Getting started

Install

Install from Conda-Forge:

conda install -c conda-forge thermoml-io

The Conda-Forge package includes the dependencies for YAML and Parquet export. For pandas integration, add pandas to the same environment:

conda install -c conda-forge thermoml-io pandas

Alternatively, install the core package from PyPI:

pip install thermoml-io

With PyPI, install every exporter and pandas integration through the package extras:

pip install "thermoml-io[export,pandas]"

Load one source

from thermoml_io import load_thermoml_url

document = load_thermoml_url(
    "https://trc.nist.gov/ThermoML/10.1016/j.fluid.2015.07.026.xml"
)

print(document.citation.normalized_doi)
print(document.schema_version)
print(document.provenance.sha256)

The loader accepts HTTPS only, applies a configurable size limit, and records the source URL, checksum, and UTC retrieval time. It does not persist the XML.

Local files and in-memory bytes use parse_thermoml:

from thermoml_io import parse_thermoml

document = parse_thermoml("path/to/user-supplied.xml")

Validate against a pinned schema

Pass an explicit local schema to parse_thermoml(..., schema=...). The package does not silently fetch the live NIST XSD because that URL can change over time. Record the schema checksum in the surrounding research workflow.

Build a collection

from thermoml_io import ThermoMLCollection

collection = ThermoMLCollection((document,))

The collection can search multiple publications while preserving the source document for every result.