Why is specifying the extension required when reading files?

[ghuls]

Why is specifying the extension required when reading files?

When the wrong one is specified, an empty dataframe is returned.

In [41]: fasta = session.read_fasta_file("Ns.fa", options=bb.FASTAReadOptions()).to_polars()

In [42]: fasta
Out[42]: 
shape: (0, 3)
┌─────┬─────────────┬──────────┐
│ id  ┆ description ┆ sequence │
│ --- ┆ ---         ┆ ---      │
│ str ┆ str         ┆ str      │
╞═════╪═════════════╪══════════╡
└─────┴─────────────┴──────────┘

In [43]: fasta = session.read_fasta_file("Ns.fa", options=bb.FASTAReadOptions(file_extension="fa").to_polars()

In [44]: fasta
Out[44]: 
shape: (1, 3)
┌──────┬─────────────┬─────────────────────────────────┐
│ id   ┆ description ┆ sequence                        │
│ ---  ┆ ---         ┆ ---                             │
│ str  ┆ str         ┆ str                             │
╞══════╪═════════════╪═════════════════════════════════╡
│ chr1 ┆ null        ┆ NNNNNNNNNNNNNNNNNNNNNNNNNNNNNN… │
└──────┴─────────────┴─────────────────────────────────┘

In [47]: fasta = session.read_fasta_file("/home/luna.kuleuven.be/u0079808/Downloads/Ns.fa", options=bb.FASTAReadOptions(file_extension="fasta",)).to_polars()

In [48]: fasta
Out[48]: 
shape: (0, 3)
┌─────┬─────────────┬──────────┐
│ id  ┆ description ┆ sequence │
│ --- ┆ ---         ┆ ---      │
│ str ┆ str         ┆ str      │
╞═════╪═════════════╪══════════╡
└─────┴─────────────┴──────────┘

It would be nice if the extension and compression would be automatically be detected by checking the extension(s) if not explicitly set.

Also at the moment: fasta_sequence_data_type option can't be set as FastaSequenceDataType is not exposed.

[tshauck]

Thanks for bringing this up... it's setup to potentially treat many files as one table (e.g. a directory of FASTAs), which is why if you're not using .fasta it needs to be specified. But you're right that can/should be improved to automatically detect in the single file case.

I'll look into it this afternoon and try to resolve at least for FASTA files (along w/ compression).

[tshauck]

This should work now[1] for FASTA and FASTQ, except for the last one example in that if the user explicitly passes an option for which extension to use biobear will respect that. I've made #154 to continue to make file path options inferable ordered in rough priority. If there's a particular one you're interested in, happy to do that first. Thanks again for raising this!

[1]: on main, will release shortly