fix: 支持解析工作流受管上传文件

- 优先通过上传记录验证后的读取器加载文件内容

- 保留普通存储与远端地址校验回退路径

- 补充来源读取和解析桥接回归测试
This commit is contained in:
2026-08-10 23:54:25 +08:00
parent 1c0fbfa5ff
commit 402c0f16b8
3 changed files with 145 additions and 55 deletions

View File

@@ -1,5 +1,6 @@
package tech.easyflow.ai.document.support; package tech.easyflow.ai.document.support;
import org.springframework.beans.factory.annotation.Autowired;
import org.springframework.beans.factory.annotation.Qualifier; import org.springframework.beans.factory.annotation.Qualifier;
import org.slf4j.Logger; import org.slf4j.Logger;
import org.slf4j.LoggerFactory; import org.slf4j.LoggerFactory;
@@ -7,12 +8,14 @@ import org.springframework.stereotype.Component;
import org.springframework.util.StringUtils; import org.springframework.util.StringUtils;
import tech.easyflow.ai.document.exception.DocumentParseBridgeException; import tech.easyflow.ai.document.exception.DocumentParseBridgeException;
import tech.easyflow.ai.document.model.DocumentSourceRef; import tech.easyflow.ai.document.model.DocumentSourceRef;
import tech.easyflow.ai.easyagentsflow.upload.WorkflowApiUploadedFileReader;
import tech.easyflow.common.filestorage.FileStorageService; import tech.easyflow.common.filestorage.FileStorageService;
import tech.easyflow.common.filestorage.utils.PathGeneratorUtil; import tech.easyflow.common.filestorage.utils.PathGeneratorUtil;
import java.io.IOException; import java.io.IOException;
import java.io.InputStream; import java.io.InputStream;
import java.net.URLConnection; import java.net.URLConnection;
import java.util.Optional;
/** /**
* 文档源加载器。 * 文档源加载器。
@@ -29,9 +32,29 @@ public class DocumentSourceLoader {
LoggerFactory.getLogger(DocumentSourceLoader.class); LoggerFactory.getLogger(DocumentSourceLoader.class);
private final FileStorageService fileStorageService; private final FileStorageService fileStorageService;
private final WorkflowApiUploadedFileReader uploadedFileReader;
public DocumentSourceLoader(@Qualifier("default") FileStorageService fileStorageService) { /**
* 创建文档源加载器。
*
* @param fileStorageService 默认文件存储服务
* @param uploadedFileReader 已验证的工作流 API 上传文件读取器
*/
@Autowired
public DocumentSourceLoader(
@Qualifier("default") FileStorageService fileStorageService,
WorkflowApiUploadedFileReader uploadedFileReader) {
this.fileStorageService = fileStorageService; this.fileStorageService = fileStorageService;
this.uploadedFileReader = uploadedFileReader;
}
/**
* 创建不启用工作流 API 上传识别的加载器,供同包隔离测试使用。
*
* @param fileStorageService 文件存储服务
*/
DocumentSourceLoader(FileStorageService fileStorageService) {
this(fileStorageService, null);
} }
/** /**
@@ -80,7 +103,7 @@ public class DocumentSourceLoader {
private LoadedDocumentSource loadFromFilePath(DocumentSourceRef sourceRef, long maxBytes) { private LoadedDocumentSource loadFromFilePath(DocumentSourceRef sourceRef, long maxBytes) {
String fileName = resolveFileName(sourceRef); String fileName = resolveFileName(sourceRef);
try (InputStream inputStream = fileStorageService.readStream(sourceRef.getFilePath())) { try (InputStream inputStream = openStoredValue(sourceRef.getFilePath(), maxBytes)) {
byte[] contentBytes = DocumentInputStreamSupport.readBytes(inputStream, maxBytes); byte[] contentBytes = DocumentInputStreamSupport.readBytes(inputStream, maxBytes);
logSizeMismatch(sourceRef, contentBytes.length); logSizeMismatch(sourceRef, contentBytes.length);
return buildLoadedSource( return buildLoadedSource(
@@ -105,7 +128,7 @@ public class DocumentSourceLoader {
String remoteUrl, String remoteUrl,
long maxBytes) { long maxBytes) {
String fileName = resolveFileName(sourceRef); String fileName = resolveFileName(sourceRef);
try (InputStream inputStream = DocumentInputStreamSupport.openRemote(remoteUrl, maxBytes)) { try (InputStream inputStream = openRemoteValue(remoteUrl, maxBytes)) {
byte[] contentBytes = DocumentInputStreamSupport.readBytes(inputStream, maxBytes); byte[] contentBytes = DocumentInputStreamSupport.readBytes(inputStream, maxBytes);
logSizeMismatch(sourceRef, contentBytes.length); logSizeMismatch(sourceRef, contentBytes.length);
return buildLoadedSource( return buildLoadedSource(
@@ -122,6 +145,54 @@ public class DocumentSourceLoader {
} }
} }
/**
* 优先打开经过上传记录验证的受管文件,再回退普通存储路径。
*
* @param filePath 文件路径
* @param maxBytes 最大允许读取字节数
* @return 受限输入流
* @throws IOException 文件无法读取时抛出
*/
private InputStream openStoredValue(String filePath, long maxBytes) throws IOException {
Optional<InputStream> managed = openManagedValue(filePath);
if (managed.isPresent()) {
return DocumentInputStreamSupport.limit(managed.get(), maxBytes);
}
return DocumentInputStreamSupport.limit(
fileStorageService.readStream(filePath),
maxBytes);
}
/**
* 优先打开经过上传记录验证的受管 URL再执行普通公网 URL 校验与下载。
*
* @param remoteUrl 远端 URL
* @param maxBytes 最大允许读取字节数
* @return 受限输入流
* @throws IOException 文件无法读取时抛出
*/
private InputStream openRemoteValue(String remoteUrl, long maxBytes) throws IOException {
Optional<InputStream> managed = openManagedValue(remoteUrl);
if (managed.isPresent()) {
return DocumentInputStreamSupport.limit(managed.get(), maxBytes);
}
return DocumentInputStreamSupport.openRemote(remoteUrl, maxBytes);
}
/**
* 尝试打开受管工作流上传文件。
*
* @param filePath 文件路径
* @return 受管输入流;测试未配置读取器或普通路径时为空
* @throws IOException 受管文件校验或读取失败时抛出
*/
private Optional<InputStream> openManagedValue(String filePath) throws IOException {
if (uploadedFileReader == null) {
return Optional.empty();
}
return uploadedFileReader.openVerified(filePath);
}
private LoadedDocumentSource buildLoadedSource(String fileName, String contentType, Long size, byte[] contentBytes) { private LoadedDocumentSource buildLoadedSource(String fileName, String contentType, Long size, byte[] contentBytes) {
LoadedDocumentSource loadedSource = new LoadedDocumentSource(); LoadedDocumentSource loadedSource = new LoadedDocumentSource();
loadedSource.setFileName(fileName); loadedSource.setFileName(fileName);

