Debt/met 5132 dates normalization cleanup part 4 (#622)

* MET-5132: Update naming of numeric spaces variant

* MET-5132: Prepare first sample of tests

* MET-5132: More tests

* MET-5132: Restructure with enum MontNameDateExtractor

* MET-5132: Reusable DatePartsIndices

* MET-5132: Cleanup

* MET-5132: Add spaces clean and trim

* MET-5132: Cleanup

* MET-5132: First split the range part

* MET-5132: Refactor PatternBcAdDateExtractor with tests

* MET-5132: PatternBcAdRangeDateExtractor cleanup

* MET-5132: Reuse range code for dates

* MET-5132: Reuse range code for BriefRangeDateExtractor

* MET-5132: Simplify hierarchy for ranges

* MET-5132: Add sample tests for PatternLongNegativeYearDateExtractor and adapt edtf builder

* MET-5132: Refactor LongNegativeYearDateExtractor with a separate range class reusing already existent code

* MET-5132: Split century extraction to numeric, roman, roman range reusing code

* MET-5132: Centralize sanitization operation for all extractors.

* MET-5132: Add tests for generic properties

* MET-5132: Reuse test code

* MET-5132: Repackage

* MET-5132: Split EdtfDateExtractor to handle ranges separately with code reuse

* MET-5132: Centralize date qualification overwriting

* MET-5132: Simplify date qualification overwriting

* MET-5132: Update code after answers from rnd

* MET-5132: Cleanup

* MET-5132: PatternFormattedFullDateDateExtractor cleanup

* MET-5132: Add millisecond support

* MET-5132: Process review

* MET-5132: Process review 2

metis-normalization/src/main/java/eu/europeana/normalization/dates/DateNormalizationExtractorMatchId.java

@@ -19,7 +19,7 @@ public enum DateNormalizationExtractorMatchId {
   NUMERIC_ALL_VARIANTS_XX("numeric date (various separators and unknown parts)"),
   NUMERIC_RANGE_ALL_VARIANTS("numeric date interval (various separators)"),
   NUMERIC_RANGE_ALL_VARIANTS_XX("numeric date interval (various separators and unknown parts)"),
-  YYYY_MM_DD_SPACES("numeric date (whitespace separators)");
+  NUMERIC_SPACES_VARIANT("numeric date (whitespace separators)");
 
   final String label;
 

metis-normalization/src/main/java/eu/europeana/normalization/dates/edtf/AbstractEdtfDate.java

@@ -1,5 +1,7 @@
 package eu.europeana.normalization.dates.edtf;
 
+import java.util.Set;
+
 /**
  * An abstract class that contains the template that an EDTF date with compliance level 1 should implement.
  * <p>See more in the specification of <a href="https://www.loc.gov/standards/datetime/">EDTF</a></p>
@@ -17,11 +19,18 @@ protected AbstractEdtfDate(String label) {
     this.label = label;
   }
 
+  /**
+   * Add the date qualification, mainly used for pre-sanitized values.
+   *
+   * @param dateQualification the date qualification
+   */
+  public abstract void addQualification(DateQualification dateQualification);
+
   public String getLabel() {
     return label;
   }
 
-  public abstract DateQualification getDateQualification();
+  public abstract Set<DateQualification> getDateQualifications();
 
   public abstract boolean isOpen();
 

metis-normalization/src/main/java/eu/europeana/normalization/dates/edtf/DateQualification.java

@@ -1,39 +1,59 @@
 package eu.europeana.normalization.dates.edtf;
 
-import java.util.Arrays;
+import java.util.EnumSet;
+import java.util.Set;
 import java.util.regex.Pattern;
 
 /**
  * Date qualification characters according to <a href="https://www.loc.gov/standards/datetime/">Extended Date/Time Format (EDTF)
  * Specification</a>
  */
 public enum DateQualification {
+  UNCERTAIN, APPROXIMATE;
 
