#! /usr/bin/env guile
!#

(use-modules (rnrs programs)
             (rnrs io ports)
             (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 (blank-p str)
  (if (string-blank? str) #f str))


(define-transformer gn:set->gn:dataset
  (tables (Species
           (inner-join InbredSet "ON InbredSet.SpeciesId = Species.Id")
           (inner-join PublishFreeze "ON PublishFreeze.InbredSetId = InbredSet.Id"))
          "WHERE PublishFreeze.public > 0 AND Species.Name != 'monkey' GROUP BY Species.Name, PublishFreeze.ShortName")
  (triples (string->identifier "set" (field InbredSet Name InbredSetName) #:separator "_")
    (multiset gnt:has_phenotype_data
              (map (cut string->identifier "dataset" <> #:separator "_")
                   (string-split
                    (field ("GROUP_CONCAT(PublishFreeze.Name SEPARATOR ',')"
                            dataset_name))
                    #\,)))))

(define-transformer gn:dataset->gn:trait
  (tables (PublishXRef
           (left-join InbredSet "ON InbredSet.InbredSetId = PublishXRef.InbredSetId")
           (inner-join PublishFreeze "ON PublishFreeze.InbredSetId = InbredSet.Id")
           (left-join Publication "ON Publication.Id = PublishXRef.PublicationId")
           (left-join Phenotype "ON Phenotype.Id = PublishXRef.PhenotypeId"))
          "WHERE InbredSet.public > 0")
  (triples (string->identifier "dataset" (field PublishFreeze Name) #:separator "_")
    (set gnt:has_phenotype_trait
         (let ((post-abbrev (blank-p (field Phenotype Post_publication_abbreviation)))
               (pre-abbrev (blank-p (field Phenotype Pre_publication_abbreviation)))
               (post-desc (blank-p (field Phenotype Post_publication_description)))
               (pre-desc (blank-p (field Phenotype Post_publication_description))))
           (string->identifier
            "trait"
            (format #f "~a_~a" (field PublishFreeze Name)
                    (or post-abbrev pre-abbrev post-desc pre-desc))
            #:separator "_")))))

(define-transformer gnc:phenotype->gn:phenotype
  (tables (Phenotype))
  (triples "gnc:phenotype"
    (set skos:member
         (let ((post-abbrev (blank-p (field Phenotype Post_publication_abbreviation)))
               (pre-abbrev (blank-p (field Phenotype Pre_publication_abbreviation)))
               (post-desc (blank-p (field Phenotype Post_publication_description)))
               (pre-desc (blank-p (field Phenotype Post_publication_description))))
           (string->identifier
            "phenotype"
            (or post-abbrev pre-abbrev post-desc pre-desc)
            #:separator "_")))))

(define-transformer gn:phenotype->metadata
  (tables (Phenotype))
  (triples (let ((post-abbrev (blank-p (field Phenotype Post_publication_abbreviation)))
                 (pre-abbrev (blank-p (field Phenotype Pre_publication_abbreviation)))
                 (post-desc (blank-p (field Phenotype Post_publication_description)))
                 (pre-desc (blank-p (field Phenotype Post_publication_description))))
             (string->identifier
              "phenotype"
              (or post-abbrev pre-abbrev post-desc pre-desc)
              #:separator "_"))
    (set rdf:type 'gnc:phenotype)
    ;; All phenotypes have a post-publication description
    (set dct:description
         (sanitize-rdf-string
          (field Phenotype Post_publication_description)))
    ;; All phenotypes have a post-publication abbreviation
    (set gnt:abbreviation (sanitize-rdf-string (field Phenotype Post_publication_abbreviation)))
    (set gnt:has_lab_code (field Phenotype Lab_code))
    (set gnt:submitter
         (sanitize-rdf-string (field Phenotype Submitter)))
    (set dct:contributor (sanitize-rdf-string (field Phenotype Owner)))
    (set skos:member
         (let ((post-abbrev (blank-p (field Phenotype Post_publication_abbreviation)))
               (pre-abbrev (blank-p (field Phenotype Pre_publication_abbreviation)))
               (post-desc (blank-p (field Phenotype Post_publication_description)))
               (pre-desc (blank-p (field Phenotype Post_publication_description))))
           (string->identifier
            "phenotype"
            (or post-abbrev pre-abbrev post-desc pre-desc)
            #:separator "_")))))

