taxprofiler/subworkflows/local/db_check.nf

//
// Check input samplesheet and get read channels
//

include { DATABASE_CHECK } from '../../modules/local/database_check'
include { UNTAR          } from '../../modules/nf-core/modules/untar/main'

workflow DB_CHECK {
    take:
    dbsheet // file: /path/to/dbsheet.csv

    main:

    // TODO: make database sheet check
    // Checks:
    // 1) no duplicates,
    // 2) dbs with no special arguments does not have quotes, e.g. just `,,` and NOT `,"",`
    parsed_samplesheet = DATABASE_CHECK ( dbsheet )
        .csv
        .splitCsv ( header:true, sep:',' )
        .dump(tag: "db_split_csv_out")
        .map { create_db_channels(it) }
        .dump(tag: "db_channel_prepped")

    ch_dbs_for_untar = parsed_samplesheet
        .branch {
            untar: it[1].toString().endsWith(".tar.gz") && it[0]['tool'] != 'centrifuge'
            skip: true
        }

    // TODO Filter to only run UNTAR on DBs of tools actually using?
    // TODO make optional whether to save
    UNTAR ( ch_dbs_for_untar.untar )

    ch_final_dbs = ch_dbs_for_untar.skip.mix( UNTAR.out.untar )

    emit:
    dbs = ch_final_dbs                        // channel: [ val(meta), [ db ] ]
    versions = DATABASE_CHECK.out.versions // channel: [ versions.yml ]
}

def create_db_channels(LinkedHashMap row) {
    def meta = [:]
    meta.tool             = row.tool
    meta.db_name          = row.db_name
    meta.db_params        = row.db_params

    def array = []
    if (!file(row.db_path, type: 'dir').exists()) {
        exit 1, "ERROR: Please check input samplesheet -> database could not be found!\n${row.db_path}"
    }
    array = [ meta, file(row.db_path) ]

    return array
}