fix: fixed fn to_hashmap so it can avoid an error when @CO tag appeared

[shohei-kojima]

When a header contains @CO line(s), it may issue an error at the line 88 of 4d99455 (let cap = TAG_RE.captures(part).unwrap();) because @CO line may not have values of "([A-Za-z][A-Za-z0-9]):([ -~]+)".

E.g.
ftp://ftp.1000genomes.ebi.ac.uk/vol1/ftp/data_collections/illumina_platinum_pedigree/data/CEU/NA12878/alignment/NA12878.alt_bwamem_GRCh38DH.20150706.CEU.illumina_platinum_ped.cram

Header lines:

samtools view -H NA12878.alt_bwamem_GRCh38DH.20150706.CEU.illumina_platinum_ped.cram | tail -n 5

@CO     $known_indels_file(s) = ftp://ftp.1000genomes.ebi.ac.uk/vol1/ftp/technical/reference/GRCh38_reference_genome/other_mapping_resources/Mills_and_1000G_gold_standard.indels.b38.primary_assembly.vcf.gz
@CO     $known_indels_file(s) = ftp://ftp.1000genomes.ebi.ac.uk/vol1/ftp/technical/reference/GRCh38_reference_genome/other_mapping_resources/ALL.wgs.1000G_phase3.GRCh38.ncbi_remapper.20150424.shapeit2_indels.vcf.gz
@CO     FASTQ=ERR194147_1.fastq.gz
@CO     FASTQ=ERR194147_2.fastq.gz
@CO     FASTQ=ERR194147.fastq.gz

Error:

thread 'main' panicked at 'called `Option::unwrap()` on a `None` value', /home/kooojiii/.cargo/registry/src/garden.eu.org-1ecc6299db9ec823/rust-htslib-0.39.5/src/bam/header.rs:88:49
stack backtrace:
   0: rust_begin_unwind
             at /rustc/fe5b13d681f25ee6474be29d748c65adcd91f69e/library/std/src/panicking.rs:584:5
   1: core::panicking::panic_fmt
             at /rustc/fe5b13d681f25ee6474be29d748c65adcd91f69e/library/core/src/panicking.rs:143:14
   2: core::panicking::panic
             at /rustc/fe5b13d681f25ee6474be29d748c65adcd91f69e/library/core/src/panicking.rs:48:5
   3: rust_htslib::bam::header::Header::to_hashmap

[veldsla]

to_hashmap is indeed wrong. It's also undocumented. The current return type will not be suitable for the CO lines so I guess skipping them would be a valid workaround for this. Maybe add the documentation that this function returns an incomplete map?

It could also be convenient to add a comments() function to complete the functionality?

[shohei-kojima]

Added the documentations and get_comments() function.

[veldsla]

get_comments can avoid the unwrap by using String::from_utf8_lossy. It can also be implemented without allocations (if valid utf-8) by returning an Iterator

Also I think the get_ prefix is commonly used for functions consuming self so the function signature could look something like this:

fn comments(&self) -> impl Iterator<Item=Cow<str>>

The function body is only a few lines. Do you want to have a go?

[shohei-kojima]

Thank you for revising it!
I agree, returning an Iterator is better. I changed it so.

[veldsla]

Isn't the header is already split on newlines in the struct? This would make the flat_map and split unnecessary. I think you only need the filter and map!

[shohei-kojima]

Unfortunately, no. When I generated a Header struct from from_template function, records in the Header struct contained non-split headers. So I needed to split it.
Like this:
[ [ "@hd\tVN:1.5\tGO:none\tSO:coordinate\n@PG\tID:MarkDuplicates\n...." ] ]

[veldsla]

You are correct. Only pushed records are appended to the Vec

[veldsla]

So you need to run rustfmt and apparently the PR title needs to be formatted for the release tool. The clippy warnings appear to be for other code.

[coveralls]

Pull Request Test Coverage Report for Build 2910319725

• 1 of 3 (33.33%) changed or added relevant lines in 1 file are covered.
• No unchanged relevant lines lost coverage.
• Overall coverage decreased (-0.01%) to 94.668%

---

Changes Missing Coverage	Covered Lines	Changed/Added Lines	%
src/bam/header.rs	1	3	33.33%

Totals
Change from base Build 2710471876:	-0.01%
Covered Lines:	11647
Relevant Lines:	12303

---

💛 - Coveralls

[shohei-kojima]

I added #[allow(clippy::read_zero_byte_vec)] to avoid the clippy error, but I am not sure if this is OK. Please review this.

[veldsla]

Looks good to me. Not sure if the tbx::mod change is within the scope of this PR, but it does please clippy.

Now to find someone who can actually merge it 😄 @rust-bio/mergers ?

[pmarks]

This is a breaking change because existing users of to_hashmap will no longer get comments in the output (assuming they were reading BAM files that didn't hit this bug). So I think we should change the title of the PR to feat!: before merging and that will do the right thing. See e.g. #306 for an example - that PR led to a "Breaking Change" entry in the changelog. @veldsla @shohei-kojima sound good?

[veldsla]

You never got comments in the hashmap. It used to panic when a comment was present.

[pmarks]

Ah right, the regex requires the colon. Cool.