[New] Initial implementation of the xls files support in TabularModule

s-pipes-modules/module-tabular/pom.xml

@@ -92,5 +92,11 @@
             <artifactId>junit-jupiter-params</artifactId>
             <scope>test</scope>
         </dependency>
+        <dependency>
+            <groupId>org.apache.poi</groupId>
+            <artifactId>poi</artifactId>
+            <version>3.11</version>
+            <scope>compile</scope>
+        </dependency>
     </dependencies>
 </project>

s-pipes-modules/module-tabular/src/main/java/cz/cvut/spipes/modules/TabularModule.java

@@ -16,6 +16,7 @@
 import cz.cvut.spipes.modules.util.BNodesTransformer;
 import cz.cvut.spipes.modules.util.HTML2TSVConvertor;
 import cz.cvut.spipes.modules.util.JopaPersistenceUtils;
+import cz.cvut.spipes.modules.util.XLS2TSVConvertor;
 import cz.cvut.spipes.registry.StreamResource;
 import cz.cvut.spipes.registry.StreamResourceRegistry;
 import cz.cvut.spipes.util.JenaUtils;
@@ -104,6 +105,12 @@ public class TabularModule extends AbstractModule {
     private final Property P_SOURCE_RESOURCE_URI = getSpecificParameter("source-resource-uri");
     private final Property P_SKIP_HEADER = getSpecificParameter("skip-header");
     private final Property P_PROCESS_HTML_FILE = getSpecificParameter("process-html-file");
+    private final Property P_PROCESS_XLS_FILE = getSpecificParameter("process-xls-file");
+
+    /**
+    Optional parameter that indicates that only specific single sheet should be converted
+     */
+    private final Property P_PROCESS_SPECIFIC_SHEET_IN_XLS_FILE = getSpecificParameter("process-specific-sheet-in-xls-file");
 
     //sml:replace
     private boolean isReplace;
@@ -126,6 +133,12 @@ public class TabularModule extends AbstractModule {
     //:process-html-file
     private boolean processHTMLFile;
 
+    //:process-xls-file
+    private boolean processXLSFile;
+
+    //:process-specific-sheet-in-xls-file
+    private int processSpecificSheetInXLSFile;
+
     //:output-mode
     private Mode outputMode;
 
@@ -146,6 +159,7 @@ public class TabularModule extends AbstractModule {
      * Represents the table schema that was used to describe the table
      */
     private TableSchema tableSchema;
+    private int numberOfSheets;
 
     /**
      * Default charset to process input file.
@@ -161,6 +175,14 @@ ExecutionContext executeSelf() {
             setDelimiter('\t');
         }
 
+        if(processXLSFile) {
+            XLS2TSVConvertor xlsConvertor = new XLS2TSVConvertor();
+            numberOfSheets = xlsConvertor.getNumberOfSheets(sourceResource); // Currently isn't used
+            LOG.debug("Number of sheets:{}",numberOfSheets);
+            setSourceResource(xlsConvertor.convertToTSV(sourceResource));
+            setDelimiter('\t');
+        }
+
         BNodesTransformer bNodesTransformer = new BNodesTransformer();
         Model inputModel = bNodesTransformer.convertBNodesToNonBNodes(executionContext.getDefaultModel());
         boolean hasInputSchema = false;
@@ -368,6 +390,8 @@ public void loadConfiguration() {
         delimiter = getPropertyValue(P_DELIMITER, getDefaultDelimiterSupplier());
         skipHeader = getPropertyValue(P_SKIP_HEADER, false);
         processHTMLFile = getPropertyValue(P_PROCESS_HTML_FILE, false);
+        processXLSFile = getPropertyValue(P_PROCESS_XLS_FILE, false);
+        processSpecificSheetInXLSFile = getPropertyValue(P_PROCESS_SPECIFIC_SHEET_IN_XLS_FILE, 0);
         acceptInvalidQuoting = getPropertyValue(P_ACCEPT_INVALID_QUOTING, false);
         quoteCharacter = getPropertyValue(P_QUOTE_CHARACTER, getDefaultQuoteCharacterSupplier(delimiter));
         dataPrefix = getEffectiveValue(P_DATE_PREFIX).asLiteral().toString();
@@ -538,6 +562,9 @@ public void setSkipHeader(boolean skipHeader) {
     public void setProcessHTMLFile(boolean processHTMLFile) {
         this.processHTMLFile = processHTMLFile;
     }
+    public void setProcessXLSFile(boolean processXLSFile) {
+        this.processXLSFile = processXLSFile;
+    }
 
     private String[] getHeaderFromSchema(Model inputModel, String[] header, boolean hasInputSchema) {
         if (hasInputSchema) {

s-pipes-modules/module-tabular/src/main/java/cz/cvut/spipes/modules/util/XLS2TSVConvertor.java

@@ -0,0 +1,51 @@
+package cz.cvut.spipes.modules.util;
+
+import cz.cvut.spipes.registry.StreamResource;
+import cz.cvut.spipes.registry.StringStreamResource;
+import org.apache.poi.hssf.usermodel.HSSFWorkbook;
+import org.apache.poi.ss.usermodel.Cell;
+import org.apache.poi.ss.usermodel.Row;
+import org.apache.poi.ss.usermodel.Sheet;
+import org.apache.poi.ss.usermodel.Workbook;
+
+import java.io.ByteArrayInputStream;
+import java.io.IOException;
+
+/**
+ * Module for converting tabular data from XLS to TSV.
+ * Currently, converts only first sheet.
+ */
+public class XLS2TSVConvertor {
+    public StringStreamResource convertToTSV(StreamResource streamResource){
+        try {
+            Workbook workbook = new HSSFWorkbook(new ByteArrayInputStream(streamResource.getContent()));
+            Sheet sheet = workbook.getSheetAt(0);
+
+            StringBuilder tsvStringBuilder = new StringBuilder();
+            for (Row row : sheet) {
+                for (Cell cell : row) {
+                    tsvStringBuilder.append(cell.toString().replace('\t', ' '));
+                    tsvStringBuilder.append('\t');
+                }
+                tsvStringBuilder.deleteCharAt(tsvStringBuilder.length() - 1);
+                tsvStringBuilder.append('\n');
+            }
+            return new StringStreamResource(
+                    streamResource.getUri(),
+                    tsvStringBuilder.toString().getBytes(),
+                    "text/tsv"
+            );
+        }
+        catch (IOException e) {
+            throw new RuntimeException(e);
+        }
+    }
+
+    public int getNumberOfSheets(StreamResource streamResource){
+        try {
+            return new HSSFWorkbook(new ByteArrayInputStream(streamResource.getContent())).getNumberOfSheets();
+        } catch (IOException e) {
+            throw new RuntimeException(e);
+        }
+    }
+}

s-pipes-modules/module-tabular/src/test/java/cz/cvut/spipes/modules/TabularModuleTest.java

@@ -62,6 +62,20 @@ void executeWithSimpleTransformation() throws URISyntaxException, IOException {
         assertTrue(outputContext.getDefaultModel().size() > 0);
     }
 
+     @Test
+     void executeWithSimpleTransformationXls() throws URISyntaxException, IOException {
+         module.setSourceResource(
+                 StreamResourceUtils.getStreamResource(
+                         "http://test-file",
+                         getFilePath("countries.xls"))
+         );
+         module.setProcessXLSFile(true);
+
+         ExecutionContext outputContext = module.executeSelf();
+
+         assertTrue(outputContext.getDefaultModel().size() > 0);
+     }
+
     @Test
      void executeWithDuplicateColumnsThrowsResourceNotUniqueException()
             throws URISyntaxException, IOException {