001 /**
002 * Licensed to the Apache Software Foundation (ASF) under one
003 * or more contributor license agreements. See the NOTICE file
004 * distributed with this work for additional information
005 * regarding copyright ownership. The ASF licenses this file
006 * to you under the Apache License, Version 2.0 (the
007 * "License"); you may not use this file except in compliance
008 * with the License. You may obtain a copy of the License at
009 *
010 * http://www.apache.org/licenses/LICENSE-2.0
011 *
012 * Unless required by applicable law or agreed to in writing, software
013 * distributed under the License is distributed on an "AS IS" BASIS,
014 * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
015 * See the License for the specific language governing permissions and
016 * limitations under the License.
017 */
018
019 package org.apache.hadoop.fs;
020
021 import java.io.IOException;
022 import java.net.URI;
023 import java.net.URISyntaxException;
024 import java.util.regex.Pattern;
025
026 import org.apache.avro.reflect.Stringable;
027 import org.apache.commons.lang.StringUtils;
028 import org.apache.hadoop.HadoopIllegalArgumentException;
029 import org.apache.hadoop.classification.InterfaceAudience;
030 import org.apache.hadoop.classification.InterfaceStability;
031 import org.apache.hadoop.conf.Configuration;
032
033 /** Names a file or directory in a {@link FileSystem}.
034 * Path strings use slash as the directory separator. A path string is
035 * absolute if it begins with a slash.
036 */
037 @Stringable
038 @InterfaceAudience.Public
039 @InterfaceStability.Stable
040 public class Path implements Comparable {
041
042 /** The directory separator, a slash. */
043 public static final String SEPARATOR = "/";
044 public static final char SEPARATOR_CHAR = '/';
045
046 public static final String CUR_DIR = ".";
047
048 public static final boolean WINDOWS
049 = System.getProperty("os.name").startsWith("Windows");
050
051 /**
052 * Pre-compiled regular expressions to detect path formats.
053 */
054 private static final Pattern hasUriScheme =
055 Pattern.compile("^[a-zA-Z][a-zA-Z0-9+-.]+:");
056 private static final Pattern hasDriveLetterSpecifier =
057 Pattern.compile("^/?[a-zA-Z]:");
058
059 private URI uri; // a hierarchical uri
060
061 /**
062 * Pathnames with scheme and relative path are illegal.
063 * @param path to be checked
064 */
065 void checkNotSchemeWithRelative() {
066 if (toUri().isAbsolute() && !isUriPathAbsolute()) {
067 throw new HadoopIllegalArgumentException(
068 "Unsupported name: has scheme but relative path-part");
069 }
070 }
071
072 void checkNotRelative() {
073 if (!isAbsolute() && toUri().getScheme() == null) {
074 throw new HadoopIllegalArgumentException("Path is relative");
075 }
076 }
077
078 public static Path getPathWithoutSchemeAndAuthority(Path path) {
079 // This code depends on Path.toString() to remove the leading slash before
080 // the drive specification on Windows.
