Add replaced by information

aodn · Oct 15, 2024 · 30b009d · 30b009d
1 parent b98b96d
commit 30b009d
Show file tree

Hide file tree

Showing 4 changed files with 288 additions and 120 deletions.
diff --git a/ardcvocabs/src/main/java/au/org/aodn/ardcvocabs/model/VocabModel.java b/ardcvocabs/src/main/java/au/org/aodn/ardcvocabs/model/VocabModel.java
@@ -22,6 +22,8 @@ public class VocabModel {
     protected List<String> altLabels;
     @JsonProperty("is_latest_label")
     protected Boolean isLatestLabel;
+    @JsonProperty("replaced_by")
+    protected String replacedBy;
     protected String definition;
     protected String about;
     protected List<VocabModel> broader;

diff --git a/ardcvocabs/src/main/java/au/org/aodn/ardcvocabs/service/ArdcVocabServiceImpl.java b/ardcvocabs/src/main/java/au/org/aodn/ardcvocabs/service/ArdcVocabServiceImpl.java
@@ -17,6 +17,8 @@
 import java.util.function.Function;
 import java.util.stream.StreamSupport;
 import java.util.stream.Collectors;
+import java.util.regex.Matcher;
+import java.util.regex.Pattern;
 
 public class ArdcVocabServiceImpl implements ArdcVocabService {
 
@@ -28,7 +30,7 @@ public class ArdcVocabServiceImpl implements ArdcVocabService {
     protected RestTemplate restTemplate;
     protected RetryTemplate retryTemplate;
 
-    protected Function<JsonNode, String> extractValueField(String key) {
+    protected Function<JsonNode, String> extractSingleText(String key) {
         return (node) -> {
             JsonNode labelNode = node.get(key);
             if (labelNode != null) {
@@ -43,7 +45,7 @@ protected Function<JsonNode, String> extractValueField(String key) {
             return null;
         };
     }
-    protected Function<JsonNode, List<String>> extractMultipleValueFields(String key) {
+    protected Function<JsonNode, List<String>> extractMultipleTexts(String key) {
         return (node) -> {
             JsonNode labelNode = node.get(key);
             if (labelNode != null && labelNode.isArray()) {
@@ -57,13 +59,31 @@ protected Function<JsonNode, List<String>> extractMultipleValueFields(String key
     }
 
     // Reusing the utility methods for specific labels
-    protected Function<JsonNode, String> label = extractValueField("prefLabel");
-    protected Function<JsonNode, String> displayLabel = extractValueField("displayLabel");
-    protected Function<JsonNode, List<String>> hiddenLabels = extractMultipleValueFields("hiddenLabel");
-    protected Function<JsonNode, List<String>> altLabels = extractMultipleValueFields("altLabel");
-    protected Function<JsonNode, String> about = extractValueField("_about");
-    protected Function<JsonNode, String> definition = extractValueField("definition");
-    protected Function<JsonNode, Boolean> isLatestLabel = (node) -> !(node.has("isReplacedBy") || (node.has("scopeNote") && extractValueField("scopeNote").apply(node).contains("no longer exists")));
+    protected Function<JsonNode, String> label = extractSingleText("prefLabel");
+    protected Function<JsonNode, String> displayLabel = extractSingleText("displayLabel");
+    protected Function<JsonNode, List<String>> hiddenLabels = extractMultipleTexts("hiddenLabel");
+    protected Function<JsonNode, List<String>> altLabels = extractMultipleTexts("altLabel");
+    protected Function<JsonNode, String> about = extractSingleText("_about");
+    protected Function<JsonNode, String> definition = extractSingleText("definition");
+    protected Function<JsonNode, Boolean> isLatestLabel = (node) -> !(node.has("isReplacedBy") || (node.has("scopeNote") && extractSingleText("scopeNote").apply(node).toLowerCase().contains("no longer exists")));
+    protected Function<JsonNode, Boolean> isReplacedBy = (node) -> node.has("isReplacedBy") && node.has("scopeNote") && extractSingleText("scopeNote").apply(node).toLowerCase().contains("replaced by");
+
+    private String extractReplacedVocabUri(String scopeNote) {
+        String regex = "Replaced by (https?://[\\w./-]+)";
+        Pattern pattern = Pattern.compile(regex);
+        Matcher matcher = pattern.matcher(scopeNote);
+
+        if (matcher.find()) {
+            String result = matcher.group(1);
+            if (result.endsWith(".")) {
+                result = result.substring(0, result.length() - 1);
+            }
+            return result;
+        }
+
+        return null;
+    }
+
     protected BiFunction<JsonNode, String, Boolean> isNodeValid = (node, item) -> node != null && !node.isEmpty() && node.has(item) && !node.get(item).isEmpty();
 
     public ArdcVocabServiceImpl(RestTemplate restTemplate, RetryTemplate retryTemplate) {
@@ -95,6 +115,10 @@ protected VocabModel buildVocabByResourceUri(String vocabUri, String vocabApiBas
                         .isLatestLabel(isLatestLabel.apply(target))
                         .build();
 
+                if (!vocab.getIsLatestLabel() && isReplacedBy.apply(target)) {
+                    vocab.setReplacedBy(extractReplacedVocabUri(extractSingleText("scopeNote").apply(target)));
+                }
+
                 List<VocabModel> narrowerNodes = new ArrayList<>();
                 if (isNodeValid.apply(target, "narrower")) {
                     for (JsonNode j : target.get("narrower")) {
@@ -175,6 +199,10 @@ protected Map<String, List<VocabModel>> getVocabLeafNodes(String vocabApiBase, V
                                             .isLatestLabel(isLatestLabel.apply(target))
                                             .build();
 
+                                    if (!vocab.getIsLatestLabel() && isReplacedBy.apply(target)) {
+                                        vocab.setReplacedBy(extractReplacedVocabUri(extractSingleText("scopeNote").apply(target)));
+                                    }
+
                                     List<VocabModel> vocabNarrower = new ArrayList<>();
                                     if(target.has("narrower") && !target.get("narrower").isEmpty()) {
                                         for(JsonNode currentNode : target.get("narrower")) {