Add experiment_name argument to OpenEphysBinaryRecordingExtractor

[h-mayorquin]

The current block_index parameter exposes a Neo implementation detail that most users won't understand. Since Open Ephys organizes data into named experiment folders (experiment1, experiment2, etc.), I've added an experiment_name parameter that lets users specify experiments directly by folder name (e.g., experiment_name="experiment2"). I've also improved error messages to list available experiments instead of just mentioning block_index.

I add the experiment_name parameter to OpenEphysBinaryRecordingExtractor with full backward compatibility. I deprecate block_index with a FutureWarning (removal in 0.105.0) and add a get_available_experiments() class method for discovery. My implementation passes experiment_names=[experiment_name] to Neo to filter efficiently. I've enhanced the docstring with the Open Ephys binary format and how it maps to SpikeInterface concepts.

I'm also deprecating experiment_names (plural). This parameter was added to Neo in PR #1159 to handle inconsistent stream configurations across experiments, but this doesn't apply to SpikeInterface since we only load one experiment at a time. The new experiment_name (singular) is clearer for our single-block loading model and achieves the same goal of filtering out incompatible streams so the data can be loaded.

[EDIT] After discussion below, we are keeping block index.

[alejoe91]

Thanks @h-mayorquin

Please let's not deprecate block_index though! We have convenient functions to get the number of blocks and stream names for NEO extractors. These can be used to traverse the blocks/streams and are angnostic to the specific reader. Eg.

from spikeinterface.extractors import recording_extractor_full_dict

reader_type = "openephysbinary" # or any other neo
reader_kwargs = {"folder_path": "path-to-folder"}
num_blocks = se.get_neo_num_blocks(reader_type)
stream_names, stream_ids = se.get_neo_streams(reader_type)

for block_index in range(num_blocks):
    for stream_name in stream_names:
        recording = recording_extractor_full_dict[reader_type](**reader_kwargs)

[alejoe91]

LGTM, but let's not deprecate block_index pls :)

[h-mayorquin]

Undid the deprecation. I was thinking that only this extractor uses the block index but I realize some others do:

MaxwellRecordingExtractor, MCSRawRecordingExtractor NixRecordingExtractor,TdtRecordingExtractor

So now the logic is that you can use either but not both.

[alejoe91]

Undid the deprecation. I was thinking that only this extractor uses the block index but I realize some others do:

MaxwellRecordingExtractor, MCSRawRecordingExtractor NixRecordingExtractor,TdtRecordingExtractor

So now the logic is that you can use either but not both.

Perfect!

[alejoe91]

maybe add settings_2.xml which is generated for experiment2?

[h-mayorquin]

Done.

[chrishalcrow]

Hello, think we'll need to pin neo to >=0.14.3 so that _parse_folder_structure is exposed. Agree?

[h-mayorquin]

Hello, think we'll need to pin neo to >=0.14.3 so that _parse_folder_structure is exposed. Agree?

Agree. I should not have used a private internal detail of neo as well : /