fixed import of tags

gbif · Jun 10, 2024 · 1aab822 · 1aab822
1 parent 1e917d0
commit 1aab822
Show file tree

Hide file tree

Showing 3 changed files with 8 additions and 8 deletions.
diff --git a/vocabulary-importer/README.md b/vocabulary-importer/README.md
@@ -94,12 +94,10 @@ This tool can be run via command line by sending these params:
 * *--csvDelimiter, -d  (Optional)*
 
     Delimiter of the CSV files for concepts and hidden labels. If not specified it uses a comma (",").
-    Note that some delimiters may need to be escaped for Java, e.g.: \\\\|
 
 * *--listDelimiter, -ld (Optional)*
 
-    Delimiter to specify multiple values in the alternative labels. If not specified it uses a pipe ("|").
-    Note that some delimiters may need to be escaped for Java, e.g.: \\\\|
+    Delimiter to specify multiple values in the alternative labels. If not specified it uses a pipe ("|")
 
 * *--apiUrl, -a*
 

diff --git a/vocabulary-importer/src/main/java/org/gbif/vocabulary/importer/Fields.java b/vocabulary-importer/src/main/java/org/gbif/vocabulary/importer/Fields.java
@@ -28,5 +28,5 @@ public class Fields {
   static final String TAGS = "TAGS";
 
   static final List<String> CONCEPT_FIELDS =
-      Arrays.asList(CONCEPT, PARENT, SAME_AS_URIS, EXTERNAL_DEFINITIONS);
+      Arrays.asList(CONCEPT, PARENT, SAME_AS_URIS, EXTERNAL_DEFINITIONS, TAGS);
 }
diff --git a/vocabulary-importer/src/main/java/org/gbif/vocabulary/importer/VocabularyImporter.java b/vocabulary-importer/src/main/java/org/gbif/vocabulary/importer/VocabularyImporter.java
@@ -46,6 +46,7 @@
 import java.util.Optional;
 import java.util.Set;
 import java.util.function.Function;
+import java.util.regex.Pattern;
 import java.util.stream.Collectors;
 import java.util.stream.Stream;
 
@@ -566,7 +567,8 @@ private ConceptData parseConceptFields(
     if (headersIndexes.containsKey(SAME_AS_URIS)
         && !Strings.isNullOrEmpty(values[headersIndexes.get(SAME_AS_URIS)])) {
       Set<URI> sameAsUris = new HashSet<>();
-      String[] urisValues = values[headersIndexes.get(SAME_AS_URIS)].split(listDelimiter);
+      String[] urisValues =
+          values[headersIndexes.get(SAME_AS_URIS)].split(Pattern.quote(listDelimiter));
       for (String uri : urisValues) {
         try {
           sameAsUris.add(URI.create(uri.trim()));
@@ -582,7 +584,7 @@ private ConceptData parseConceptFields(
     if (headersIndexes.containsKey(EXTERNAL_DEFINITIONS)
         && !Strings.isNullOrEmpty(values[headersIndexes.get(SAME_AS_URIS)])) {
       Set<URI> externalDefinitions = new HashSet<>();
-      String[] externalDefsValues = values[headersIndexes.get(SAME_AS_URIS)].split(listDelimiter);
+      String[] externalDefsValues = values[headersIndexes.get(SAME_AS_URIS)].split(Pattern.quote(listDelimiter));
       for (String definition : externalDefsValues) {
         try {
           externalDefinitions.add(URI.create(definition.trim()));
@@ -597,7 +599,7 @@ private ConceptData parseConceptFields(
     // tags
     if (headersIndexes.containsKey(TAGS)
         && !Strings.isNullOrEmpty(values[headersIndexes.get(TAGS)])) {
-      String[] tagsValues = values[headersIndexes.get(TAGS)].split(listDelimiter);
+      String[] tagsValues = values[headersIndexes.get(TAGS)].split(Pattern.quote(listDelimiter));
       conceptData.tags =
           Arrays.stream(tagsValues)
               .filter(s -> !Strings.isNullOrEmpty(s))
@@ -628,7 +630,7 @@ private static List<Label> parseAlternativeLabels(
     List<Label> alternativeLabels = new ArrayList<>();
     alternativeLabelsIndexes.forEach(
         (lang, index) ->
-            Stream.of(values[index].split(listDelimiter))
+            Stream.of(values[index].split(Pattern.quote(listDelimiter)))
                 .filter(s -> !s.isEmpty())
                 .map(String::trim)
                 .forEach(