Skip to content

Examples

Thirteen runnable Python scripts live in examples/; each is self-contained, writes its session under a temporary directory, and prints what it does. MATLAB equivalents of the core operations are shown in the MATLAB guide and exercised by matlab/test_mef3io.m.

Script Shows
01_write_and_read.py Writing float data (NaN = gap, precision inference), whole-channel and windowed reads, read_raw
02_int32_primitive.py The primitive path: int32 counts + conversion factor, validity mask, bit-exact round trip
03_appending.py In-segment append, reopening a session, gaps, forced new segments, rejected conflicts
04_segment_map.py Reader.segments() / toc(): locating data across day-long gaps, targeted windowed reads
05_annotations.py Records at session and channel level; Note/SyLg/EDFA with durations
06_encryption.py Two-level passwords and what each level unlocks
07_legacy_mef_tools_style.py from mef3io import MefReader, MefWriter as a mef_tools drop-in
08_legacy_compatibility.py Legacy-vs-new write/read timing and value comparison (needs mef3io[test] + scipy/tqdm)
09_encryption_replicability.py The full encryption matrix: both writers × plain/encrypted × both readers, plus L1/L2/wrong-password probes (needs mef3io[test])
10_tar_archive.py Single-file sessions: archive_session, reading the tar in place, writer rejection, standard-tar interop
11_legacy_rto_sessions.py Legacy recording-time-offset ("anonymized") sessions: the date recovered from plain bytes, true-UTC reads in mef3io and pymef, consistent appends/channels/notes, the refusals (needs mef3io[test]; see De-identification)
12_edf_to_mef.py Converting EDF, EDF+ or BDF to MEF 3.0, with a check of the result (needs pyedflib: pip install "mef3io[edf]"; see the how-to)
13_anonymize.py De-identified copies: anonymize.check before copying (a refused copy name and its code), anonymize_session with one channel left out and the dates moved to start on 1 January 2000, a copy that keeps the real dates (the default), the verification and its notes, the copy read back (Python and C++ only; see De-identification)

The numbers produced by 08/09 (and the MATLAB/Python twins benchmarks/bindings_benchmark.py + matlab/benchmark_mef3io.m) are aggregated in Performance & legacy comparison.