-  NO_QUALIFICATION(""),
-  UNCERTAIN("?"),
-  APPROXIMATE("~"),
-  UNCERTAIN_APPROXIMATE("%");
-
-  public static final Pattern CHECK_QUALIFICATION_PATTERN = Pattern.compile("^[^\\?~%]*([\\?~%]?)$");
-  private final String character;
-
-  DateQualification(String character) {
-    this.character = character;
-  }
+  private static final String UNCERTAIN_CHARACTER = "?";
+  private static final String APPROXIMATE_CHARACTER = "~";
+  private static final String UNCERTAIN_APPROXIMATE_CHARACTER = "%";
+  private static final String CHARACTERS_REGEX = UNCERTAIN_CHARACTER + APPROXIMATE_CHARACTER + UNCERTAIN_APPROXIMATE_CHARACTER;
+  public static final Pattern PATTERN = Pattern.compile("^[^" + CHARACTERS_REGEX + "]*([" + CHARACTERS_REGEX + "])$");
 
   /**
-   * Get the enum value based on the character provided.
-   * <p>It will return a matched enum value or {@link #NO_QUALIFICATION}.</p>
+   * Get the enum values based on the character provided.
+   * <p>It will return an empty set or the set with the applicable qualifications.</p>
    *
    * @param character the provided character
    * @return the enum value
    */
-  public static DateQualification fromCharacter(String character) {
-    return Arrays.stream(DateQualification.values()).filter(value -> value.character.equals(character)).findFirst().orElse(
-        NO_QUALIFICATION);
+  public static Set<DateQualification> fromCharacter(String character) {
+    final Set<DateQualification> dateQualifications = EnumSet.noneOf(DateQualification.class);
+    if (UNCERTAIN_APPROXIMATE_CHARACTER.equals(character)) {
+      dateQualifications.add(DateQualification.UNCERTAIN);
+      dateQualifications.add(DateQualification.APPROXIMATE);
+    } else if (UNCERTAIN_CHARACTER.equals(character)) {
+      dateQualifications.add(DateQualification.UNCERTAIN);
+    } else if (APPROXIMATE_CHARACTER.equals(character)) {
+      dateQualifications.add(DateQualification.APPROXIMATE);
+    }
+    return dateQualifications;
   }
 
-  public String getCharacter() {
+  /**
+   * Get the string representation based on the provided date qualifications.
+   *
+   * @param dateQualifications the date qualifications
+   * @return the string representation
+   */
+  public static String getCharacterFromQualifications(Set<DateQualification> dateQualifications) {
+    final String character;
+    if (dateQualifications.contains(UNCERTAIN) && dateQualifications.contains(APPROXIMATE)) {
+      character = UNCERTAIN_APPROXIMATE_CHARACTER;
+    } else if (dateQualifications.contains(UNCERTAIN)) {
+      character = UNCERTAIN_CHARACTER;
+    } else if (dateQualifications.contains(APPROXIMATE)) {
+      character = APPROXIMATE_CHARACTER;
+    } else {
+      character = "";
+    }
     return character;
   }
 }

metis-normalization/src/main/java/eu/europeana/normalization/dates/edtf/InstantEdtfDate.java

@@ -3,7 +3,6 @@
 import static eu.europeana.normalization.dates.edtf.DateBoundaryType.DECLARED;
 import static eu.europeana.normalization.dates.edtf.DateBoundaryType.OPEN;
 import static eu.europeana.normalization.dates.edtf.DateBoundaryType.UNKNOWN;
-import static eu.europeana.normalization.dates.edtf.DateQualification.NO_QUALIFICATION;
 import static eu.europeana.normalization.dates.edtf.InstantEdtfDateBuilder.THRESHOLD_4_DIGITS_YEAR;
 import static eu.europeana.normalization.dates.edtf.Iso8601Parser.ISO_8601_MINIMUM_YEAR_DIGITS;
 import static java.lang.Math.abs;
@@ -19,7 +18,9 @@
 import java.time.Year;
 import java.time.YearMonth;
 import java.time.temporal.TemporalAccessor;
+import java.util.EnumSet;
 import java.util.Objects;
+import java.util.Set;
 import org.slf4j.Logger;
 import org.slf4j.LoggerFactory;
 
