cobra-array 0.2.2

About

cobra-array is a backend-agnostic array utility library that unifies array conversion, context control, and cross-library operations across NumPy/PyTorch-style ecosystems.

• Python: 3.9+
• Runtime deps: array-api-compat (>= 1.11.2)

Features

• 🚀 Ackend-agnostic API: Work with different array backends through one consistent interface.
• 🚀 Context-driven unification: Automatically align namespace, dtype, and device via context managers and decorators.
• 🚀 Compatibility wrappers: CompatArray and CompatNamespace provide a clean, consistent layer over native backend behavior.

Installation

Install from PyPI

pip install cobra-array

Quick Start

• Basic conversions:

  import numpy as np
  from cobra_array.convert import to_numpy, to_tensor, to_list
  
  data = [[1, 2], [3, 4]]
  
  arr_np = to_numpy(data, dtype=np.float32)  # numpy.ndarray float32
  
  arr_torch = to_tensor(data, device="cpu")
  
  back_to_list = to_list(arr_np)  # [[1.0, 2.0], [3.0, 4.0]]
  

• Context-based conversion:

  import numpy as np
  from cobra_array import array_context, as_context, context_spec
  
  with array_context(xp="numpy", dtype=np.float32, device="cpu"):
      x = as_context([1, 2, 3])
      y = as_context(np.array([4, 5]))
      spec = context_spec()
  

• Auto-unify function arguments:

  import numpy as np
  from cobra_array import unify_args
  
  @unify_args(ref=0, unify_dtype=True, unify_device=True, arraylike_only=True)
  def add_and_mean(a, b):
      c = a + b
      return c.mean()
  
  out = add_and_mean(np.array([1, 2, 3]), [4, 5, 6])
  

• Default backend strategy:

  from cobra_array.default import as_default, default_spec
  
  spec = default_spec()
  
  x = as_default([1, 2, 3], unify_dtype=True, unify_device=True)
  

Requirements

• Python >= 3.9
• array-api-compat >= 1.11.2

License

See LICENSE in the repository.

Links

• Homepage/Repo
• Issues