Draft of cached regex references for testing

core/utils/type-utils/src/main/java/datawave/data/type/util/MACAddress.java

@@ -9,6 +9,11 @@ public class MACAddress implements Serializable, Comparable<MACAddress> {
 
     private static final long serialVersionUID = 4366259028581959024L;
 
+    private static final String DOT_REGEX = "\\.";
+
+    private static final String MAC_REGEX_1 = "^[0-9a-fA-F]+$";
+    private static final String MAC_REGEX_2 = "^([0-9a-fA-F]+[^0-9a-fA-F])+[0-9a-fA-F]+$";
+
     /**
      * String representation of the MAC address
      */
@@ -74,7 +79,7 @@ public String toNormalizedString(String sep, int groupingSize) {
 
         if (!this.separator.equals("")) {
             String sepRegex = new String(this.separator);
-            if (this.separator.matches("\\.")) {
+            if (this.separator.matches(DOT_REGEX)) {
                 sepRegex = "\\" + sepRegex;
             }
             returnAddress = returnAddress.replaceAll(this.separator, "");
@@ -92,7 +97,7 @@ public String toNormalizedString(String sep, int groupingSize) {
         int totalStringLength = MAC_ADDRESS_LENGTH + groups - 1;
         int digitCount = 1;
         String sepRegex = new String(sep);
-        if (sepRegex.matches("\\.")) {
+        if (sepRegex.matches(DOT_REGEX)) {
             sepRegex = "\\" + sepRegex;
         }
         // populate replacement as a regex to properly format / separate the hex digits
@@ -131,7 +136,7 @@ public static MACAddress parse(String addr, String sep, int groupingSize, boolea
             if (groupingSize < 1 || groupingSize > MAC_ADDRESS_LENGTH) {
                 throw new IllegalArgumentException("Grouping size must be between 1 and " + MAC_ADDRESS_LENGTH + ", inclusive.");
             }
-            if (sep.matches("\\.")) {
+            if (sep.matches(DOT_REGEX)) {
                 sep = "\\" + sep;
             }
             String[] digits;
@@ -223,10 +228,10 @@ public static MACAddress parse(String addr, String sep) {
      *             if unable to parse a MAC address
      */
     public static MACAddress parse(String addr) {
-        if (addr.matches("^[0-9a-fA-F]+$")) {
+        if (addr.matches(MAC_REGEX_1)) {
             return parse(addr, "", MAC_ADDRESS_LENGTH, true);
-        } else if (addr.matches("^([0-9a-fA-F]+[^0-9a-fA-F])+[0-9a-fA-F]+$")) {
-            String[] pieces = addr.split("[^0-9a-fA-F]");
+        } else if (addr.matches(MAC_REGEX_2)) {
+            String[] pieces = addr.split(MAC_REGEX_1);
             int groupingSize = MAC_ADDRESS_LENGTH / pieces.length;
             String sep = String.valueOf(addr.charAt(groupingSize));
             return parse(addr, sep, groupingSize, true);

core/utils/type-utils/src/main/java/datawave/data/type/util/PhoneNumber.java

@@ -10,6 +10,26 @@ public class PhoneNumber implements Serializable, Comparable<PhoneNumber> {
     private String originalPhoneNumber = "";
     private String normalizedPhoneNumber = "";
 
