Merge pull request #82 from realrolfje/feature/partial-character-anon…

…ymizer-cleanup Feature/partial character anonymizer cleanup
realrolfje · Feb 15, 2020 · 9f6fcf1 · 9f6fcf1
2 parents cd90fcd + 8849702
commit 9f6fcf1
Show file tree

Hide file tree

Showing 16 changed files with 455 additions and 186 deletions.
diff --git a/create_release.sh b/create_release.sh
@@ -37,9 +37,9 @@ git push --follow-tags
 # Sign the zip file
 gpg -ab --default-key 45E2A5E085182DC26EFEF6E796BB2760490D54DD target/anonimatron*.zip
 
-echo "The file to upload for this release is:"
+echo "The files to upload for this release are:"
 echo
-ls -l target/anonimatron*.zip
+ls -l target/anonimatron*.zip*
 echo
 echo "When creating a release, github automatically adds src zip"
 echo "files, you only need to upload the binary."

diff --git a/docs/_config.yml b/docs/_config.yml
@@ -9,6 +9,7 @@ exclude: [ruby, Gemfile, Gemfile.lock, serve_local_jekyll_site.sh]
 github:
   zip_url: https://github.com/realrolfje/anonimatron/releases/latest
   repository_url: https://github.com/realrolfje/anonimatron/tree/master
+  documentation_url: /documentation
 
 plugins:
   - jekyll-mentions

diff --git a/docs/_layouts/default.html b/docs/_layouts/default.html
@@ -22,14 +22,17 @@
   <body>
     <header class="page-header" role="banner">
       <div class="logo">
-          <h1 class="project-name">{{ site.title | default: site.github.repository_name }}</h1>
+        <h1 class="project-name"><a href="/" style="color: inherit">{{ site.title | default: site.github.repository_name }}</a></h1>
           <h2 class="project-tagline">{{ page.description | default: site.description | default: site.github.project_tagline }}</h2>
-        {% if site.github.is_project_page %}
+      {% if site.github.is_project_page %}
         <a href="{{ site.github.repository_url }}" class="btn github">View on GitHub</a>
       {% endif %}
       {% if site.show_downloads %}
         <a href="{{ site.github.zip_url }}" class="btn zip">Download .zip</a>
       {% endif %}
+        {% if site.github.is_project_page %}
+        <a href="{{ site.github.documentation_url }}" class="btn documentation">Documentation</a>
+      {% endif %}
     </div>
   </header>
 

diff --git a/docs/assets/css/book-flat-30.png b/docs/assets/css/book-flat-30.png
diff --git a/docs/assets/css/style.scss b/docs/assets/css/style.scss
@@ -17,6 +17,13 @@
     padding-left: 47px;
 }
 
+.btn.documentation {
+    background-image: url(book-flat-30.png);
+    background-repeat: no-repeat;
+    background-position: 10px;
+    padding-left: 47px;
+}
+
 .logo {
     background-image: url(anonimatron-logo-a.png);
     background-repeat: no-repeat;

