public DocumentAffiliation(String id, String index, String rawText) {
this.id = id;
this.index = ContentCleaner.clean(index);
if (rawText.matches(".+ \\([^\\(\\)]*\\)$")) {
rawText = StringUtils.reverse(rawText).replaceFirst("\\)", "").replaceFirst("\\( ", " ,");
rawText = StringUtils.reverse(rawText);
}
this.rawText = MetadataTools.cleanAndNormalize(rawText);
this.tokens = new ArrayList<Token<AffiliationLabel>>();
}
/**
* parse IPV6 ARPA domain to IP bytes.
*
* @param name an ARPA string.
* @return ARPA.
*/
private static NetworkInBytes parseIpV6Arpa(String name) {
LOGGER.debug("parseIp6Arpa, name:" + name);
String arpa = StringUtils.removeEndIgnoreCase(name, CHAR_DOT + DomainUtil.IPV6_ARPA_SUFFIX);
arpa = StringUtils.remove(arpa, CHAR_DOT);
String ip = StringUtils.reverse(arpa);
String fullIp = StringUtils.rightPad(ip, 32, '0');
byte[] startIpBytes = DatatypeConverter.parseHexBinary(fullIp);
int networkMask = StringUtils.length(arpa) * 4;
IPv6Address fromByteArray = IPv6Address.fromByteArray(startIpBytes);
IPv6Network network =
IPv6Network.fromAddressAndMask(
fromByteArray, IPv6NetworkMask.fromPrefixLength(networkMask));
NetworkInBytes result =
new NetworkInBytes(
IpVersion.V6, network.getFirst().toByteArray(), network.getLast().toByteArray());
return result;
}
private String getHash(String data) {
String reversedHash = StringUtils.reverse(data);
String reversedHashMd5 = cryptoUtils.getDigest(reversedHash);
return reversedHashMd5;
}
public static void identifyDuplicateJars(
List<File> jarFileListInDistribution,
File distributionVersion,
HashSet<String> distributionDuplicateJarList,
ArrayList<String> unidentifiedVersionJars) {
Iterator<File> itJarList = jarFileListInDistribution.iterator();
ArrayList<String> tempArr = new ArrayList<String>();
ArrayList<File> pathListForAddedJarToJarVersions = new ArrayList<File>();
HashMap<String, String> jarVersions = new HashMap<String, String>();
StringBuilder builder = new StringBuilder();
Pattern numeric = Pattern.compile("[^0-9_.-]");
Pattern nonNumeric = Pattern.compile("[a-zA-Z]");
while (itJarList.hasNext()) {
File jarFilePath = itJarList.next();
String jarName = (jarFilePath).getName();
if (!jarFilePath
.getAbsolutePath()
.contains(
distributionVersion.getName().replaceAll(".zip", "")
+ File.separator
+ TEMP_DIRECTORY
+ File.separator)) {
for (int letter = jarName.length() - 1; letter >= 0; letter--) {
char singleChar = jarName.charAt(letter);
Matcher matcher = numeric.matcher(Character.toString(singleChar));
// Find all matches
if (!matcher.find()) {
// Get the matching string
builder.append(singleChar);
} else if (nonNumeric.matcher(Character.toString(singleChar)).find()) {
if (builder.length() > 1) {
tempArr.add(builder.toString());
builder.setLength(0);
} else {
builder.setLength(0);
}
}
}
int max;
int previousMax = 0;
String[] version = new String[1];
for (String element : tempArr) {
max = element.length();
if (max > previousMax) {
previousMax = max;
version[0] = element;
}
}
tempArr.clear();
if (version[0] != null) {
String jar = jarName.split((StringUtils.reverse(version[0])))[0];
if (jar.length() >= 2) {
if (jarVersions.containsKey(jar)) {
if (!jarVersions.get(jar).equals(jarName.split(jar)[1])) {
// removing patches - plugins duplication
if (distributionDuplicateJarList.toString().contains(jarName)) {
for (String itemDistributionDuplicateJarList : distributionDuplicateJarList) {
if (itemDistributionDuplicateJarList.contains(jarName)
&& (itemDistributionDuplicateJarList.contains("patches")
|| itemDistributionDuplicateJarList.contains("plugins"))) {
if (!(jarFilePath.getAbsolutePath().contains("patches")
|| jarFilePath.getAbsolutePath().contains("plugins"))) {
distributionDuplicateJarList.add(jarFilePath.getAbsolutePath());
}
}
}
} else {
distributionDuplicateJarList.add(jarFilePath.getAbsolutePath());
}
for (File pathListForAddedJarToJarVersion : pathListForAddedJarToJarVersions) {
String path = pathListForAddedJarToJarVersion.toString();
if (path.contains(jar + jarVersions.get(jar))) {
distributionDuplicateJarList.add(path);
break;
}
}
}
} else {
jarVersions.put(jar, jarName.split(jar)[1]);
pathListForAddedJarToJarVersions.add(jarFilePath);
}
} else {
log.info("Unable to identify the version " + jar);
unidentifiedVersionJars.add(jarFilePath.getAbsolutePath());
}
} else {
jarVersions.put(jarName, null);
pathListForAddedJarToJarVersions.add(jarFilePath);
}
}
}
}