081 Path newPath = path.isUriPathAbsolute() ?
082 new Path(null, null, path.toUri().getPath()) :
083 path;
084 return newPath;
085 }
086
087 /** Resolve a child path against a parent path. */
088 public Path(String parent, String child) {
089 this(new Path(parent), new Path(child));
090 }
091
092 /** Resolve a child path against a parent path. */
093 public Path(Path parent, String child) {
094 this(parent, new Path(child));
095 }
096
097 /** Resolve a child path against a parent path. */
098 public Path(String parent, Path child) {
099 this(new Path(parent), child);
100 }
101
102 /** Resolve a child path against a parent path. */
103 public Path(Path parent, Path child) {
104 // Add a slash to parent's path so resolution is compatible with URI's
105 URI parentUri = parent.uri;
106 String parentPath = parentUri.getPath();
107 if (!(parentPath.equals("/") || parentPath.isEmpty())) {
108 try {
109 parentUri = new URI(parentUri.getScheme(), parentUri.getAuthority(),
110 parentUri.getPath()+"/", null, parentUri.getFragment());
111 } catch (URISyntaxException e) {
112 throw new IllegalArgumentException(e);
113 }
114 }
115 URI resolved = parentUri.resolve(child.uri);
116 initialize(resolved.getScheme(), resolved.getAuthority(),
117 resolved.getPath(), resolved.getFragment());
118 }
119
120 private void checkPathArg( String path ) throws IllegalArgumentException {
121 // disallow construction of a Path from an empty string
122 if ( path == null ) {
123 throw new IllegalArgumentException(
124 "Can not create a Path from a null string");
125 }
126 if( path.length() == 0 ) {
127 throw new IllegalArgumentException(
128 "Can not create a Path from an empty string");
129 }
130 }
131
132 /** Construct a path from a String. Path strings are URIs, but with
133 * unescaped elements and some additional normalization. */
134 public Path(String pathString) throws IllegalArgumentException {
135 checkPathArg( pathString );
136
137 // We can't use 'new URI(String)' directly, since it assumes things are
138 // escaped, which we don't require of Paths.
139
140 // add a slash in front of paths with Windows drive letters
141 if (hasWindowsDrive(pathString) && pathString.charAt(0) != '/') {
142 pathString = "/" + pathString;
143 }
144
145 // parse uri components
146 String scheme = null;
147 String authority = null;
148
149 int start = 0;
150
151 // parse uri scheme, if any
152 int colon = pathString.indexOf(':');
153 int slash = pathString.indexOf('/');
154 if ((colon != -1) &&
155 ((slash == -1) || (colon < slash))) { // has a scheme
156 scheme = pathString.substring(0, colon);
157 start = colon+1;
158 }
159
160 // parse uri authority, if any
161 if (pathString.startsWith("//", start) &&
162 (pathString.length()-start > 2)) { // has authority
163 int nextSlash = pathString.indexOf('/', start+2);
164 int authEnd = nextSlash > 0 ? nextSlash : pathString.length();
165 authority = pathString.substring(start+2, authEnd);
166 start = authEnd;
167 }
168
169 // uri path is the rest of the string -- query & fragment not supported
170 String path = pathString.substring(start, pathString.length());
171
172 initialize(scheme, authority, path, null);
173 }
174
175 /**
176 * Construct a path from a URI
177 */
178 public Path(URI aUri) {
179 uri = aUri.normalize();
180 }
181
182 /** Construct a Path from components. */
183 public Path(String scheme, String authority, String path) {
184 checkPathArg( path );
185
186 // add a slash in front of paths with Windows drive letters
187 if (hasWindowsDrive(path) && path.charAt(0) != '/') {
188 path = "/" + path;
189 }
190
191 // add "./" in front of Linux relative paths so that a path containing
192 // a colon e.q. "a:b" will not be interpreted as scheme "a".
193 if (!WINDOWS && path.charAt(0) != '/') {
194 path = "./" + path;
195 }
196
197 initialize(scheme, authority, path, null);
198 }
199
200 private void initialize(String scheme, String authority, String path,
201 String fragment) {
202 try {
203 this.uri = new URI(scheme, authority, normalizePath(scheme, path), null, fragment)
204 .normalize();
205 } catch (URISyntaxException e) {
206 throw new IllegalArgumentException(e);
207 }
208 }
209
210 /**
211 * Merge 2 paths such that the second path is appended relative to the first.
212 * The returned path has the scheme and authority of the first path. On
213 * Windows, the drive specification in the second path is discarded.
214 *
215 * @param path1 Path first path
216 * @param path2 Path second path, to be appended relative to path1
217 * @return Path merged path
218 */
219 public static Path mergePaths(Path path1, Path path2) {
220 String path2Str = path2.toUri().getPath();
221 if(hasWindowsDrive(path2Str)) {
222 path2Str = path2Str.substring(path2Str.indexOf(':')+1);
223 }
224 return new Path(path1 + path2Str);
225 }
226
227 /**
228 * Normalize a path string to use non-duplicated forward slashes as
229 * the path separator and remove any trailing path separators.