+    private static final String DIGIT_STRING = "^\\d+$";
+
+    private static final String ISBN_REGEX = "^\\d\\d\\d([ \\-])\\d\\d\\1\\d\\d\\d\\d$";
+
+    private static final String SSN_REGEX = "^[12]\\d\\d\\d ?- ?[12]\\d\\d\\d$";
+
+    private static final String YEAR_RANGE_REGEX = "^(19|20)\\d\\d([\\-\\. ])[01]\\d\\2[0-3]\\d$";
+
+    private static final String YYYY_MM_DD_REGEX = "^(19|20)\\d\\d[01]\\d[0-3]\\d$";
+
+    private static final String YYYYMMDD_REGEX = "^(19|20)\\d\\d([\\-\\. ])?[0-3]\\d\\2[01]\\d$";
+
+    private static final String YYYYDDMM_REGEX = "^(19|20)\\d\\d[0-3]\\d[01]\\d$";
+
+    private static final String DD_MM_YYYY_REGEX = "^[0-3]\\d([\\-\\.])[01]\\d\\1(19|20)\\d\\d ([0-1]\\d|2[0-4])$";
+
+    private static final String HHMM_REGEX_1 = "^[0-3]\\d([\\-\\.])[1-9]\\1(19|20)\\d\\d ([0-1]\\d|2[0-4])$";
+
+    private static final String YYYY_JJJ_REGEX = "^(19|20)\\d\\d([\\-\\. ])([0-2]\\d\\d|3[0-5]\\d|36[0-6])$";
+
     /**
      * A valid phone number must contain at least 7 digits.
      */
@@ -100,7 +120,7 @@ private static PhoneNumber isValid(String number) {
          * This normalizer is just worrying about stripping punctuation from phone numbers, so if this is a string of digits, just return instead of doing the
          * other checks.
          */
-        if (number.matches("^\\d+$")) {
+        if (number.matches(DIGIT_STRING)) {
             return new PhoneNumber(number);
         }
 
@@ -210,23 +230,23 @@ private static PhoneNumber isValid(String number) {
 
         if (data[start] != '+' && isISBN(s) && (spaceCount > 0 || dashCount > 0 || dotCount > 0) && (openCount + closCount) == 0) {
             throw new IllegalArgumentException("Looks like an ISBN");
-        } else if (number.matches("^\\d\\d\\d([ \\-])\\d\\d\\1\\d\\d\\d\\d$")) {
+        } else if (number.matches(ISBN_REGEX)) {
             throw new IllegalArgumentException(number + " looks like a SSN");
-        } else if (number.matches("^[12]\\d\\d\\d ?- ?[12]\\d\\d\\d$")) {
+        } else if (number.matches(SSN_REGEX)) {
             throw new IllegalArgumentException(number + " looks like a year range");
-        } else if (number.matches("^(19|20)\\d\\d([\\-\\. ])[01]\\d\\2[0-3]\\d$")) {
+        } else if (number.matches(YEAR_RANGE_REGEX)) {
             throw new IllegalArgumentException(number + " looks like a yyyy mm dd date");
-        } else if (number.matches("^(19|20)\\d\\d[01]\\d[0-3]\\d$")) {
+        } else if (number.matches(YYYY_MM_DD_REGEX)) {
             throw new IllegalArgumentException(number + " looks like a yyyymmdd date");
-        } else if (number.matches("^(19|20)\\d\\d([\\-\\. ])?[0-3]\\d\\2[01]\\d$")) {
+        } else if (number.matches(YYYYMMDD_REGEX)) {
             throw new IllegalArgumentException(number + " looks like a yyyy dd mm date");
-        } else if (number.matches("^(19|20)\\d\\d[0-3]\\d[01]\\d$")) {
+        } else if (number.matches(YYYYDDMM_REGEX)) {
             throw new IllegalArgumentException(number + " looks like a yyyyddmm date");
-        } else if (number.matches("^[0-3]\\d([\\-\\.])[01]\\d\\1(19|20)\\d\\d ([0-1]\\d|2[0-4])$")) {
+        } else if (number.matches(DD_MM_YYYY_REGEX)) {
             throw new IllegalArgumentException(number + " looks like a dd-mm-yyyy hh:mm date");
-        } else if (number.matches("^[0-3]\\d([\\-\\.])[1-9]\\1(19|20)\\d\\d ([0-1]\\d|2[0-4])$")) {
+        } else if (number.matches(HHMM_REGEX_1)) {
             throw new IllegalArgumentException(number + " looks like a dd-mm-yyyy hh:mm date");
-        } else if (number.matches("^(19|20)\\d\\d([\\-\\. ])([0-2]\\d\\d|3[0-5]\\d|36[0-6])$")) {
+        } else if (number.matches(YYYY_JJJ_REGEX)) {
             throw new IllegalArgumentException(number + " looks like a yyyy jjj date");
         }
 
@@ -247,7 +267,7 @@ private static String basicPhoneNumberCheck(String number) {
          * This normalizer is just worrying about stripping punctuation from phone numbers, so if this is a string of digits, just return instead of doing the
          * other checks.
          */
-        if (number.matches("^\\d+$")) {
+        if (number.matches(DIGIT_STRING)) {
             return number;
         }
 

warehouse/ingest-core/src/main/java/datawave/util/flag/config/FlagMakerConfig.java

