org.apache.cxf.common.util.URIParserUtil Maven / Gradle / Ivy
Go to download
Show more of this group Show more artifacts with this name
Show all versions of cxf-bundle-minimal Show documentation
Show all versions of cxf-bundle-minimal Show documentation
Apache CXF Minimal Bundle Jar
The newest version!
/**
* Licensed to the Apache Software Foundation (ASF) under one
* or more contributor license agreements. See the NOTICE file
* distributed with this work for additional information
* regarding copyright ownership. The ASF licenses this file
* to you under the Apache License, Version 2.0 (the
* "License"); you may not use this file except in compliance
* with the License. You may obtain a copy of the License at
*
* http://www.apache.org/licenses/LICENSE-2.0
*
* Unless required by applicable law or agreed to in writing,
* software distributed under the License is distributed on an
* "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY
* KIND, either express or implied. See the License for the
* specific language governing permissions and limitations
* under the License.
*/
package org.apache.cxf.common.util;
import java.io.File;
import java.io.UnsupportedEncodingException;
import java.net.MalformedURLException;
import java.net.URI;
import java.net.URISyntaxException;
import java.net.URL;
import java.util.ArrayList;
import java.util.Arrays;
import java.util.HashSet;
import java.util.List;
import java.util.Set;
import java.util.StringTokenizer;
import org.apache.cxf.common.classloader.ClassLoaderUtils;
public final class URIParserUtil {
private static final Set KEYWORDS = new HashSet(Arrays
.asList(new String[] {"abstract", "boolean", "break", "byte", "case", "catch", "char", "class",
"const", "continue", "default", "do", "double", "else", "extends", "final",
"finally", "float", "for", "goto", "if", "implements", "import", "instanceof",
"int", "interface", "long", "native", "new", "package", "private", "protected",
"public", "return", "short", "static", "strictfp", "super", "switch",
"synchronized", "this", "throw", "throws", "transient", "try", "void",
"volatile", "while", "true", "false", "null", "assert", "enum"}));
private static final String EXCLUDED_CHARS = "<>\"{}|\\^`";
private static final String HEX_DIGITS = "0123456789abcdef";
private URIParserUtil() {
// complete
}
private static boolean isExcluded(char ch) {
return ch <= 0x20 || ch >= 0x7F || EXCLUDED_CHARS.indexOf(ch) != -1;
}
public static URL[] pathToURLs(String path) {
StringTokenizer st = new StringTokenizer(path, File.pathSeparator);
URL[] urls = new URL[st.countTokens()];
int count = 0;
while (st.hasMoreTokens()) {
File file = new File(st.nextToken());
URL url = null;
try {
url = file.toURI().toURL();
} catch (MalformedURLException e) {
e.printStackTrace();
}
if (url != null) {
urls[count++] = url;
}
}
if (urls.length != count) {
URL[] tmp = new URL[count];
System.arraycopy(urls, 0, tmp, 0, count);
urls = tmp;
}
return urls;
}
public static String parsePackageName(String namespace, String defaultPackageName) {
String packageName = (defaultPackageName != null && defaultPackageName.trim().length() > 0)
? defaultPackageName : null;
if (packageName == null) {
packageName = getPackageName(namespace);
}
return packageName;
}
public static String getPackageName(String nameSpaceURI) {
int idx = nameSpaceURI.indexOf(':');
String scheme = "";
if (idx >= 0) {
scheme = nameSpaceURI.substring(0, idx);
if ("http".equalsIgnoreCase(scheme) || "urn".equalsIgnoreCase(scheme)) {
nameSpaceURI = nameSpaceURI.substring(idx + 1);
}
}
List tokens = tokenize(nameSpaceURI, "/: ");
if (tokens.size() == 0) {
return "cxf";
}
if (tokens.size() > 1) {
String lastToken = tokens.get(tokens.size() - 1);
idx = lastToken.lastIndexOf('.');
if (idx > 0) {
//lastToken = lastToken.substring(0, idx);
lastToken = lastToken.replace('.', '_');
tokens.set(tokens.size() - 1, lastToken);
}
}
String domain = tokens.get(0);
idx = domain.indexOf(':');
if (idx >= 0) {
domain = domain.substring(0, idx);
}
List r = reverse(tokenize(domain, "urn".equals(scheme) ? ".-" : "."));
if ("www".equalsIgnoreCase(r.get(r.size() - 1))) {
// remove leading www
r.remove(r.size() - 1);
}
// replace the domain name with tokenized items
tokens.addAll(1, r);
tokens.remove(0);
// iterate through the tokens and apply xml->java name algorithm
for (int i = 0; i < tokens.size(); i++) {
// get the token and remove illegal chars
String token = tokens.get(i);
token = removeIllegalIdentifierChars(token);
// this will check for reserved keywords
if (containsReservedKeywords(token)) {
token = '_' + token;
}
tokens.set(i, token.toLowerCase());
}
// concat all the pieces and return it
return combine(tokens, '.');
}
public static String getNamespace(String packageName) {
if (packageName == null || packageName.length() == 0) {
return null;
}
StringTokenizer tokenizer = new StringTokenizer(packageName, ".");
String[] tokens;
if (tokenizer.countTokens() == 0) {
tokens = new String[0];
} else {
tokens = new String[tokenizer.countTokens()];
for (int i = tokenizer.countTokens() - 1; i >= 0; i--) {
tokens[i] = tokenizer.nextToken();
}
}
StringBuilder namespace = new StringBuilder("http://");
