"""Utility functions to copy raw data files.
When writing BIDS datasets, we often move and/or rename raw data files. several
original data formats have properties that restrict such operations. That is,
moving/renaming raw data files naively might lead to broken files, for example
due to internal pointers that are not being updated.
"""
# Authors: The MNE-BIDS developers
# SPDX-License-Identifier: BSD-3-Clause
import os
import os.path as op
import re
import shutil as sh
from pathlib import Path
import numpy as np
from mne.io import anonymize_info, read_raw_bdf, read_raw_brainvision, read_raw_edf
from mne.utils import logger, verbose
from scipy.io import loadmat, savemat
from mne_bids.path import BIDSPath, _mkdir_p, _parse_ext
from mne_bids.utils import _check_anonymize, _get_mrk_meas_date, warn
def _copytree(src, dst, **kwargs):
"""See: https://github.com/jupyterlab/jupyterlab/pull/5150."""
try:
sh.copytree(src, dst, **kwargs)
except sh.Error as error:
# `copytree` throws an error if copying to + from NFS even though
# the copy is successful (see https://bugs.python.org/issue24564)
if "[Errno 22]" not in str(error) or not op.exists(dst):
raise
def _get_brainvision_encoding(vhdr_file):
"""Get the encoding of .vhdr and .vmrk files.
Parameters
----------
vhdr_file : str
Path to the header file.
Returns
-------
enc : str
Encoding of the .vhdr file to pass it on to open() function
either 'UTF-8' (default) or whatever encoding scheme is specified
in the header.
"""
with open(vhdr_file, "rb") as ef:
enc = ef.read()
if enc.find(b"Codepage=") != -1:
enc = enc[enc.find(b"Codepage=") + 9 :]
enc = enc.split()[0]
enc = enc.decode()
src = "(read from header)"
else:
enc = "UTF-8"
src = "(default)"
logger.debug(f"Detected file encoding: {enc} {src}.")
return enc
def _get_brainvision_paths(vhdr_path):
"""Get the .eeg/.dat and .vmrk file paths from a BrainVision header file.
Parameters
----------
vhdr_path : str
Path to the header file.
Returns
-------
paths : tuple
Paths to the .eeg/.dat file at index 0 and the .vmrk file at index 1 of
the returned tuple.
"""
fname, ext = _parse_ext(vhdr_path)
if ext != ".vhdr":
raise ValueError(f'Expecting file ending in ".vhdr", but got {ext}')
# Header file seems fine
# extract encoding from brainvision header file, or default to utf-8
enc = _get_brainvision_encoding(vhdr_path)
# ..and read it
with open(vhdr_path, encoding=enc) as f:
lines = f.readlines()
# Try to find data file .eeg/.dat
eeg_file_match = re.search(r"DataFile=(.*\.(eeg|dat))", " ".join(lines))
if not eeg_file_match:
raise ValueError(f"Could not find a .eeg or .dat file link in {vhdr_path}")
else:
eeg_file = eeg_file_match.groups()[0]
# Try to find marker file .vmrk
vmrk_file_match = re.search(r"MarkerFile=(.*\.vmrk)", " ".join(lines))
if not vmrk_file_match:
raise ValueError(f"Could not find a .vmrk file link in {vhdr_path}")
else:
vmrk_file = vmrk_file_match.groups()[0]
# Make sure we are dealing with file names as is customary, not paths
# Paths are problematic when copying the files to another system. Instead,
# always use the file name and keep the file triplet in the same directory
assert os.sep not in eeg_file
assert os.sep not in vmrk_file
# Assert the paths exist
head, tail = op.split(vhdr_path)
eeg_file_path = op.join(head, eeg_file)
vmrk_file_path = op.join(head, vmrk_file)
assert op.exists(eeg_file_path)
assert op.exists(vmrk_file_path)
# Return the paths
return (eeg_file_path, vmrk_file_path)
[docs]
def copyfile_ctf(src, dest):
"""Copy and rename CTF files to a new location.
Parameters
----------
src : path-like
Path to the source raw .ds folder.
dest : path-like
Path to the destination of the new bids folder.