@@ -39,6 +39,8 @@ public class FlagMakerConfig {
 
     public static final String DEFAULT_FILE_PATTERN = "2*/*/*/*";
 
+    public static final String FOLDER_DATE_PATTERN = "(simple|date|folderdate)";
+
     // a list of file patterns.
     @XmlElement(name = "filePattern")
     private List<String> filePatterns = new ArrayList<>();
@@ -289,7 +291,7 @@ public void validate() {
             throw new IllegalArgumentException(prefix + "Default Max Flags must be set.");
         }
 
-        if (this.distributorType == null || !this.distributorType.matches("(simple|date|folderdate)")) {
+        if (this.distributorType == null || !this.distributorType.matches(FOLDER_DATE_PATTERN)) {
             throw new IllegalArgumentException(
                             "Invalid Distributor type provided: " + this.distributorType + ". Must be one of the following: simple|date|folderdate");
         }

warehouse/ingest-core/src/test/java/datawave/ingest/mapreduce/handler/edge/EdgeHandlerTestUtil.java

@@ -34,6 +34,7 @@ public class EdgeHandlerTestUtil {
 
     public static final Text edgeTableName = new Text(TableName.EDGE);
     public static final String NB = "\u0000";
+    public static final String SHARD_KEY = "\\d{8}_\\d+";
 
     public static ListMultimap<String,String[]> edgeKeyResults = ArrayListMultimap.create();
     public static ListMultimap<String,String> edgeValueResults = ArrayListMultimap.create();
@@ -45,7 +46,7 @@ public static boolean isDocumentKey(Key k) {
     }
 
     public static boolean isShardKey(Key k) {
-        return k.getRow().toString().matches("\\d{8}_\\d+");
+        return k.getRow().toString().matches(SHARD_KEY);
     }
 
     public static void processEvent(Multimap<String,NormalizedContentInterface> eventFields, ExtendedDataTypeHandler<Text,BulkIngestKey,Value> edgeHandler,

warehouse/ingest-core/src/test/java/datawave/ingest/mapreduce/handler/shard/content/BoundedOffsetQueueTest.java

@@ -8,6 +8,8 @@
  */
 public class BoundedOffsetQueueTest {
 
+    private static final String OFFSET_REGEX = "Cannot compare a key that has no offsets to be found";
+
     @Test
     public void testAddOffset() {
 
@@ -58,7 +60,7 @@ public void testOffsetListComparatorLogic() {
 
             Assert.assertTrue(String.format(
                             "BoundedOffsetQueue.OffsetListComparator threw the expected exception, however it did not have the correct message: %s", msg),
-                            msg.matches("Cannot compare a key that has no offsets to be found"));
+                            msg.matches(OFFSET_REGEX));
         }
 
         TermAndZone taz = new TermAndZone(token);

warehouse/ingest-csv/src/test/java/datawave/ingest/csv/mr/handler/ColumnBasedHandlerTestUtil.java

@@ -46,6 +46,7 @@ public class ColumnBasedHandlerTestUtil {
     public static final Text shardReverseIndexTableName = new Text(TableName.SHARD_RINDEX);
     public static final Text edgeTableName = new Text(TableName.EDGE);
     public static final String NB = "\u0000";
+    public static final String SHARD_KEY_PATTERN = "\\d{8}_\\d+";
 
     private static Logger log = Logger.getLogger(ColumnBasedHandlerTestUtil.class);
 
@@ -54,7 +55,7 @@ public static boolean isDocumentKey(Key k) {
     }
 
     public static boolean isShardKey(Key k) {
-        return k.getRow().toString().matches("\\d{8}_\\d+");
+        return k.getRow().toString().matches(SHARD_KEY_PATTERN);
     }
 
     public static InputSplit getSplit(String file) throws URISyntaxException {

warehouse/ingest-json/src/test/java/datawave/ingest/json/mr/handler/ColumnBasedHandlerTestUtil.java

