#! /usr/bin/env guile
!#

(use-modules (rnrs programs)
             (rnrs io ports)
             (srfi srfi-1)
             (srfi srfi-26)
             (ice-9 match)
             (ice-9 regex)
             (dump strings)
             (dump sql)
             (dump triples)
             (dump special-forms))



(define %connection-settings
  (call-with-input-file (list-ref (command-line) 1)
    read))



(define (remap-species-identifiers str)
  "This procedure remaps identifiers to standard binominal. Obviously this should
   be sorted by correcting the database!"
  (match str
    ["Fly (Drosophila melanogaster dm6)" "Drosophila melanogaster"]
    ["Oryzias latipes (Japanese medaka)" "Oryzias latipes"]
    ["Macaca mulatta" "Macaca nemestrina"]
    ["Bat (Glossophaga soricina)" "Glossophaga soricina"]
    [str str]))

(define-dump dump-genofreeze
  (tables (GenoFreeze
           (left-join InfoFiles "ON InfoFiles.InfoPageName = GenoFreeze.Name")
           (left-join InbredSet "ON GenoFreeze.InbredSetId = InbredSet.InbredSetId"))
          "WHERE GenoFreeze.public > 0 AND GenoFreeze.confidentiality < 1 AND InfoFiles.InfoPageName IS NULL")
  (triples
      (string->identifier
       ""
       (regexp-substitute/global
        #f "[^A-Za-z0-9:]"
        (regexp-substitute/global
         #f "[^A-Za-z0-9:]"
         (field GenoFreeze Name)
         'pre "_" 'post)
        'pre "_" 'post)
       #:separator ""
       #:proc string-capitalize-first)
    (set rdf:type 'gnc:genotypeDataset)
    (set rdfs:label (field GenoFreeze Name))
    (set skos:prefLabel (field GenoFreeze FullName))
    (set skos:altLabel (field GenoFreeze ShortName))
    (set dct:created (annotate-field
                      (field GenoFreeze CreateTime)
                      '^^xsd:date))
    (set gnt:belongsToInbredSet
         (string->identifier
            "inbredSet" (field InbredSet Name)
            #:separator ""
            #:proc string-capitalize-first))))


(define-dump dump-genotypes
  (tables (Geno
           (left-join Species "USING (SpeciesId)")))
  (schema-triples
   (gnc:genotype a skos:Concept)
   (gnc:genotype
    skos:description
    "This is a set of controlled terms that are used to describe a given genotype")
   (gnt:chr a owl:ObjectProperty)
   (gnt:chr skos:description "This resource is located on a given chromosome")
   (gnt:chr rdfs:domain gnc:genotype)
   (gnt:mb a owl:ObjectProperty)
   (gnt:mb skos:definition "The size of this resource in Mb")
   (gnt:mb rdfs:domain gnc:genotype)
   (gnt:mbMm8 a owl:ObjectProperty)
   (gnt:mbMm8 skos:definition "TODO")
   (gnt:mbMm8 rdfs:domain gnc:genotype)
   (gnt:mb2016 a owl:ObjectProperty)
   (gnt:mb2016 skos:definition "TODO")
   (gnt:mb2016 rdfs:domain gnc:genotype)
   (gnt:hasSequence a owl:ObjectProperty)
   (gnt:hasSequence skos:definition "This resource has a given sequence")
   (gnt:hasSequence rdfs:domain gnc:genotype)
   (gnt:hasSource a owl:ObjectProperty)
   (gnt:hasSource rdfs:domain gnc:genotype)
   (gnt:hasSource skos:definition "This resource was obtained from this given source")
   (gnt:hasAltSourceName a owl:ObjectProperty)
   (gnt:hasAltSourceName rdfs:domain gnc:genotype)
   (gnt:hasAltSourceName
    skos:definition
    "The alternative name this resource was obtained from")
   (gnt:chrNum a owl:ObjectProperty)
   (gnt:chrNum rdfs:domain gnc:genotype)
   (gnt:chrNum skos:definition "The chromosome number for this resource")
   (gnt:chrNum skos:definition "The chromosome number for this resource"))
  (triples
      (string->identifier
       ""
       (regexp-substitute/global
        #f "[^A-Za-z0-9:]"
        (field Geno Name)
        'pre "_" 'post)
       #:separator ""
       #:proc string-capitalize-first)
    (set rdf:type 'gnc:genotype)
    (set skos:prefLabel (sanitize-rdf-string (field Geno Name)))
    (set gnt:chr (field Geno Chr))
    (set gnt:mb (annotate-field
                 (field ("IFNULL(Geno.Mb, '')" Mb)) '^^xsd:double))
    (set gnt:mbMm8 (annotate-field (field ("IFNULL(Geno.Mb_mm8, '')" Mb_mm8))
                                   '^^xsd:double))
    (set gnt:mb2016
         (annotate-field (field ("IFNULL(Geno.Mb_2016, '')" Mb_2016))
                         '^^xsd:double))
    (set gnt:hasSequence (field Geno Sequence))
    (set gnt:hasSource (field Geno Source))
    ;; Only dump Source2 if it differs from Source
    (set gnt:hasAltSourceName
         (field ("IF((Source2 = Source), NULL, Source2)"
                 Source2)))
    (set gnt:belongsToSpecies
         (string->identifier
          "" (remap-species-identifiers (field Species Fullname))
          #:separator ""
          #:proc string-capitalize-first))
    (set gnt:chrNum
         (annotate-field
          (field Geno chr_num)
          '^^xsd:int))
    (set rdfs:comments (field Geno Comments))))



(dump-with-documentation
 (name "Genotype Metadata")
 (connection %connection-settings)
 (table-metadata? #f)
 (prefixes
  '(("dct:" "<http://purl.org/dc/terms/>")
    ("gn:" "<http://genenetwork.org/id/>")
    ("gnc:" "<http://genenetwork.org/category/>")
    ("gnt:" "<http://genenetwork.org/term/>")
    ("rdf:" "<http://www.w3.org/1999/02/22-rdf-syntax-ns#>")
    ("rdfs:" "<http://www.w3.org/2000/01/rdf-schema#>")
    ("owl:" "<http://www.w3.org/2002/07/owl#>")
    ("skos:" "<http://www.w3.org/2004/02/skos/core#>")
    ("xsd:" "<http://www.w3.org/2001/XMLSchema#>")))
 (inputs
  (list dump-genofreeze
        dump-genotypes))
 (outputs
  '(#:documentation "./docs/dump-genotype.md"
    #:rdf "/export/data/genenetwork-virtuoso/dump-genotype.ttl")))