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

Fedict · Jul 11, 2024 · f335478 · f335478
1 parent 43aa09b
commit f335478
Show file tree

Hide file tree

Showing 4 changed files with 19 additions and 33 deletions.
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
@@ -55,7 +55,9 @@
 import com.opencsv.exceptions.CsvValidationException;
 
 /**
- * Basic (experimental) CSV on the Web Parser
+ * Basic (experimental) CSV on the Web parser.
+ *
+ * Currently only "minimal mode" is supported
  *
  * @author Bart Hanssens
  * @see <a href="https://w3c.github.io/csvw/primer/">CSV on the Web Primer</a>
@@ -107,9 +109,8 @@ public synchronized void parse(InputStream in, String baseURI)
 	@Override
 	public void parse(Reader reader, String baseURI)
 			throws IOException, RDFParseException, RDFHandlerException {
-		Model metadata = parseMetadata(null, reader, baseURI);
 
-		clear();
+		Model metadata = parseMetadata(null, reader, baseURI);
 	}
 
 	/**
@@ -217,16 +218,24 @@ private CellParser getCellParser(Model metadata, Resource column) {
 
 		CellParser parser = CellParserFactory.create(datatype);
 
-		Models.getPropertyString(metadata, column, CSVW.LANG).ifPresent(v -> parser.setLang(v));
-		getFormat(metadata, column).ifPresent(v -> parser.setFormat(v.stringValue()));
-
 		Models.getPropertyString(metadata, column, CSVW.NAME)
 				.ifPresentOrElse(v -> parser.setName(v),
 						() -> new RDFParseException("Metadata file does not contain name for column " + column));
 
 		Models.getPropertyString(metadata, column, CSVW.DEFAULT).ifPresent(v -> parser.setDefaultValue(v));
 		Models.getPropertyString(metadata, column, CSVW.REQUIRED)
 				.ifPresent(v -> parser.setIsRequired(Boolean.parseBoolean(v)));
+
+		// only useful for strings
+		Models.getPropertyString(metadata, column, CSVW.LANG).ifPresent(v -> parser.setLang(v));
+
+		// only useful for numeric
+		Models.getPropertyString(metadata, column, CSVW.DECIMAL_CHAR).ifPresent(v -> parser.setDecimalChar(v));
+		Models.getPropertyString(metadata, column, CSVW.GROUP_CHAR).ifPresent(v -> parser.setGroupChar(v));
+
+		// mostly for date formats
+		Models.getPropertyString(metadata, column, CSVW.FORMAT).ifPresent(v -> parser.setFormat(v));
+
 		Models.getPropertyString(metadata, column, CSVW.VALUE_URL).ifPresent(v -> parser.setValueURL(v));
 
 		// use a property from a vocabulary as predicate, or create a property relative to the namespace of the CSV
@@ -238,7 +247,7 @@ private CellParser getCellParser(Model metadata, Resource column) {
 	}
 
 	/**
-	 * Get IRI of base or derived datatype
+	 * Get name of base or derived datatype
 	 *
 	 * @param metadata
 	 * @param column
@@ -263,26 +272,6 @@ private IRI getDatatypeIRI(Model metadata, Resource column) {
 		return XSD.valueOf(datatype.stringValue().toUpperCase()).getIri();
 	}
 
-	/**
-	 * Get name of the generic datatype or more specific datatype
-	 *
-	 * @param metadata
-	 * @param column
-	 * @return
-	 */
-	private Optional<Value> getFormat(Model metadata, Resource column) {
-		Optional<Value> val = Models.getProperty(metadata, column, CSVW.DATATYPE);
-		if (val.isPresent()) {
-			Value datatype = val.get();
-			// derived datatype
-			if (datatype.isBNode()) {
-				Optional<Value> fmt = Models.getProperty(metadata, (Resource) datatype, CSVW.FORMAT);
-				val = Models.getProperty(metadata, (Resource) fmt.get(), CSVW.BASE);
-			}
-		}
-		return val;
-	}
-
 	/**
 	 * Get "about" URL template, to be used to create the subject of the triples
 	 *
@@ -376,7 +365,7 @@ private CSVReader getCSVReader(Model metadata, Resource table, Reader reader) {
 			Models.getPropertyString(metadata, (Resource) dialect.get(), CSVW.DELIMITER)
 					.ifPresent(v -> parserBuilder.withSeparator(v.charAt(0)));
 			Models.getPropertyString(metadata, (Resource) dialect.get(), CSVW.HEADER)
-					.ifPresent(v -> builder.withSkipLines(v.equalsIgnoreCase("false") ? 1 : 0));
+					.ifPresent(v -> builder.withSkipLines(v.equalsIgnoreCase("false") ? 0 : 1));
 			Models.getPropertyString(metadata, (Resource) dialect.get(), CSVW.QUOTE_CHAR)
 					.ifPresent(v -> parserBuilder.withQuoteChar(v.charAt(0)));
 		}

diff --git a/core/rio/csvw/src/main/java/org/eclipse/rdf4j/rio/csvw/parsers/CellParserDate.java b/core/rio/csvw/src/main/java/org/eclipse/rdf4j/rio/csvw/parsers/CellParserDate.java
@@ -28,6 +28,7 @@ public class CellParserDate extends CellParser {
 	@Override
 	public void setFormat(String format) {
 		super.setFormat(format);
+		System.err.println(format);
 		formatter = DateTimeFormatter.ofPattern(format);
 	}
 

diff --git a/core/rio/csvw/src/main/java/org/eclipse/rdf4j/rio/csvw/parsers/CellParserString.java b/core/rio/csvw/src/main/java/org/eclipse/rdf4j/rio/csvw/parsers/CellParserString.java
@@ -10,10 +10,6 @@
  *******************************************************************************/
 package org.eclipse.rdf4j.rio.csvw.parsers;
 
-import java.util.Set;
-
-import org.eclipse.rdf4j.model.IRI;
-import org.eclipse.rdf4j.model.Namespace;
 import org.eclipse.rdf4j.model.Value;
 import org.eclipse.rdf4j.model.util.Values;
 

diff --git a/core/rio/csvw/src/test/resources/painters-metadata.json b/core/rio/csvw/src/test/resources/painters-metadata.json
@@ -21,7 +21,7 @@
            "base": "date", 
            "format": "d/M/yyyy"
         } },
-      { "name": "maried",
+      { "name": "married",
         "datatype": {
             "base": "boolean",
             "format": "Yes|No"