@@ -45,6 +45,7 @@ public class ColumnBasedHandlerTestUtil {
     public static final Text shardReverseIndexTableName = new Text(TableName.SHARD_RINDEX);
     public static final Text edgeTableName = new Text("edge");
     public static final String NB = "\u0000";
+    public static final String SHARD_KEY_PATTERN = "\\d{8}_\\d+";
 
     private static Logger log = Logger.getLogger(ColumnBasedHandlerTestUtil.class);
 
@@ -53,7 +54,7 @@ public static boolean isDocumentKey(Key k) {
     }
 
     public static boolean isShardKey(Key k) {
-        return k.getRow().toString().matches("\\d{8}_\\d+");
+        return k.getRow().toString().matches(SHARD_KEY_PATTERN);
     }
 
     public static InputSplit getSplit(String file) throws URISyntaxException {

warehouse/query-core/src/main/java/datawave/query/jexl/visitors/JexlFormattedStringBuildingVisitor.java

@@ -30,6 +30,7 @@
  */
 public class JexlFormattedStringBuildingVisitor extends JexlStringBuildingVisitor {
     protected static final String NEWLINE = System.getProperty("line.separator");
+    public static final String AND_OR_PATTERN = "^([)]+ (&&|\\|\\|) )$";
 
     public JexlFormattedStringBuildingVisitor() {
         super(false);
@@ -103,7 +104,7 @@ private static boolean containsOnly(String str, char ch) {
      * @return boolean
      */
     private static boolean closeParensFollowedByAndOr(String str) {
-        return str.matches("^([)]+ (&&|\\|\\|) )$");
+        return str.matches(AND_OR_PATTERN);
     }
 
     /**

warehouse/query-core/src/test/java/datawave/query/jexl/JexlPatternCacheTest.java

@@ -10,25 +10,35 @@
 
 public class JexlPatternCacheTest {
 
+    public static final String ASTERISK_WORD = ".*word.*";
+    public static final String BLA_WORD_BLA = "bla word bla";
+    public static final String BLA_PATTERN_2 = "bla\nbla word bla\n bla";
+    public static final String BLA_PATTERN_3 = "(\\s|.)*word(\\s|.)*";
+
+    public static final String FOOBAR_PATTERN = "foobar";
+    public static final String BAR_PATTERN_LOWER = "bar";
+    public static final String BAR_PATTERN_UPPER = "BAR";
+    public static final String FOOBAR_PATTERN_SPACED = "foo\nbar";
+
     @Test
     public void testDotAll() {
-        Pattern p = JexlPatternCache.getPattern(".*word.*");
-        assertTrue(p.matcher("bla word bla").matches());
-        assertTrue(p.matcher("bla\nbla word bla\n bla").matches());
-        p = JexlPatternCache.getPattern("(\\s|.)*word(\\s|.)*");
-        assertTrue(p.matcher("bla\nbla word bla\n bla").matches());
+        Pattern p = JexlPatternCache.getPattern(ASTERISK_WORD);
+        assertTrue(p.matcher(BLA_WORD_BLA).matches());
+        assertTrue(p.matcher(BLA_PATTERN_2).matches());
+        p = JexlPatternCache.getPattern(BLA_PATTERN_3);
+        assertTrue(p.matcher(BLA_PATTERN_2).matches());
     }
 
     /**
      * Verify that {@link JexlPatternCache#getPattern(String)} will return a new {@link Pattern} that has case-insensitive and multiline matching.
      */
     @Test
     public void testRetrievingNewPattern() {
-        Pattern pattern = JexlPatternCache.getPattern("bar");
-        assertFalse(pattern.matcher("foobar").matches());
-        assertTrue(pattern.matcher("bar").matches());
-        assertTrue(pattern.matcher("BAR").matches());
-        assertTrue(pattern.matcher("foo\nbar").find());
+        Pattern pattern = JexlPatternCache.getPattern(BAR_PATTERN_LOWER);
+        assertFalse(pattern.matcher(FOOBAR_PATTERN).matches());
+        assertTrue(pattern.matcher(BAR_PATTERN_LOWER).matches());
+        assertTrue(pattern.matcher(BAR_PATTERN_UPPER).matches());
+        assertTrue(pattern.matcher(FOOBAR_PATTERN_SPACED).find());
     }
 
     /**