4380 Validation / normalization code for new Correlation Attributes

Core/src/org/sleuthkit/autopsy/centralrepository/datamodel/CorrelationAttributeNormalizer.java

@@ -25,14 +25,19 @@ import org.apache.commons.validator.routines.DomainValidator;
 import org.apache.commons.validator.routines.EmailValidator;
 
 /**
- * Provides functions for normalizing data by attribute type before insertion or querying.
+ * Provides functions for normalizing data by attribute type before insertion or
+ * querying.
  */
 final public class CorrelationAttributeNormalizer {
 
+    //common seperators that may be removed for normalizing
+    private static final String SEPERATORS_REGEX = "[\\s-:]";
+
     /**
      * This is a utility class - no need for constructing or subclassing, etc...
      */
-    private CorrelationAttributeNormalizer() { }
+    private CorrelationAttributeNormalizer() {
+    }
 
     /**
      * Normalize the data. Converts text to lower case, and ensures that the
@@ -64,15 +69,15 @@ final public class CorrelationAttributeNormalizer {
             case CorrelationAttributeInstance.USBID_TYPE_ID:
                 return normalizeUsbId(data);
             case CorrelationAttributeInstance.SSID_TYPE_ID:
-                return data;
+                return verifySsid(data);
             case CorrelationAttributeInstance.MAC_TYPE_ID:
-                return data;
+                return normalizeMac(data);
             case CorrelationAttributeInstance.IMEI_TYPE_ID:
-                return data;
+                return normalizeImei(data);
             case CorrelationAttributeInstance.IMSI_TYPE_ID:
-                return data;
+                return normalizeImsi(data);
             case CorrelationAttributeInstance.ICCID_TYPE_ID:
-                return data;
+                return normalizeIccid(data);
             default:
                 final String errorMessage = String.format(
                         "Validator function not found for attribute type: %s",
@@ -82,8 +87,8 @@ final public class CorrelationAttributeNormalizer {
     }
 
     /**
-     * Validate the data.  Converts text to lower case, and ensures that the
-     * data is a valid string of the format expected given the attributeType.
+     * Validate the data. Converts text to lower case, and ensures that the data
+     * is a valid string of the format expected given the attributeType.
      *
      * @param attributeTypeId correlation type of data
      * @param data            data to normalize
@@ -120,7 +125,8 @@ final public class CorrelationAttributeNormalizer {
     }
 
     /**
-     * Verify there are no slashes or invalid domain name characters (such as '?' or \: ). Normalize to lower case.
+     * Verify there are no slashes or invalid domain name characters (such as
+     * '?' or \: ). Normalize to lower case.
      */
     private static String normalizeDomain(String data) throws CorrelationAttributeNormalizationException {
         DomainValidator validator = DomainValidator.getInstance(true);
@@ -137,7 +143,8 @@ final public class CorrelationAttributeNormalizer {
     }
 