diff --git a/docs/documentation/index.md b/docs/documentation/index.md
@@ -0,0 +1,26 @@
+# Anonimatron documentation
+
+For more information on how Anonimatron works and runs, check our [GitHub readme.md](https://github.com/realrolfje/anonimatron/blob/master/README.md).
+
+## Available Anonymizers
+
+Anonimatron comes with the following default anonymizers. Please start Anonimatron with the `-configexample`
+parameter to see how these are configured:
+
+| Name                              | Type                | Input            | Output                                                                   |
+|:----------------------------------|:--------------------|:-----------------|:-------------------------------------------------------------------------|
+| CharacterStringAnonymizer         | RANDOMCHARACTERS    | Any string       | A-Z, same length                                                         |
+| CharacterStringPrefetchAnonymizer | RANDOMCHARACTERS    | Any string       | Characters from all input data, same length                              |
+| CountryCodeAnonymizer             | COUNTRY_CODE        | Any string       | ISO 3166-1 alpha 2 or alpha 3 code                                       |
+| DateAnonymizer                    | DATE                | Valid date       | Date between 31 days before and 32 days after the input date             |
+| DigitStringAnonymizer             | RANDOMDIGITS        | Any string       | 0-9, same length, optional mask                                          | 
+| DutchBankAccountAnononymizer      | DUTCHBANKACCOUNT    | Any string       | 11 proof number, minimal 9 digits                                        | 
+| DutchBSNAnononymizer              | BURGERSERVICENUMMER | Number or string | Valid Dutch "Burger Service Nummer" or "SOFI Nummer" as number or string |
+| DutchZipCodeAnonymizer            | DUTCH_ZIP_CODE      | Any string       | Valid Dutch zip/postal code                                              |
+| ElvenNameGenerator                | ELVEN_NAME          | Any string       | Pronounceable elven name, 2 to 5 syllables                               | 
+| EmailAddressAnonymizer            | EMAIL_ADDRESS       | Any string       | Valid email address in the domain "@example.com"                         |
+| IbanAnonymizer                    | IBAN                | Any string       | Valid International Bank Account Number                                  |
+| RomanNameGenerator                | ROMAN_NAME          | Any string       | Pronounceable Roman name, 2 to 5 syllables                               | 
+| StringAnonymizer                  | STRING              | Any string       | Random hexadecimal string                                                |
+| UkPostCodeAnonymizer              | UK_POST_CODE        | Any string       | Valid Uk Post code                                                       |
+| UUIDAnonymizer                    | UUID                | Any string       | A random UUID                                                            |
diff --git a/resources/documentation/VERSION_INFO.TXT b/resources/documentation/VERSION_INFO.TXT
@@ -1,3 +1,14 @@
+*** Version 1.12
+- Added mask parameter to DigitStringAnonymizer
+  (fixes issue #74, thanks to Balogh Tamás and Bartlomiej Komendarczuk)
+- Added UK Postalcode Anonymizer
+- Optimized Random number generators (less instantiations)
+- Fixed override problem with password, userid and jdbcurl
+  (pull request #69, thanks to Stephan Schrader)
+- Security, performance and readability fixes, as suggested by SonarCube,
+  see https://sonarcloud.io/dashboard?id=realrolfje_anonimatron)
+- Integrated documentation in the main development branch
+
 *** Version 1.11
 - Add numeric BSN support, closes issue #31
 - Ability to provide database connection parameters through

diff --git a/src/main/java/com/rolfje/anonimatron/anonymizer/AbstractElevenProofAnonymizer.java b/src/main/java/com/rolfje/anonimatron/anonymizer/AbstractElevenProofAnonymizer.java
@@ -81,10 +81,10 @@ protected int[] generate11ProofNumber(int numberOfDigits) {
 	}
 
 	protected int[] getRandomDigits(int numberOfDigits) {
-		int[] elevenProof = new int[numberOfDigits];
-		for (int i = 0; i < elevenProof.length; i++) {
-			elevenProof[i] = (int) Math.floor(Math.random() * 10);
+		int[] randomDigits = new int[numberOfDigits];
+		for (int i = 0; i < randomDigits.length; i++) {
+			randomDigits[i] = (int) Math.floor(Math.random() * 10);
 		}
-		return elevenProof;
+		return randomDigits;
 	}
 }
diff --git a/src/main/java/com/rolfje/anonimatron/anonymizer/DigitStringAnonymizer.java b/src/main/java/com/rolfje/anonimatron/anonymizer/DigitStringAnonymizer.java
@@ -3,28 +3,100 @@
 import com.rolfje.anonimatron.synonyms.StringSynonym;
 import com.rolfje.anonimatron.synonyms.Synonym;
 
+import java.util.Map;
+
 public class DigitStringAnonymizer extends AbstractElevenProofAnonymizer {
 