230 * @param scheme Supplies the URI scheme. Used to deduce whether we
231 * should replace backslashes or not.
232 * @param path Supplies the scheme-specific part
233 * @return Normalized path string.
234 */
235 private static String normalizePath(String scheme, String path) {
236 // Remove double forward slashes.
237 path = StringUtils.replace(path, "//", "/");
238
239 // Remove backslashes if this looks like a Windows path. Avoid
240 // the substitution if it looks like a non-local URI.
241 if (WINDOWS &&
242 (hasWindowsDrive(path) ||
243 (scheme == null) ||
244 (scheme.isEmpty()) ||
245 (scheme.equals("file")))) {
246 path = StringUtils.replace(path, "\\", "/");
247 }
248
249 // trim trailing slash from non-root path (ignoring windows drive)
250 int minLength = hasWindowsDrive(path) ? 4 : 1;
251 if (path.length() > minLength && path.endsWith("/")) {
252 path = path.substring(0, path.length()-1);
253 }
254
255 return path;
256 }
257
258 private static boolean hasWindowsDrive(String path) {
259 return (WINDOWS && hasDriveLetterSpecifier.matcher(path).find());
260 }
261
262 /**
263 * Determine whether a given path string represents an absolute path on
264 * Windows. e.g. "C:/a/b" is an absolute path. "C:a/b" is not.
265 *
266 * @param pathString Supplies the path string to evaluate.
267 * @param slashed true if the given path is prefixed with "/".
268 * @return true if the supplied path looks like an absolute path with a Windows
269 * drive-specifier.
270 */
271 public static boolean isWindowsAbsolutePath(final String pathString,
272 final boolean slashed) {
273 int start = (slashed ? 1 : 0);
274
275 return
276 hasWindowsDrive(pathString) &&
277 pathString.length() >= (start + 3) &&
278 ((pathString.charAt(start + 2) == SEPARATOR_CHAR) ||
279 (pathString.charAt(start + 2) == '\\'));
280 }
281
282 /** Convert this to a URI. */
283 public URI toUri() { return uri; }
284
285 /** Return the FileSystem that owns this Path. */
286 public FileSystem getFileSystem(Configuration conf) throws IOException {
287 return FileSystem.get(this.toUri(), conf);
288 }
289
290 /**
291 * Is an absolute path (ie a slash relative path part)
292 * AND a scheme is null AND authority is null.
293 */
294 public boolean isAbsoluteAndSchemeAuthorityNull() {
295 return (isUriPathAbsolute() &&
296 uri.getScheme() == null && uri.getAuthority() == null);
297 }
298
299 /**
300 * True if the path component (i.e. directory) of this URI is absolute.
301 */
302 public boolean isUriPathAbsolute() {
303 int start = hasWindowsDrive(uri.getPath()) ? 3 : 0;
304 return uri.getPath().startsWith(SEPARATOR, start);
305 }
306
307 /** True if the path component of this URI is absolute. */
308 /**
309 * There is some ambiguity here. An absolute path is a slash
310 * relative name without a scheme or an authority.
311 * So either this method was incorrectly named or its
312 * implementation is incorrect. This method returns true
313 * even if there is a scheme and authority.
