Using FILE_ABSOLUTEPATH and Original-Filename to populate FILEXT

src/main/java/emissary/core/constants/Parameters.java

@@ -11,6 +11,7 @@ public class Parameters {
     public static final String INPUT_FILENAME = "INPUT_FILENAME";
     public static final String ORIGINAL_FILENAME = "Original-Filename";
     public static final String SUMMARY = "SUMMARY";
+    public static final String FILE_ABSOLUTEPATH = "FILE_ABSOLUTEPATH";
 
     // Common parameter prefixes
 

src/main/java/emissary/output/DropOffUtil.java

@@ -33,11 +33,14 @@
 import java.util.Set;
 import java.util.SimpleTimeZone;
 import java.util.UUID;
+import java.util.regex.Matcher;
+import java.util.regex.Pattern;
 import javax.annotation.Nullable;
 
 import static emissary.core.Form.PREFIXES_LANG;
 import static emissary.core.Form.TEXT;
 import static emissary.core.Form.UNKNOWN;
+import static emissary.core.constants.Parameters.FILE_ABSOLUTEPATH;
 import static emissary.core.constants.Parameters.ORIGINAL_FILENAME;
 
 public class DropOffUtil {
@@ -82,6 +85,8 @@ public class DropOffUtil {
     private static final String DEFAULT_EVENT_DATE_TO_NOW = "DEFAULT_EVENT_DATE_TO_NOW";
     protected boolean defaultEventDateToNow = true;
 
+    private final Pattern JUST_FILENAME_PATTERN = Pattern.compile("(^|\\\\|/)[^\\\\/]+$");
+
     /**
      * Create with the default configuration
      */
@@ -969,25 +974,32 @@ public Date extractEventDateFrom(final IBaseDataObject d, final boolean lastReso
     }
 
     /**
-     * Extracts from the provided {@link IBaseDataObject} the last file extension from each value in the "Original-Filename"
-     * parameter, if that value contains "." before its last character. If one or more file extensions are extracted, the
-     * IBaseDataObject's "FILEXT" parameter is set as the unique set of extracted file extensions, converted to lowercase.
+     * Extracts from the provided {@link IBaseDataObject} the last file extension from each filename, if that value contains
+     * "." before its last few characters. If one or more file extensions are extracted, the IBaseDataObject's "FILEXT"
+     * parameter is set as the unique set of extracted file extensions, converted to lowercase.
      *
      * @param p IBaseDataObject to process
      *
      */
     void extractUniqueFileExtensions(IBaseDataObject p) {
-        if (p.hasParameter(ORIGINAL_FILENAME)) {
-            final Set<String> extensions = new HashSet<>();
-            for (Object filename : p.getParameter(ORIGINAL_FILENAME)) {
-                final String fn = (String) filename;
-                if (StringUtils.isNotEmpty(fn) && fn.lastIndexOf('.') > -1) {
-                    final int pos = fn.lastIndexOf('.') + 1;
-                    if (pos < fn.length()) {
-                        final String fext = fn.substring(pos).toLowerCase();
-                        if (fext.length() > 0 && fext.length() <= this.maxFilextLen) {
-                            extensions.add(fext);
-                        }
+        List<String> filenames = getBestFilenames(p);
+        final Set<String> extensions = new HashSet<>();
+        for (String filename : filenames) {
+
+            // if what we have is a full filepath, extract just the filename (text after the file path separator)
+            Matcher matcher = JUST_FILENAME_PATTERN.matcher(filename);
+            String justFilename = filename;
+            if (matcher.find()) {
+                justFilename = matcher.group(0);
+            }
+
+            // get the text after the last period
+            if (justFilename.lastIndexOf('.') > -1) {
+                final int pos = justFilename.lastIndexOf('.') + 1;
+                if (pos < justFilename.length()) {
+                    final String fext = justFilename.substring(pos).toLowerCase();
+                    if (fext.length() > 0 && fext.length() <= this.maxFilextLen) {
+                        extensions.add(fext);
                     }
                 }
             }
@@ -997,6 +1009,31 @@ void extractUniqueFileExtensions(IBaseDataObject p) {
         }
     }
 
+    /**
+     * Checks the Original-Filename and FILE_ABSOLUTEPATH for the filename of the object. Returns a list with of the
+     * non-empty strings found in these fields. If nothing is found in either field, return an empty list.
+     *
+     * @param d The IBDO
+     * @return The list of filenames found in the field Original-Filename or FILE_ABSOLUTEPATH
+     */
+    public static List<String> getBestFilenames(IBaseDataObject d) {
+        String[] fieldsToTry = {ORIGINAL_FILENAME, FILE_ABSOLUTEPATH};
+
+        List<String> filenames = new ArrayList<>();
+
+        for (String ibdoField : fieldsToTry) {
+            if (d.hasParameter(ibdoField)) {
+                for (Object filename : d.getParameter(ibdoField)) {
+                    String stringFileName = (String) filename;
+                    if (StringUtils.isNotEmpty(stringFileName)) {
+                        filenames.add(stringFileName);
+                    }
+                }
+            }
+        }
+        return filenames;
+    }
+
     /**
      * Process metadata before doing any output
      *

src/test/java/emissary/output/DropOffUtilTest.java

@@ -23,6 +23,7 @@
 import static emissary.core.Form.TEXT;
 import static emissary.core.Form.UNKNOWN;
 import static emissary.core.constants.Parameters.EVENT_DATE;
+import static emissary.core.constants.Parameters.FILE_ABSOLUTEPATH;
 import static emissary.core.constants.Parameters.FILE_DATE;
 import static emissary.core.constants.Parameters.ORIGINAL_FILENAME;
 import static org.junit.jupiter.api.Assertions.assertEquals;
@@ -38,6 +39,7 @@
 class DropOffUtilTest extends UnitTest {
     private DropOffUtil util = null;
     private IBaseDataObject payload = null;
+    static final String FILEXT = "FILEXT";
 
