View Javadoc
1   /*
2    * #%L
3    * wcm.io
4    * %%
5    * Copyright (C) 2014 wcm.io
6    * %%
7    * Licensed under the Apache License, Version 2.0 (the "License");
8    * you may not use this file except in compliance with the License.
9    * You may obtain a copy of the License at
10   *
11   *      http://www.apache.org/licenses/LICENSE-2.0
12   *
13   * Unless required by applicable law or agreed to in writing, software
14   * distributed under the License is distributed on an "AS IS" BASIS,
15   * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
16   * See the License for the specific language governing permissions and
17   * limitations under the License.
18   * #L%
19   */
20  package io.wcm.handler.url.impl;
21  
22  import java.net.URI;
23  import java.net.URISyntaxException;
24  import java.util.regex.Matcher;
25  import java.util.regex.Pattern;
26  
27  import org.apache.commons.lang3.StringUtils;
28  import org.apache.commons.lang3.Strings;
29  import org.apache.sling.api.SlingHttpServletRequest;
30  import org.apache.sling.api.resource.ResourceResolver;
31  import org.jetbrains.annotations.NotNull;
32  import org.jetbrains.annotations.Nullable;
33  
34  import io.wcm.sling.commons.util.Escape;
35  
36  /**
37   * Utility methods for externalizing URLs.
38   */
39  final class Externalizer {
40  
41    private Externalizer() {
42      // static util methods only
43    }
44  
45    /**
46     * Externalizes an URL by applying Sling Mapping. Hostname and scheme are not added because they are added by the
47     * link handler depending on site URL configuration and secure/non-secure mode. URLs that are already externalized
48     * remain untouched.
49     * @param url Unexternalized URL (without scheme or hostname)
50     * @param resolver Resource resolver
51     * @param request Request
52     * @return Exernalized URL without scheme or hostname, but with short URLs (if configured in Sling Mapping is
53     *         configured), and the path is URL-encoded if it contains special chars.
54     */
55    public static @Nullable String externalizeUrl(@NotNull String url, @NotNull ResourceResolver resolver, @Nullable SlingHttpServletRequest request) {
56      return externalizeUrlWithSlingMapping(url, resolver, request, false);
57    }
58  
59    /**
60     * Externalizes a URL by applying Sling Mapping. Hostname and scheme will be added. URLs that are already externalized
61     * remain untouched.
62     * @param url non-externalized URL (without scheme or hostname)
63     * @param resolver Resource resolver
64     * @param request Request
65     * @return Externalized URL with scheme or hostname, short URLs (if configured in Sling Mapping),
66     *         and the path is URL-encoded if it contains special chars.
67     */
68    public static @Nullable String externalizeUrlWithHost(@NotNull String url, @NotNull ResourceResolver resolver, @Nullable SlingHttpServletRequest request) {
69      return externalizeUrlWithSlingMapping(url, resolver, request, true);
70    }
71  
72    @SuppressWarnings("java:S112") // allow runtime exception
73    private static @Nullable String externalizeUrlWithSlingMapping(@NotNull String url, @NotNull ResourceResolver resolver,
74        @Nullable SlingHttpServletRequest request, boolean keepHost) {
75  
76      // apply externalization only path part
77      String path = url;
78  
79      // split off query string or fragment that may be appended to the URL
80      String urlRemainder = null;
81      int urlRemainderPos = StringUtils.indexOfAny(path, '?', '#');
82      if (urlRemainderPos >= 0) {
83        urlRemainder = path.substring(urlRemainderPos);
84        path = path.substring(0, urlRemainderPos);
85      }
86  
87      // apply reverse mapping based on current sling mapping configuration for current request
88      // e.g. to support a host-based prefix stripping mapping configuration configured at /etc/map
89  
90      // please note: the sling map method does a lot of things:
91      // 1. applies reverse mapping depending on the sling mapping configuration
92      // (this can even add a hostname if defined in sling mapping configuration)
93      // 2. applies namespace mangling (e.g. replace jcr: with _jcr_)
94      // 3. adds webapp context path if required
95      // 4. url-encodes the whole url
96      if (request != null) {
97        path = resolver.map(request, path);
98      }
99      else {
100       path = resolver.map(path);
101     }
102 
103     if (!keepHost) {
104       // remove scheme and hostname (probably added by sling mapping), but leave path in escaped form
105       try {
106         path = new URI(path).getRawPath();
107         // replace %2F back to / for better readability
108         path = Strings.CS.replace(path, "%2F", "/");
109       }
110       catch (URISyntaxException ex) {
111         throw new RuntimeException("Sling map method returned invalid URI: " + path, ex);
112       }
113     }
114 
115     // build full URL again
116     if (path == null) {
117       return null;
118     }
119     else {
120       return path + (urlRemainder != null ? urlRemainder : "");
121     }
122   }
123 
124   /**
125    * Externalizes an URL without applying Sling Mapping. Instead the servlet context path is added and sling namespace
126    * mangling is applied manually.
