From ce0c8e5dafa1e074cda0aa4562f9a7da37fb67de Mon Sep 17 00:00:00 2001
From: Bart Hanssens <bart.hanssens@bosa.fgov.be>
Date: Thu, 11 Jul 2024 16:12:28 +0100
Subject: [PATCH] GH-5058: additional parser code (WIP)

---
 .../eclipse/rdf4j/rio/csvw/CSVWParser.java    | 34 +++++++++++++++----
 .../rio/csvw/parsers/CellParserDate.java      |  3 +-
 core/rio/csvw/src/test/resources/painters.csv |  2 +-
 3 files changed, 30 insertions(+), 9 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
index 3eafd1e952..190c1249d6 100644
--- 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
@@ -234,7 +234,7 @@ private CellParser getCellParser(Model metadata, Resource column) {
 		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));
+		getFormat(metadata, column).ifPresent(v -> parser.setFormat(v));
 
 		Models.getPropertyString(metadata, column, CSVW.VALUE_URL).ifPresent(v -> parser.setValueURL(v));
 
@@ -272,6 +272,24 @@ private IRI getDatatypeIRI(Model metadata, Resource column) {
 		return XSD.valueOf(datatype.stringValue().toUpperCase()).getIri();
 	}
 
+	/**
+	 * Get format string
+	 *
+	 * @param metadata
+	 * @param column
+	 * @return
+	 */
+	private Optional<String> getFormat(Model metadata, Resource column) {
+		Optional<Value> val = Models.getProperty(metadata, column, CSVW.DATATYPE);
+		if (val.isPresent() && val.get().isBNode()) {
+			val = Models.getProperty(metadata, (Resource) val.get(), CSVW.FORMAT);
+			if (val.isPresent() && val.get().isLiteral()) {
+				return Optional.of(val.get().stringValue());
+			}
+		}
+		return Optional.empty();
+	}
+
 	/**
 	 * Get "about" URL template, to be used to create the subject of the triples
 	 *
@@ -359,18 +377,20 @@ private void parseCSV(Model metadata, RDFHandler handler, URI csvFile, CellParse
 	private CSVReader getCSVReader(Model metadata, Resource table, Reader reader) {
 		CSVParserBuilder parserBuilder = new CSVParserBuilder();
 		CSVReaderBuilder builder = new CSVReaderBuilder(reader);
+		builder.withSkipLines(1);
 
-		Optional<Value> dialect = Models.getProperty(metadata, table, CSVW.DIALECT);
-		if (dialect.isPresent()) {
-			Models.getPropertyString(metadata, (Resource) dialect.get(), CSVW.DELIMITER)
+		Optional<Value> val = Models.getProperty(metadata, table, CSVW.DIALECT);
+		if (val.isPresent()) {
+			Resource dialect = (Resource) val.get();
+			Models.getPropertyString(metadata, dialect, CSVW.DELIMITER)
 					.ifPresent(v -> parserBuilder.withSeparator(v.charAt(0)));
-			Models.getPropertyString(metadata, (Resource) dialect.get(), CSVW.HEADER)
+			Models.getPropertyString(metadata, dialect, CSVW.HEADER)
 					.ifPresent(v -> builder.withSkipLines(v.equalsIgnoreCase("false") ? 0 : 1));
-			Models.getPropertyString(metadata, (Resource) dialect.get(), CSVW.QUOTE_CHAR)
+			Models.getPropertyString(metadata, dialect, CSVW.QUOTE_CHAR)
 					.ifPresent(v -> parserBuilder.withQuoteChar(v.charAt(0)));
 		}
 
-		return new CSVReaderBuilder(reader).withCSVParser(parserBuilder.build()).build();
+		return builder.withCSVParser(parserBuilder.build()).build();
 	}
 
 	/**
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
index bafee06f24..fb8d491ef8 100644
--- 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
@@ -28,7 +28,7 @@ public class CellParserDate extends CellParser {
 	@Override
 	public void setFormat(String format) {
 		super.setFormat(format);
-		System.err.println(format);
+		System.err.println("format = " + format);
 		formatter = DateTimeFormatter.ofPattern(format);
 	}
 
@@ -39,6 +39,7 @@ public Value parse(String cell) {
 		if (formatter != null) {
 			s = DateTimeFormatter.ISO_DATE.format(formatter.parse(s));
 		}
+		System.err.println("date = " + s);
 		return Values.literal(s, dataType);
 	}
 
diff --git a/core/rio/csvw/src/test/resources/painters.csv b/core/rio/csvw/src/test/resources/painters.csv
index 4c3a87048c..096a2eb690 100644
--- a/core/rio/csvw/src/test/resources/painters.csv
+++ b/core/rio/csvw/src/test/resources/painters.csv
@@ -1,4 +1,4 @@
-"wikidata_id","first_name","last_name,country_id","country_name_nl","country_name_en","date_of_birth","married","languages"
+"wikidata_id","first_name","last_name","country_id","country_name_nl","country_name_en","date_of_birth","married","languages"
 "Q5582","Vincent","van Gogh","Q29999","Nederland","The Netherlands","30/3/1853","No","dutch french"
 "Q164712","Paul","Delvaux","Q31","België","Belgium","23/9/1897","Yes","french"
 "Q46408","Georgia","O'Keeffe","Q30","Verenigde Staten","United States","15/11/1887","Yes","english"