See issue over here: neilotoole/sq#506
The root cause of the detection issue is the h2non/filetype library's msooxml() detection function, which assumes a specific ZIP entry ordering that varies between XLSX creators.
The h2non/filetype library's OOXML detection (msooxml() function) uses
fragile heuristics that assume specific ZIP entry ordering. XLSX files from
different tools (excelize, Excel, Google Sheets) have varying internal
structures, and the 4KB magic-number approach doesn't reliably detect all
variants.
See issue over here: neilotoole/sq#506