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

Fedict · Jul 12, 2024 · 6c44c26 · 6c44c26
1 parent b7c58ec
commit 6c44c26
Show file tree

Hide file tree

Showing 8 changed files with 112 additions and 114 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
@@ -84,6 +84,9 @@ public class CSVW {
 	/** csvw:name */
 	public static final IRI NAME;
 
+	/** csvw:null */
+	public static final IRI NULL;
+
 	/** csvw:propertyUrl */
 	public static final IRI PROPERTY_URL;
 
@@ -95,7 +98,7 @@ public class CSVW {
 
 	/** csvw:skipRows */
 	public static final IRI SKIP_ROWS;
-	
+
 	/** csvw:tableSchema */
 	public static final IRI TABLE_SCHEMA;
 
@@ -105,6 +108,9 @@ public class CSVW {
 	/** csvw:title */
 	public static final IRI TITLE;
 
+	/** csvw:trim */
+	public static final IRI TRIM;
+
 	/** csvw:url */
 	public static final IRI URL;
 
@@ -130,13 +136,15 @@ public class CSVW {
 		HEADER_ROW_COUNT = Vocabularies.createIRI(NAMESPACE, "headerRowCount");
 		LANG = Vocabularies.createIRI(NAMESPACE, "lang");
 		NAME = Vocabularies.createIRI(NAMESPACE, "name");
+		NULL = Vocabularies.createIRI(NAMESPACE, "null");
 		PROPERTY_URL = Vocabularies.createIRI(NAMESPACE, "propertyUrl");
 		QUOTE_CHAR = Vocabularies.createIRI(NAMESPACE, "quoteChar");
 		REQUIRED = Vocabularies.createIRI(NAMESPACE, "required");
 		SKIP_ROWS = Vocabularies.createIRI(NAMESPACE, "skipRows");
 		TABLE_SCHEMA = Vocabularies.createIRI(NAMESPACE, "tableSchema");
 		TABLES = Vocabularies.createIRI(NAMESPACE, "tables");
 		TITLE = Vocabularies.createIRI(NAMESPACE, "title");
+		TRIM = Vocabularies.createIRI(NAMESPACE, "trim");
 		URL = Vocabularies.createIRI(NAMESPACE, "url");
 		VALUE_URL = Vocabularies.createIRI(NAMESPACE, "valueUrl");
 		VIRTUAL = Vocabularies.createIRI(NAMESPACE, "virtual");

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
@@ -225,7 +225,7 @@ private CellParser getCellParser(Model metadata, Resource 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)));
+				.ifPresent(v -> parser.setRequired(Boolean.parseBoolean(v)));
 		Models.getPropertyString(metadata, column, CSVW.VIRTUAL)
 				.ifPresent(v -> parser.setVirtual(Boolean.parseBoolean(v)));
 
@@ -239,12 +239,15 @@ private CellParser getCellParser(Model metadata, Resource column) {
 		// mostly for date formats
 		getFormat(metadata, column).ifPresent(v -> parser.setFormat(v));
 
-		Models.getPropertyString(metadata, column, CSVW.VALUE_URL).ifPresent(v -> parser.setValueURL(v));
+		Models.getPropertyString(metadata, column, CSVW.TRIM)
+				.ifPresent(v -> parser.setVirtual(Boolean.parseBoolean(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
 		Optional<String> propertyURL = Models.getPropertyString(metadata, column, CSVW.PROPERTY_URL);
 		String s = propertyURL.isPresent() ? propertyURL.get() : "_local:" + parser.getName();
-		parser.setPropertyURL(metadata.getNamespaces(), s);
+		parser.setPropertyIRI(metadata.getNamespaces(), s);
 
 		return parser;
 	}

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
@@ -22,66 +22,77 @@
  * @author Bart Hanssens
  */
 public abstract class CellParser {
-	protected String name;
-	protected IRI dataType;
-	protected String lang;
-	protected String defaultValue;
-	protected boolean isRequired;
-	protected boolean isVirtual;
-	protected IRI propertyIRI;
-	protected String valueUrl;
-	protected String format;
-	protected String decimalChar;
-	protected String groupChar;
-	protected String separator;
+	private String name;
+	private IRI dataType;
+	private String lang;
+	private String defaultValue;
+	private String nullValue;
+	private boolean required;
+	private boolean virtual = false;
+	private IRI propertyIRI;
+	private String valueUrl;
+	private String format;
+	private String decimalChar = ".";
+	private String groupChar;
+	private String separator;
+	private boolean trim = true;
+
+	public String getName() {
+		return name;
+	}
 