View File

@@ -221,7 +221,10 @@ public class DocumentParseBridgeServiceImplTest {
pptxDocumentParseService, pptxDocumentParseService,
xlsxDocumentParseService, xlsxDocumentParseService,
parseService, parseService,
new DocumentSourceLoader(new InMemoryFileStorageService())); new DocumentSourceLoader(
new InMemoryFileStorageService(),
org.mockito.Mockito.mock(
tech.easyflow.ai.easyagentsflow.upload.WorkflowApiUploadedFileReader.class)));
} }
/** /**
@@ -292,7 +295,10 @@ public class DocumentParseBridgeServiceImplTest {
private long maxBytes; private long maxBytes;
private RecordingDocumentSourceLoader() { private RecordingDocumentSourceLoader() {
super(new InMemoryFileStorageService()); super(
new InMemoryFileStorageService(),
org.mockito.Mockito.mock(
tech.easyflow.ai.easyagentsflow.upload.WorkflowApiUploadedFileReader.class));
} }
/** /**

View File

@@ -2,6 +2,8 @@ package tech.easyflow.ai.document.support;
import org.junit.Assert; import org.junit.Assert;
import org.junit.Test; import org.junit.Test;
import org.mockito.Mockito;
import tech.easyflow.ai.easyagentsflow.upload.WorkflowApiUploadedFileReader;
import tech.easyflow.ai.document.exception.DocumentParseBridgeException; import tech.easyflow.ai.document.exception.DocumentParseBridgeException;
import tech.easyflow.ai.document.model.DocumentSourceRef; import tech.easyflow.ai.document.model.DocumentSourceRef;
import tech.easyflow.common.filestorage.FileStorageService; import tech.easyflow.common.filestorage.FileStorageService;
@@ -9,9 +11,8 @@ import tech.easyflow.common.filestorage.FileStorageService;
import java.io.ByteArrayInputStream; import java.io.ByteArrayInputStream;
import java.io.IOException; import java.io.IOException;
import java.io.InputStream; import java.io.InputStream;
import java.net.InetSocketAddress;
import java.nio.charset.StandardCharsets; import java.nio.charset.StandardCharsets;
import com.sun.net.httpserver.HttpServer; import java.util.Optional;
/** /**
* {@link DocumentSourceLoader} 单元测试。 * {@link DocumentSourceLoader} 单元测试。
@@ -54,54 +55,69 @@ public class DocumentSourceLoaderTest {
} }
/** /**
* 验证 filePath 远端 URL 时不会误走存储读取 * 验证普通 filePath 远端 URL 仍拒绝访问回环地址
*/ */
@Test @Test
public void shouldPreferRemoteDownloadWhenFilePathIsRemoteUrl() throws IOException { public void shouldRejectLoopbackWhenRemoteUrlIsNotManagedUpload() {
DocumentSourceLoader loader = new DocumentSourceLoader(new FailingFileStorageService()); DocumentSourceLoader loader = new DocumentSourceLoader(new FailingFileStorageService());
HttpServer server = HttpServer.create(new InetSocketAddress(0), 0);
byte[] body = "demo-pdf".getBytes(StandardCharsets.UTF_8);
server.createContext("/demo.pdf", exchange -> {
exchange.sendResponseHeaders(200, body.length);
exchange.getResponseBody().write(body);
exchange.close();
});
server.start();
try {
DocumentSourceRef sourceRef = new DocumentSourceRef(); DocumentSourceRef sourceRef = new DocumentSourceRef();
sourceRef.setFileName("demo.pdf"); sourceRef.setFileName("demo.pdf");
sourceRef.setFilePath("http://127.0.0.1:" + server.getAddress().getPort() + "/demo.pdf"); sourceRef.setFilePath("http://127.0.0.1:39000/demo.pdf");
try {
loader.load(sourceRef);
Assert.fail("expected DocumentParseBridgeException");
} catch (DocumentParseBridgeException exception) {
