#! /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:resource_classification_scheme a skos:ConceptScheme)
   (gnc:resource_classification_scheme skos:prefLabel "GeneNetwork Classification Scheme For Resources which are either defines as a dataset, an inbred group, or a species.")
   (gnc:resource_classification_scheme xkos:numberOfLevels "3")
   (gnc:resource_classification_scheme xkos:levels "( gnc:dataset_type gnc:set gnc:species )")
   (gnc:dataset_type a xkos:ClassificationLevel)
   (gnc:dataset_type skos:prefLabel "The Type of a Dataset which can be a ProbeSet, Genotype, or Phenotype")
   (gnc:dataset_type xkos:depth "1")
   (gnc:dataset_type skos:member gnc:probeset)
   (gnc:dataset_type skos:member gnc:genotype)
   (gnc:dataset_type 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))
                             #:separator "_"
                             #:proc string-downcase))))

(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)
          #:separator "_"
          #:proc (lambda (x) x)))))

(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:short_name a owl:ObjectProperty)
   (gnt:short_name rdfs:domain gnc:species)
   (gnt:short_name skos:definition "The short name of a given resource")
   (gnt:belongs_to_species a rdf:property)
   (gnt:belongs_to_species rdf:comment "This resource given to this species")
   (gnt:belongs_to_species rdf:label "belongsToSpecies"))
  (triples
      (string->identifier "" (remap-species-identifiers (field Species Fullname))
                          #:separator "_"
                          #:proc string-downcase)
    (set skos:inScheme 'gnc:resource_classification_scheme)
    (set rdfs:label (remap-species-identifiers (field Species Fullname)))
    (set skos:prefLabel (field Species MenuName))
    (set skos:altLabel (field Species SpeciesName))
    (set gnt:short_name (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:genetic_type a owl:ObjectProperty)
   (gnt:genetic_type 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:mapping_method a owl:ObjectProperty)
   (gnt:mapping_method rdfs:domain gnc:set)
   (gnt:belongs_to_group a rdf:property)
   (gnt:belongs_to_group rdf:comment "This resource given to this group")
   (gnt:belongs_to_group rdf:label "belongs_to_group"))
  (triples (string->identifier
            "set" (field InbredSet Name InbredSetName)
            #:separator "_"
            #:proc (lambda (x) x))
    (set skos:inScheme 'gnc:resource_classification_scheme)
    (set rdfs:label (field InbredSet FullName))
    (set skos:prefLabel (field InbredSet Name InbredSetName))
    (set gnt:genetic_type (field InbredSet GeneticType))
    (set gnt:family (field InbredSet Family))
    (set gnt:mapping_method (field MappingMethod Name))
    (set gnt:code (field InbredSet InbredSetCode))
    (set xkos:generalizes
         (string->identifier "" (remap-species-identifiers (field Species Fullname))
                             #:separator "_"
                             #:proc string-downcase))))



(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))))