See Also
--------
copyfile_brainvision
copyfile_bti
copyfile_edf
copyfile_eeglab
copyfile_kit
"""
_copytree(src, dest)
# list of file types to rename
file_types = (
".acq",
".eeg",
".dat",
".hc",
".hist",
".infods",
".bak",
".meg4",
".newds",
".res4",
)
# Consider CTF files that are split having consecutively numbered extensions
extra_ctf_file_types = tuple(
f".{i}_meg4" for i in range(1, 21)
) # cap at 20 is arbitrary
file_types += extra_ctf_file_types
# Rename files in dest with the name of the dest directory
fnames = [f for f in os.listdir(dest) if f.endswith(file_types)]
bids_folder_name = op.splitext(op.split(dest)[-1])[0]
for fname in fnames:
ext = op.splitext(fname)[-1]
os.replace(op.join(dest, fname), op.join(dest, bids_folder_name + ext))
[docs]
def copyfile_kit(src, dest, subject_id, session_id, task, run, _init_kwargs):
"""Copy and rename KIT files to a new location.
Parameters
----------
src : path-like
Path to the source raw .con or .sqd folder.
dest : path-like
Path to the destination of the new bids folder.
subject_id : str | None
The subject ID. Corresponds to "sub".
session_id : str | None
The session identifier. Corresponds to "ses".
task : str | None
The task identifier. Corresponds to "task".
run : int | None
The run number. Corresponds to "run".
_init_kwargs : dict
Extract information of marker and headpoints
See Also
--------
copyfile_brainvision
copyfile_bti
copyfile_ctf
copyfile_edf
copyfile_eeglab
"""
# create parent directories in case it does not exist yet
_mkdir_p(op.dirname(dest))
# KIT data requires the marker file to be copied over too
sh.copyfile(src, dest)
data_path = op.split(dest)[0]
datatype = "meg"
if "mrk" in _init_kwargs and _init_kwargs["mrk"] is not None:
hpi = _init_kwargs["mrk"]
acq_map = dict()
if isinstance(hpi, list):
if _get_mrk_meas_date(hpi[0]) > _get_mrk_meas_date(hpi[1]):
raise ValueError("Markers provided in incorrect order.")
_, marker_ext = _parse_ext(hpi[0])
acq_map = dict(zip(["pre", "post"], hpi))
else:
_, marker_ext = _parse_ext(hpi)
acq_map[None] = hpi
for key, value in acq_map.items():
marker_path = BIDSPath(
subject=subject_id,
session=session_id,
task=task,
run=run,
acquisition=key,
suffix="markers",
extension=marker_ext,
datatype=datatype,
)
sh.copyfile(value, op.join(data_path, marker_path.basename))
for acq in ["elp", "hsp"]:
if acq in _init_kwargs and _init_kwargs[acq] is not None:
position_file = _init_kwargs[acq]
task, run, acq = None, None, acq.upper()
position_ext = ".pos"
position_path = BIDSPath(
subject=subject_id,
session=session_id,
task=task,
run=run,
acquisition=acq,
suffix="headshape",
extension=position_ext,
datatype=datatype,
)
sh.copyfile(position_file, op.join(data_path, position_path.basename))
def _replace_file(fname, pattern, replace):
"""Overwrite file, replacing end of lines matching pattern with replace."""
new_content = []
for line in open(fname):
match = re.match(pattern, line)
if match:
line = match.group()[: -len(replace)] + replace + "\n"
new_content.append(line)
with open(fname, "w", encoding="utf-8") as fout:
fout.writelines(new_content)
def _anonymize_brainvision(vhdr_file, date):
"""Anonymize vmrk and vhdr files in place using `date` datetime object."""
_, vmrk_file = _get_brainvision_paths(vhdr_file)
# Go through VMRK
pattern = re.compile(r"^Mk\d+=New Segment,.*,\d+,\d+,\d+,\d{20}$")
replace = date.strftime("%Y%m%d%H%M%S%f")
_replace_file(vmrk_file, pattern, replace)
# Go through VHDR
pattern = re.compile(r"^Impedance \[kOhm\] at \d\d:\d\d:\d\d :$")
replace = f"at {date.strftime('%H:%M:%S')} :"
_replace_file(vhdr_file, pattern, replace)
[docs]
@verbose
def copyfile_brainvision(vhdr_src, vhdr_dest, anonymize=None, verbose=None):
"""Copy a BrainVision file triplet to a new location and repair links.
