From 397745b554e03fa2df0784c0c48ac43d01428980 Mon Sep 17 00:00:00 2001
From: Munyoki Kilyungi
Date: Tue, 27 Aug 2024 15:23:16 +0300
Subject: Replace blank-nodes for gn-genewiki-entries with unique identifiers.

Signed-off-by: Munyoki Kilyungi <me@bonfacemunyoki.com>
---
 examples/generif.scm | 116 ++++++++++++++++++++++-----------------------------
 1 file changed, 49 insertions(+), 67 deletions(-)

diff --git a/examples/generif.scm b/examples/generif.scm
index 11235e0..894b766 100755
--- a/examples/generif.scm
+++ b/examples/generif.scm
@@ -49,7 +49,9 @@
            (left-join Species "ON Species.SpeciesId = GeneRIF.SpeciesId")
            (left-join GeneRIFXRef "ON GeneRIFXRef.GeneRIFId = GeneRIF.Id")
            (left-join GeneCategory "ON GeneRIFXRef.GeneCategoryId = GeneCategory.Id"))
-          "WHERE GeneRIF.display > 0 AND GeneRIF.comment IS NOT NULL GROUP BY GeneRIF.Id, GeneRIF.versionId, GeneRIF.symbol, GeneRIF.SpeciesId, GeneRIF.createtime, GeneRIF.reason")
+          "WHERE GeneRIF.display > 0 AND GeneRIF.comment IS NOT NULL
+GROUP BY GeneRIF.Id, GeneRIF.versionId, GeneRIF.symbol, GeneRIF.SpeciesId,
+GeneRIF.createtime, GeneRIF.reason")
   (schema-triples
    (gnc:GeneWikiEntry a rdfs:Class)
    (gnc:GNWikiEntry rdfs:subClassOf gnc:GeneWikiEntry)
@@ -61,73 +63,53 @@
    (gnt:reason skos:definition "The reason why this resource was modified")
    (gnc:GNWikiEntry rdfs:comment "Represents GeneRIF Entries entered from GeneNetwork")
    (gnt:geneSymbol rdfs:domain gnc:GNWikiEntry))
+  ;; Here we use the Id and VersionId to uniquely identify comments.
+  ;; We could use blank-nodes here; however, querying blank nodes
+  ;; E.g. getting the latest versionId is very complicated.  Prefer
+  ;; normal triplets over blank-nodes.
   (triples
-      (string->identifier
-       "symbol"
-       (regexp-substitute/global
-        #f "[^A-Za-z0-9:]"
-        (field GeneRIF symbol)
-        'pre "_" 'post)
-       #:proc (lambda (x) x))
-    (set rdfs:comment
-         (let* ((generif-comment (sanitize-rdf-string (field GeneRIF comment)))
-                (create-time (field GeneRIF createtime EntryCreateTime))
-                (pmid (field GeneRIF PubMed_ID PMID))
-                (web-url (field GeneRIF weburl))
-                (species (string->identifier
-                          ""
-                          (remap-species-identifiers (field Species Fullname))
-                          #:separator ""
-                          #:proc string-capitalize-first))
-                (version-id (field GeneRIF versionId))
-                (identifier (field GeneRIF Id))
-                (initial (sanitize-rdf-string (field GeneRIF initial)))
-                (reason (field GeneRIF reason))
-                (email (sanitize-rdf-string (field GeneRIF email)))
-                (category
-                 (field ("GROUP_CONCAT(DISTINCT GeneCategory.Name SEPARATOR '; ')"
-                         GeneCategory))))
-           (string->symbol
-            (string-append
-             "[ "
-             (format #f "rdf:type gnc:GNWikiEntry ; ")
-             (if (string? species)
-                 ""
-                 (format #f "gnt:belongsToSpecies ~a ; "
-                         species))
-             (format #f "rdfs:comment ~s^^xsd:string ; "
-                     generif-comment)
-             (if (string? create-time)
-                 ""
-                 (format #f "dct:created ~s^^xsd:datetime ; "
-                         (time-unix->string
-                          create-time "~5")))
-             (if (and (string? pmid) (not (string-null? pmid)))
-                 (format #f
-                         "~{dct:references pubmed:~a ; ~}"
-                         (string-split pmid #\space))
-                 "")
-             (if (string-blank? email)
-                 ""
-                 (format #f "foaf:mbox ~s ; " email))
-             (format #f "dct:identifier ~s ; " identifier)
-             (format #f "dct:hasVersion \"~s\"^^xsd:int ; " version-id)
-             (if (string-blank? reason)
-                 ""
-                 (format #f "gnt:reason ~s ; " reason))
-             (if (or (null? initial)
-                      (string-blank? initial))
-                 "" (format #f "gnt:initial ~s ; " initial))
-             (if (string-blank? category)
-                 ""
-                 (format #f
-                         "gnt:belongsToCategory ~s ; "
-                         category))
-             (if (and (string? web-url) (not (string-null? web-url)))
-                 (format #f "foaf:homepage ~s ; "
-                         web-url)
-                 "")
-             " ] "))))))
+      (format #f "gn:wiki-~a-~a"
+              (field GeneRIF Id)
+              (field GeneRIF versionId))
+    (set rdfs:comment (sanitize-rdf-string (field GeneRIF comment)))
+    (set rdf:type 'gnc:GNWikiEntry)
+    (set gnt:symbol
+         (string->identifier
+          "symbol"
+          (regexp-substitute/global #f "[^A-Za-z0-9:]"
+                                    (field GeneRIF symbol)
+                                    'pre "_" 'post)
+          #:proc (lambda (x) x)))
+    (set dct:created
+         (let ((create-time (field GeneRIF createtime EntryCreateTime)))
+           (if (string? create-time)
+               ""
+               (annotate-field
+                (time-unix->string
+                 create-time
+                 "~5")
+                '^^xsd:datetime))))
+    (multiset dct:references
+              (string-split (field GeneRIF PubMed_ID PMID)
+                            #\space))
+    (set foaf:homepage (field GeneRIF weburl))
+    (set gnt:belongsToSpecies (string->identifier
+                               ""
+                               (remap-species-identifiers (field Species Fullname))
+                               #:separator ""
+                               #:proc string-capitalize-first))
+    (set dct:hasVersion (annotate-field (format #f "~s" (field GeneRIF versionId))
+                                        '^^xsd:int))
+    (set dct:identifier (annotate-field (format #f "~s" (field GeneRIF Id))
+                                        '^^xsd:int))
+    (set gnt:initial (sanitize-rdf-string (field GeneRIF initial)))
+    (set gnt:reason (field GeneRIF reason))
+    (set foaf:mbox (sanitize-rdf-string (field GeneRIF email)))
+    (multiset gnt:belongsToCategory
+              (string-split
+               (field ("GROUP_CONCAT(DISTINCT GeneCategory.Name SEPARATOR ';')"
+                       GeneCategory))
+               #\;))))
 
 (define-transformer ncbi-genewiki-entries
   (tables (GeneRIF_BASIC
-- 
cgit v1.2.3