The data extracted from Bruker files is sometimes background data, sometimes sample data. In most of our testing, it's sample data, but in a small number of files, we're seeing background data being extracted.
Using the brukeropusreader Python library shows that sample and background data exist in the files in question.
Via the brukeropusreader Python library, sample data exists in the block whose key is "AB" (absorbance presumably) and background data exists in the data block whose key is "ScSm".
One possibility is that the order in which data blocks are stored is non-deterministic and opusreader is always reading the Nth block. Another is that the wrong block is identified.