The Translate Toolkit implementation of UTX can correctly:
Adjustments and not implemented features where the spec is not clear:
We do not implement the ”#.” comment as we need clarity on this
The ”<space>” override for no part of speech is not implemented
The
spec calls for 2 header lines, while examples in the field have 2-3 lines. We can read as many as supplied but assume the last header line is the column titles
We remove # from all field line entries, some examples in the field have #tgt
as a column name