Incorrect validation messages for FASTA sequence upload errors

[corneliusroemer]

The backend currently responds not very helpfully when the FASTA upload file from a submitting user is not correct. We should make the validation more precise. Maybe we could try to use an existing FASTA parsing library, if a good one exists, rather than hand roll our own.

Example error at the moment:

Metadata file contains 2 submissionIds that are not present in the sequence file: custom0, custom3; Sequence file contains 1 submissionIds that are not present in the metadata file: custom0

For this fasta tsv and metadata tsv

>custom1
ACTG
>custom2
ACTG
                    >custom3
ACTG
>custom4
ACTG
ACTG
>custom7
                                            
>custom8
ACTG
>custom9
>custom6


>custom5
ACTG
>custom0                       
ACTG

submissionId	date	region	country	division	host
custom4	2020-12-03	Europe	Switzerland	Zürich	Homo sapiens
custom0	2020-12-26	Europe	Switzerland	Bern	Homo sapiens
custom1	2020-12-15	Europe	Switzerland	Schaffhausen	Homo sapiens
custom2	2020-12-02	Europe	Switzerland	Bern	Homo sapiens
custom6	2020-12-16	Europe	Switzerland	Aargau
custom3		Europe	Switzerland	Schaffhausen	Homo sapiens
custom5	2020-12-23	Europe	Switzerland	Basel-Land	Homo sapiens
custom7	2XXXXX	Europe	Switzerland	Sankt Gallen	Homo sapiens
custom8	2020-12-16	Europe	Switzerland	Aargau	Homo sapiens
custom9	2020-12-01	Europe	Switzerland	Basel-Stadt	Homo sapiens

[corneliusroemer]

Some other edge cases to try:

• Emojis in submissionId

[chaoran-chen]

Interesting, what would be the correct behavior? Is this an invalid fasta file or should the whitespace be ignored?

[theosanderson]

IMO this is an invalid file. To me this is relatively low priority though nice to have in due course.

[corneliusroemer]

It's an invalid fasta and it's a bug that we provide an incorrect error message to submitters, though as Theo says not high priority bug

I think per fasta spec (there's not a single one, but I think there are few ambiguities) this is invalid but of course better to double check.