Simplify raw_to_bids

doc/api.rst

@@ -13,7 +13,17 @@ MNE BIDS (:py:mod:`mne_bids.mne_bids`):
 .. autosummary::
    :toctree: generated/
 
-   raw_to_bids
+   write_raw_bids
+   make_bids_filename
+
+IO (:py:mod:`mne_bids.io`):
+
+.. current_module:: mne_bids.io
+
+.. autosummary::
+   :toctree: generated/
+
+   read_raw
 
 Utils (:py:mod:`mne_bids.utils`):
 
@@ -24,5 +34,4 @@ Utils (:py:mod:`mne_bids.utils`):
 
    print_dir_tree
    make_bids_folders
-   make_bids_filename
    make_dataset_description

examples/convert_ds000117.py

@@ -24,7 +24,8 @@
 # Let us import ``mne_bids``
 
 import os.path as op
-from mne_bids import raw_to_bids
+from mne_bids import raw_to_bids, make_bids_filename
+from mne_bids.io import read_raw
 from mne_bids.datasets import fetch_faces_data
 from mne_bids.utils import print_dir_tree
 
@@ -63,13 +64,17 @@
 for subject_id in subject_ids:
     subject = 'sub%03d' % subject_id
     for run in runs:
-        raw_file = op.join(data_path, repo, subject, 'MEG',
-                           'run_%02d_raw.fif' % run)
+        fname = op.join(data_path, repo, subject, 'MEG',
+                        'run_%02d_raw.fif' % run)
+
+        raw = read_raw(fname)
+        bids_fname = make_bids_filename(
+            subject_id='%02d' % subject_id, session='01', run=run,
+            task='VisualFaces', suffix='meg.fif')
 
         # Make it BIDS compatible
-        raw_to_bids(subject_id='%02d' % subject_id, session_id='01', run=run,
-                    task='VisualFaces', raw_file=raw_file,
-                    event_id=event_id, output_path=output_path, overwrite=True)
+        raw_to_bids(raw, output_path, bids_fname, event_id=event_id,
+                    overwrite=True)
 
 ###############################################################################
 # Now let's see the structure of the BIDS folder we created.

examples/convert_mne_sample.py

@@ -19,7 +19,8 @@
 
 import os.path as op
 from mne.datasets import sample
-from mne_bids import raw_to_bids
+from mne_bids import write_raw_bids, make_bids_filename
+from mne_bids.io import read_raw
 from mne_bids.utils import print_dir_tree
 
 ###############################################################################
@@ -29,17 +30,18 @@
 event_id = {'Auditory/Left': 1, 'Auditory/Right': 2, 'Visual/Left': 3,
             'Visual/Right': 4, 'Smiley': 5, 'Button': 32}
 
-raw_file = op.join(data_path, 'MEG', 'sample', 'sample_audvis_raw.fif')
+fname = op.join(data_path, 'MEG', 'sample', 'sample_audvis_raw.fif')
 events_data = op.join(data_path, 'MEG', 'sample', 'sample_audvis_raw-eve.fif')
 output_path = op.join(data_path, '..', 'MNE-sample-data-bids')
 
 ###############################################################################
 # Finally, we specify the raw_file and events_data
 
-raw_to_bids(subject_id='01', run='01', session_id='01', task='audiovisual',
-            raw_file=raw_file, events_data=events_data,
-            output_path=output_path, event_id=event_id,
-            overwrite=True)
+raw = read_raw(fname)
+bids_fname = make_bids_filename(subject='01', run='01', session='01',
+                                task='audiovisual', suffix='meg.fif')
+write_raw_bids(raw, bids_fname, output_path, events_data=events_data,
+               event_id=event_id, overwrite=True)
 
 ###############################################################################
 # Now let's see the structure of the BIDS folder we created.

examples/make_eeg_bids.py

@@ -25,7 +25,7 @@
 from mne.datasets import eegbci
 from mne.io import read_raw_edf
 
-from mne_bids import raw_to_bids
+from mne_bids import write_raw_bids, make_bids_filename
 from mne_bids.utils import print_dir_tree
 
 ###############################################################################
@@ -103,7 +103,7 @@
 # * output_path
 #
 # ... as you can see in the docstring:
-print(raw_to_bids.__doc__)
+print(write_raw_bids.__doc__)
 