Assert.assertEquals("source_load_failed", exception.getCode());
Assert.assertTrue(exception.getCause() instanceof java.net.UnknownHostException);
}
}
/**
* 验证已通过上传记录校验的内网存储 URL 会走恢复句柄读取。
*
* @throws IOException 测试流创建失败时抛出
*/
@Test
public void shouldLoadVerifiedManagedUploadBeforeRemoteAddressGuard() throws IOException {
WorkflowApiUploadedFileReader uploadedFileReader =
Mockito.mock(WorkflowApiUploadedFileReader.class);
DocumentSourceLoader loader = new DocumentSourceLoader(
new FailingFileStorageService(),
uploadedFileReader);
String fileUrl = "http://127.0.0.1:39000/easyflow/attachment/"
+ "workflow-api-upload/0123456789abcdef0123456789abcdef/file.pdf";
byte[] body = "demo-pdf".getBytes(StandardCharsets.UTF_8);
Mockito.when(uploadedFileReader.openVerified(fileUrl))
.thenReturn(Optional.of(new ByteArrayInputStream(body)));
DocumentSourceRef sourceRef = new DocumentSourceRef();
sourceRef.setFileName("demo.pdf");
sourceRef.setFilePath(fileUrl);
LoadedDocumentSource loadedSource = loader.load(sourceRef); LoadedDocumentSource loadedSource = loader.load(sourceRef);
Assert.assertEquals("demo.pdf", loadedSource.getFileName()); Assert.assertEquals("demo.pdf", loadedSource.getFileName());
Assert.assertArrayEquals(body, loadedSource.getContentBytes()); Assert.assertArrayEquals(body, loadedSource.getContentBytes());
} finally {
server.stop(0);
}
} }
/** /**
* 验证远端响应声明值不可信时仍按实际读取字节数拒绝超限内容。 * 验证受管上传文件仍按实际读取字节数拒绝超限内容。
*
* @throws IOException 测试服务启动失败时抛出
*/ */
@Test @Test
public void shouldRejectRemoteContentThatExceedsActualByteLimit() throws IOException { public void shouldRejectManagedContentThatExceedsActualByteLimit() throws IOException {
DocumentSourceLoader loader = new DocumentSourceLoader(new FailingFileStorageService()); WorkflowApiUploadedFileReader uploadedFileReader =
HttpServer server = HttpServer.create(new InetSocketAddress(0), 0); Mockito.mock(WorkflowApiUploadedFileReader.class);
DocumentSourceLoader loader = new DocumentSourceLoader(
new FailingFileStorageService(),
uploadedFileReader);
byte[] body = "123456789".getBytes(StandardCharsets.UTF_8); byte[] body = "123456789".getBytes(StandardCharsets.UTF_8);
server.createContext("/oversized.pdf", exchange -> { String fileUrl = "http://127.0.0.1:39000/easyflow/attachment/"
exchange.sendResponseHeaders(200, 0); + "workflow-api-upload/0123456789abcdef0123456789abcdef/oversized.pdf";
exchange.getResponseBody().write(body); Mockito.when(uploadedFileReader.openVerified(fileUrl))
exchange.close(); .thenReturn(Optional.of(new ByteArrayInputStream(body)));
});
server.start();
try {
DocumentSourceRef sourceRef = new DocumentSourceRef(); DocumentSourceRef sourceRef = new DocumentSourceRef();
sourceRef.setFileName("oversized.pdf"); sourceRef.setFileName("oversized.pdf");
sourceRef.setFilePath( sourceRef.setFilePath(fileUrl);
"http://127.0.0.1:" + server.getAddress().getPort() + "/oversized.pdf");
sourceRef.setSize(1L); sourceRef.setSize(1L);
try { try {
@@ -112,9 +128,6 @@ public class DocumentSourceLoaderTest {
Assert.assertTrue(exception.getCause() Assert.assertTrue(exception.getCause()
instanceof DocumentInputStreamSupport.SizeLimitExceededException); instanceof DocumentInputStreamSupport.SizeLimitExceededException);
} }
} finally {
server.stop(0);
}
} }
/** /**