fastqrab-steps 0.9.1

use crate::transformations::prelude::*;
use fastqrab_config::{
    dna::HitRegion, fileformats::PhredEncoding, tpd_adapt_bstring, tpd_adapt_dna_bstring_plus_n,
};

/// add a fixed prefix to the start of reads
#[derive(Clone, JsonSchema)]
#[tpd]
#[derive(Debug)]
pub struct Prefix {
    #[schemars(with = "String")]
    #[tpd(adapt_in_verify(String))]
    segment: SegmentIndex,

    #[schemars(with = "String")]
    #[tpd(with = "tpd_adapt_dna_bstring_plus_n")]
    pub seq: BString,
    //we don't check the quality. It's on you if you
    //write non phred values in there
    #[schemars(with = "String")]
    #[tpd(with = "tpd_adapt_bstring")]
    pub qual: BString,

    pub encoding: PhredEncoding,
    #[tpd(alias = "if_label")]
    if_tag: Option<ConditionalTagLabel>,
}

impl VerifyIn<PartialConfig> for PartialPrefix {
    fn verify(
        &mut self,
        parent: &PartialConfig,
        _options: &VerifyOptions,
    ) -> std::result::Result<(), ValidationFailure>
    where
        Self: Sized + toml_pretty_deser::Visitor,
    {
        self.segment.validate_segment(parent);
        if let Some(seq) = self.seq.as_ref()
            && let Some(qual) = self.qual.as_ref()
            && seq.len() != qual.len()
        {
            let spans = vec![
                (self.seq.span(), format!("{} characters", seq.len())),
                (self.qual.span(), format!("{} characters", qual.len())),
            ];
            self.seq.state = TomlValueState::Custom { spans };
            self.seq.help = Some("'seq' and 'qual' must be the same length".to_string());
        }
        self.encoding.or(PhredEncoding::Sanger);

        if let Some(encoding) = self.encoding.as_ref() {
            let (lower, upper) = encoding.limits();
            self.qual.verify(|v| {
                if v.iter().all(|&x| x >= lower && x <= upper) {
                    Ok(())
                } else {
                    Err(ValidationFailure::new(
                        format!(
                            "Quality values must be in the range ({lower}..{upper}) ('{encoding}')"
                        ),
                        None,
                    ))
                }
            });
        } // cov:excl-line
        Ok(())
    }
}

impl TagUser for PartialTaggedVariant<PartialPrefix> {
    fn get_tag_usage(
        &mut self,
        _tags_available: &IndexMap<TagLabel, TagMetadata>,
        _segment_order: &[String],
    ) -> Option<TagUsageInfo<'_>> {
        if let Some(inner) = self.toml_value.value.as_mut() {
            Some(TagUsageInfo {
                used_tags: vec![inner.if_tag.to_used_tag(&[])],
                must_see_all_tags: true,
                ..Default::default()
            })
        } else {
            None // cov:excl-line
        }
    }
}

impl Step for Prefix {
    //to modify location tags

    fn apply(
        &self,
        mut block: FastQBlocksCombined,
        _input_info: &InputInfo,
        _demultiplex_info: &OptDemultiplex,
    ) -> anyhow::Result<(FastQBlocksCombined, bool)> {
        let condition = self
            .if_tag
            .as_ref()
            .map(|tag| get_bool_vec_from_tag(&block, tag));

        block.apply_in_place_wrapped(
            self.segment,
            |read| read.prefix(&self.seq, &self.qual),
            condition.as_deref(),
        );
        let prefix_len = self.seq.len();

        block.filter_tag_locations(
            self.segment,
            |location: HitRegion, _pos, _seq, _read_len: usize| -> NewLocation {
                {
                    NewLocation::New(HitRegion {
                        start: location.start + prefix_len,
                        len: location.len,
                        segment_index: location.segment_index,
                    })
                }
            },
            condition.as_deref(),
        );

        Ok((block, true))
    }
}