 2000.into(), 127.into(), 128.into()], vec![13, 128, 11, 129, 130, 13, 131, 98, 132, 129, 127, 133], HashMap::from([(0.into(), 134.into()), (128.into(), 133.into()), (222.into(), 132.into()), (500.into(), 128.into()), (501.into(), 131.into()), (2000.into(), 129.into()), (2001.into(), 130.into())]))]

can you break long lines here? (no autoformatter for attributes..)

Code quote:

#[rstest]
#[case(Vec::new(), Vec::new(), HashMap::from([(0.into(), 128.into())]))]
#[case(vec![Felt::from(&*L2_ADDRESS_UPPER_BOUND)], vec![128], HashMap::from([(0.into(), 129.into()), (Felt::from(&*L2_ADDRESS_UPPER_BOUND), 128.into())]))]
#[case(vec![2000.into(), 1999999999.into(), 3000.into(), 2000.into()], vec![128, 129, 130, 128], HashMap::from([(0.into(), 131.into()), (2000.into(), 128.into()), (3000.into(), 130.into()), (1999999999.into(), 129.into())]))]
#[case(Vec::from_iter((0..128).map(Felt::from)), (0..128).collect::<Vec<_>>(), HashMap::from_iter([(0.into(), 128.into())]))]
#[case(Vec::from_iter((0..129).map(Felt::from)), (0..129).collect::<Vec<_>>(), HashMap::from_iter([(0.into(), 129.into()), (128.into(), 128.into())]))]
#[case(vec![13.into(), 500.into(), 11.into(), 2000.into(), 2001.into(), 13.into(), 501.into(), 98.into(), 222.into(), 2000.into(), 127.into(), 128.into()], vec![13, 128, 11, 129, 130, 13, 131, 98, 132, 129, 127, 133], HashMap::from([(0.into(), 134.into()), (128.into(), 133.into()), (222.into(), 132.into()), (500.into(), 128.into()), (501.into(), 131.into()), (2000.into(), 129.into()), (2001.into(), 130.into())]))]

---

crates/starknet_os/src/tests/aliases.rs line 80 at r1 (raw file):

    let unique_keys: HashSet<Felt> = HashSet::from_iter(
        keys.iter()
            .filter(|key| key >= &&INITIAL_AVAILABLE_ALIAS)

this is not in the original test, why is it needed here?

Code quote:

.filter(|key| key >= &&INITIAL_AVAILABLE_ALIAS)

---

crates/starknet_os/src/tests/aliases.rs line 82 at r1 (raw file):

            .filter(|key| key >= &&INITIAL_AVAILABLE_ALIAS)
            .copied()
            .chain([*ALIAS_COUNTER_STORAGE_KEY.key()]),

this is not in the original test, why is it needed here?

Code quote:

.chain([*ALIAS_COUNTER_STORAGE_KEY.key()]),

---

crates/starknet_os/src/tests/aliases.rs line 90 at r1 (raw file):

            (unique_keys.len()) * DICT_ACCESS_SIZE
            ])),
        EndpointArg::Pointer(PointerArg::Array(vec![Felt::ZERO; keys.len()])),

Suggestion:

// Aliases per-key ptr.
EndpointArg::Pointer(PointerArg::Array(vec![Felt::ZERO; keys.len()])),

---

crates/starknet_os/src/tests/aliases.rs line 95 at r1 (raw file):

    let keys_arg = EndpointArg::Pointer(PointerArg::Array(keys));
    let explicit_args = vec![n_keys_arg, keys_arg];
    let alias_contract_address: ContractAddress = Felt::TWO.try_into().unwrap();

don't we have a named constant for this?
maybe about time we have one... but maybe already defined in the blockifier?

Code quote:

Felt::TWO

---

crates/starknet_os/src/tests/aliases.rs line 129 at r1 (raw file):

            let new_value = aliases_storage_updates[i * DICT_ACCESS_SIZE + new_value_offset];
            actual_alias_storage.insert(key, new_value);
        }

maybe this should be a util - dictaccess vector to hashmap. non-blocking

Code quote:

        let n_aliases = felt_to_usize(n_aliases).unwrap();
        assert!(aliases_storage_updates.len() % DICT_ACCESS_SIZE == 0);
        assert!(aliases_storage_updates.len() / DICT_ACCESS_SIZE == n_aliases);
        let key_offset = 0;
        let new_value_offset = 2;
        for i in 0..n_aliases {
            let key = aliases_storage_updates[i * DICT_ACCESS_SIZE + key_offset];
            let new_value = aliases_storage_updates[i * DICT_ACCESS_SIZE + new_value_offset];
            actual_alias_storage.insert(key, new_value);
        }