     @BeforeEach
     public void createUtil() {
@@ -576,7 +578,6 @@ void testGetFileType() {
     @Test
     void testExtractUniqueFileExtensions() {
         // these should be constants
-        final String FILEXT = "FILEXT";
         DropOffUtil util = new DropOffUtil();
 
         final IBaseDataObject bdo = new BaseDataObject();
@@ -601,6 +602,41 @@ void testExtractUniqueFileExtensions() {
         assertTrue(fileExts.contains("mp3"), "FILEXT values should contain \"mp3\"");
     }
 
+    @Test
+    void testExtractFullFilepaths() {
+        DropOffUtil util = new DropOffUtil();
+
+        final IBaseDataObject ibdo = new BaseDataObject();
+        ibdo.setParameter(FILE_ABSOLUTEPATH, "D:\\Users\\jdoe\\Documents\\Taxes 2023.csv");
+        util.extractUniqueFileExtensions(ibdo);
+        assertEquals("csv", ibdo.getStringParameter(FILEXT), "FILEXT should be extracted");
+        ibdo.setParameter(FILEXT, "");
+
+        ibdo.setParameter(FILE_ABSOLUTEPATH, "D:\\Users\\jdoe\\interesting.folder\\a.table");
+        util.extractUniqueFileExtensions(ibdo);
+        assertEquals("table", ibdo.getStringParameter(FILEXT), "FILEXT should be extracted");
+        ibdo.setParameter(FILEXT, "");
+
+        ibdo.setParameter(FILE_ABSOLUTEPATH, "/paper.abc.zzz");
+        util.extractUniqueFileExtensions(ibdo);
+        assertEquals("zzz", ibdo.getStringParameter(FILEXT), "FILEXT should be extracted");
+        ibdo.setParameter(FILEXT, "");
+
+        ibdo.setParameter(FILE_ABSOLUTEPATH, "flowers.456.123");
+        util.extractUniqueFileExtensions(ibdo);
+        assertEquals("123", ibdo.getStringParameter(FILEXT), "FILEXT should be extracted");
+        ibdo.setParameter(FILEXT, "");
+
+        ibdo.setParameter(FILE_ABSOLUTEPATH, "/home/jdoe/SHARED_D.IR/cat.mov");
+        util.extractUniqueFileExtensions(ibdo);
+        assertEquals("mov", ibdo.getStringParameter(FILEXT), "FILEXT should be extracted");
+        ibdo.setParameter(FILEXT, "");
+
+        ibdo.setParameter(FILE_ABSOLUTEPATH, "/home/jdoe/SHARED_D.IR/cat");
+        util.extractUniqueFileExtensions(ibdo);
+        assertEquals("", ibdo.getStringParameter(FILEXT), "FILEXT should not be extracted if there is no period in the filename");
+    }
+
     @Test
     void testCleanSpecPath() {
         assertEquals("/this/is/fine", util.cleanSpecPath("/this/is/fine"));
@@ -609,6 +645,28 @@ void testCleanSpecPath() {
         assertEquals("/this/./././/./is/fine", util.cleanSpecPath("/this/....../../..//./is/fine"));
     }
 
+    @Test
+    void testGetBestFilename() {
+        IBaseDataObject ibdo = new BaseDataObject();
+        List<String> bestFilenames;
+
+        ibdo.setParameter(ORIGINAL_FILENAME, "");
+        ibdo.setParameter(FILE_ABSOLUTEPATH, "");
+        bestFilenames = DropOffUtil.getBestFilenames(ibdo);
+        assertEquals(0, bestFilenames.size(), "No filename should have been found");
+
+        ibdo.setParameter(FILE_ABSOLUTEPATH, "theOtherFile.csv");
+        bestFilenames = DropOffUtil.getBestFilenames(ibdo);
+        assertEquals(1, bestFilenames.size(), "There should be one filename extracted");
+        assertEquals("theOtherFile.csv", bestFilenames.get(0), "The FILE_ABSOLUTEPATH should have been extracted");
+
+        ibdo.setParameter(ORIGINAL_FILENAME, "file.docx");
+        bestFilenames = DropOffUtil.getBestFilenames(ibdo);
+        assertEquals(2, bestFilenames.size(), "There should be two filenames extracted");
+        assertEquals("file.docx", bestFilenames.get(0), "The Original-Filename should have been extracted");
+        assertEquals("theOtherFile.csv", bestFilenames.get(1), "The Original-Filename should have been extracted");
+    }
+
     private void setupMetadata(IBaseDataObject bdo, String fieldValue, DropOffUtil.FileTypeCheckParameter fileTypeCheckParameter) {
         bdo.clearParameters();
         bdo.putParameter(fileTypeCheckParameter.getFieldName(), fieldValue);