Added methods to test strings for urls and to extract important portions of the url

Former-commit-id: be5878b61d1ce226a4d802b431cfa6c0d32252d1
This commit is contained in:
Jeremy Long
2013-06-23 22:38:45 -04:00
parent f3c026f278
commit 6016370515

View File

@@ -0,0 +1,104 @@
/*
* This file is part of Dependency-Check.
*
* Dependency-Check is free software: you can redistribute it and/or modify it
* under the terms of the GNU General Public License as published by the Free
* Software Foundation, either version 3 of the License, or (at your option) any
* later version.
*
* Dependency-Check is distributed in the hope that it will be useful, but
* WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
* FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License for more
* details.
*
* You should have received a copy of the GNU General Public License along with
* Dependency-Check. If not, see http://www.gnu.org/licenses/.
*
* Copyright (c) 2013 Jeremy Long. All Rights Reserved.
*/
package org.owasp.dependencycheck.utils;
import java.net.MalformedURLException;
import java.net.URL;
import java.util.ArrayList;
import java.util.List;
import java.util.regex.Pattern;
/**
*
* @author Jeremy Long (jeremy.long@gmail.com)
*/
public final class UrlStringUtils {
/**
* Private constructor for a utility class.
*/
private UrlStringUtils() {
}
/**
* A regular expression to test if a string contains a URL.
*/
private static final Pattern CONTAINS_URL_TEST = Pattern.compile("^.*(ht|f)tps?://.*$", Pattern.MULTILINE | Pattern.CASE_INSENSITIVE);
/**
* A regular expression to test if a string is a URL.
*/
private static final Pattern IS_URL_TEST = Pattern.compile("^(ht|f)tps?://.*", Pattern.CASE_INSENSITIVE);
/**
* Tests if the text provided contains a URL. This is somewhat limited
* search in that it only looks for (ftp|http|https)://
*
* @param text the text to search
* @return true if the text contains a url, otherwise false
*/
public static boolean containsUrl(String text) {
return CONTAINS_URL_TEST.matcher(text).matches();
}
/**
* Tests if the given text is url.
*
* @param text the string to test
* @return returns true if the text is a url, otherwise false
*/
public static boolean isUrl(String text) {
return IS_URL_TEST.matcher(text).matches();
}
/**
* <p>Takes a URL, in String format, and adds the important parts of the URL
* to a list of strings.</p>
* <p>Example, given the following input:</p>
* <code>"https://www.somedomain.com/path1/path2/file.php?id=439"</code>
* <p>The function would return:</p>
* <code>{"somedomain", "path1", "path2", "file"}</code>
*
* @param text a URL
* @return importantParts a list of the important parts of the URL
* @throws MalformedURLException thrown if the URL is malformed
*/
public static List<String> extractImportantUrlData(String text) throws MalformedURLException {
final ArrayList<String> importantParts = new ArrayList<String>();
final URL url = new URL(text);
final String[] domain = url.getHost().split("\\.");
//add the domain except www and the tld.
for (int i = 0; i < domain.length - 1; i++) {
final String sub = domain[i];
if (!"www".equalsIgnoreCase(sub)) {
importantParts.add(sub);
}
}
final String document = url.getPath();
final String[] pathParts = document.split("[\\//]");
for (int i = 0; i < pathParts.length - 2; i++) {
if (!pathParts[i].isEmpty()) {
importantParts.add(pathParts[i]);
}
}
if (!pathParts[pathParts.length - 1].isEmpty()) {
final String fileNameNoExt = pathParts[pathParts.length - 1].replaceAll("\\..*{0,5}$", "");
importantParts.add(fileNameNoExt);
}
return importantParts;
}
}