@@ -38,11 +39,13 @@ public final class InstantEdtfDate extends AbstractEdtfDate implements Comparabl
   private Month month;
   private LocalDate yearMonthDay;
   private YearPrecision yearPrecision;
-  private DateQualification dateQualification = NO_QUALIFICATION;
+  private Set<DateQualification> dateQualifications = EnumSet.noneOf(DateQualification.class);
   private DateBoundaryType dateBoundaryType = DECLARED;
 
   /**
    * Restricted constructor by provided {@link InstantEdtfDateBuilder}.
+   * <p>All fields apart from {@link #dateQualifications} are strictly contained in the constructor. The date qualifications can
+   * be further extended to, for example, add an approximate qualification for a date that was sanitized.</p>
    *
    * @param instantEdtfDateBuilder the builder with all content verified
    */
@@ -51,13 +54,18 @@ public final class InstantEdtfDate extends AbstractEdtfDate implements Comparabl
     year = instantEdtfDateBuilder.getYearObj();
     month = instantEdtfDateBuilder.getMonthObj();
     yearMonthDay = instantEdtfDateBuilder.getYearMonthDayObj();
-    dateQualification = instantEdtfDateBuilder.getDateQualification();
+    dateQualifications = instantEdtfDateBuilder.getDateQualifications();
   }
 
   private InstantEdtfDate(DateBoundaryType dateBoundaryType) {
     this.dateBoundaryType = dateBoundaryType;
   }
 
+  @Override
+  public void addQualification(DateQualification dateQualification) {
+    this.dateQualifications.add(dateQualification);
+  }
+
   /**
    * Create an {@link DateBoundaryType#UNKNOWN} instant.
    *
@@ -188,7 +196,7 @@ public Integer getCentury() {
     int centuryDivision = year.getValue() / YearPrecision.CENTURY.getDuration();
     int centuryModulo = year.getValue() % YearPrecision.CENTURY.getDuration();
     //For case 1900 it is 19th. For case 1901 it is 20th century
-    return centuryModulo == 0 ? centuryDivision : centuryDivision + 1;
+    return (centuryModulo == 0) ? centuryDivision : (centuryDivision + 1);
   }
 
   /**
@@ -230,7 +238,7 @@ public String toString() {
       stringBuilder.append(
           ofNullable(yearMonthDay).map(LocalDate::getDayOfMonth).map(decimalFormat::format).map(d -> "-" + d).orElse(""));
     }
-    stringBuilder.append(dateQualification.getCharacter());
+    stringBuilder.append(DateQualification.getCharacterFromQualifications(dateQualifications));
     return stringBuilder.toString();
   }
 
@@ -256,13 +264,13 @@ public boolean equals(Object o) {
     }
     InstantEdtfDate that = (InstantEdtfDate) o;
     return yearPrecision == that.yearPrecision && Objects.equals(year, that.year) && Objects.equals(month,
-        that.month) && Objects.equals(yearMonthDay, that.yearMonthDay) && dateQualification == that.dateQualification
+        that.month) && Objects.equals(yearMonthDay, that.yearMonthDay) && dateQualifications == that.dateQualifications
         && dateBoundaryType == that.dateBoundaryType;
   }
 
   @Override
   public int hashCode() {
-    return Objects.hash(yearPrecision, year, month, yearMonthDay, dateQualification, dateBoundaryType);
+    return Objects.hash(yearPrecision, year, month, yearMonthDay, dateQualifications, dateBoundaryType);
   }
 
   public Year getYear() {
@@ -281,8 +289,8 @@ public YearPrecision getYearPrecision() {
     return yearPrecision;
   }
 
-  public DateQualification getDateQualification() {
-    return dateQualification;
+  public Set<DateQualification> getDateQualifications() {
+    return EnumSet.copyOf(dateQualifications);
   }
 
   public DateBoundaryType getDateBoundaryType() {