The BrainVision file format consists of three files:
.vhdr, .eeg/.dat, and .vmrk
The .eeg/.dat and .vmrk files associated with the .vhdr file will be
given names as in `vhdr_dest` with adjusted extensions. Internal file
pointers will be fixed.
Parameters
----------
vhdr_src : path-like
The source path of the .vhdr file to be copied.
vhdr_dest : path-like
The destination path of the .vhdr file.
anonymize : dict | None
If None (default), no anonymization is performed.
If dict, data will be anonymized depending on the keys provided with
the dict: `daysback` is a required key, `keep_his` is an optional key.
`daysback` : int
Number of days by which to move back the recording date in time.
In studies with multiple subjects the relative recording date
differences between subjects can be kept by using the same number
of `daysback` for all subject anonymizations. `daysback` should be
great enough to shift the date prior to 1925 to conform with BIDS
anonymization rules.
`keep_his` : bool
By default (False), all subject information next to the recording
date will be overwritten as well. If True, keep subject information
apart from the recording date.
%(verbose)s
See Also
--------
mne.io.anonymize_info
copyfile_bti
copyfile_ctf
copyfile_edf
copyfile_eeglab
copyfile_kit
"""
# Get extension of the brainvision file
fname_src, ext_src = _parse_ext(vhdr_src)
fname_dest, ext_dest = _parse_ext(vhdr_dest)
if ext_src != ext_dest:
raise ValueError(
f"Need to move data with same extension, "
f' but got "{ext_src}" and "{ext_dest}"'
)
eeg_file_path, vmrk_file_path = _get_brainvision_paths(vhdr_src)
# extract encoding from brainvision header file, or default to utf-8
enc = _get_brainvision_encoding(vhdr_src)
# raise warning if binary file has .dat extension
if ".dat" in eeg_file_path:
warn(
"The file extension of your binary EEG data file is .dat, while "
"the expected extension for raw data is .eeg. "
"This might imply it's preprocessed or processed data: "
"We copied the files and changed the extension to .eeg, "
"but please ensure that this is actually BIDS compatible data!"
)
# Copy data .eeg/.dat ... no links to repair
sh.copyfile(eeg_file_path, fname_dest + ".eeg")
# Write new header and marker files, fixing the file pointer links
# For that, we need to replace an old "basename" with a new one
# assuming that all .eeg/.dat, .vhdr, .vmrk share one basename
__, basename_src = op.split(fname_src)
assert op.split(eeg_file_path)[-1] in [basename_src + ".eeg", basename_src + ".dat"]
assert basename_src + ".vmrk" == op.split(vmrk_file_path)[-1]
__, basename_dest = op.split(fname_dest)
search_lines = [
"DataFile=" + basename_src + ".eeg",
"DataFile=" + basename_src + ".dat",
"MarkerFile=" + basename_src + ".vmrk",
]
with open(vhdr_src, encoding=enc) as fin:
with open(vhdr_dest, "w", encoding=enc) as fout:
for line in fin.readlines():
if line.strip() in search_lines:
line = line.replace(basename_src, basename_dest)
fout.write(line)
with open(vmrk_file_path, encoding=enc) as fin:
with open(fname_dest + ".vmrk", "w", encoding=enc) as fout:
for line in fin.readlines():
if line.strip() in search_lines:
line = line.replace(basename_src, basename_dest)
fout.write(line)
if anonymize is not None:
raw = read_raw_brainvision(vhdr_src, preload=False, verbose=0)
daysback, keep_his, _ = _check_anonymize(anonymize, raw, ".vhdr")
raw.info = anonymize_info(raw.info, daysback=daysback, keep_his=keep_his)
_anonymize_brainvision(fname_dest + ".vhdr", date=raw.info["meas_date"])
for ext in [".eeg", ".vhdr", ".vmrk"]:
_, fname = os.path.split(fname_dest + ext)
dirname = op.dirname(op.realpath(vhdr_dest))
logger.info(f'Created "{fname}" in "{dirname}".')
if anonymize:
logger.info("Anonymized all dates in VHDR and VMRK.")