(define-transformer gn:trait->gn:phenotype
  (tables (PublishXRef
           (left-join InbredSet "ON InbredSet.InbredSetId = PublishXRef.InbredSetId")
           (inner-join PublishFreeze "ON PublishFreeze.InbredSetId = InbredSet.Id")
           (left-join Publication "ON Publication.Id = PublishXRef.PublicationId")
           (left-join Phenotype "ON Phenotype.Id = PublishXRef.PhenotypeId"))
          "WHERE InbredSet.public > 0")
  (triples (let ((post-abbrev (blank-p (field Phenotype Post_publication_abbreviation)))
                 (pre-abbrev (blank-p (field Phenotype Pre_publication_abbreviation)))
                 (post-desc (blank-p (field Phenotype Post_publication_description)))
                 (pre-desc (blank-p (field Phenotype Post_publication_description))))
             (string->identifier
              "trait"
              (format #f "~a_~a" (field PublishFreeze Name)
                      (or post-abbrev pre-abbrev post-desc pre-desc))
              #:separator "_"))
    (set rdf:type 'gnc:phenotype_trait)
    (set gnt:has_strain (string->identifier "set" (field InbredSet Name InbredSetName) #:separator "_"))
    (set owl:equivalentClass
         (field ("CONCAT(IFNULL(InbredSet.InbredSetCode, PublishXRef.InbredSetId), '_', PublishXRef.Id)"
                 Phenotype)))
    (set dct:references
         (let ((pmid (field
                      ("IF(Publication.PubMed_ID IS NULL, '', CONVERT(Publication.PubMed_Id, INT))"
                       pmid)))
               (publication-id (field Publication Id)))
           (if (string-null? pmid)
               (string->identifier "unpublished"
                                   (number->string publication-id))
               (ontology 'pubmed: pmid))))
    (set gnt:has_phenotype
         (let ((post-abbrev (blank-p (field Phenotype Post_publication_abbreviation)))
               (pre-abbrev (blank-p (field Phenotype Pre_publication_abbreviation)))
               (post-desc (blank-p (field Phenotype Post_publication_description)))
               (pre-desc (blank-p (field Phenotype Post_publication_description))))
           (string->identifier
            "phenotype"
            (or post-abbrev pre-abbrev post-desc pre-desc)
            #:separator "_")))
    (set gnt:mean (annotate-field (field ("IFNULL(PublishXRef.mean, '')" mean))
                                  '^^xsd:double))
    (set gnt:locus
         (string->identifier
          ""
          (regexp-substitute/global
           #f "[^A-Za-z0-9:]"
           (sanitize-rdf-string (field PublishXRef Locus))
           'pre "_" 'post)
          #:separator ""
          #:proc string-capitalize-first))
    (set gnt:lod_score (annotate-field
                        (field ("IFNULL((PublishXRef.LRS/4.604), '')" lrs))
                        '^^xsd:double))
    (set gnt:additive
         (annotate-field (field ("IFNULL(PublishXRef.additive, '')" additive))
                         '^^xsd:double))
    (set gnt:sequence (annotate-field (field PublishXRef Sequence) '^^xsd:integer))
    (set rdfs:comment (sanitize-rdf-string (field PublishXRef comments)))))



(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 "Phenotypes Metadata")
   (connection %connection-settings)
   (table-metadata? #f)
   (prefixes
    '(("dct:" "<http://purl.org/dc/terms/>")
      ("gn:" "<http://rdf.genenetwork.org/v1/id/>")
      ("owl:" "<http://www.w3.org/2002/07/owl#>")
      ("gnc:" "<http://rdf.genenetwork.org/v1/category/>")
      ("gnt:" "<http://rdf.genenetwork.org/v1/term/>")
      ("sdmx-measure:" "<http://purl.org/linked-data/sdmx/2009/measure#>")
      ("skos:" "<http://www.w3.org/2004/02/skos/core#>")
      ("rdf:" "<http://www.w3.org/1999/02/22-rdf-syntax-ns#>")
      ("rdfs:" "<http://www.w3.org/2000/01/rdf-schema#>")
      ("xsd:" "<http://www.w3.org/2001/XMLSchema#>")
      ("qb:" "<http://purl.org/linked-data/cube#>")
      ("xkos:" "<http://rdf-vocabulary.ddialliance.org/xkos#>")
      ("pubmed:" "<http://rdf.ncbi.nlm.nih.gov/pubmed/>")))
   (inputs
    (list gn:set->gn:dataset
          gn:dataset->gn:trait
          gnc:phenotype->gn:phenotype
          gn:phenotype->metadata
          gn:trait->gn:phenotype))
   (outputs
    `(#:documentation ,documentation
      #:rdf ,output))))