Adding enhanced error handling for invalid column names in coordinate sources

CHANGELOG.md

@@ -20,6 +20,7 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0
 - Allow hierarchic grid structure for JointGridContainer [#768](https://github.com/ie3-institute/PowerSystemDataModel/issues/768)
 - Adding SQL id coordinate sources (``IdCoordinateSource``) [#689](https://github.com/ie3-institute/PowerSystemDataModel/issues/689)
 - Added some standard asset types to documentation [#642](https://github.com/ie3-institute/PowerSystemDataModel/issues/642)
+- Enhanced error handling of coordinate sources with invalid column names [#670](https://github.com/ie3-institute/PowerSystemDataModel/issues/670)
 
 ### Fixed
 - Fixed wrong rated power unit hint [#804](https://github.com/ie3-institute/PowerSystemDataModel/issues/804)

src/main/java/edu/ie3/datamodel/io/factory/timeseries/CosmoIdCoordinateFactory.java

@@ -53,4 +53,9 @@ public String getLatField() {
   public String getLonField() {
     return LONG_GEO;
   }
+
+  @Override
+  public List<String> getFieldNames() {
+    return List.of(TID, COORDINATE_ID, "long_geo", "lat_geo", "long_rot", "lat_rot");
+  }
 }

src/main/java/edu/ie3/datamodel/io/factory/timeseries/IconIdCoordinateFactory.java

@@ -50,4 +50,9 @@ public String getLatField() {
   public String getLonField() {
     return LONG;
   }
+
+  @Override
+  public List<String> getFieldNames() {
+    return List.of(COORDINATE_ID, LONG, LAT, "coordinate_type");
+  }
 }

src/main/java/edu/ie3/datamodel/io/factory/timeseries/IdCoordinateFactory.java

@@ -5,8 +5,11 @@
 */
 package edu.ie3.datamodel.io.factory.timeseries;
 
+import edu.ie3.datamodel.exceptions.InvalidColumnNameException;
 import edu.ie3.datamodel.io.factory.Factory;
 import edu.ie3.datamodel.io.factory.SimpleFactoryData;
+import java.util.List;
+import java.util.Set;
 import org.apache.commons.lang3.tuple.Pair;
 import org.locationtech.jts.geom.Point;
 
@@ -28,4 +31,27 @@ protected IdCoordinateFactory() {
 
   /** @return the field id for the coordinate longitude */
   public abstract String getLonField();
+
+  /** @return all fields */
+  public abstract List<String> getFieldNames();
+
+  public void checkForInvalidColumnNames(Set<String> columnNames) {
+    List<String> validColumnNames = getFieldNames();
+
+    if (!columnNames.containsAll(validColumnNames)) {
+      throw new InvalidColumnNameException(
+          "The provided column names "
+              + columnNames
+              + " does not match the expected column names "
+              + validColumnNames
+              + "!");
+    }
+
+    if (columnNames.size() != validColumnNames.size()) {
+      log.warn(
+          "The provided row has more column names than expected. Provided: {}, expected: {}.",
+          columnNames,
+          validColumnNames);
+    }
+  }
 }

src/main/java/edu/ie3/datamodel/io/factory/timeseries/SqlIdCoordinateFactory.java

@@ -60,6 +60,11 @@ public String getLonField() {
         "this is not supported by " + SqlIdCoordinateFactory.class + "!");
   }
 
+  @Override
+  public List<String> getFieldNames() {
+    return List.of(COORDINATE_ID, COORDINATE);
+  }
+
   public String getCoordinateField() {
     return COORDINATE;
   }

src/main/java/edu/ie3/datamodel/io/source/csv/CsvIdCoordinateSource.java

@@ -162,6 +162,9 @@ protected Stream<Map<String, String>> buildStreamWithFieldsToAttributesMap() {
     try (BufferedReader reader = dataSource.connector.initIdCoordinateReader()) {
       final String[] headline = dataSource.parseCsvRow(reader.readLine(), dataSource.csvSep);
 
+      // checking the column names of csv source
+      factory.checkForInvalidColumnNames(Set.of(headline));
+
       // by default try-with-resources closes the reader directly when we leave this method (which
       // is wanted to avoid a lock on the file), but this causes a closing of the stream as well.
       // As we still want to consume the data at other places, we start a new stream instead of

src/main/java/edu/ie3/datamodel/io/source/sql/SqlIdCoordinateSource.java

@@ -53,6 +53,9 @@ public SqlIdCoordinateSource(
     String dbPointColumnName =
         dataSource.getDbColumnName(factory.getCoordinateField(), coordinateTableName);
 
