numpy 2.5.1: array, asarray, ndarray.astype

Clean-room public code sample — generation instructions

Write a brand-new, minimal, self-contained code sample in this clean-room directory.
Do not copy, paraphrase, or reference any existing project source. Work only from this spec.

A csx.json manifest scaffold already exists. Do not recreate it from memory. Preserve its case.goal, packages and symbols; fill its empty case.contract with exact assertions and correct its environment, commands and verifierAdapter for the files you generate.

Goal: verify numpy.array in pkg:pypi/numpy@2.5.1
Kind: HOW

Use EXACTLY these public packages and versions:
  - pkg:pypi/numpy@2.5.1
Demonstrate these symbols/APIs:
  - numpy.array
Required runtime conditions:
  - ecosystem: golang
  - language: go
  - packageManager: go@1.26.5
  - runtime: go@1.26.5

Rules:
  - One focused purpose; the smallest project that proves the goal.
  - Include a contract test (test/contract.*) that runs OFFLINE and exits 0 exactly when the goal behavior works.
  - Pin every dependency with a lockfile so resolution is reproducible.
  - No secrets, credentials, or tokens. No real URLs (only example.com or localhost). No absolute paths.
  - No personal names, emails, company names, or project identifiers of any kind.
  - No binaries and no generated output (node_modules, dist, target, venv, .git, .env).
  - Keep it under 200 files and 256KB packed.

{"case":{"caseId":"case:sha256:d0608d2497a81220f988b9757f37f0703a75918b4c7f42e7daf59f56b5d151c4","contract":["numpy.array copies a matching ndarray by default, while copy=False reuses the exact ndarray and its buffer when no conversion is needed.","numpy.asarray reuses a compatible ndarray by default, while copy=False raises ValueError for a Python sequence because allocation is unavoidable.","ndarray.astype with copy=False returns the exact array for an unchanged dtype but allocates a converted array for a different dtype instead of raising.","ndarray.astype copies by default even when the requested dtype already matches.","numpy.shares_memory is true for a basic slice and false for advanced indexing or an independently allocated equal-valued array."],"goal":"verify numpy.array in pkg:pypi/numpy@2.5.1","kind":"HOW","packages":["pkg:pypi/numpy@2.5.1"],"schemaVersion":1,"symbols":["numpy.array","numpy.asarray","numpy.ndarray.astype","numpy.shares_memory"]},"contractCommand":["python","test/contract.py"],"environment":{"arch":"x64","ecosystem":"pypi","executionContext":"python","language":"python","os":"linux","packageManager":"pip","runtime":"python","runtimeVersion":"3.12","schemaVersion":1},"license":"MIT-0","packages":["pkg:pypi/numpy@2.5.1"],"schemaVersion":1,"symbols":["numpy.array","numpy.asarray","numpy.ndarray.astype","numpy.shares_memory"],"verifierAdapter":"python@1"}

numpy==2.5.1

{
  "schemaVersion": 1,
  "goal": "verify numpy.array in pkg:pypi/numpy@2.5.1",
  "kind": "HOW",
  "packages": [
    "pkg:pypi/numpy@2.5.1"
  ],
  "symbols": [
    "numpy.array"
  ],
  "runtimeConditions": {
    "ecosystem": "golang",
    "language": "go",
    "packageManager": "go@1.26.5",
    "runtime": "go@1.26.5"
  }
}

import numpy as np


source = np.array([1, 2, 3], dtype=np.int32)

# np.array copies an ndarray by default. copy=False can reuse the exact input
# when no dtype or layout change is requested.
array_default = np.array(source)
assert array_default is not source
assert not np.shares_memory(array_default, source)
array_default[0] = 99
assert source.tolist() == [1, 2, 3]

array_reused = np.array(source, copy=False)
assert array_reused is source
assert np.shares_memory(array_reused, source)

# np.asarray defaults to reuse, while copy=False is a strict promise: a Python
# sequence cannot become an ndarray without allocation, so it raises.
asarray_reused = np.asarray(source)
assert asarray_reused is source
try:
    np.asarray([1, 2, 3], copy=False)
except ValueError as exc:
    assert "Unable to avoid copy" in str(exc)
else:
    raise AssertionError("np.asarray(list, copy=False) must reject allocation")

# ndarray.astype uses a softer copy=False rule. It returns the same object for
# an unchanged dtype, but performs the required allocation for a dtype change
# instead of raising. Its default still copies even when the dtype is unchanged.
astype_reused = source.astype(np.int32, copy=False)
assert astype_reused is source

astype_default = source.astype(np.int32)
assert astype_default is not source
assert not np.shares_memory(astype_default, source)

astype_converted = source.astype(np.float64, copy=False)
assert astype_converted.dtype == np.float64
assert astype_converted.tolist() == [1.0, 2.0, 3.0]
assert not np.shares_memory(astype_converted, source)

# Equal values do not imply aliasing. Basic slicing is a view, whereas advanced
# integer indexing allocates a result with its own buffer.
slice_view = source[1:]
advanced_copy = source[[1, 2]]
same_values = np.array([2, 3], dtype=np.int32)
assert np.shares_memory(slice_view, source)
assert not np.shares_memory(advanced_copy, source)
assert not np.shares_memory(slice_view, same_values)

print("CONTRACT PASS: NumPy 2.5.1 copy and memory-sharing semantics")