Merge pull request #34 from ontodev/existential-blank-nodes

Existential blank nodes in LDTab
ontodev · Nov 12, 2024 · 85ae634 · 85ae634
2 parents 8d4f94f + 87e5548
commit 85ae634
Show file tree

Hide file tree

Showing 2 changed files with 61 additions and 3 deletions.
diff --git a/src/ldtab/thick_rdf.clj b/src/ldtab/thick_rdf.clj
@@ -137,6 +137,7 @@
                 (translate-predicate-map (dissoc x "meta") prefix-2-base model)))))
     bnode))
 
+
 (defn parse-json
   [json]
   (let [success (try
@@ -145,6 +146,7 @@
         success (or (map? success)
                     (coll? success)
                     (seq? success))]
+
     (if success
       (cs/parse-string json)
       json)))
@@ -154,6 +156,35 @@
   (and (string? input)
        (str/starts-with? input "<wiring:blanknode")))
 
+(defn blanknode-triple-map
+  [blanknode-triples]
+  (cs/generate-string
+    (into {}
+          (map (fn [{:keys [predicate object datatype]}]
+                 [predicate [{"object" (parse-json object),
+                              "datatype" datatype}]])
+               blanknode-triples))))
+
+
+(defn merge-existential-blanknodes
+  "Merge thin triples belonging to the same existential blank nodes into a 'raw' LDTab triple."
+  [triples]
+  (let [blanknodes (filter #(is-wiring-blanknode (:subject %)) triples)
+        blanknode-2-triples (group-by :subject blanknodes)
+        complex-blanknodes (into {} (filter (fn [[k v]] (> (count v) 1)) blanknode-2-triples))
+        triples (remove #(contains? complex-blanknodes (:subject %)) triples)
+        raw-blank-nodes (map (fn [[k v]] {:assertion (:assertion (first v)),
+                                          :retraction (:retraction (first v)),
+                                          :graph (:graph (first v)),
+                                          :subject k,
+                                          :predicate "unknown",
+                                          :object (blanknode-triple-map v),
+                                          :datatype "_JSONMAP",
+                                          :annotation (:annotation (first v))})
+                             complex-blanknodes)
+        final (concat triples raw-blank-nodes)]
+    final))
+
 (defn thick-2-rdf-model ^Model
   [thick-triple prefixes]
   (let [;{:keys [assertion retraction graph s p o datatype annotation]} thick-triple 
@@ -178,7 +209,8 @@
 
 (defn triples-2-rdf-model-stream
   [thick-triples prefixes output]
-  (let [out-stream (io/output-stream output)
+  (let [thick-triples (merge-existential-blanknodes thick-triples)
+        out-stream (io/output-stream output)
         model (set-prefix-map (ModelFactory/createDefaultModel) prefixes)
         prefix-map (.lock model)
         writer-stream (StreamRDFWriter/getWriterStream out-stream RDFFormat/TURTLE_BLOCKS)]

diff --git a/src/ldtab/thin2thick.clj b/src/ldtab/thin2thick.clj
@@ -141,6 +141,30 @@
                            datatype))
      :else "ERROR")))
 
+
+(defn existential-blanknode-2-triples
+  [existential-blanknode]
+  ;(print "existblanknode: " existential-blanknode)
+  (let [blanknode (:subject existential-blanknode)
+        object (:object existential-blanknode)
+        datatype (:datatype existential-blanknode)
+        triples (if (= datatype "_JSONMAP")
+                  (map (fn [[k v]] {:subject blanknode,
+                                    :predicate k,
+                                    :object (get (first v) "object"),
+                                    :datatype (get (first v) "datatype")}) object)
+                  [existential-blanknode])]
+  ;(print "translated: " triples)
+    triples))
+
+(defn split-existential-blanknode-encoding
+  [triples]
+  (let [existential-blanknodes (filter (fn [x] (is-wiring-blanknode (:subject x))) triples)
+        triples (remove (fn [x] (is-wiring-blanknode (:subject x))) triples)
+        existential-blanknode-triples (mapcat existential-blanknode-2-triples existential-blanknodes)
+        triples (concat existential-blanknode-triples triples)]
+    triples))
+
 (defn encode-object
   "Given a triple t = [s p o] and a map from subject nodes to its triples,
   returns predicate map for the o"
@@ -286,7 +310,8 @@
          rdf-lists (map rdf-list/encode-rdf-list annotations)
          sorted (map sort-json rdf-lists)
          hashed (map hash-existential-subject-blanknode sorted)
-         normalised (map #(cs/parse-string (cs/generate-string %)) hashed)];TODO: stringify keys - this is a (probably an inefficient?) workaround 
+         split (split-existential-blanknode-encoding hashed)
+         normalised (map #(cs/parse-string (cs/generate-string %)) split)];TODO: stringify keys - this is a (probably an inefficient?) workaround 
      normalised))
   ([triples iri2prefix]
    (let [raw-thick-triples (thin-2-thick-raw triples iri2prefix)
@@ -301,5 +326,6 @@
          rdf-lists (map rdf-list/encode-rdf-list annotations)
          sorted (map sort-json rdf-lists)
          hashed (map hash-existential-subject-blanknode sorted)
-         normalised (map #(cs/parse-string (cs/generate-string %)) hashed)];TODO: stringify keys - this is a (probably an inefficient?) workaround 
+         split (split-existential-blanknode-encoding hashed)
+         normalised (map #(cs/parse-string (cs/generate-string %)) split)];TODO: stringify keys - this is a (probably an inefficient?) workaround 
      normalised)))