Azure Blob多线程压缩文件报错：no current ZIP entry

我在Azure Blob中压缩文件时，单线程常规压缩能正常生成压缩包，但改用CompletableFuture多线程处理时抛出异常。

代码示例 #

try (ZipOutputStream zipStream = new ZipOutputStream(outputStream)) {
        List<CompletableFuture<Void>> futures = new ArrayList<>();

        for (String singleFileName : fileNames) {
            String singleFilePath = azureFilePath + "/" + singleFileName.trim();
            BlobClient blobClient = blobContainerClient.getBlobClient(singleFilePath);

            CompletableFuture<Void> future = CompletableFuture.runAsync(() -> {
                try (ByteArrayOutputStream fileOutputStream = new ByteArrayOutputStream()) {
                    if (blobClient.exists()) {
                        // Download the file to a separate output stream
                        blobClient.download(fileOutputStream);

                        // Add the file to the zip stream
                        zipStream.putNextEntry(new ZipEntry(singleFileName));
                        zipStream.write(fileOutputStream.toByteArray());
                      // zipStream.closeEntry();
                    } else {
                        throw new RuntimeException("File not found: " + singleFilePath);
                    }
                } catch (IOException e) {
                    throw new RuntimeException("Error processing file: " + singleFileName, e);
                }finally {
                    try {
                        zipStream.closeEntry();
                    } catch (IOException e) {
                        // Handle any potential exception when closing the zip entry
                        e.printStackTrace();
                    }
                }
            });

            futures.add(future);
        }
        CompletableFuture<Void> allOf = CompletableFuture.allOf(futures.toArray(new CompletableFuture[0]));
        allOf.join();
    }

    // Upload the zip file to the same location
    BlobClient zipFileClient = blobContainerClient.getBlobClient(azureFilePath + "/" + zipFileName);
    zipFileClient.upload(new ByteArrayInputStream(outputStream.toByteArray()), outputStream.size(),true);

异常信息 #

java.util.zip.ZipException: no current ZIP entry
at java.base/java.util.zip.ZipOutputStream.write(ZipOutputStream.java:343) ~[na:na]
at java.base/java.io.FilterOutputStream.write(FilterOutputStream.java:108) ~[na:na]
at main.java.com.azure.filezip.ApiControllerzip.lambda$uploadZip$0(ApiControllerzip.java:78) ~[classes/:na]

解决建议 #

• 核心原因：ZipOutputStream并非线程安全
  ZipOutputStream的putNextEntry、write、closeEntry等方法没有线程同步机制，多线程同时操作会导致内部状态混乱——比如线程A刚调用putNextEntry，线程B就覆盖了当前entry，线程A再执行write时就找不到对应的entry，直接抛出异常。

• 可行的优化方案：

  1. 多线程下载+单线程写入Zip流
     将文件下载与Zip写入解耦，用多线程并行下载Blob文件到内存，所有文件下载完成后，由主线程统一写入ZipOutputStream，既保留下载的性能优势，又避免多线程操作Zip流的冲突。调整后示例代码：

     try (ZipOutputStream zipStream = new ZipOutputStream(outputStream)) {
         List<CompletableFuture<Map.Entry<String, byte[]>>> futures = new ArrayList<>();
     
         for (String singleFileName : fileNames) {
             String singleFilePath = azureFilePath + "/" + singleFileName.trim();
             BlobClient blobClient = blobContainerClient.getBlobClient(singleFilePath);
     
             CompletableFuture<Map.Entry<String, byte[]>> future = CompletableFuture.supplyAsync(() -> {
                 try (ByteArrayOutputStream fileOutputStream = new ByteArrayOutputStream()) {
                     if (blobClient.exists()) {
                         blobClient.download(fileOutputStream);
                         return new AbstractMap.SimpleEntry<>(singleFileName, fileOutputStream.toByteArray());
                     } else {
                         throw new RuntimeException("File not found: " + singleFilePath);
                     }
                 } catch (IOException e) {
                     throw new RuntimeException("Error processing file: " + singleFileName, e);
                 }
             });
     
             futures.add(future);
         }
     
         // 等待所有文件下载完成，逐个写入Zip流
         for (CompletableFuture<Map.Entry<String, byte[]>> future : futures) {
             Map.Entry<String, byte[]> entry = future.join();
             zipStream.putNextEntry(new ZipEntry(entry.getKey()));
             zipStream.write(entry.getValue());
             zipStream.closeEntry();
         }
     }
     
     // 上传Zip文件到Blob
     BlobClient zipFileClient = blobContainerClient.getBlobClient(azureFilePath + "/" + zipFileName);
     zipFileClient.upload(new ByteArrayInputStream(outputStream.toByteArray()), outputStream.size(), true);
     

  2. 加全局锁强制串行写入（不推荐）
     如果一定要在多线程中写入Zip流，可给zipStream加全局锁，但这样会让写入操作串行化，失去多线程的性能优势，仅作为应急方案：

     // 在lambda中操作Zip流时加锁
     synchronized (zipStream) {
         zipStream.putNextEntry(new ZipEntry(singleFileName));
         zipStream.write(fileOutputStream.toByteArray());
         zipStream.closeEntry();
     }
     

  3. 内存优化（针对大文件场景）
     如果处理的文件较大，用ByteArrayOutputStream会占用过多内存，建议下载到临时文件，再从临时文件写入ZipOutputStream，避免内存溢出。

内容的提问来源于stack exchange，提问作者Mani