scitex-io 0.2.0

Universal scientific data I/O with plugin registry

scitex-io

Universal scientific data I/O with plugin registry

Full Documentation · pip install scitex-io

---

Problem

Three problems recur in every scientific Python project:

1. Format fragmentation. Loading a CSV requires pandas.read_csv(), an HDF5 file requires h5py.File(), a NumPy array requires numpy.load(). Each format demands its own library, its own API, and its own boilerplate. Operating systems solved this decades ago — double-click any file and the OS dispatches to the right application. Python has no equivalent.

2. Hard-coded parameters scattered across scripts. Sample rates, thresholds, model hyperparameters, plot dimensions — magic numbers buried in code, duplicated across files, impossible to track or share. Changing one parameter means grepping through the entire project.

3. Figures without provenance. A saved PNG has no record of the code, parameters, or session that produced it. Months later, reproducing a figure means reverse-engineering which script with which settings generated it.

Solution

scitex-io addresses all three:

• save()/load() — One interface for 30+ formats with automatic extension-based dispatch. A plugin registry lets you add custom formats without modifying the library.
• load_configs() — Loads all YAML files from a config/ directory into a single DotDict with dot-notation access. Parameters are version-controlled, centralized, and separate from code.
• embed_metadata()/read_metadata() — Embeds provenance (timestamps, session IDs, parameters) directly into image and PDF files. The figure carries its own history.

Supported Formats (30+)

Category	Extensions
Spreadsheet	.csv, .tsv, .xlsx, .xls
Scientific	.npy, .npz, .mat, .hdf5, .h5, .zarr
Serialization	.pkl, .pickle, .pkl.gz, .joblib
ML/DL	.pth, .pt, .cbm
Config	.json, .yaml, .yml
Documents	.txt, .md, .pdf, .docx, .tex
Images	.png, .jpg, .jpeg, .gif, .tiff, .tif, .svg
Media	.mp4
Web	.html
Bibliography	.bib

Installation

Requires Python >= 3.9.

pip install scitex-io

For MCP server support:

pip install scitex-io[mcp]

SciTeX users: pip install scitex already includes scitex-io.

Quickstart

Save and Load

from scitex_io import save, load

# Universal save/load — format auto-detected from extension
import pandas as pd
df = pd.DataFrame({"x": [1, 2, 3], "y": [4, 5, 6]})
save(df, "data.csv")
loaded = load("data.csv")

# 30+ formats work the same way
import numpy as np
save(np.array([1, 2, 3]), "data.npy")
save({"key": "value"}, "config.yaml")
save({"nested": [1, 2]}, "data.json")

Project Configuration

Hard-coded parameters belong in config files, not in code. Use UPPER_CASE keys — Python's convention for constants — to signal that these are user-defined values:

project/
  config/
    PATHS.yaml          # DATA_DIR: /data/experiment_01
    PREPROCESS.yaml     # SAMPLE_RATE: 1000, BANDPASS: [0.5, 40]
    MODEL.yaml          # HIDDEN_DIM: 256, DROPOUT: 0.3
    PLOT.yaml           # FIGSIZE: [180, 60], DPI: 300
    IS_DEBUG.yaml       # IS_DEBUG: true

from scitex_io import load_configs

CONFIG = load_configs()          # loads ./config/*.yaml
CONFIG.PATHS.DATA_DIR            # "/data/experiment_01"
CONFIG.PREPROCESS.SAMPLE_RATE    # 1000
CONFIG.MODEL.HIDDEN_DIM          # 256

# Debug mode: DEBUG_ prefixed keys override their counterparts
# In MODEL.yaml: { HIDDEN_DIM: 256, DEBUG_HIDDEN_DIM: 32 }
CONFIG = load_configs(IS_DEBUG=True)
CONFIG.MODEL.HIDDEN_DIM          # 32 (debug value promoted)

Returns a DotDict — a nested dictionary with dot-notation access. Parameters become version-controlled, shareable, and separate from code.