314 */
315 public boolean isAbsolute() {
316 return isUriPathAbsolute();
317 }
318
319 /**
320 * @return true if and only if this path represents the root of a file system
321 */
322 public boolean isRoot() {
323 return getParent() == null;
324 }
325
326 /** Returns the final component of this path.*/
327 public String getName() {
328 String path = uri.getPath();
329 int slash = path.lastIndexOf(SEPARATOR);
330 return path.substring(slash+1);
331 }
332
333 /** Returns the parent of a path or null if at root. */
334 public Path getParent() {
335 String path = uri.getPath();
336 int lastSlash = path.lastIndexOf('/');
337 int start = hasWindowsDrive(path) ? 3 : 0;
338 if ((path.length() == start) || // empty path
339 (lastSlash == start && path.length() == start+1)) { // at root
340 return null;
341 }
342 String parent;
343 if (lastSlash==-1) {
344 parent = CUR_DIR;
345 } else {
346 int end = hasWindowsDrive(path) ? 3 : 0;
347 parent = path.substring(0, lastSlash==end?end+1:lastSlash);
348 }
349 return new Path(uri.getScheme(), uri.getAuthority(), parent);
350 }
351
352 /** Adds a suffix to the final name in the path.*/
353 public Path suffix(String suffix) {
354 return new Path(getParent(), getName()+suffix);
355 }
356
357 @Override
358 public String toString() {
359 // we can't use uri.toString(), which escapes everything, because we want
360 // illegal characters unescaped in the string, for glob processing, etc.
361 StringBuilder buffer = new StringBuilder();
362 if (uri.getScheme() != null) {
363 buffer.append(uri.getScheme());
364 buffer.append(":");
365 }
366 if (uri.getAuthority() != null) {
367 buffer.append("//");
368 buffer.append(uri.getAuthority());
369 }
370 if (uri.getPath() != null) {
371 String path = uri.getPath();
372 if (path.indexOf('/')==0 &&
373 hasWindowsDrive(path) && // has windows drive
374 uri.getScheme() == null && // but no scheme
375 uri.getAuthority() == null) // or authority
376 path = path.substring(1); // remove slash before drive
377 buffer.append(path);
378 }
379 if (uri.getFragment() != null) {
380 buffer.append("#");
381 buffer.append(uri.getFragment());
382 }
383 return buffer.toString();
384 }
385
386 @Override
387 public boolean equals(Object o) {
388 if (!(o instanceof Path)) {
389 return false;
390 }
391 Path that = (Path)o;
392 return this.uri.equals(that.uri);
393 }
394
395 @Override
396 public int hashCode() {
397 return uri.hashCode();
398 }
399
400 @Override
401 public int compareTo(Object o) {
402 Path that = (Path)o;
403 return this.uri.compareTo(that.uri);
404 }
405
406 /** Return the number of elements in this path. */
407 public int depth() {
408 String path = uri.getPath();
409 int depth = 0;
410 int slash = path.length()==1 && path.charAt(0)=='/' ? -1 : 0;
411 while (slash != -1) {
412 depth++;
413 slash = path.indexOf(SEPARATOR, slash+1);
414 }
415 return depth;
416 }
417
418 /**
419 * Returns a qualified path object.
420 *
421 * Deprecated - use {@link #makeQualified(URI, Path)}
422 */
423 @Deprecated
424 public Path makeQualified(FileSystem fs) {
425 return makeQualified(fs.getUri(), fs.getWorkingDirectory());
426 }
427
428 /** Returns a qualified path object. */
429 @InterfaceAudience.LimitedPrivate({"HDFS", "MapReduce"})
430 public Path makeQualified(URI defaultUri, Path workingDir ) {
431 Path path = this;
432 if (!isAbsolute()) {
433 path = new Path(workingDir, this);
434 }
435
436 URI pathUri = path.toUri();
437
438 String scheme = pathUri.getScheme();
439 String authority = pathUri.getAuthority();
440 String fragment = pathUri.getFragment();
441
442 if (scheme != null &&
443 (authority != null || defaultUri.getAuthority() == null))
444 return path;
445
446 if (scheme == null) {
447 scheme = defaultUri.getScheme();
448 }
449
450 if (authority == null) {
451 authority = defaultUri.getAuthority();
452 if (authority == null) {
453 authority = "";
454 }
455 }
456
457 URI newUri = null;
458 try {
459 newUri = new URI(scheme, authority ,
460 normalizePath(scheme, pathUri.getPath()), null, fragment);
461 } catch (URISyntaxException e) {
462 throw new IllegalArgumentException(e);
463 }
464 return new Path(newUri);
465 }
466 }