java - 我需要 ByteArrayInputStream 的读取缓冲区吗

Question

我有一个 JPA AttributeConverter，用于将 String 转换为 gzipped byte[] 并返回。

转换方法相当简单:

public byte[] convertToDatabaseColumn(String attribute) {
    try (ByteArrayOutputStream baos = new ByteArrayOutputStream();
         GZIPOutputStream gos = new GZIPOutputStream(baos)) {

        gos.write(attribute.getBytes(StandardCharsets.UTF_8));
        gos.finish();
        gos.flush();

        return baos.toByteArray();
    } catch (IOException e) {
        throw new UncheckedIOException(e);
    }
}

我的问题是关于转换方法:

public String convertToEntityAttribute(byte[] dbData) {
    try (GZIPInputStream gis = new GZIPInputStream(new ByteArrayInputStream(dbData));
         ByteArrayOutputStream baos = new ByteArrayOutputStream()) {

        byte[] buffer = new byte[1024];

        int len;
        while ((len = gis.read(buffer)) > 0) {
            baos.write(buffer, 0, len);
        }

        return new String(baos.toByteArray(), StandardCharsets.UTF_8);
    } catch (IOException e) {
        throw new UncheckedIOException(e);
    }
}

鉴于字节数组dbData已经在内存中，那么缓冲区中还有什么点吗？直接逐字节读入 baos 完全跳过 buffer 不是更“性能”吗？

如果读取方法正在进行底层操作系统调用，则缓冲区是有意义的，但它不在这里......

Answer 1

Given that the byte array dbData is already in memory, is there any point in the buffer?

缓冲区用于提高性能。与一次读取一个字节相比，它们通常确实可以提高性能，这是这里唯一的选择。

Is it not more "performant" to read byte-by-byte straight into the baos skipping the buffer altogether?

您的 baos 中的字节已被压缩，如果您可以读取这些字节，则您不会使用 GZIPInputStream。

<小时/>

如果您想要效率和简单性，我建议您直接从 ByteArrayInput/OutputStream 读取/写入，而不使用 byte[]。

public byte[] convertToDatabaseColumn(String text) {
    try (ByteArrayOutputStream baos = new ByteArrayOutputStream();
         Writer out = new OutputStreamWriter(
                      new GZIPOutputStream(baos), StandardCharsets.UTF_8))) {
        out.write(text);
        out.close();    
        return baos.toByteArray();
    } catch (IOException e) {
        throw new UncheckedIOException(e);
    }
}

public String convertToEntityAttribute(byte[] dbData) {
    try (Reader reader = new InputStreamReader(
                         new GZIPInputStream(new ByteArrayInputStream(dbData)),
                         StandardCharsets.UTF_8) {

        char[] chars = new char[512];
        StringBuilder sb = new StringBuilder();
        for (int len; (len = reader.read(chars)) > 0;)
            sb.append(chars, 0, len);

        return sb.toString();
    } catch (IOException e) {
        throw new UncheckedIOException(e);
    }
}

为了简化这一点，假设您的字符串不包含换行符，您可以这样做

public static byte[] convertToDatabaseColumn(String text) throws IOException {
    try (ByteArrayOutputStream baos = new ByteArrayOutputStream();
         Writer out = new OutputStreamWriter(
                 new GZIPOutputStream(baos), StandardCharsets.UTF_8)) {
        out.write(text);
        out.write("\n");
        out.close();
        return baos.toByteArray();
    } catch (IOException e) {
        throw new UncheckedIOException(e);
    }
}

public static String convertToEntityAttribute(byte[] dbData) throws IOException {
    try (BufferedReader br = new BufferedReader(
            new InputStreamReader(
                    new GZIPInputStream(new ByteArrayInputStream(dbData)),
                    StandardCharsets.UTF_8))) {

        return br.readLine();
    } catch (IOException e) {
        throw new UncheckedIOException(e);
    }
}

public static void main(String[] args) throws IOException {
    byte[] bytes = convertToDatabaseColumn("Hello world, 0123456789 0123456789");
    System.out.println(convertToEntityAttribute(bytes));
}