[docs]
def copyfile_edf(src, dest, anonymize=None):
"""Copy an EDF, EDF+, or BDF file to a new location, optionally anonymize.
.. warning:: EDF/EDF+/BDF files contain two fields for recording dates:
A generic "startdate" field that supports only 2-digit years,
and a "Startdate" field as part of the "local recording
identification", which supports 4-digit years.
If you want to anonymize your file, MNE-BIDS will set the
"startdate" field to 85 (i.e., 1985), the earliest possible
date for that field. However, the "Startdate" field in the
file's "local recording identification" and the date in the
session's corresponding ``scans.tsv`` will be set correctly
according to the argument provided to the ``anonymize``
parameter. Note that it is possible that not all EDF/EDF+/BDF
reading software parses the accurate recording date, and
that for some reading software, the wrong year (1985) may
be parsed.
Parameters
----------
src : path-like
The source path of the .edf or .bdf file to be copied.
dest : path-like
The destination path of the .edf or .bdf file.
anonymize : dict | None
If None (default), no anonymization is performed.
If dict, data will be anonymized depending on the keys provided with
the dict: `daysback` is a required key, `keep_his` is an optional key.
`daysback` : int
Number of days by which to move back the recording date in time.
In studies with multiple subjects the relative recording date
differences between subjects can be kept by using the same number
of `daysback` for all subject anonymizations. `daysback` should be
great enough to shift the date prior to 1925 to conform with BIDS
anonymization rules. Due to limitations of the EDF/BDF format, the
year of the anonymized date will always be set to 1985 in the
'startdate' field of the file. The correctly-shifted year will be
written to the 'local recording identification' region of the
file header, which may not be parsed by all EDF/EDF+/BDF reader
software.
`keep_his` : bool
By default (False), all subject information next to the recording
date will be overwritten as well. If True, keep subject information
apart from the recording date. Participant names and birthdates
will always be anonymized if present, regardless of this setting.
See Also
--------
mne.io.anonymize_info
copyfile_brainvision
copyfile_bti
copyfile_ctf
copyfile_eeglab
copyfile_kit
"""
# Ensure source & destination extensions are the same
fname_src, ext_src = _parse_ext(src)
fname_dest, ext_dest = _parse_ext(dest)
if ext_src.lower() != ext_dest.lower():
raise ValueError(
f"Need to move data with same extension, "
f' but got "{ext_src}" and "{ext_dest}"'
)
if ext_dest in [".EDF", ".BDF"]:
warn(
"Upper-case extension for EDF/BDF files is not supported "
"in BIDS. Converting destination extension to lower-case."
)
ext_dest = ext_dest.lower()
dest = Path(dest).with_suffix(ext_dest)
# Copy data prior to any anonymization
sh.copyfile(src, dest)
# Anonymize EDF/BDF data, if requested
if anonymize is not None:
if ext_src in [".bdf", ".BDF"]:
raw = read_raw_bdf(dest, preload=False, verbose=0)
elif ext_src in [".edf", ".EDF"]:
raw = read_raw_edf(dest, preload=False, verbose=0)
else:
raise ValueError(f"Unsupported file type ({ext_src})")
# Get subject info, recording info, and recording date
with open(dest, "rb") as f:
f.seek(8) # id_info field starts 8 bytes in
id_info = f.read(80).decode("ascii").rstrip()
rec_info = f.read(80).decode("ascii").rstrip()
# Parse metadata from file
if len(id_info) == 0 or len(id_info.split(" ")) != 4:
id_info = "X X X X"
if len(rec_info) == 0 or len(rec_info.split(" ")) != 5:
rec_info = "Startdate X X X X"
pid, sex, birthdate, name = id_info.split(" ")
start_date, admin_code, tech, equip = rec_info.split(" ")[1:5]
# Try to anonymize the recording date
daysback, keep_his, _ = _check_anonymize(anonymize, raw, ".edf")
anonymize_info(raw.info, daysback=daysback, keep_his=keep_his)
start_date = "01-JAN-1985"
meas_date = "01.01.85"
# Anonymize ID info and write to file
if keep_his:
# Always remove participant birthdate and name to be safe
id_info = [pid, sex, "X", "X"]
rec_info = ["Startdate", start_date, admin_code, tech, equip]
else:
id_info = ["0", "X", "X", "X"]
rec_info = ["Startdate", start_date, "X", "mne-bids_anonymize", "X"]
with open(dest, "r+b") as f:
f.seek(8) # id_info field starts 8 bytes in
f.write(bytes(" ".join(id_info).ljust(80), "ascii"))
f.write(bytes(" ".join(rec_info).ljust(80), "ascii"))
f.write(bytes(meas_date, "ascii"))
[docs]
def copyfile_eeglab(src, dest):
"""Copy an EEGLAB file to a new location.
