`np.asarray(buf)` returns incorrect number of samples for multichannel buffers

[ashNotKetchup]

Issue: np.asarray(buf) returns incorrect number of samples for multichannel buffers

Summary

When using np.asarray(buf) on a multichannel buffer~, the resulting NumPy array does not reflect the actual number of samples expected from interleaved audio.

Reproduction

def get_samples_multichannel(buffer_name: str):
    channel_count = 2
    length = 100

    buf = api.Buffer(buffer_name, "", channels=channel_count)
    buf.framecount = length

    t = np.linspace(0, 1, length, endpoint=False, dtype=np.float32)
    xs = np.array(
        [np.sin(t * 2 * math.pi * 5) * (-1 if ch % 2 else 1)
         for ch in range(channel_count)]).T
    flattened_data = xs.reshape(-1)
    buf.set_samples(flattened_data)

    buf = api.get_buffer(buffer_name)
    api.post(f"Accessed buffer '{buffer_name}' with {buf.n_samples} samples and {buf.channelcount} dimensions")

    xs = np.asarray(buf)
    xs = np.array(xs, dtype=np.float32)
    api.post(f"Converted buffer samples to numpy array with shape: {xs.shape} and dtype: {xs.dtype}")

[''get_samples_multichannel('myBufferName')"] -> [s to_py]

Expected Output

A 2-channel, 100-frame buffer should return 200 samples when accessed as a NumPy array — either in interleaved or (frames, channels) format.

Actual Output

Accessed buffer 'myBufferName' with 200 samples and 2 dimensions **Good and expected!**
Converted buffer samples to numpy array with shape: (200,) and dtype: float32 *

• np.asarray(buf) silently returns a flat array of unclear dimensionality

Suggestion

If possible:

• Clarify np.asarray(buf) behaviour in the pyjs docs.
• Consider a helper function or .to_numpy() API that guarantees correct multichannel extraction.
• Emit a warning if np.asarray(buf) is used on a multichannel buffer, since it may silently return incomplete data.

[shakfu]

Thanks for your helpful bug report. Let's we what can done to resolve the issue.

[shakfu]

@ashNotKetchup

If you are impatient, I pushed some fixes to the py project where I have been experimenting with the py external's code. I will probably migrate some of these to the py-js project when I have some spare time.