String dot = "";
for (int i = 0; i < tokens.length; i++) {
if (i == 1) {
dot = ".";
}
namespace.append(dot + tokens[i]);
}
namespace.append('/');
return namespace.toString();
}
private static List tokenize(String str, String sep) {
StringTokenizer tokens = new StringTokenizer(str, sep);
List r = new ArrayList();
while (tokens.hasMoreTokens()) {
r.add(tokens.nextToken());
}
return r;
}
private static String removeIllegalIdentifierChars(String token) {
StringBuilder newToken = new StringBuilder();
for (int i = 0; i < token.length(); i++) {
char c = token.charAt(i);
if (i == 0 && !Character.isJavaIdentifierStart(c)) {
// prefix an '_' if the first char is illegal
newToken.append("_" + c);
} else if (!Character.isJavaIdentifierPart(c)) {
// replace the char with an '_' if it is illegal
newToken.append('_');
} else {
// add the legal char
newToken.append(c);
}
}
return newToken.toString();
}
private static String combine(List r, char sep) {
StringBuilder buf = new StringBuilder(r.get(0));
for (int i = 1; i < r.size(); i++) {
buf.append(sep);
buf.append(r.get(i));
}
return buf.toString();
}
private static List reverse(List a) {
List r = new ArrayList();
for (int i = a.size() - 1; i >= 0; i--) {
r.add(a.get(i));
}
return r;
}
public static boolean containsReservedKeywords(String token) {
return KEYWORDS.contains(token);
}
public static String escapeChars(String s) {
StringBuilder b = new StringBuilder(s.length());
for (int x = 0; x < s.length(); x++) {
char ch = s.charAt(x);
if (isExcluded(ch)) {
try {
byte[] bytes = Character.toString(ch).getBytes("UTF-8");
for (int y = 0; y < bytes.length; y++) {
b.append("%");
b.append(HEX_DIGITS.charAt((bytes[y] & 0xFF) >> 4));
b.append(HEX_DIGITS.charAt(bytes[y] & 0x0F));
}
} catch (UnsupportedEncodingException e) {
//should not happen
}
} else {
b.append(ch);
}
}
return b.toString();
}
public static String normalize(final String uri) {
URL url = null;
String result = null;
try {
url = new URL(uri);
result = escapeChars(url.toURI().normalize().toString().replace("\\", "/"));
} catch (MalformedURLException e1) {
try {
if (uri.startsWith("classpath:")) {
url = ClassLoaderUtils.getResource(uri.substring(10), URIParserUtil.class);
return url != null ? url.toExternalForm() : uri;
}
File file = new File(uri);
if (file.exists()) {
return file.toURI().normalize().toString();
}
String f = null;
if (uri.indexOf(":") != -1 && !uri.startsWith("/")) {
f = "file:/" + uri;
} else {
f = "file:" + uri;
}
url = new URL(f);
return escapeChars(url.toString().replace("\\", "/"));
} catch (Exception e2) {
return escapeChars(uri.replace("\\", "/"));
}
} catch (URISyntaxException e) {
result = escapeChars(url.toString().replace("\\", "/"));
}
return result;
}
public static String getAbsoluteURI(final String arg) {
if (arg == null) {
return null;
}
try {
URI uri = new URI(arg);
if ("file".equalsIgnoreCase(uri.getScheme())) {
if (!uri.isOpaque()) {
return uri.normalize().toString();
}
return new File("").toURI().resolve(uri.getPath()).toString();
} else {
return normalize(arg);
}
} catch (Exception e2) {
return normalize(arg);
}
}
public static String relativize(String base, String toBeRelativized) throws URISyntaxException {
if (base == null || toBeRelativized == null) {
return null;
}
return relativize(new URI(base), new URI(toBeRelativized));
}
/**
* This is a custom implementation for doing what URI.relativize(URI uri) should be
* doing but is not actually doing when URI roots do not fully match.
* See http://bugs.java.com/bugdatabase/view_bug.do?bug_id=6226081
*
* @param baseURI The base URI
* @param toBeRelativizedURI The URI to be realivized
* @return The string value of the URI you'd expect to get as result
* of calling baseURI.relativize(toBeRelativizedURI).
* null is returned if the parameters are null or are not
* both absolute or not absolute.
* @throws URISyntaxException
*/
public static String relativize(URI baseURI, URI toBeRelativizedURI) throws URISyntaxException {
if (baseURI == null || toBeRelativizedURI == null) {
return null;
}
if (baseURI.isAbsolute() ^ toBeRelativizedURI.isAbsolute()) {
return null;
}
final String base = baseURI.getSchemeSpecificPart();
final String toBeRelativized = toBeRelativizedURI.getSchemeSpecificPart();
final int l1 = base.length();
final int l2 = toBeRelativized.length();
if (l1 == 0) {
return toBeRelativized;
}
int slashes = 0;
StringBuilder sb = new StringBuilder();
boolean differenceFound = false;
for (int i = 0; i < l1; i++) {
char c = base.charAt(i);
if (i < l2) {
if (!differenceFound && c == toBeRelativized.charAt(i)) {
sb.append(c);
} else {
differenceFound = true;
if (c == '/') {
slashes++;
}
}
} else {
if (c == '/') {
slashes++;
}
}
}
String rResolved = new URI(getRoot(sb.toString())).relativize(new URI(toBeRelativized)).toString();
StringBuilder relativizedPath = new StringBuilder();
for (int i = 0; i < slashes; i++) {
relativizedPath.append("../");
}
relativizedPath.append(rResolved);
return relativizedPath.toString();
}
private static String getRoot(String uri) {
int idx = uri.lastIndexOf('/');
if (idx == uri.length() - 1) {
return uri;
} else if (idx == -1) {
return "";
} else {
return uri.substring(0, idx + 1);
}
}
}