Merge pull request #34 from realrolfje/feature/transient-synonyms

Feature/transient synonyms

create_release.sh

@@ -34,6 +34,9 @@ git commit -m "Update version to $2"
 
 git push --follow-tags
 
+# Sign the zip file
+gpg -ab --default-key D6584B937338F123 target/anonimatron*.zip
+
 echo "The file to upload for this release is:"
 echo
 ls -l target/anonimatron*.zip

pom.xml

@@ -4,7 +4,7 @@
 
     <groupId>com.rolfje.anonimatron</groupId>
     <artifactId>anonimatron</artifactId>
-    <version>1.9.4-SNAPSHOT</version>
+    <version>1.10.0-SNAPSHOT</version>
     <packaging>jar</packaging>
 
     <name>anonimatron</name>

src/main/java/com/rolfje/anonimatron/anonymizer/AbstractNameGenerator.java

@@ -407,17 +407,13 @@ && hatesPreviousVocals(sur.get(c)) || last == 2
 	}
 
 	@Override
-	public Synonym anonymize(Object from, int size) {
-
-		StringSynonym synonym = new StringSynonym();
-		synonym.setFrom(from);
-
-		if (from != null){
-			synonym.setTo(getName(size));
-		}
-
-		synonym.setType(getType());
-		return synonym;
+	public Synonym anonymize(Object from, int size, boolean shortlived) {
+		return new StringSynonym(
+				getType(),
+				(String) from,
+				from == null ? null : getName(size),
+				shortlived
+		);
 	}
 
 	private String getName(int size) {

src/main/java/com/rolfje/anonimatron/anonymizer/Anonymizer.java

@@ -6,37 +6,37 @@
 
 /**
  * Provides functionality for consitently anonymizing a piece of data.
- * 
+ * <p>
  * Implementations of this interface must make sure that anonymization is done
  * in a reproducable manner. That is, if A transforms into B, it has to
  * consistently do so on each and every call.
- * 
+ * <p>
  * By doing this, anonimatron can guarantee that data is transformed
  * consistently accross all tables of the database, and referential constraints
  * can be re-enforced after anonymization.
  */
 public interface Anonymizer {
 
-	/**
-	 * 
-	 * @return The ID or name of this anonymizer, as used in the XML
-	 *         configuration file. This is generally something along the lines
-	 *         of "LASTNAME" or "UNEVENNUMBER". Please see the output of the
-	 *         -configexample command line option when running Anonimatron.
-	 */
-	String getType();
+    /**
+     * @return The ID or name of this anonymizer, as used in the XML
+     * configuration file. This is generally something along the lines
+     * of "LASTNAME" or "UNEVENNUMBER". Please see the output of the
+     * -configexample command line option when running Anonimatron.
+     */
+    String getType();
 
-	/**
-	 * Anonymizes the given data into a non-tracable, non-reversible synonym,
-	 * and does it consistently, so that A always translates to B.
-	 * 
-	 * @param from
-	 *            the data to be anonymized, usually passed in as a
-	 *            {@link String}, {@link Integer}, {@link Date} or other classes
-	 *            which can be stored in a single JDBC database column.
-	 * @param size the optional maximum size of the generated value
-	 * @return a {@link Synonym}
-	 */
-	Synonym anonymize(Object from, int size);
+    /**
+     * Anonymizes the given data into a non-tracable, non-reversible synonym,
+     * and does it consistently, so that A always translates to B.
+     *
+     * @param from       the data to be anonymized, usually passed in as a
+     *                   {@link String}, {@link Integer}, {@link Date} or other classes
+     *                   which can be stored in a single JDBC database column.
+     * @param size       the optional maximum size of the generated value
+     * @param shortlived indicates that the generated synonym must have the
+     *                   {@link Synonym#isShortLived()} boolean set
+     * @return a {@link Synonym}
+     */
+    Synonym anonymize(Object from, int size, boolean shortlived);
 
 }

src/main/java/com/rolfje/anonimatron/anonymizer/AnonymizerService.java

@@ -1,5 +1,6 @@
 package com.rolfje.anonimatron.anonymizer;
 
