如何防止 javax.xml.transform.Transformer 用 UTF-8 编码中的数字字符引用替换非 BMP Unicode 字符？

Question

背景

我想编写一个包含 UTF-8 编码的非 BMP 字符的 XML 文件。

问题

使用以下代码，生成的 XML 文件将非 BMP Unicode 字符替换为数字字符引用。

package xml;

import java.io.File;

import javax.xml.parsers.DocumentBuilder;
import javax.xml.parsers.DocumentBuilderFactory;
import javax.xml.parsers.ParserConfigurationException;
import javax.xml.transform.Transformer;
import javax.xml.transform.TransformerConfigurationException;
import javax.xml.transform.TransformerException;
import javax.xml.transform.TransformerFactory;
import javax.xml.transform.dom.DOMSource;
import javax.xml.transform.stream.StreamResult;

import org.w3c.dom.Document;
import org.w3c.dom.Element;

public class XMLClass {

    static String names[] = {"𠀋一郎", "𠮷野","辻󠄀","👨‍👩‍👦"};

    public static void main(String[] args) {

        DocumentBuilder documentBuilder = null;
        try {
            documentBuilder = DocumentBuilderFactory.newInstance().newDocumentBuilder();
        } catch (ParserConfigurationException e) {
            e.printStackTrace();
        }
        Document document = documentBuilder.newDocument();
        document.setXmlStandalone(true);

        Element list = document.createElement("list");
        document.appendChild(list);
        for (int i = 0; i < names.length; i++) {
            Element name = (Element) document.createElement("name").cloneNode(false);
            list.appendChild(name);
            name.appendChild(document.createTextNode(names[i]));
        }

        File file = new File("NameList.xml");
        write(file, document);
    }


    public static boolean write(File file, Document document) {

        Transformer transformer = null;
        try {
             TransformerFactory transformerFactory = TransformerFactory.newInstance();
             transformer = transformerFactory.newTransformer();
        } catch (TransformerConfigurationException e) {
             e.printStackTrace();
             return false;
        }

        transformer.setOutputProperty("indent", "yes");
        // non-BMP characters written in characters (no numeric character reference style)
        // when you set encoding UTF-16
        transformer.setOutputProperty("encoding", "UTF-8");
        transformer.setOutputProperty("{http://xml.apache.org/xalan}indent-amount", "2");

        try {
             transformer.transform(new DOMSource(document), new StreamResult(
                       file));
        } catch (TransformerException e) {
             e.printStackTrace();
             return false;
        }

        return true;
   }
}

我的期望是：

<?xmlversion="1.0"encoding="UTF-8"?><list>
<name>𠀋一郎</name>
<name>𠮷野</name>
<name>辻󠄀</name>
<name>👨‍👩‍👦</name>
</list>

但是我得到的是：

<?xmlversion="1.0"encoding="UTF-8"?><list>
<name>&#131083;一郎</name>
<name>&#134071;野</name>
<name>辻&#917760;</name>
<name>&#128104;‍&#128105;‍&#128102;</name>
</list>

问题

当我指定使用 UTF-8 编码时，如何防止

javax.xml.transform.Transformer

将非 BMP Unicode 字符替换为数字字符引用？

Answer 1

这不是一个漂亮的解决方案，但是......您可以写入管道而不是文件，然后在不同的线程中读取该管道并自行修复数字实体，然后将该固定版本发送到文件：

private static void writeWithoutNumericEntities(Reader xml,
                                                File file) {
    StringBuilder entity = new StringBuilder();

    try (Writer out = Files.newBufferedWriter(file.toPath())) {
        int c;
        while ((c = xml.read()) >= 0) {
            if (c == '&') {
                c = xml.read();
                if (c == '#') {
                    entity.setLength(0);
                    int codepoint;

                    c = xml.read();
                    if (c == 'x' || c == 'X') {
                        while ((c = xml.read()) != ';') {
                            entity.append((char) c);
                        }
                        codepoint = Integer.parseInt(entity.toString(), 16);
                    } else {
                        entity.setLength(0);
                        do {
                            entity.append((char) c);
                        } while ((c = xml.read()) != ';');
                        codepoint = Integer.parseInt(entity.toString());
                    }

                    out.write(Character.toChars(codepoint));
                } else {
                    out.write('&');
                    out.write(c);
                }
            } else {
                out.write(c);
            }
        }
    } catch (IOException e) {
        throw new UncheckedIOException(e);
    }
}

public static boolean write(File file, Document document) {

    Transformer transformer = null;
    try {
        TransformerFactory transformerFactory = TransformerFactory.newInstance();
        transformer = transformerFactory.newTransformer();

        transformer.setOutputProperty("indent", "yes");
        // non-BMP characters written in characters (no numeric character reference style)
        // when you set encoding UTF-16
        transformer.setOutputProperty("encoding", "UTF-8");
        transformer.setOutputProperty("{http://xml.apache.org/xalan}indent-amount", "2");

        Thread fixXmlThread;
        try (PipedWriter destination = new PipedWriter()) {
            PipedReader xmlWithNumericEntities = new PipedReader(destination);
            fixXmlThread = Thread.startVirtualThread(
                () -> writeWithoutNumericEntities(xmlWithNumericEntities, file));

            transformer.transform(new DOMSource(document), new StreamResult(destination));
        }

        fixXmlThread.join();
    } catch (TransformerException | IOException | InterruptedException e) {
         e.printStackTrace();
         return false;
    }

    return true;
}

这不考虑注释和 CDATA 部分，它们可能包含会破坏上述代码的文本。 不过，可以对其进行修改以考虑注释和 CDATA。