Supported Formats
scitex-io supports 30+ file formats out of the box. Format is determined automatically from the file extension.
Format Table
Category |
Extensions |
Save |
Load |
Backend |
|---|---|---|---|---|
Spreadsheet |
|
Yes |
Yes |
pandas |
Excel |
|
Yes |
Yes |
openpyxl / xlrd |
NumPy |
|
Yes |
Yes |
numpy |
HDF5 |
|
Yes |
Yes |
h5py |
Zarr |
|
Yes |
Yes |
zarr |
MATLAB |
|
Yes |
Yes |
scipy.io |
Pickle |
|
Yes |
Yes |
pickle / gzip |
Joblib |
|
Yes |
Yes |
joblib |
PyTorch |
|
Yes |
Yes |
torch |
CatBoost |
|
Yes |
Yes |
catboost |
JSON |
|
Yes |
Yes |
json (stdlib) |
YAML |
|
Yes |
Yes |
PyYAML |
Text |
|
Yes |
Yes |
built-in open() |
|
No |
Yes |
pdfplumber / PyPDF2 |
|
Word |
|
No |
Yes |
python-docx |
Images |
|
Yes |
Yes |
PIL / matplotlib |
Video |
|
Yes |
No |
matplotlib.animation |
HTML |
|
Yes |
No |
built-in |
LaTeX |
|
Yes |
No |
built-in |
BibTeX |
|
Yes |
Yes |
bibtexparser |
SQLite |
|
No |
Yes |
sqlite3 (stdlib) |
XML |
|
No |
Yes |
xml.etree (stdlib) |
Optuna |
(study objects) |
Yes |
Yes |
optuna |
EEG |
|
No |
Yes |
mne |
Two-Tier Registry
┌──────────────────────────────────────────┐
│ Format Lookup Order │
├──────────────────────────────────────────┤
│ 1. User-registered handlers (priority) │
│ 2. Built-in handlers (fallback) │
└──────────────────────────────────────────┘
User handlers always take precedence. This means you can:
Override a built-in format (e.g., use a custom CSV parser)
Extend with new formats (e.g.,
.parquet,.feather)Swap backends without changing calling code
Optional Dependencies
Not all backends are installed by default. If a format’s backend is missing,
save()/load() will raise an ImportError with installation instructions.
>>> load("data.hdf5")
ImportError: h5py is required for HDF5 support.
Install with: pip install h5py