+    // checking the column names of sql source
+    factory.checkForInvalidColumnNames(Set.of(dbIdColumnName, dbPointColumnName));
+
     // setup queries
     this.basicQuery = createBaseQueryString(dataSource.schemaName, coordinateTableName);
     this.queryForPoint = createQueryForPoint(dbIdColumnName);

src/test/groovy/edu/ie3/datamodel/io/factory/timeseries/CosmoIdCoordinateFactoryTest.groovy

@@ -5,6 +5,8 @@
  */
 package edu.ie3.datamodel.io.factory.timeseries
 
+import edu.ie3.datamodel.exceptions.FactoryException
+import edu.ie3.datamodel.exceptions.InvalidColumnNameException
 import edu.ie3.datamodel.io.factory.SimpleFactoryData
 import edu.ie3.util.geo.GeoUtils
 import org.apache.commons.lang3.tuple.Pair
@@ -96,4 +98,16 @@ class CosmoIdCoordinateFactoryTest extends Specification {
       assert it.value.equalsExact(expectedPair.value, 1E-6)
     }
   }
+
+  def "Throws an exception if an invalid column name is given" () {
+    given:
+    def invalidColumnNames = ["id", "lat", "lon"] as Set<String>
+    when:
+    factory.checkForInvalidColumnNames(invalidColumnNames)
+
+    then:
+    Exception ex = thrown()
+    ex.class == InvalidColumnNameException
+    ex.message == "The provided column names [id, lat, lon] does not match the expected column names [tid, id, long_geo, lat_geo, long_rot, lat_rot]!"
+  }
 }

src/test/groovy/edu/ie3/datamodel/io/factory/timeseries/IconIdCoordinateFactoryTest.groovy

@@ -5,6 +5,8 @@
  */
 package edu.ie3.datamodel.io.factory.timeseries
 
+import edu.ie3.datamodel.exceptions.FactoryException
+import edu.ie3.datamodel.exceptions.InvalidColumnNameException
 import edu.ie3.datamodel.io.factory.SimpleFactoryData
 import edu.ie3.util.geo.GeoUtils
 import org.apache.commons.lang3.tuple.Pair
@@ -82,4 +84,16 @@ class IconIdCoordinateFactoryTest extends Specification {
       assert it.value.equalsExact(expectedPair.value, 1E-6)
     }
   }
+
+  def "Throws an exception if an invalid column name is given" () {
+    given:
+    def invalidColumnNames = ["id", "lat", "lon"] as Set<String>
+    when:
+    factory.checkForInvalidColumnNames(invalidColumnNames)
+
+    then:
+    Exception ex = thrown()
+    ex.class == InvalidColumnNameException
+    ex.message == "The provided column names [id, lat, lon] does not match the expected column names [id, longitude, latitude, coordinate_type]!"
+  }
 }

src/test/groovy/edu/ie3/datamodel/io/source/sql/SqlIdCoordinateSourceIT.groovy

@@ -5,6 +5,7 @@
  */
 package edu.ie3.datamodel.io.source.sql
 
+import edu.ie3.datamodel.exceptions.InvalidColumnNameException
 import edu.ie3.datamodel.io.connectors.SqlConnector
 import edu.ie3.datamodel.io.factory.timeseries.SqlIdCoordinateFactory
 import edu.ie3.test.helper.TestContainerHelper
@@ -187,4 +188,16 @@ class SqlIdCoordinateSourceIT extends Specification implements TestContainerHelp
       expectedValues.contains(coordinateDistance.coordinateB)
     }
   }
+
+  def "Throws an exception if an invalid column name is given" () {
+    given:
+    def invalidColumnNames = ["id", "lat", "lon"] as Set<String>
+    when:
+    source.factory.checkForInvalidColumnNames(invalidColumnNames)
+
+    then:
+    Exception ex = thrown()
+    ex.class == InvalidColumnNameException
+    ex.message == "The provided column names [id, lat, lon] does not match the expected column names [id, coordinate]!"
+  }
 }