Metadata Embedding

Embed provenance into figures so they carry their own history:

from scitex_io import embed_metadata, read_metadata, has_metadata

# Embed metadata into an image
embed_metadata("figure.png", {
    "experiment": "exp_042",
    "model": "resnet50",
    "accuracy": 0.94,
    "timestamp": "2026-03-11",
})

# Read it back — months later, from the file alone
meta = read_metadata("figure.png")
print(meta["experiment"])    # "exp_042"

# Check if a file has embedded metadata
has_metadata("figure.png")   # True

Supports PNG (tEXt chunks), JPEG (EXIF), SVG (XML metadata), and PDF (Info Dictionary).

Custom Format Registration

from scitex_io import register_saver, register_loader, save, load

@register_saver(".custom")
def save_custom(obj, path, **kwargs):
    with open(path, "w") as f:
        f.write(str(obj))

@register_loader(".custom")
def load_custom(path, **kwargs):
    with open(path) as f:
        return f.read()

save("hello", "data.custom")
assert load("data.custom") == "hello"

Three Interfaces

Python API

from scitex_io import save, load, list_formats, register_saver, register_loader
from scitex_io import load_configs, DotDict
from scitex_io import embed_metadata, read_metadata, has_metadata

save(obj, "path.ext")        # Save any object
data = load("path.ext")      # Load any file
fmts = list_formats()        # Show all registered formats
cfg  = load_configs()        # Load ./config/*.yaml as DotDict
embed_metadata("fig.png", d) # Embed provenance into figure

Full API reference

CLI Commands

scitex-io --help-recursive          # Show all commands
scitex-io info                      # Show registered formats
scitex-io configs                   # Load and display project configs
scitex-io configs -d ./my_configs   # Custom config directory
scitex-io configs --json            # Output as JSON
scitex-io list-python-apis -vv      # List Python APIs with signatures
scitex-io version                   # Show version
scitex-io mcp start                 # Start MCP server
scitex-io mcp doctor                # Check MCP health
scitex-io mcp list-tools -vv        # List MCP tools with parameters

Full CLI reference

MCP Server — for AI Agents

AI agents can save, load, and discover formats autonomously.

Tool	Description
io_list_formats	List all registered save/load formats
io_load	Load data from any supported format
io_save	Save data to any supported format
io_load_configs	Load YAML project configurations
io_register_info	Show how to register custom formats

scitex-io mcp start

Full MCP specification

Lint Rules

Detected by scitex-linter when this package is installed.

Rule	Severity	Message
STX-IO001	warning	np.save() detected — use stx.io.save() for provenance tracking
STX-IO002	warning	np.load() detected — use stx.io.load() for provenance tracking
STX-IO003	warning	pd.read_csv() detected — use stx.io.load() for provenance tracking
STX-IO004	warning	.to_csv() detected — use stx.io.save() for provenance tracking
STX-IO005	warning	pickle.dump() detected — use stx.io.save() for provenance tracking
STX-IO006	warning	json.dump() detected — use stx.io.save() for provenance tracking
STX-IO007	warning	.savefig() detected — use stx.io.save(fig, path) for metadata embedding

Part of SciTeX

scitex-io is part of SciTeX. When used inside the SciTeX framework, I/O is seamless:

import scitex

@scitex.session
def main(CONFIG=scitex.INJECTED):
    data = scitex.io.load("input.csv")     # auto-tracked by clew
    result = process(data)
    scitex.io.save(result, "output.csv")   # auto-tracked by clew
    return 0

scitex.io delegates to scitex_io — they share the same API and registry.

The SciTeX ecosystem follows the Four Freedoms for researchers:

Four Freedoms for Research

0. The freedom to run your research anywhere — your machine, your terms.
1. The freedom to study how every step works — from raw data to final manuscript.
2. The freedom to redistribute your workflows, not just your papers.
3. The freedom to modify any module and share improvements with the community.

AGPL-3.0 — because research infrastructure deserves the same freedoms as the software it runs on.

---