This independent study turns an explicitly labeled Word procedure into JSON and CSV while preserving a reference to each source paragraph. It uses a fictional two-step work instruction, not a client document.
The Python parser records the source file hash, title, optional safety notes and consecutive steps. Missing or out-of-order labels stop the run. Tables, images and automatic Word lists are rejected rather than silently dropped. Formula-like values are guarded in CSV, while JSON retains their original text. Existing output files cannot be overwritten.
The parser passed 28 unit tests. Four separate command-line checks covered Unicode, leading zeros, quoted values and preservation when an output folder already exists. The public study shows the source paragraphs beside the recorded result, with the parser, tests and both output files available to download.
View the study and download the source:
https://sandersaxbuilds.github.io/selected-work/sop-extraction-study.html
This supports the demonstrated template only. A real project would first need an anonymized document and agreed output mapping. Native Word rendering and any ION MES integration remain outside this demonstration.
0 likes · 2 views · Oct 17, 2026
Fictional Word-to-JSON and CSV study with paragraph references, refusal checks and a tested Python parser. Template-specific, no MES connection.