-	/**
-	 * @param name
-	 */
 	public void setName(String name) {
 		this.name = name;
 	}
 
-	/**
-	 * @return name
-	 */
-	public String getName() {
-		return name;
+	public IRI getDataType() {
+		return dataType;
 	}
 
-	/**
-	 * @param dataType
-	 */
 	public void setDataType(IRI dataType) {
 		this.dataType = dataType;
 	}
 
-	/**
-	 * Set language code
-	 *
-	 * @param lang language code
-	 */
+	public String getLang() {
+		return lang;
+	}
+
 	public void setLang(String lang) {
 		this.lang = lang;
 	}
 
-	/**
-	 * @param defaultValue the defaultValue to set
-	 */
+	public String getDefaultValue() {
+		return defaultValue;
+	}
+
 	public void setDefaultValue(String defaultValue) {
 		this.defaultValue = defaultValue;
 	}
 
-	/**
-	 * @param isRequired the isRequired to set
-	 */
-	public void setIsRequired(boolean isRequired) {
-		this.isRequired = isRequired;
+	public String getNullValue() {
+		return nullValue;
+	}
+
+	public void setNullValue(String nullValue) {
+		this.nullValue = nullValue;
+	}
+
+	public boolean isRequired() {
+		return required;
+	}
+
+	public void setRequired(boolean isRequired) {
+		this.required = isRequired;
+	}
+
+	public boolean isVirtual() {
+		return virtual;
+	}
+
+	public void setVirtual(boolean isVirtual) {
+		this.virtual = isVirtual;
 	}
 
-	/**
-	 * @return the propertyUrl as IRI
-	 */
 	public IRI getPropertyIRI() {
 		return propertyIRI;
 	}
@@ -92,7 +103,7 @@ public IRI getPropertyIRI() {
 	 * @param namespaces  set of namespaces
 	 * @param propertyUrl the propertyUrl to set
 	 */
-	public void setPropertyURL(Set<Namespace> namespaces, String propertyUrl) {
+	public void setPropertyIRI(Set<Namespace> namespaces, String propertyUrl) {
 		this.propertyIRI = Values.iri(namespaces, propertyUrl);
 	}
 
@@ -101,92 +112,76 @@ public void setPropertyURL(Set<Namespace> namespaces, String propertyUrl) {
 	 *
 	 * @param propertyUrl the propertyUrl to set
 	 */
-	public void setPropertyURL(String propertyUrl) {
+	public void setPropertyIRI(String propertyUrl) {
 		this.propertyIRI = Values.iri("", propertyUrl);
 	}
 
-	/**
-	 * @return the valueUrl
-	 */
-	public String getValueURL() {
+	public String getValueUrl() {
 		return valueUrl;
 	}
 
-	/**
-	 * @param valueUrl the valueUrl to set
-	 */
-	public void setValueURL(String valueUrl) {
+	public void setValueUrl(String valueUrl) {
 		this.valueUrl = valueUrl;
 	}
 
-	/**
-	 * @return the separator
-	 */
-	public String getSeparator() {
-		return separator;
+	public String getFormat() {
+		return format;
 	}
 
-	/**
-	 * @param separator the separator to set
-	 */
-	public void setSeparator(String separator) {
-		this.separator = separator;
+	public void setFormat(String format) {
+		this.format = format;
 	}
 
-	/**
-	 * @return the decimal character
-	 */
 	public String getDecimalChar() {
 		return decimalChar;
 	}
 
-	/**
-	 * @param decimalChar the decimal character to set
-	 */
 	public void setDecimalChar(String decimalChar) {
 		this.decimalChar = decimalChar;
 	}
 
-	/**
-	 * @return the group character
-	 */
 	public String getGroupChar() {
 		return groupChar;
 	}
 