If the EEGLAB ``.set`` file comes with an accompanying ``.fdt`` binary file
that contains the actual data, this function will copy this file, too, and
update all internal pointers in the new ``.set`` file.
Parameters
----------
src : path-like
Path to the source raw .set file.
dest : path-like
Path to the destination of the new .set file.
See Also
--------
copyfile_brainvision
copyfile_bti
copyfile_ctf
copyfile_edf
copyfile_kit
"""
# Get extension of the EEGLAB file
_, ext_src = _parse_ext(src)
fname_dest, ext_dest = _parse_ext(dest)
if ext_src != ext_dest:
raise ValueError(
f"Need to move data with same extension but got {ext_src}, {ext_dest}"
)
# Load the EEG struct
# NOTE: do *not* simplify cells, because this changes the underlying
# structure and potentially breaks re-reading of the file
uint16_codec = None
eeg = loadmat(
file_name=src,
simplify_cells=False,
appendmat=False,
uint16_codec=uint16_codec,
mat_dtype=True,
)
oldstyle = False
if "EEG" in eeg:
eeg = eeg["EEG"]
oldstyle = True
has_fdt_link = False
try:
# If the data field is a string, it points to a .fdt file in src dir
if isinstance(eeg["data"][0, 0][0], str):
has_fdt_link = True
except IndexError:
pass
if has_fdt_link:
fdt_fname = eeg["data"][0, 0][0]
assert fdt_fname.endswith(".fdt"), f"Unexpected fdt name: {fdt_fname}"
head, _ = op.split(src)
fdt_path = op.join(head, fdt_fname)
# Copy the .fdt file and give it a new name
fdt_name_new = fname_dest + ".fdt"
sh.copyfile(fdt_path, fdt_name_new)
# Now adjust the pointer in the .set file
# NOTE: Clunky numpy code is to match MATLAB structure for "savemat"
_, tail = op.split(fdt_name_new)
new_value = np.empty((1, 1), dtype=object)
new_value[0, 0] = np.atleast_1d(np.array(tail))
eeg["data"] = new_value
# Save the EEG dictionary as a Matlab struct again
mdict = dict(EEG=eeg) if oldstyle else eeg
savemat(file_name=dest, mdict=mdict, appendmat=False)
else:
# If no .fdt file, simply copy the .set file, no modifications
# necessary
sh.copyfile(src, dest)
[docs]
def copyfile_bti(raw, dest):
"""Copy BTi data.
Parameters
----------
raw : mne.io.Raw
An MNE-Python raw object of BTi data.
dest : path-like
Destination to copy the BTi data to.
See Also
--------
copyfile_brainvision
copyfile_ctf
copyfile_edf
copyfile_eeglab
copyfile_kit
"""
os.makedirs(dest, exist_ok=True)
for key, val in (
("pdf_fname", None),
("config_fname", "config"),
("head_shape_fname", "hs_file"),
):
keyfile = raw._raw_extras[0].get(key)
# Keep name of pdf file
if key == "pdf_fname":
val = op.basename(keyfile)
# If no headshape file present, cannot copy it
if key == "head_shape_fname" and keyfile is None:
continue
sh.copyfile(keyfile, op.join(dest, val))