-	@Override
-	public String getType() {
-		return "RANDOMDIGITS";
-	}
+    final static String PARAMETER = "mask";
+
+    @Override
+    public String getType() {
+        return "RANDOMDIGITS";
+    }
+
+    @Override
+    public Synonym anonymize(Object from, int size, boolean shortlived, Map<String, String> parameters) {
+        if (parameters == null || !parameters.containsKey(PARAMETER)) {
+            throw new UnsupportedOperationException("Please provide '" + PARAMETER + "' with a digit mask in the form 111******, where only stars are replaced with random characters.");
+        }
+        return anonymizeMasked(from, size, shortlived, parameters.get(PARAMETER));
+    }
+
+    @Override
+    public Synonym anonymize(Object from, int size, boolean shortlived) {
+		if (from == null) {
+			return new StringSynonym(
+					getType(),
+					null,
+					null,
+					shortlived
+			);
+		}
+
+        int length = from.toString().length();
+
+        int[] digits = getRandomDigits(length);
+        String to = digitsAsString(digits);
+
+        StringSynonym stringSynonym = new StringSynonym(
+                getType(),
+                (String) from,
+                to,
+                shortlived
+        );
 
-	@Override
-	public Synonym anonymize(Object from, int size, boolean shortlived) {
-		int length = from.toString().length();
+        return stringSynonym;
+    }
 
-		int[] digits = getRandomDigits(length);
-		String to = digitsAsString(digits);
+    /**
+     * Anonymizes a string with digits into a new string of digits where only a part
+     * of the original digits are changed. You can use this to mask out a certain part
+     * of the original string, like an area code of a phone number, or the bank number
+     * of a credit card number.
+     * <p>
+     * A mask looks like "111***", where digits remain original, and stars (or other
+     * characters) are replaced with random digits. An example:
+     * <p>
+     * Original number: 555-1234
+     * Mask:            1111**** (note that the '-' is also untouched by placing a digit there)
+     * New number     : 555-9876
+     * <p>
+     * Note that if the mask is shorter than the input string, the output string will
+     * have random digits where there is no mask. In short: only the locations where
+     * there is a digit in the mask are left untouched and will be in the output.
+     *
+     * @param from
+     * @param size
+     * @param shortlived
+     * @param mask
+     * @return
+     */
+    private Synonym anonymizeMasked(Object from, int size, boolean shortlived, String mask) {
+        if (from == null) {
+            return new StringSynonym(
+                    getType(),
+                    null,
+                    null,
+                    shortlived
+            );
+        }
 
-		StringSynonym stringSynonym = new StringSynonym(
-				getType(),
-				(String) from,
-				to,
-				shortlived
-		);
+        char[] fromChars = from.toString().toCharArray();
+        char[] toChars = new char[fromChars.length];
 
-		return stringSynonym;
-	}
+        for (int i = 0; i < fromChars.length; i++) {
+            if (i < mask.length() && Character.isDigit(mask.charAt(i))) {
+                toChars[i] = fromChars[i];
+            } else {
+                toChars[i] = Character.forDigit((int) Math.floor(Math.random() * 10), 10);
+            }
+        }
 
+        return new StringSynonym(
+                getType(),
+                String.valueOf(fromChars),
+                String.valueOf(toChars),
+                shortlived
+        );
+    }
 }
diff --git a/src/main/java/com/rolfje/anonimatron/anonymizer/DutchBSNAnononymizer.java b/src/main/java/com/rolfje/anonimatron/anonymizer/DutchBSNAnononymizer.java
@@ -8,7 +8,7 @@
 import java.math.BigInteger;
 
 /**
- * Generates valid Dutch "Burger Service Nummer" or "SOFI Nummer",a social
+ * Generates valid Dutch "Burger Service Nummer" or "SOFI Nummer", a social
  * security number uniquely identifying civilians to the governement.
  * <p>
  * A BSN Number is 9 digits long, can not start with 3 zeroes, the first digit

diff --git a/src/main/java/com/rolfje/anonimatron/anonymizer/PartialCharacterAnonymizer.java b/src/main/java/com/rolfje/anonimatron/anonymizer/PartialCharacterAnonymizer.java