127    * Hostname and scheme are not added because they are added by the link handler depending on site URL configuration
128    * and secure/non-secure mode. URLs that are already externalized remain untouched.
129    * @param url Unexternalized URL (without scheme or hostname)
130    * @param request Request
131    * @return Exernalized URL without scheme or hostname, the path is URL-encoded if it contains special chars.
132    */
133   public static @NotNull String externalizeUrlWithoutMapping(@NotNull String url, @Nullable SlingHttpServletRequest request) {
134 
135     // apply externalization only path part
136     String path = url;
137 
138     // split off query string or fragment that may be appended to the URL
139     String urlRemainder = null;
140     int urlRemainderPos = StringUtils.indexOfAny(path, '?', '#');
141     if (urlRemainderPos >= 0) {
142       urlRemainder = path.substring(urlRemainderPos);
143       path = path.substring(0, urlRemainderPos);
144     }
145 
146     // apply namespace mangling (e.g. replace jcr: with _jcr_)
147     path = mangleNamespaces(path);
148 
149     // add webapp context path
150     if (request != null) {
151       path = StringUtils.defaultString(request.getContextPath()) + path; //NOPMD
152     }
153 
154     // url-encode path
155     path = Escape.urlEncode(path);
156     path = Strings.CS.replace(path, "+", "%20");
157     // replace %2F back to / for better readability
158     path = Strings.CS.replace(path, "%2F", "/");
159 
160     // build full URL again
161     return path + (urlRemainder != null ? urlRemainder : "");
162   }
163 
164   /*
165    * Detect as externalized:
166    * - everything staring with protocol and a colon is handled as externalized (http:, tel:, mailto:, javascript: etc.)
167    * - everything starting with // or # is handles as exteranlized
168    * - all other strings handles as not externalized
169    */
170   private static final Pattern EXTERNALIZED_PATTERN = Pattern.compile("^([^/]+:|//|#).+?");
171 
172   /**
173    * Checks if the given URL is already externalized.
174    * For this check some heuristics are applied.
175    * @param url URL
176    * @return true if path is already externalized.
177    */
178   public static boolean isExternalized(@NotNull String url) {
179     return EXTERNALIZED_PATTERN.matcher(url).matches();
180   }
181 
182   /**
183    * Checks if the given URL can be externalize, that means seems to be an content path that needs externalization.
184    * @param url URL
185    * @return true if url seems to be a path than needs externaliziation
186    */
187   public static boolean isExternalizable(@NotNull String url) {
188     return Strings.CS.startsWith(url, "/");
189   }
190 
191   private static final String MANGLED_NAMESPACE_PREFIX = "/_";
192   private static final String MANGLED_NAMESPACE_SUFFIX = "_";
193   private static final char NAMESPACE_SEPARATOR = ':';
194   private static final Pattern NAMESPACE_PATTERN = Pattern.compile("/([^:/]+):");
195 
196   /**
197    * Mangle the namespaces in the given path for usage in sling-based URLs.
198    *
199    * <p>
200    * Example: /path/jcr:content to /path/_jcr_content
201    * </p>
202    *
203    * @param path Path to mangle
204    * @return Mangled path
205    */
206   public static @NotNull String mangleNamespaces(@NotNull String path) {
207     if (!StringUtils.contains(path, NAMESPACE_SEPARATOR)) {
208       return path;
209     }
210     Matcher matcher = NAMESPACE_PATTERN.matcher(path);
211     StringBuffer sb = new StringBuffer();
212     while (matcher.find()) {
213       String replacement = MANGLED_NAMESPACE_PREFIX + matcher.group(1) + MANGLED_NAMESPACE_SUFFIX;
214       matcher.appendReplacement(sb, replacement);
215     }
216     matcher.appendTail(sb);
217     return sb.toString();
218   }
219 
220 }