#! /usr/bin/env guile
!#

(use-modules (srfi srfi-1)
             (srfi srfi-26)
             (ice-9 getopt-long)
             (ice-9 match)
             (ice-9 regex)
             (transform strings)
             (transform sql)
             (transform triples)
             (transform special-forms))



(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]))

;; Classification Scheme
(define-transformer classification-scheme-species
  (tables (Species))
  (schema-triples
   (gnc:ResourceClassificationScheme a skos:ConceptScheme)
   (gnc:ResourceClassificationScheme skos:prefLabel "GeneNetwork Classification Scheme For Resources")
   (gnc:ResourceClassificationScheme xkos:numberOfLevels "3")
   (gnc:ResourceClassificationScheme xkos:levels "( gnc:DatasetType gnc:Set gnc:Species )")
   (gnc:DatasetType a xkos:ClassificationLevel)
   (gnc:DatasetType skos:prefLabel "The Type of a Dataset which can be a ProbeSet, Genotype, or Phenotype")
   (gnc:DatasetType xkos:depth "1")
   (gnc:DatasetType skos:member gnc:Probeset)
   (gnc:DatasetType skos:member gnc:Genotype)
   (gnc:DatasetType skos:member gnc:Phenotype)
   (gnc:Probeset skos:prefLabel "mRNA Assay Datasets")
   (gnc:Probeset skos:altLabel "ProbeSet")
   (gnc:Genotype skos:prefLabel "Genotype")
   (gnc:Genotype skos:altLabel "DNA Markers and SNPs")
   (gnc:Phenotype skos:prefLabel "Phenotype")
   (gnc:Phenotype skos:altLabel "Traits and Cofactors")
   (gnc:Species a xkos:ClassificationLevel)
   (gnc:Species skos:prefLabel "The species in which this resource belongs")
   (gnc:Species xkos:depth "3")
   (gnc:Species xkos:specializes gnc:Set))
  (triples "gnc:Species"
    (set skos:member (string->identifier "" (remap-species-identifiers (field Species Fullname))))))

(define-transformer classification-scheme-set
  (tables (InbredSet))
  (schema-triples
   (gnc:Set a xkos:ClassificationLevel)
   (gnc:Set skos:prefLabel "The Type of Set, Ie InbredSet/OutbredSet that a resource can belong to")
   (gnc:Set xkos:depth "2")
   (gnc:Set xkos:generalizes gnc:Species))
  (triples "gnc:Set"
    (set skos:member
         (string->identifier
          "set" (field InbredSet Name InbredSetName)))))

(define-transformer species
  (tables (Species))
  (schema-triples
   (gnt:family a owl:ObjectProperty)
   (gnt:family rdfs:domain gnc:Species)
   (gnt:family skos:definition "This resource belongs to this family")
   (gnt:shortName a owl:ObjectProperty)
   (gnt:shortName rdfs:domain gnc:Species)
   (gnt:shortName skos:definition "The short name of a given resource"))
  (triples
      (string->identifier "" (remap-species-identifiers (field Species Fullname)))
    (set skos:inScheme 'gnc:ResourceClassificationScheme)
    (set rdfs:label (remap-species-identifiers (field Species Fullname)))
    (set skos:prefLabel (field Species MenuName))
    (set skos:altLabel (field Species SpeciesName))
    (set gnt:shortName (field Species Name))
    (set gnt:family (field Species Family))
    (set skos:notation (ontology
                         'taxon:
                         (field Species TaxonomyId)))))

(define-transformer inbred-set
  (tables (InbredSet
           (left-join Species "ON InbredSet.SpeciesId=Species.Id")
           (left-join MappingMethod
                      "ON InbredSet.MappingMethodId=MappingMethod.Id")))
  (schema-triples
   (gnt:geneticType a owl:ObjectProperty)
   (gnt:geneticType rdfs:domain gnc:set)
   (gnt:code a owl:ObjectProperty)
   (gnt:code rdfs:domain gnc:set)
   ;; Already defined as an owl prop in species
   (gnt:family rdfs:domain gnc:Set)
   (gnt:mappingMethod a owl:ObjectProperty)
   (gnt:mappingMethod rdfs:domain gnc:set))
  (triples (string->identifier
            "set" (field InbredSet Name InbredSetName))
    (set skos:inScheme 'gnc:ResourceClassificationScheme)
    (set rdfs:label (field InbredSet FullName))
    (set skos:prefLabel (field InbredSet Name InbredSetName))
    (set gnt:geneticType (field InbredSet GeneticType))
    (set gnt:family (field InbredSet Family))
    (set gnt:mappingMethod (field MappingMethod Name))
    (set gnt:code (field InbredSet InbredSetCode))
    (set xkos:generalizes
         (string->identifier "" (remap-species-identifiers (field Species Fullname))
                             #:separator ""
                             #:proc string-capitalize-first))))



(let* ((option-spec
        '((settings (single-char #\s) (value #t))
          (output (single-char #\o) (value #t))
          (documentation (single-char #\d) (value #t))))
       (options (getopt-long (command-line) option-spec))
       (settings (option-ref options 'settings #f))
       (output (option-ref options 'output #f))
       (documentation (option-ref options 'documentation #f))
       (%connection-settings
        (call-with-input-file settings
          read)))

  (with-documentation
   (name "Species Metadata")
   (connection %connection-settings)
   (table-metadata? #f)
   (prefixes
    '(("gn:" "<http://genenetwork.org/id/>")
      ("gnc:" "<http://genenetwork.org/category/>")
      ("owl:" "<http://www.w3.org/2002/07/owl#>")
      ("gnt:" "<http://genenetwork.org/term/>")
      ("skos:" "<http://www.w3.org/2004/02/skos/core#>")
      ("xkos:" "<http://rdf-vocabulary.ddialliance.org/xkos#>")
      ("rdf:" "<http://www.w3.org/1999/02/22-rdf-syntax-ns#>")
      ("rdfs:" "<http://www.w3.org/2000/01/rdf-schema#>")
      ("taxon:" "<http://purl.uniprot.org/taxonomy/>")))
   (inputs
    (list classification-scheme-species
          classification-scheme-set
          species
          inbred-set))
   (outputs
    `(#:documentation ,documentation
      #:rdf ,output))))