What you want is the bracketed notation the forensic field uses, and STRnaming is the tool that generates it algorithmically from sequence (lusSTR wraps it if you'd rather run a batch through a CLI). TRF won't do it, since it reports one consensus period and a copy number rather than decomposing a compound repeat into blocks.

There's a bigger problem with the markers in your table though. DYF387S1 and DYS399S1 are multi-copy - that's what the "multi (2)" and "multi (3)" column is telling you - so two or three loci amplify together and a single Sanger read is a superposition of two or three different alleles. You can't recover block structure from that trace whatever software you point at it. It's also why STRait Razor, FDSTools and lusSTR are all built around NGS amplicon data: separating the copies needs reads that each came from one molecule.

If you're stuck with Sanger, the workable route is cloning the amplicon and sequencing individual colonies so each read comes from one copy. Tedious, but it does give you something STRnaming can actually parse.