 ###############################################################################
 # We loaded 'S001R02.edf', which corresponds to subject 1 in the second task.
@@ -129,8 +129,10 @@
 trial_type = {'rest': 0, 'imagine left fist': 1, 'imagine right fist': 2}
 
 # Now convert our data to be in a new BIDS dataset.
-raw_to_bids(subject_id=subject_id, task=task, raw_file=raw_file,
-            output_path=output_path, kind=kind, event_id=trial_type)
+bids_fname = make_bids_filename(subject=subject_id, task=task,
+                                suffix='eeg.edf')
+write_raw_bids(raw_file, bids_fname, output_path, kind=kind,
+               event_id=trial_type)
 
 ###############################################################################
 # What does our fresh BIDS directory look like?
@@ -162,16 +164,11 @@
                                 'S{:03}R{:02}.edf'.format(subj_idx, task_idx))
         raw = read_raw_edf(edf_path, preload=True)
 
-        # `kind` and `trial_type` were already defined above
-        raw_to_bids(subject_id='{:03}'.format(subj_idx),
-                    task=task_names[task_idx],
-                    run=run_mapping[task_idx],
-                    raw_file=raw,
-                    output_path=output_path,
-                    kind=kind,
-                    event_id=trial_type,
-                    overwrite=False
-                    )
+        bids_fname = make_bids_filename(
+            subject='{:03}'.format(subj_idx), task=task_names[task_idx],
+            run=run_mapping[task_idx], suffix='eeg.edf')
+        write_raw_bids(raw, bids_fname, output_path, kind=kind,
+                       event_id=trial_type, overwrite=False)
 
 ###############################################################################
 # Step 3: Check and compare with standard

mne_bids/__init__.py

@@ -3,6 +3,6 @@
 __version__ = '0.1.dev0'
 
 
-from .mne_bids import raw_to_bids  # noqa
+from .mne_bids import write_raw_bids  # noqa
 from .utils import make_bids_folders, make_bids_filename  # noqa
 from .config import BIDS_VERSION  # noqa

mne_bids/io.py

@@ -32,8 +32,8 @@ def _parse_ext(raw_fname, verbose=False):
     return fname, ext
 
 
-def _read_raw(raw_fname, electrode=None, hsp=None, hpi=None, config=None,
-              montage=None, verbose=None):
+def read_raw(raw_fname, electrode=None, hsp=None, hpi=None, config=None,
+             montage=None, verbose=None):
     """Read a raw file into MNE, making inferences based on extension."""
     fname, ext = _parse_ext(raw_fname)
 

mne_bids/mne_bids.py

@@ -19,7 +19,6 @@
 from mne.io.pick import channel_type
 from mne.io import BaseRaw
 from mne.channels.channels import _unit2human
-from mne.externals.six import string_types
 
 from datetime import datetime
 from warnings import warn
@@ -29,8 +28,8 @@
                     make_dataset_description, _write_json,
                     _read_events, _mkdir_p, age_on_date,
                     copyfile_brainvision, copyfile_eeglab,
-                    _infer_eeg_placement_scheme)
-from .io import (_parse_ext, _read_raw, ALLOWED_EXTENSIONS)
+                    _infer_eeg_placement_scheme, _parse_bids_filename)
+from .io import (_parse_ext, ALLOWED_EXTENSIONS)
 
 
 ALLOWED_KINDS = ['meg', 'eeg', 'ieeg']
@@ -485,30 +484,20 @@ def _sidecar_json(raw, task, manufacturer, fname, kind, eeg_ref=None,
     return fname
 
 
-def raw_to_bids(subject_id, task, raw_file, output_path, session_id=None,
-                acquisition=None, run=None, kind='meg', events_data=None,
-                event_id=None, hpi=None, electrode=None, hsp=None,
-                eeg_ref=None, eeg_gnd=None, config=None, overwrite=True,
-                verbose=True):
+def write_raw_bids(raw, bids_fname, output_path, kind='meg', events_data=None,
+                   event_id=None, eeg_ref=None, eeg_gnd=None, overwrite=True,
+                   verbose=True):
     """Walk over a folder of files and create BIDS compatible folder.
 
