eclipse-rdf4jGH-5058: additional parser code (WIP)

Fedict · Jul 15, 2024 · 2b1d57f · 2b1d57f
1 parent 15e3321
commit 2b1d57f
Show file tree

Hide file tree

Showing 4 changed files with 71 additions and 11 deletions.
diff --git a/core/model-vocabulary/src/main/java/org/eclipse/rdf4j/model/vocabulary/CSVW.java b/core/model-vocabulary/src/main/java/org/eclipse/rdf4j/model/vocabulary/CSVW.java
@@ -122,6 +122,9 @@ public class CSVW {
 	/** csvw:skipRows */
 	public static final IRI SKIP_ROWS;
 
+	/** csvw:suppressOutput */
+	public static final IRI SUPPRESS_OUTPUT;
+
 	/** csvw:tableSchema */
 	public static final IRI TABLE_SCHEMA;
 
@@ -173,6 +176,7 @@ public class CSVW {
 		ROWNUM = Vocabularies.createIRI(NAMESPACE, "rownum");
 		SKIP_COLUMNS = Vocabularies.createIRI(NAMESPACE, "skipColumns");
 		SKIP_ROWS = Vocabularies.createIRI(NAMESPACE, "skipRows");
+		SUPPRESS_OUTPUT = Vocabularies.createIRI(NAMESPACE, "suppressOutput");
 		TABLE_SCHEMA = Vocabularies.createIRI(NAMESPACE, "tableSchema");
 		TABLES = Vocabularies.createIRI(NAMESPACE, "tables");
 		TITLE = Vocabularies.createIRI(NAMESPACE, "title");

diff --git a/core/rio/csvw/src/main/java/org/eclipse/rdf4j/rio/csvw/CSVWParser.java b/core/rio/csvw/src/main/java/org/eclipse/rdf4j/rio/csvw/CSVWParser.java
@@ -234,6 +234,8 @@ private CellParser getCellParser(Model metadata, Resource column) {
 				.ifPresent(v -> parser.setRequired(Boolean.parseBoolean(v)));
 		Models.getPropertyString(metadata, column, CSVW.VIRTUAL)
 				.ifPresent(v -> parser.setVirtual(Boolean.parseBoolean(v)));
+		Models.getPropertyString(metadata, column, CSVW.SUPPRESS_OUTPUT)
+				.ifPresent(v -> parser.setVirtual(Boolean.parseBoolean(v)));
 
 		// only useful for strings
 		Models.getPropertyString(metadata, column, CSVW.LANG).ifPresent(v -> parser.setLang(v));
@@ -417,7 +419,9 @@ private void parseCSV(Model metadata, RDFHandler handler, URI csvFile, CellParse
 
 					IRI predicate = cellParsers[i].getPropertyIRI();
 					val = cellParsers[i].parse(cells[i]);
-					handler.handleStatement(Statements.statement(subject, predicate, val, null));
+					if (!cellParsers[i].isSuppressed()) {
+						handler.handleStatement(Statements.statement(subject, predicate, val, null));
+					}
 				}
 				line++;
 			}

diff --git a/core/rio/csvw/src/main/java/org/eclipse/rdf4j/rio/csvw/parsers/CellParser.java b/core/rio/csvw/src/main/java/org/eclipse/rdf4j/rio/csvw/parsers/CellParser.java
@@ -34,6 +34,7 @@ public abstract class CellParser {
 	private String defaultValue;
 	private String nullValue;
 	private boolean required;
+	private String aboutUrl;
 	private IRI propertyIRI;
 	private String valueUrl;
 	private String format;
@@ -42,8 +43,11 @@ public abstract class CellParser {
 	private String separator;
 	private boolean trim = true;
 	private boolean virtual = false;
-	private String[] propPlaceholder;
-	private String[] valPlaceholder;
+	private boolean suppressed = false;
+
+	private String[] aboutPlaceholders;
+	private String[] propPlaceholders;
+	private String[] valPlaceholders;
 
 	public String getName() {
 		return name;
@@ -131,10 +135,6 @@ public void setVirtual(boolean virtual) {
 		this.virtual = virtual;
 	}
 
-	public IRI getPropertyIRI() {
-		return propertyIRI;
-	}
-
 	/**
 	 * Extract placeholders (if any)
 	 *
@@ -154,6 +154,34 @@ private String[] extractPlaceholders(String template) {
 		return null;
 	}
 
+	/**
+	 * Get aboutURL
+	 *
+	 * @return
+	 */
+	public String getAboutUrl() {
+		return aboutUrl;
+	}
+
+	/**
+	 * Set aboutUrl
+	 *
+	 * @param aboutUrl
+	 */
+	public void setAboutUrl(String aboutUrl) {
+		this.aboutUrl = aboutUrl;
+		this.aboutPlaceholders = extractPlaceholders(aboutUrl);
+	}
+
+	/**
+	 * Get propertyUrl as IRI
+	 *
+	 * @return
+	 */
+	public IRI getPropertyIRI() {
+		return propertyIRI;
+	}
+
 	/**
 	 * Set property URL (predicate IRI)
 	 *
@@ -189,7 +217,7 @@ public String getValueUrl() {
 	 */
 	public void setValueUrl(String valueUrl) {
 		this.valueUrl = valueUrl;
-		this.valPlaceholder = extractPlaceholders(valueUrl);
+		this.valPlaceholders = extractPlaceholders(valueUrl);
 	}
 
 	/**
@@ -282,6 +310,24 @@ public void setTrim(boolean trim) {
 		this.trim = trim;
 	}
 
+	/**
+	 * Set if output needs to be suppressed
+	 *
+	 * @return
+	 */
+	public boolean isSuppressed() {
+		return suppressed;
+	}
+
+	/**
+	 * Set if output needs to be suppressed
+	 *
+	 * @param suppressed
+	 */
+	public void setSuppressed(boolean suppressed) {
+		this.suppressed = suppressed;
+	}
+
 	/**
 	 * Get the (possibly trimmed) value or default value
 	 *

diff --git a/core/rio/csvw/src/test/resources/painters-metadata.json b/core/rio/csvw/src/test/resources/painters-metadata.json
@@ -11,8 +11,7 @@
       { "name": "last_name",
         "propertyUrl": "schema:familyName"},
       { "name": "country_id",
-        "propertyUrl": "schema:nationality",
-        "valueUrl": "https://www.wikidata.org/wiki/{country_id}"},
+        "suppressOutput": "true" },
       { "name": "country_name_nl",
         "lang": "nl" },
       { "name": "country_name_en",
@@ -32,7 +31,14 @@
         "separator": " " },
       { "virtual": true,
         "propertyUrl": "rdf:type",
-        "valueUrl": "schema:Person" }
+        "valueUrl": "schema:Person" },
+      { "virtual": true,
+        "propertyUrl": "schema:nationality",
+        "valueUrl": "https://www.wikidata.org/wiki/{country_id}" },
+      { "virtual": true,
+        "aboutUrl": "https://www.wikidata.org/wiki/{country_id}",
+        "propertyUrl": "rdf:type",
+        "valueUrl": "schema:Country" }
     ],
    "primaryKey": "wikidata_id"
   }