     /**
-     *  Verify that there is an '@' and no invalid characters. Should normalize to lower case.
+     * Verify that there is an '@' and no invalid characters. Should normalize
+     * to lower case.
      */
     private static String normalizeEmail(String data) throws CorrelationAttributeNormalizationException {
         EmailValidator validator = EmailValidator.getInstance(true, true);
@@ -167,4 +174,137 @@ final public class CorrelationAttributeNormalizer {
         //TODO replace with correct usb id validation at a later date
         return data;
     }
+
+    /**
+     * Verify the wireless network name is valid
+     *
+     * SSIDs for wireless networks can be at most 32 characters, are case
+     * sensitive, and allow special characters.
+     *
+     * @param data The string to normalize and validate
+     *
+     * @return the unmodified data if the data was a valid length to be an SSID
+     *
+     * @throws CorrelationAttributeNormalizationException if the data was not a
+     *                                                    valid SSID
+     */
+    private static String verifySsid(String data) throws CorrelationAttributeNormalizationException {
+        if (data.length() <= 32) {
+            return data;
+        } else {
+            throw new CorrelationAttributeNormalizationException("Name provided was longer than the maximum valid SSID (32 characters). Name: " + data);
+        }
+    }
+
+    /**
+     * Verify the ICCID (Integrated Circuit Card Identifier) number and
+     * normalize format.
+     *
+     * E.118 defines as up to 22 digits long including luhn check digit while
+     * GSM Phase 1 defines it as a 20 digit operator specific structure. They
+     * begin with 89 which is the ISO 7812 Major Industry Identifier for
+     * telecommunication, followed by a contry code of 1-3 digits as definted by
+     * ITU-T E.164, followed by issuer identifier 1-4 digits, followed by 1 luhn
+     * checksum digit (sometimes omitted). The hexidecimal digit F is used as
+     * filler when necessary in GSM Phase 1 specification.
+     *
+     * 18 digits appears to be the shortest ICCID in use.
+     *
+     * @param data The string to normalize and validate
+     *
+     * @return the data with common number seperators removed and lower cased if
+     *         the data was determined to be a possible ICCID
+     *
+     * @throws CorrelationAttributeNormalizationException if the data was not a
+     *                                                    valid ICCID
+     */
+    private static String normalizeIccid(String data) throws CorrelationAttributeNormalizationException {
+        final String validIccidRegex = "^([8][9][f0-9]{17,22})$";
+        final String iccidWithoutSeperators = data.toLowerCase().replaceAll(SEPERATORS_REGEX, "");
+        if (iccidWithoutSeperators.matches(validIccidRegex)) {
+            return iccidWithoutSeperators;
+        } else {
+            throw new CorrelationAttributeNormalizationException("Data provided was not a valid ICCID. : " + data);
+        }
+    }
+
+    /**
+     * Verify the IMSI (International mobile subscriber identity) number and
+     * normalize format.
+     *
+     * First 3 digits Mobile Country Code 2-3 digits Mobile Network Code Up to
+     * 10 digits for mobile subscriber identification number MSIN
+     *
+     * Length will be 14 or 15 digits total
+     *
+     * @param data The string to normalize and validate
+     *
+     * @return the data with common number seperators removed if the data was
+     *         determined to be a possible IMSI
+     *
+     * @throws CorrelationAttributeNormalizationException if the data was not a
+     *                                                    valid IMSI
+     */
+    private static String normalizeImsi(String data) throws CorrelationAttributeNormalizationException {
+        final String validImsiRegex = "^[0-9]{14,15}$";
+        final String imsiWithoutSeperators = data.replaceAll(SEPERATORS_REGEX, "");
+        if (imsiWithoutSeperators.matches(validImsiRegex)) {
+            return imsiWithoutSeperators;
+        } else {
+            throw new CorrelationAttributeNormalizationException("Data provided was not a valid Imsi. : " + data);
+        }
+    }
+
+    /**
+     * Verify the MAC (media access control) address and normalize format.
+     *
+     * A 12 or 16 Hexadecimal digits long depending on standard (Possible
+     * standards EUI-48, MAC-48, EUI-64)
+     *
+     * @param data The string to normalize and validate
+     *
+     * @return the data with common number seperators removed and lowercased if
+     *         the data was determined to be a possible MAC
+     *
+     * @throws CorrelationAttributeNormalizationException if the data was not a
+     *                                                    valid MAC
+     */
+    private static String normalizeMac(String data) throws CorrelationAttributeNormalizationException {
+        final String validMacRegex = "^([a-f0-9]{12}|[a-f0-9]{16})$";
+        final String macWithoutSeperators = data.toLowerCase().replaceAll(SEPERATORS_REGEX, "");
+        if (macWithoutSeperators.matches(validMacRegex)) {
+            return macWithoutSeperators;
+        } else {
+            throw new CorrelationAttributeNormalizationException("Data provided was not a valid Imsi. : " + data);
+        }
+    }
+
+    /**
+     * Verify the IMEI (International Mobile Equipment Identity) number and
+     * normalize format.
+     *
+     * 14 to 16 digits digits 1 through 6 are TAC (Type Allocation Code) digits
+     * 7 and 8 are also part of the TAC in phones made in 2003 or later digits 7
+     * and 8 are FAC (Final Assembly Code) in phones made prior to 2003 digits 9
+     * through 14 are the serial number digits 15 and 16 if present represent an
+     * optional luhn checksum (or software version number when dealing with an
+     * IMEI software version)
+     *
+     * @param data The string to normalize and validate
+     *
+     * @return the data with common number seperators removed if the data was
+     *         determined to be a possible IMEI
+     *
+     * @throws CorrelationAttributeNormalizationException if the data was not a
+     *                                                    valid IMEI
+     */
+    private static String normalizeImei(String data) throws CorrelationAttributeNormalizationException {
+        final String validImeiRegex = "^[0-9]{14,16}$";
+        final String imeiWithoutSeperators = data.replaceAll(SEPERATORS_REGEX, "");
+        if (imeiWithoutSeperators.matches(validImeiRegex)) {
+            return imeiWithoutSeperators;
+        } else {
+            throw new CorrelationAttributeNormalizationException("Data provided was not a valid Imsi. : " + data);
+        }
+    }
 }