+import com.rolfje.anonimatron.configuration.Column;
 import com.rolfje.anonimatron.synonyms.NullSynonym;
 import com.rolfje.anonimatron.synonyms.Synonym;
 import org.apache.log4j.Logger;
@@ -78,24 +79,32 @@ public Set<String> getDefaultAnonymizerTypes() {
 		return Collections.unmodifiableSet(defaultTypeMapping.keySet());
 	}
 
-	public Synonym anonymize(String type, Object from, int size) {
+	public Synonym anonymize(Column column, Object from) {
 		if (from == null) {
-			return new NullSynonym(type);
+			return new NullSynonym(column.getType());
 		}
 
-		// Hash from here.
-		String hashedFrom = new Hasher("secretsalt").base64Hash(from);
-
 		// Find for regular type
-		Synonym synonym = synonymCache.get(type, from);
+		Synonym synonym = getSynonym(column, from);
+
 		if (synonym == null) {
-			// Fallback for default type
-			synonym = synonymCache.get(defaultTypeMapping.get(type), from);
+            Anonymizer anonymizer = getAnonymizer(column.getType());
+            synonym = anonymizer.anonymize(from, column.getSize(), column.isShortLived());
+
+			synonymCache.put(synonym);
+		}
+		return synonym;
+	}
+
+	private Synonym getSynonym(Column c, Object from) {
+		if (c.isShortLived()){
+			return null;
 		}
 
+		Synonym synonym = synonymCache.get(c.getType(), from);
 		if (synonym == null) {
-			synonym = getAnonymizer(type).anonymize(from, size);
-			synonymCache.put(synonym);
+			// Fallback for default type
+			synonym = synonymCache.get(defaultTypeMapping.get(c.getType()), from);
 		}
 		return synonym;
 	}

src/main/java/com/rolfje/anonimatron/anonymizer/CharacterStringAnonymizer.java

@@ -17,29 +17,30 @@ protected String getDefaultCharacterString() {
 		return "ABCDEFGHIJKLMNOPQRSTUVWXYZ";
 	}
 
-
-
 	@Override
 	public String getType() {
 		return "RANDOMCHARACTERS";
 	}
 
 	@Override
-	public Synonym anonymize(Object from, int size) {
-		int length = from.toString().length();
+	public Synonym anonymize(Object from, int size, boolean shortlived) {
+		String fromString = from.toString();
 		Random r = new Random();
 
+		int length = fromString.length();
 		StringBuilder sb = new StringBuilder();
 		for (int i = 0; i < length; i++) {
 			sb.append(CHARS.charAt(r.nextInt(CHARS.length())));
 		}
 
 		String to = sb.toString();
 
-		StringSynonym stringSynonym = new StringSynonym();
-		stringSynonym.setFrom(from);
-		stringSynonym.setType(getType());
-		stringSynonym.setTo(to);
+		StringSynonym stringSynonym = new StringSynonym(
+				getType(),
+				fromString,
+				to,
+				shortlived
+		);
 
 		return stringSynonym;
 	}

src/main/java/com/rolfje/anonimatron/anonymizer/CharacterStringPrefetchAnonymizer.java

@@ -32,12 +32,12 @@ public void prefetch(Object sourceData) {
 	}
 
 	@Override
-	public Synonym anonymize(Object from, int size) {
+	public Synonym anonymize(Object from, int size, boolean shortlived) {
 		if (CHARS.length() < 1) {
 			LOG.warn("No characters were collected during prefetch. Using the default set '" + getDefaultCharacterString() + "'.");
 			CHARS = getDefaultCharacterString();
 		}
-		return super.anonymize(from, size);
+		return super.anonymize(from, size, shortlived);
 	}
 
 	@Override

src/main/java/com/rolfje/anonimatron/anonymizer/CountryCodeAnonymizer.java

@@ -23,7 +23,7 @@ public String getType() {
 	}
 
 	@Override
-	public Synonym anonymize(Object from, int size) {
+	public Synonym anonymize(Object from, int size, boolean shortlived) {
 
 		if (size < 2) {
 			throw new UnsupportedOperationException("Can not produce country codes of one character.");
@@ -50,10 +50,9 @@ else if (size == 2) {
 		return new StringSynonym(
 				getType(),
 				from.toString(),
-				country
+				country,
+				shortlived
 		);
-
-
 	}
 
 	public static String padRight(String s, int n) {

src/main/java/com/rolfje/anonimatron/anonymizer/DateAnonymizer.java

@@ -14,9 +14,10 @@ class DateAnonymizer implements Anonymizer {
 	private Set<Date> generatedDates = new HashSet<Date>();
 
