set maximum length of YEAR values to 5 digits; fixes #27819

Author: Keelhaul

goobi-viewer-indexer/src/main/java/io/goobi/viewer/indexer/helper/DateTools.java

@@ -84,7 +84,10 @@ private DateTools() {
      * @should parse year ranges correctly
      * @should parse single years correctly
      * @should throw IllegalArgumentException if normalizeYearMinDigits less than 1
+     * @should ignore values with too many digits
      */
+    private static final int MAX_YEAR_DIGITS = 5;
+
     static List<PrimitiveDate> normalizeDate(String dateString, int normalizeYearMinDigits) {
         if (normalizeYearMinDigits < 1) {
             throw new IllegalArgumentException("normalizeYearMinDigits must be at least 1");
@@ -159,7 +162,7 @@ static List<PrimitiveDate> normalizeDate(String dateString, int normalizeYearMin
             while (m.find()) {
                 try {
                     String sub = dateString.substring(m.start(), m.end());
-                    if (sub.length() >= normalizeYearMinDigits) {
+                    if (sub.length() >= normalizeYearMinDigits && sub.length() <= MAX_YEAR_DIGITS) {
                         int year = Integer.parseInt(sub);
                         ret.add(new PrimitiveDate(year, null, null));
                     }
@@ -175,7 +178,8 @@ static List<PrimitiveDate> normalizeDate(String dateString, int normalizeYearMin
         while (m.find()) {
             try {
                 String sub = dateString.substring(m.start(), m.end());
-                if (sub.length() >= (sub.charAt(0) == '-' ? (normalizeYearMinDigits + 1) : normalizeYearMinDigits)) {
+                int digitLength = sub.charAt(0) == '-' ? sub.length() - 1 : sub.length();
+                if (digitLength >= normalizeYearMinDigits && digitLength <= MAX_YEAR_DIGITS) {
                     int year = Integer.parseInt(sub);
                     ret.add(new PrimitiveDate(year, null, null));
                 }

goobi-viewer-indexer/src/test/java/io/goobi/viewer/indexer/helper/DateToolsTest.java

@@ -116,6 +116,22 @@ void normalizeDate_shouldThrowIllegalArgumentExceptionIfNormalizeYearMinDigitsLe
         Assertions.assertThrows(IllegalArgumentException.class, () -> DateTools.normalizeDate("05.08.2014", 0));
     }
 
+    /**
+     * @see DateTools#normalizeDate(String,int)
+     * @verifies ignore values with too many digits
+     */
+    @Test
+    void normalizeDate_shouldIgnoreValuesWithTooManyDigits() throws Exception {
+        // Concatenated date range "1848-1848" stripped of hyphens should not produce year 18481848
+        List<PrimitiveDate> ret = DateTools.normalizeDate("18481848", 3);
+        Assertions.assertTrue(ret.isEmpty());
+
+        // Also test in a string with other content
+        ret = DateTools.normalizeDate("18481848 1900", 3);
+        Assertions.assertEquals(1, ret.size());
+        Assertions.assertEquals(Integer.valueOf(1900), ret.get(0).getYear());
+    }
+
     /**
      * @see DateTools#normalizeDateFieldValue(String)
      * @verifies format years correctly