-	/**
-	 * @param groupChar the group character to set
-	 */
 	public void setGroupChar(String groupChar) {
 		this.groupChar = groupChar;
 	}
 
-	/**
-	 * @param format
-	 */
-	public void setFormat(String format) {
-		this.format = format;
+	public String getSeparator() {
+		return separator;
 	}
 
-	protected String getValueOrDefault(String s) {
-		if ((s == null || s.isEmpty()) && (defaultValue != null)) {
-			return defaultValue;
-		}
-		return s;
+	public void setSeparator(String separator) {
+		this.separator = separator;
 	}
 
-	/**
-	 * @return true if the column is virtual
-	 */
-	public boolean isVirtual() {
-		return this.isVirtual;
+	public boolean isTrim() {
+		return trim;
+	}
+
+	public void setTrim(boolean trim) {
+		this.trim = trim;
 	}
 
 	/**
-	 * @param isVirtual
+	 * Get the (possibly trimmed) value or default value
+	 *
+	 * @param s
+	 * @return
 	 */
-	public void setVirtual(boolean isVirtual) {
-		this.isVirtual = isVirtual;
+	protected String getValueOrDefault(String s) {
+		if ((s == null || s.isEmpty()) && (defaultValue != null)) {
+			return defaultValue;
+		}
+		if (s == null) {
+			return null;
+		}
+		if (s.equals(nullValue)) {
+			return null;
+		}
+
+		return trim ? s.trim() : s;
 	}
 
 	/**

diff --git a/core/rio/csvw/src/main/java/org/eclipse/rdf4j/rio/csvw/parsers/CellParserBoolean.java b/core/rio/csvw/src/main/java/org/eclipse/rdf4j/rio/csvw/parsers/CellParserBoolean.java
@@ -36,7 +36,7 @@ public void setFormat(String format) {
 	public Value parse(String cell) {
 		String s = getValueOrDefault(cell);
 
-		return Values.literal(valueTrue.equals(s) ? "true" : "false", dataType);
+		return Values.literal(valueTrue.equals(s) ? "true" : "false", getDataType());
 	}
 
 }
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
@@ -40,7 +40,7 @@ public Value parse(String cell) {
 			s = DateTimeFormatter.ISO_DATE.format(formatter.parse(s));
 		}
 		System.err.println("date = " + s);
-		return Values.literal(s, dataType);
+		return Values.literal(s, getDataType());
 	}
 
 }
diff --git a/core/rio/csvw/src/main/java/org/eclipse/rdf4j/rio/csvw/parsers/CellParserDecimal.java b/core/rio/csvw/src/main/java/org/eclipse/rdf4j/rio/csvw/parsers/CellParserDecimal.java
@@ -23,16 +23,16 @@ public class CellParserDecimal extends CellParser {
 	public Value parse(String cell) {
 		String s = getValueOrDefault(cell);
 
-		if (s != null && groupChar != null) {
-			s = s.replace(groupChar, "");
+		if (s != null && getGroupChar() != null) {
+			s = s.replace(getGroupChar(), "");
 		}
 
 		// always use a '.' in RDF, not the European-style ','
-		if (s != null && !decimalChar.equals(".")) {
-			s = s.replace(decimalChar, ".");
+		if (s != null && !getDecimalChar().equals(".")) {
+			s = s.replace(getDecimalChar(), ".");
 		}
 
-		return Values.literal(s, dataType);
+		return Values.literal(s, getDataType());
 	}
 
 }
diff --git a/core/rio/csvw/src/main/java/org/eclipse/rdf4j/rio/csvw/parsers/CellParserInteger.java b/core/rio/csvw/src/main/java/org/eclipse/rdf4j/rio/csvw/parsers/CellParserInteger.java
@@ -23,11 +23,10 @@ public class CellParserInteger extends CellParser {
 	public Value parse(String cell) {
 		String s = getValueOrDefault(cell);
 
-		if (s != null && groupChar != null) {
-			s = s.replace(groupChar, "");
+		if (s != null && getGroupChar() != null) {
+			s = s.replace(getGroupChar(), "");
 		}
-
-		return Values.literal(s, dataType);
+		return Values.literal(s, getDataType());
 	}
 
 }