 	@Override
-	public Synonym anonymize(Object from, int size) {
+	public Synonym anonymize(Object from, int size, boolean shortlived) {
 		DateSynonym s = new DateSynonym();
 		s.setType(TYPE);
+		s.setShortlived(shortlived);
 
 		if (from == null) {
 			s.setFrom(null);

src/main/java/com/rolfje/anonimatron/anonymizer/DigitStringAnonymizer.java

@@ -11,16 +11,18 @@ public String getType() {
 	}
 
 	@Override
-	public Synonym anonymize(Object from, int size) {
+	public Synonym anonymize(Object from, int size, boolean shortlived) {
 		int length = from.toString().length();
 
 		int[] digits = getRandomDigits(length);
 		String to = digitsAsNumber(digits);
 
-		StringSynonym stringSynonym = new StringSynonym();
-		stringSynonym.setFrom(from);
-		stringSynonym.setType(getType());
-		stringSynonym.setTo(to);
+		StringSynonym stringSynonym = new StringSynonym(
+				getType(),
+				(String) from,
+				to,
+				shortlived
+		);
 
 		return stringSynonym;
 	}

src/main/java/com/rolfje/anonimatron/anonymizer/DutchBSNAnononymizer.java

@@ -13,47 +13,52 @@
  * See http://nl.wikipedia.org/wiki/Rekeningnummer
  */
 public class DutchBSNAnononymizer extends AbstractElevenProofAnonymizer {
-	private static int LENGTH = 9;
-
-	@Override
-	public String getType() {
-		return "BURGERSERVICENUMMER";
-	}
-
-	@Override
-	public Synonym anonymize(Object from, int size) {
-		if (size < LENGTH) {
-			throw new UnsupportedOperationException(
-					"Can not generate a BSN that fits in a "
-							+ size
-							+ " character string. Must be " + LENGTH + " characters or more.");
-		}
-
-		StringSynonym s = new StringSynonym();
-		s.setType(getType());
-		s.setFrom(from);
-
-		do {
-			// Never generate identical number
-			s.setTo(generateBSN(LENGTH));
-		} while (s.getFrom().equals(s.getTo()));
-
-		return s;
-	}
-
-	String generateBSN(int numberOfDigits) {
-		// Generate random BSN number
-		int[] bsnnumber;
-
-		do {
-			bsnnumber = generate11ProofNumber(numberOfDigits);
-
-			// SOFI numbers can not start with 3 zeroes, left digit digit van
-			// not be >3
-		} while ((bsnnumber[0] > 3) && (0 != (bsnnumber[0] + bsnnumber[1] + bsnnumber[2])));
-
-		// Return the BSN
-		String result = digitsAsNumber(bsnnumber);
-		return result;
-	}
+    private static int LENGTH = 9;
+
+    @Override
+    public String getType() {
+        return "BURGERSERVICENUMMER";
+    }
+
+    @Override
+    public Synonym anonymize(Object from, int size, boolean shortlived) {
+        if (size < LENGTH) {
+            throw new UnsupportedOperationException(
+                    "Can not generate a BSN that fits in a "
+                            + size
+                            + " character string. Must be " + LENGTH + " characters or more.");
+        }
+
+        String fromString = (String) from;
+        String toString = fromString;
+
+        do {
+            // Never generate identical number
+            toString = generateBSN(LENGTH);
+        } while (fromString.equals(toString));
+
+
+        return new StringSynonym(
+                getType(),
+                fromString,
+                toString,
+                shortlived
+        );
+    }
+
+    String generateBSN(int numberOfDigits) {
+        // Generate random BSN number
+        int[] bsnnumber;
+
+        do {
+            bsnnumber = generate11ProofNumber(numberOfDigits);
+
+            // SOFI numbers can not start with 3 zeroes, left digit digit van
+            // not be >3
+        } while ((bsnnumber[0] > 3) && (0 != (bsnnumber[0] + bsnnumber[1] + bsnnumber[2])));
+
+        // Return the BSN
+        String result = digitsAsNumber(bsnnumber);
+        return result;
+    }
 }