---

crates/starknet_os/src/test_utils/utils.rs line 43 at r1 (raw file):

        hint_locals,
        None,
    )?;

for clarity of reading

Suggestion:

    let state_reader = None;
    let (actual_implicit_retdata, actual_explicit_retdata, _) = run_cairo_0_entry_point(
        runner_config,
        program_bytes,
        function_name,
        explicit_args,
        implicit_args,
        expected_explicit_retdata,
        hint_locals,
        state_reader,
    )?;

---

crates/starknet_os/src/tests/bls_field.rs line 154 at r1 (raw file):

            HashMap::new(),
            None,
        )

for clarity

Suggestion:

        let state_reader = None;
        let (_, explicit_retdata, _) = run_cairo_0_entry_point(
            &entrypoint_runner_config,
            OS_PROGRAM_BYTES,
            "starkware.starknet.core.os.data_availability.bls_field.horner_eval",
            &explicit_args,
            &implicit_args,
            &[EndpointArg::Value(ValueArg::Array(vec![Felt::ZERO, Felt::ZERO, Felt::ZERO]))],
            HashMap::new(),
            state_reader,
        )

---

crates/starknet_os/src/hint_processor/snos_hint_processor.rs line 353 at r1 (raw file):

            vec![state_reader],
        )?;
        hint_processor.execution_helpers_manager.increment_current_helper_index();

why is this needed here and not in run_os?

Code quote:

hint_processor.execution_helpers_manager.increment_current_helper_index();

---

crates/starknet_os/src/test_utils/cairo_runner.rs line 566 at r1 (raw file):

    expected_explicit_return_values: &[EndpointArg],
    hint_locals: HashMap<String, Box<dyn Any>>,
    state_reader: Option<DictStateReader>,

can this be generic?

Suggestion:

pub fn run_cairo_0_entry_point<S: StateReader>(
    runner_config: &EntryPointRunnerConfig,
    program_bytes: &[u8],
    entrypoint: &str,
    explicit_args: &[EndpointArg],
    implicit_args: &[ImplicitArg],
    expected_explicit_return_values: &[EndpointArg],
    hint_locals: HashMap<String, Box<dyn Any>>,
    state_reader: Option<S>,

[nimrod-starkware]

Reviewable status: 4 of 5 files reviewed, 9 unresolved discussions (waiting on @amosStarkware and @dorimedini-starkware)

---

crates/starknet_os/src/hint_processor/snos_hint_processor.rs line 353 at r1 (raw file):

Previously, dorimedini-starkware wrote…

why is this needed here and not in run_os?

Inrun_os it's called by a hint, and here not. A similar trick is done in python by injecting the hint scope variable aliases.

---

crates/starknet_os/src/test_utils/cairo_runner.rs line 566 at r1 (raw file):

Previously, dorimedini-starkware wrote…

can this be generic?

The function new_for_testing is not generic ATM, it's impossible to make it generic because of unwrap_or_default. we can add another constraint by S: StateReader + Default

---

crates/starknet_os/src/test_utils/utils.rs line 43 at r1 (raw file):

Previously, dorimedini-starkware wrote…

for clarity of reading

Done.

---

crates/starknet_os/src/tests/aliases.rs line 56 at r1 (raw file):

Previously, dorimedini-starkware wrote…

can you break long lines here? (no autoformatter for attributes..)

Done.

---

crates/starknet_os/src/tests/aliases.rs line 80 at r1 (raw file):

Previously, dorimedini-starkware wrote…

this is not in the original test, why is it needed here?

This calculation is for knowing the length of result.
It's needed here because the rust cairo runner should now in advance what is the expected length of the array returned.
In python we don't have to know in advance.

---

crates/starknet_os/src/tests/aliases.rs line 82 at r1 (raw file):

Previously, dorimedini-starkware wrote…

this is not in the original test, why is it needed here?

same as above. it's for calculating the length of the result.

---

crates/starknet_os/src/tests/aliases.rs line 95 at r1 (raw file):

Previously, dorimedini-starkware wrote…

don't we have a named constant for this?
maybe about time we have one... but maybe already defined in the blockifier?

We have it defined in the blockifier in a test file. It's impossible to import from there because it's gated under #[cfg(test)]

---

crates/starknet_os/src/tests/bls_field.rs line 154 at r1 (raw file):

Previously, dorimedini-starkware wrote…

for clarity

Done.

