摘要
在 Spring 中,我可以通过指定相对于我的src/main/resources/
目录的路径来访问资源。例如,如果我要求/public/index.html
,我会得到一个代表/Users/.../src/main/resources/public/index.html
的FileSystemResource
。但是,我看不到相反方向的方法。
给定一个FileSystemResource
,有没有办法找到它相对于src/main/resouces/
的路径?
例
我正在使用 PathMatchingResourcePatternResolver
获取应用中的文件资源列表。我需要的资源位于应用的src/main/resources/public/
文件夹中。
ResourcePatternResolver resolver = new PathMatchingResourcePatternResolver();
for (Resource r : resolver.getResources("/public/**")) {
// r.getURI().getString() gives me the absolute path.
}
我可以很容易地得到绝对路径,但我想要一种方法来让部分从/public/
开始,因为这就是 Spring 一开始发现它的方式。
正要问如何做到这一点,以一种比我的解决方案更不黑客的方式(加上处理在我的情况下可能发生的ClassPathResource
(。所以这是我的答案和问题:
ResourcePatternResolver.getResources
返回的资源可以是FileSystemResource
或ClassPathResource
,这取决于 glob 模式是否包含通配符。 (就我而言,glob 模式作为参数给出,应支持任何模式。所以,没有getFile
;两者唯一可用的路径信息似乎是 getDescription
.
为了确定根目录的目录深度,我在那里放置了一个参考文件"input-files.txt"。这是因为我认为根目录资源"classpath:/input-files/"在基于 jar 的类路径的情况下不需要存在。使用根深度,我可以从资源描述中删除那么多目录。
有谁知道更简单或不太黑客的解决方案?
像我的一样,它应该与 jar 和基于目录的类路径一起使用。
import java.io.IOException;
import java.nio.file.Path;
import java.util.stream.Collectors;
import java.util.stream.Stream;
import javax.inject.Inject;
import org.springframework.boot.CommandLineRunner;
import org.springframework.boot.builder.SpringApplicationBuilder;
import org.springframework.context.ApplicationContext;
import org.springframework.core.io.Resource;
public class TestClasspathRelativePath implements CommandLineRunner {
private static final String RESOURCE_ROOT = "classpath:/input-files/";
private static final String RESOURCE_ROOT_MARKER_FILE = "input-files.txt";
private static final String FILE_SEPARATOR_PATTERN = "[/\\]+";
@Inject
private ApplicationContext appContext;
public static void main(String[] args) {
new SpringApplicationBuilder()
.sources(TestClasspathRelativePath.class)
.web(false)
.build()
.run("**/*");
}
@Override
public void run(String... args) throws Exception {
for (String templateFilenamePattern : args) {
for (Resource template : appContext.getResources(RESOURCE_ROOT + templateFilenamePattern)) {
createInputFile(template);
}
}
}
private void createInputFile(Resource template) throws IOException {
// A resource whose immediate parent is the root input
// files directory, without wildcard lookup. (As of Spring 4.3.14, with
// a directory classpath entry, this is a ClassPathResource.)
Resource topLevelFile = appContext.getResource(RESOURCE_ROOT + RESOURCE_ROOT_MARKER_FILE);
// A resource whose immediate parent is the root input
// files directory, with wildcard lookup. (As of Spring 4.3.14, with a
// directory classpath entry, this is a FileSystemResource.)
Resource wildcardTopLevelFile =
// match with min description length is
// RESOURCE_ROOT/input-files.txt (probably sole match anyway)
Stream.of(appContext.getResources(RESOURCE_ROOT + "**/" + RESOURCE_ROOT_MARKER_FILE)).min(
(r1, r2) -> Integer.valueOf(r1.getDescription().length()).compareTo(Integer.valueOf(r2.getDescription().length())))
.orElseThrow(() -> new IllegalStateException("resource " + RESOURCE_ROOT + "/" + RESOURCE_ROOT_MARKER_FILE + " not found"));
// In the real program, both top level resources are computed only once and on-demand.
String targetFilename = "<target-dir>/input/" + relativize(template, topLevelFile, wildcardTopLevelFile);
System.out.println(targetFilename);
// ... read template, process, write results to targetFilename
}
/**
* Replacement for {@link Path#relativize(Path)} which should also work with
* non-file resources.
*/
private static String relativize(Resource child, Resource... topLevelFiles) {
// find the top-level file for child's type
Resource referenceFile = Stream.of(topLevelFiles)
.filter(f -> f.getClass().isInstance(child))
.findFirst()
.orElseThrow(() -> new IllegalStateException("don't know how to relativize " + child));
int rootLevel = descriptionUpToName(referenceFile).split(FILE_SEPARATOR_PATTERN, -1).length - 1;
return Stream.of(descriptionUpToName(child).split(FILE_SEPARATOR_PATTERN, -1)).skip(rootLevel)
.collect(Collectors.joining("/"));
}
/**
* Hack to strip the suffix after the name from a resource description. The
* prefix need not be stripped, because
* {@link #relativize(Resource, Resource...)} uses a safer way to get rid of
* that.
*
* @return e.g. "file [C:fooclasspathinput-filesab.txt" if the
* resource's description is "file
* [C:fooclasspathinput-filesab.txt]"
*/
private static String descriptionUpToName(Resource resource) {
String path = resource.getDescription();
int i = path.lastIndexOf(resource.getFilename());
return path.substring(0, i + resource.getFilename().length());
}
}