我的版本大致基于马特和史蒂夫的版本:

/**
 * Returns the path of one File relative to another.
 *
 * @param target the target directory
 * @param base the base directory
 * @return target's path relative to the base directory
 * @throws IOException if an error occurs while resolving the files' canonical names
 */
 public static File getRelativeFile(File target, File base) throws IOException
 {
   String[] baseComponents = base.getCanonicalPath().split(Pattern.quote(File.separator));
   String[] targetComponents = target.getCanonicalPath().split(Pattern.quote(File.separator));

   // skip common components
   int index = 0;
   for (; index < targetComponents.length && index < baseComponents.length; ++index)
   {
     if (!targetComponents[index].equals(baseComponents[index]))
       break;
   }

   StringBuilder result = new StringBuilder();
   if (index != baseComponents.length)
   {
     // backtrack to base directory
     for (int i = index; i < baseComponents.length; ++i)
       result.append(".." + File.separator);
   }
   for (; index < targetComponents.length; ++index)
     result.append(targetComponents[index] + File.separator);
   if (!target.getPath().endsWith("/") && !target.getPath().endsWith("\\"))
   {
     // remove final path separator
     result.delete(result.length() - File.separator.length(), result.length());
   }
   return new File(result.toString());
 }

Matt B的解决方案错误地获得了回溯目录的数量——它应该是基本路径的长度减去公共路径元素的数量，减去1(对于最后一个路径元素，它是一个文件名或由split生成的尾随“”)。它恰好适用于/a/b/c/和/a/x/y/，但是将参数替换为/m/n/o/a/b/c/和/m/n/o/a/x/y/，你就会发现问题。

此外，它还需要在第一个for循环中进行else中断，否则它将错误地处理碰巧具有匹配目录名的路径，例如/a/b/c/d/和/x/y/c/z——c在两个数组中的同一个槽中，但不是真正的匹配。

所有这些解决方案都缺乏处理不能相互相对化的路径的能力，因为它们具有不兼容的根，例如C:\foo\bar和D:\baz\quux。可能只是Windows上的一个问题，但值得注意。

我花在这上面的时间比我预期的要长得多，但没关系。我实际上需要这个工作，所以感谢每个人的插话，我相信这个版本也会有修正!

public static String getRelativePath(String targetPath, String basePath, 
        String pathSeparator) {

    //  We need the -1 argument to split to make sure we get a trailing 
    //  "" token if the base ends in the path separator and is therefore
    //  a directory. We require directory paths to end in the path
    //  separator -- otherwise they are indistinguishable from files.
    String[] base = basePath.split(Pattern.quote(pathSeparator), -1);
    String[] target = targetPath.split(Pattern.quote(pathSeparator), 0);

    //  First get all the common elements. Store them as a string,
    //  and also count how many of them there are. 
    String common = "";
    int commonIndex = 0;
    for (int i = 0; i < target.length && i < base.length; i++) {
        if (target[i].equals(base[i])) {
            common += target[i] + pathSeparator;
            commonIndex++;
        }
        else break;
    }

    if (commonIndex == 0)
    {
        //  Whoops -- not even a single common path element. This most
        //  likely indicates differing drive letters, like C: and D:. 
        //  These paths cannot be relativized. Return the target path.
        return targetPath;
        //  This should never happen when all absolute paths
        //  begin with / as in *nix. 
    }

    String relative = "";
    if (base.length == commonIndex) {
        //  Comment this out if you prefer that a relative path not start with ./
        //relative = "." + pathSeparator;
    }
    else {
        int numDirsUp = base.length - commonIndex - 1;
        //  The number of directories we have to backtrack is the length of 
        //  the base path MINUS the number of common path elements, minus
        //  one because the last element in the path isn't a directory.
        for (int i = 1; i <= (numDirsUp); i++) {
            relative += ".." + pathSeparator;
        }
    }
    relative += targetPath.substring(common.length());

    return relative;
}

下面是几种情况下的测试:

public void testGetRelativePathsUnixy() 
{        
    assertEquals("stuff/xyz.dat", FileUtils.getRelativePath(
            "/var/data/stuff/xyz.dat", "/var/data/", "/"));
    assertEquals("../../b/c", FileUtils.getRelativePath(
            "/a/b/c", "/a/x/y/", "/"));
    assertEquals("../../b/c", FileUtils.getRelativePath(
            "/m/n/o/a/b/c", "/m/n/o/a/x/y/", "/"));
}

public void testGetRelativePathFileToFile() 
{
    String target = "C:\\Windows\\Boot\\Fonts\\chs_boot.ttf";
    String base = "C:\\Windows\\Speech\\Common\\sapisvr.exe";