[github-actions]

Artifacts upload workflows:

• Started at 2025-07-01T13:59:07Z
• Started at 2025-07-01T14:00:41Z
• Started at 2025-07-02T05:05:55Z
• Started at 2025-07-02T06:54:07Z
• Started at 2025-07-02T07:01:38Z
• Started at 2025-07-02T07:09:23Z

[dorimedini-starkware]

Reviewed 1 of 1 files at r2, 4 of 4 files at r3, all commit messages.
Reviewable status: all files reviewed, 3 unresolved discussions (waiting on @amosStarkware and @nimrod-starkware)

---

crates/starknet_os/src/test_utils/cairo_runner.rs line 566 at r1 (raw file):

Previously, nimrod-starkware wrote…

The function new_for_testing is not generic ATM, it's impossible to make it generic because of unwrap_or_default. we can add another constraint by S: StateReader + Default

out of scope for now... I guess that in tests we'll always use the DictStateReader

---

crates/starknet_os/src/tests/aliases.rs line 95 at r1 (raw file):

Previously, nimrod-starkware wrote…

We have it defined in the blockifier in a test file. It's impossible to import from there because it's gated under #[cfg(test)]

hmmm... can you either

1. move it to a non-test, public place, or
2. add a TODO to do it?

---

crates/starknet_os/src/tests/aliases.rs line 87 at r3 (raw file):

    ])
)]
#[case(

what about this test case?

(
            [128 + 2**i for i in range(150)] + [128 + 2**i for i in range(150)],
            [128 + i for i in range(150)] + [128 + i for i in range(150)],
            [(0, 128 + 150)] + [(128 + 2**i, 128 + i) for i in range(150)],
        ),

Code quote:

)]
#[case(

[github-actions]

Benchmark movements: No major performance changes detected.

[dorimedini-starkware]

Reviewed 2 of 2 files at r4, all commit messages.
Reviewable status: all files reviewed, 3 unresolved discussions (waiting on @amosStarkware and @nimrod-starkware)

[nimrod-starkware]

Reviewable status: all files reviewed, 3 unresolved discussions (waiting on @amosStarkware and @dorimedini-starkware)

---

crates/starknet_os/src/tests/aliases.rs line 95 at r1 (raw file):

Previously, dorimedini-starkware wrote…

hmmm... can you either

1. move it to a non-test, public place, or
2. add a TODO to do it?

Done.

---

crates/starknet_os/src/tests/aliases.rs line 90 at r1 (raw file):

            (unique_keys.len()) * DICT_ACCESS_SIZE
            ])),
        EndpointArg::Pointer(PointerArg::Array(vec![Felt::ZERO; keys.len()])),

Done.

[dorimedini-starkware]

Reviewed 4 of 4 files at r5, all commit messages.
Reviewable status: all files reviewed, 1 unresolved discussion (waiting on @amosStarkware and @nimrod-starkware)

[nimrod-starkware]

Reviewable status: all files reviewed, 1 unresolved discussion (waiting on @amosStarkware and @dorimedini-starkware)

---

crates/starknet_os/src/tests/aliases.rs line 129 at r1 (raw file):

Previously, dorimedini-starkware wrote…

maybe this should be a util - dictaccess vector to hashmap. non-blocking

Done

---

crates/starknet_os/src/tests/aliases.rs line 87 at r3 (raw file):

Previously, dorimedini-starkware wrote…

what about this test case?

(
            [128 + 2**i for i in range(150)] + [128 + 2**i for i in range(150)],
            [128 + i for i in range(150)] + [128 + i for i in range(150)],
            [(0, 128 + 150)] + [(128 + 2**i, 128 + i) for i in range(150)],
        ),

Done.

[dorimedini-starkware]

Reviewed 2 of 2 files at r6, all commit messages.
Reviewable status: complete! all files reviewed, all discussions resolved (waiting on @amosStarkware)

---

crates/starknet_os/src/test_utils/utils.rs line 81 at r6 (raw file):

    }
    dict
}

does this work? I think using chunks will make this nicer

Suggestion:

    assert!(squashed_dict.len() % DICT_ACCESS_SIZE == 0, "Invalid squashed dict length");
    squashed_dict
        .iter()
        .chunks(DICT_ACCESS_SIZE)
        .map(|[key, _old_val, new_val]| (key, new_val))
        .collect()
}

[dorimedini-starkware]

Reviewed 1 of 1 files at r7, all commit messages.
Reviewable status: complete! all files reviewed, all discussions resolved (waiting on @amosStarkware)

[nimrod-starkware]

Reviewable status: complete! all files reviewed, all discussions resolved (waiting on @amosStarkware)

---

crates/starknet_os/src/test_utils/utils.rs line 81 at r6 (raw file):

Previously, dorimedini-starkware wrote…

does this work? I think using chunks will make this nicer

derefernence the chunk as [key, _old_val, new_val] doesn't work

[dorimedini-starkware]

Reviewed 1 of 1 files at r8, all commit messages.
Reviewable status: complete! all files reviewed, all discussions resolved (waiting on @amosStarkware)