J Bradley, Sharon C. Glotzer
parsnip provides a lightweight, precise, and domain agnostic interface for parsing data encoded in the Crystallographic Information File (CIF) 1.1 (Brown & McMahon, 2002), 2.0 (Bernstein et al., 2016), and Macromolecular CIF (mmCIF) (Bourne et al., 1997) formats.Designed for programmatic analysis of crystalline systems, parsnip offers a scriptable API and a suite of convenient data retrieval methods for automated lookups of structural information.Its minimal dependency set and high-level interface allows for the rapid incorporation of parsnip into existing libraries within the molecular simulation ecosystem (Ramasubramani et al., 2020).parsnip's primary functionality lies in its ability to accurately reconstruct unit cells from simulation data and from experimental data recorded with limited precision, often with only a few decimal places.Through a combination of rational and floating-point arithmetic, we achieve class-leading accuracy when reconstructing large and complex structures.parsnip's detailed processing of structural data yields better alignment with reported space group and point group symmetries than existing tools (Hester, 2006;Larsen et al., 2017;Wojdyr, 2022), providing an ideal foundation for studies centered on material design.parsnip supports a dictionary-like lookup format for both scalar and tabular data, both of which can be expanded with Unix-style wildcards to simplify complex queries.Convenient methods for parsing unit cell parameters, reconstructing particle positions, and identifying site symmetry data are exposed to streamline common workflows in materials data science.parsnip's clear documentation of conventions and units eliminates ambiguities common to interdisciplinary research.parsnip's use of NumPy structured arrays for data storage simplifies integration into Python, C, and FORTRAN libraries, resulting in a stable, scalable dependency for scientific codebases in materials research at the atomic, molecular, and colloidal scales.