     Parameters
     ----------
-    subject_id : str
-        The subject name in BIDS compatible format ('01', '02', etc.)
-    task : str
-        Name of the task the data is based on.
-    raw_file : str | instance of mne.Raw
+    raw : instance of mne.Raw
         The raw data. If a string, it is assumed to be the path to the raw data
         file. Otherwise it must be an instance of mne.Raw
-    output_path : str
-        The path of the BIDS compatible folder
-    session_id : str | None
-        The session name in BIDS compatible format.
-    acquisition : str | None
-        Acquisition parameter for the dataset.
-    run : int | None
-        The run number for this dataset.
+    bids_fname : str
+        The file path of the BIDS compatible raw file. In the case of multifile
+        systems (e.g., vhdr, .ds etc.), this is the path to a folder containing
+        all the files.
     kind : str, one of ('meg', 'eeg', 'ieeg')
         The kind of data being converted. Defaults to "meg".
     events_data : str | array | None
@@ -517,25 +506,11 @@ def raw_to_bids(subject_id, task, raw_file, output_path, session_id=None,
         inferred from the stim channel using `mne.find_events`.
     event_id : dict | None
         The event id dict used to create a 'trial_type' column in events.tsv
-    hpi : None | str | list of str
-        Marker points representing the location of the marker coils with
-        respect to the MEG Sensors, or path to a marker file.
-        If list, all of the markers will be averaged together.
-    electrode : None | str
-        Digitizer points representing the location of the fiducials and the
-        marker coils with respect to the digitized head shape, or path to a
-        file containing these points.
-    hsp : None | str | array, shape = (n_points, 3)
-        Digitizer head shape points, or path to head shape file. If more than
-        10`000 points are in the head shape, they are automatically decimated.
     eeg_ref : str
         Description of the type of reference used and (when applicable) of
         location of the reference electrode. Defaults to None.
     eeg_gnd : str
         Description  of the location of the ground electrode. Defaults to None.
-    config : str | None
-        A path to the configuration file to use if the data is from a BTi
-        system.
     overwrite : bool
         If the file already exists, whether to overwrite it.
     verbose : bool
@@ -549,26 +524,22 @@ def raw_to_bids(subject_id, task, raw_file, output_path, session_id=None,
     of the participant correctly.
 
     """
-    if isinstance(raw_file, string_types):
-        # We must read in the raw data
-        raw = _read_raw(raw_file, electrode=electrode, hsp=hsp, hpi=hpi,
-                        config=config, verbose=verbose)
-        _, ext = _parse_ext(raw_file, verbose=verbose)
-        raw_fname = raw_file
-    elif isinstance(raw_file, BaseRaw):
-        # We got a raw mne object, get back the filename if possible
-        # Assume that if no filename attr exists, it's a fif file.
-        raw = raw_file.copy()
-        if hasattr(raw, 'filenames'):
-            _, ext = _parse_ext(raw.filenames[0], verbose=verbose)
-            raw_fname = raw.filenames[0]
-        else:
-            # FIXME: How to get the filename if no filenames attribute?
-            raw_fname = 'unknown_file_name'
-            ext = '.fif'
-    else:
-        raise ValueError('raw_file must be an instance of str or BaseRaw, '
-                         'got %s' % type(raw_file))
+    if not isinstance(raw, BaseRaw):
+        raise ValueError('raw_file must be an instance of BaseRaw, '
+                         'got %s' % type(raw))
+
+    # We got a raw mne object, get back the filename if possible
+    # Assume that if no filename attr exists, it's a fif file.
+    if not hasattr(raw, 'filenames'):
+        raise ValueError('raw.filenames is missing.')
+
+    _, ext = _parse_ext(raw.filenames[0], verbose=verbose)
+    raw_fname = raw.filenames[0]
+
+    params = _parse_bids_filename(bids_fname)
+    subject_id, session_id = params['sub'], params['ses']
+    acquisition, task, run = params['acq'], params['task'], params['run']
+    fname = os.path.join(output_path, bids_fname)
 