    String relPath = FileUtils.getRelativePath(target, base, "\\");
    assertEquals("..\\..\\..\\Boot\\Fonts\\chs_boot.ttf", relPath);
}

public void testGetRelativePathDirectoryToFile() 
{
    String target = "C:\\Windows\\Boot\\Fonts\\chs_boot.ttf";
    String base = "C:\\Windows\\Speech\\Common";

    String relPath = FileUtils.getRelativePath(target, base, "\\");
    assertEquals("..\\..\\Boot\\Fonts\\chs_boot.ttf", relPath);
}

public void testGetRelativePathDifferentDriveLetters() 
{
    String target = "D:\\sources\\recovery\\RecEnv.exe";
    String base   = "C:\\Java\\workspace\\AcceptanceTests\\Standard test data\\geo\\";

    //  Should just return the target path because of the incompatible roots.
    String relPath = FileUtils.getRelativePath(target, base, "\\");
    assertEquals(target, relPath);
}

ant有一个带有getRelativePath方法的FileUtils类。我自己还没有尝试过，但是值得一试。

http://javadoc.haefelinger.it/org.apache.ant/1.7.1/org/apache/tools/ant/util/FileUtils.html getRelativePath (java.io.File java.io.File)

如果路径在JRE 1.5运行时或maven插件中不可用

package org.afc.util;

import java.io.File;
import java.util.LinkedList;
import java.util.List;

public class FileUtil {

    public static String getRelativePath(String basePath, String filePath)  {
        return getRelativePath(new File(basePath), new File(filePath));
    }

    public static String getRelativePath(File base, File file)  {

        List<String> bases = new LinkedList<String>();
        bases.add(0, base.getName());
        for (File parent = base.getParentFile(); parent != null; parent = parent.getParentFile()) {
            bases.add(0, parent.getName());
        }

        List<String> files = new LinkedList<String>();
        files.add(0, file.getName());
        for (File parent = file.getParentFile(); parent != null; parent = parent.getParentFile()) {
            files.add(0, parent.getName());
        }

        int overlapIndex = 0;
        while (overlapIndex < bases.size() && overlapIndex < files.size() && bases.get(overlapIndex).equals(files.get(overlapIndex))) {
            overlapIndex++;
        }

        StringBuilder relativePath = new StringBuilder();
        for (int i = overlapIndex; i < bases.size(); i++) {
            relativePath.append("..").append(File.separatorChar);
        }

        for (int i = overlapIndex; i < files.size(); i++) {
            relativePath.append(files.get(i)).append(File.separatorChar);
        }

        relativePath.deleteCharAt(relativePath.length() - 1);
        return relativePath.toString();
    }

}

当使用java.net.URI.relativize时，你应该注意Java错误: JDK-6226081 (URI应该能够将路径与部分根相对化)

目前，URI的relativize()方法只在一个URI是另一个URI的前缀时才会相对化URI。

这本质上意味着java.net.URI.relativize将不会创建“…这是给你的。

通过Dónal的测试，唯一的变化-如果没有公共根，它将返回目标路径(它可能已经是相对的)

import static java.util.Arrays.asList;
import static java.util.Collections.nCopies;
import static org.apache.commons.io.FilenameUtils.normalizeNoEndSeparator;
import static org.apache.commons.io.FilenameUtils.separatorsToUnix;
import static org.apache.commons.lang3.StringUtils.getCommonPrefix;
import static org.apache.commons.lang3.StringUtils.isBlank;
import static org.apache.commons.lang3.StringUtils.isNotEmpty;
import static org.apache.commons.lang3.StringUtils.join;

import java.io.File;
import java.util.ArrayList;
import java.util.List;

public class ResourceUtils {

    public static String getRelativePath(String targetPath, String basePath, String pathSeparator) {
        File baseFile = new File(basePath);
        if (baseFile.isFile() || !baseFile.exists() && !basePath.endsWith("/") && !basePath.endsWith("\\"))
            basePath = baseFile.getParent();

        String target = separatorsToUnix(normalizeNoEndSeparator(targetPath));
        String base = separatorsToUnix(normalizeNoEndSeparator(basePath));

        String commonPrefix = getCommonPrefix(target, base);
        if (isBlank(commonPrefix))
            return targetPath.replaceAll("/", pathSeparator);

        target = target.replaceFirst(commonPrefix, "");
        base = base.replaceFirst(commonPrefix, "");

        List<String> result = new ArrayList<>();
        if (isNotEmpty(base))
            result.addAll(nCopies(base.split("/").length, ".."));
        result.addAll(asList(target.replaceFirst("^/", "").split("/")));

        return join(result, pathSeparator);
    }
}