     data_path = make_bids_folders(subject=subject_id, session=session_id,
                                   kind=kind, root=output_path,
@@ -594,18 +565,6 @@ def raw_to_bids(subject_id, task, raw_file, output_path, session_id=None,
     data_meta_fname = make_bids_filename(
         subject=subject_id, session=session_id, task=task, run=run,
         acquisition=acquisition, suffix='%s.json' % kind, prefix=data_path)
-    if ext in ['.fif', '.ds', '.vhdr', '.edf', '.bdf', '.set', '.cnt']:
-        raw_file_bids = make_bids_filename(
-            subject=subject_id, session=session_id, task=task, run=run,
-            acquisition=acquisition, suffix='%s%s' % (kind, ext))
-    else:
-        raw_folder = make_bids_filename(
-            subject=subject_id, session=session_id, task=task, run=run,
-            acquisition=acquisition, suffix='%s' % kind)
-        raw_file_bids = make_bids_filename(
-            subject=subject_id, session=session_id, task=task, run=run,
-            acquisition=acquisition, suffix='%s%s' % (kind, ext),
-            prefix=raw_folder)
     events_tsv_fname = make_bids_filename(
         subject=subject_id, session=session_id, task=task,
         acquisition=acquisition, run=run, suffix='events.tsv',
@@ -636,18 +595,15 @@ def raw_to_bids(subject_id, task, raw_file, output_path, session_id=None,
                   eeg_gnd, verbose)
     _participants_tsv(raw, subject_id, "n/a", participants_fname, verbose)
     _channels_tsv(raw, channels_fname, verbose)
-    _scans_tsv(raw, os.path.join(kind, raw_file_bids), scans_fname, verbose)
+    _scans_tsv(raw, os.path.join(kind, bids_fname), scans_fname, verbose)
 
-    # set the raw file name to now be the absolute path to ensure the files
-    # are placed in the right location
-    raw_file_bids = os.path.join(data_path, raw_file_bids)
-    if os.path.exists(raw_file_bids) and not overwrite:
+    if os.path.exists(fname) and not overwrite:
         raise ValueError('"%s" already exists. Please set'
-                         ' overwrite to True.' % raw_file_bids)
-    _mkdir_p(os.path.dirname(raw_file_bids))
+                         ' overwrite to True.' % fname)
+    _mkdir_p(os.path.dirname(fname))
 
     if verbose:
-        print('Writing data files to %s' % raw_file_bids)
+        print('Writing data files to %s' % fname)
 
     if ext not in ALLOWED_EXTENSIONS:
         raise ValueError('ext must be in %s, got %s'
@@ -657,17 +613,17 @@ def raw_to_bids(subject_id, task, raw_file, output_path, session_id=None,
     # Re-save FIF files to fix the file pointer for files with multiple parts
     # This is WIP, see: https://github.com/mne-tools/mne-python/pull/5470
     if ext in ['.fif']:
-        raw.save(raw_file_bids, overwrite=overwrite)
+        raw.save(fname, overwrite=overwrite)
     # CTF data is saved in a directory
     elif ext == '.ds':
-        sh.copytree(raw_fname, raw_file_bids)
+        sh.copytree(raw_fname, fname)
     # BrainVision is multifile, copy over all of them and fix pointers
     elif ext == '.vhdr':
-        copyfile_brainvision(raw_fname, raw_file_bids)
+        copyfile_brainvision(raw_fname, fname)
     # EEGLAB .set might be accompanied by a .fdt - find out and copy it too
     elif ext == '.set':
-        copyfile_eeglab(raw_fname, raw_file_bids)
+        copyfile_eeglab(raw_fname, fname)
     else:
-        sh.copyfile(raw_fname, raw_file_bids)
+        sh.copyfile(raw_fname, fname)
 
     return output_path

mne_bids/utils.py

@@ -60,6 +60,20 @@ def _mkdir_p(path, overwrite=False, verbose=False):
             raise
 
 
+def _parse_bids_filename(fname):
+    """Get dict from BIDS fname."""
+    params = dict(sub=None, ses=None, task=None, acq=None,
+                  run=None, proc=None, recording=None, space=None)
+    entities = fname.split('_')
+    for entity in entities[:-1]:
+        key, value = entity.split('-')
+        if key not in params:
+            raise KeyError('Unexpected entity found in filename %s' % entity)
+        params[key] = value
+    params['suffix'] = entities[-1]
+    return params
+
+
 def make_bids_filename(subject=None, session=None, task=None,
                        acquisition=None, run=None, processing=None,
                        recording=None, space=None, suffix=None, prefix=None):