这是我自己问题的答案。我将各种结果的答案结合起来，编写了一个输出XML的类。

不保证它如何响应无效的XML或大型文档。

package ecb.sdw.pretty;

import org.apache.xml.serialize.OutputFormat;
import org.apache.xml.serialize.XMLSerializer;
import org.w3c.dom.Document;
import org.xml.sax.InputSource;
import org.xml.sax.SAXException;

import javax.xml.parsers.DocumentBuilder;
import javax.xml.parsers.DocumentBuilderFactory;
import javax.xml.parsers.ParserConfigurationException;
import java.io.IOException;
import java.io.StringReader;
import java.io.StringWriter;
import java.io.Writer;

/**
 * Pretty-prints xml, supplied as a string.
 * <p/>
 * eg.
 * <code>
 * String formattedXml = new XmlFormatter().format("<tag><nested>hello</nested></tag>");
 * </code>
 */
public class XmlFormatter {

    public XmlFormatter() {
    }

    public String format(String unformattedXml) {
        try {
            final Document document = parseXmlFile(unformattedXml);

            OutputFormat format = new OutputFormat(document);
            format.setLineWidth(65);
            format.setIndenting(true);
            format.setIndent(2);
            Writer out = new StringWriter();
            XMLSerializer serializer = new XMLSerializer(out, format);
            serializer.serialize(document);

            return out.toString();
        } catch (IOException e) {
            throw new RuntimeException(e);
        }
    }

    private Document parseXmlFile(String in) {
        try {
            DocumentBuilderFactory dbf = DocumentBuilderFactory.newInstance();
            DocumentBuilder db = dbf.newDocumentBuilder();
            InputSource is = new InputSource(new StringReader(in));
            return db.parse(is);
        } catch (ParserConfigurationException e) {
            throw new RuntimeException(e);
        } catch (SAXException e) {
            throw new RuntimeException(e);
        } catch (IOException e) {
            throw new RuntimeException(e);
        }
    }

    public static void main(String[] args) {
        String unformattedXml =
                "<?xml version=\"1.0\" encoding=\"UTF-8\"?><QueryMessage\n" +
                        "        xmlns=\"http://www.SDMX.org/resources/SDMXML/schemas/v2_0/message\"\n" +
                        "        xmlns:query=\"http://www.SDMX.org/resources/SDMXML/schemas/v2_0/query\">\n" +
                        "    <Query>\n" +
                        "        <query:CategorySchemeWhere>\n" +
                        "   \t\t\t\t\t         <query:AgencyID>ECB\n\n\n\n</query:AgencyID>\n" +
                        "        </query:CategorySchemeWhere>\n" +
                        "    </Query>\n\n\n\n\n" +
                        "</QueryMessage>";

        System.out.println(new XmlFormatter().format(unformattedXml));
    }

}

I have found that in Java 1.6.0_32 the normal method to pretty print an XML string (using a Transformer with a null or identity xslt) does not behave as I would like if tags are merely separated by whitespace, as opposed to having no separating text. I tried using <xsl:strip-space elements="*"/> in my template to no avail. The simplest solution I found was to strip the space the way I wanted using a SAXSource and XML filter. Since my solution was for logging I also extended this to work with incomplete XML fragments. Note the normal method seems to work fine if you use a DOMSource but I did not want to use this because of the incompleteness and memory overhead.

public static class WhitespaceIgnoreFilter extends XMLFilterImpl
{

    @Override
    public void ignorableWhitespace(char[] arg0,
                                    int arg1,
                                    int arg2) throws SAXException
    {
        //Ignore it then...
    }

    @Override
    public void characters( char[] ch,
                            int start,
                            int length) throws SAXException
    {
        if (!new String(ch, start, length).trim().equals("")) 
               super.characters(ch, start, length); 
    }
}

public static String prettyXML(String logMsg, boolean allowBadlyFormedFragments) throws SAXException, IOException, TransformerException
    {
        TransformerFactory transFactory = TransformerFactory.newInstance();
        transFactory.setAttribute("indent-number", new Integer(2));
        Transformer transformer = transFactory.newTransformer();
        transformer.setOutputProperty(OutputKeys.INDENT, "yes");
        transformer.setOutputProperty("{http://xml.apache.org/xslt}indent-amount", "4");
        StringWriter out = new StringWriter();
        XMLReader masterParser = SAXHelper.getSAXParser(true);
        XMLFilter parser = new WhitespaceIgnoreFilter();
        parser.setParent(masterParser);

        if(allowBadlyFormedFragments)
        {
            transformer.setErrorListener(new ErrorListener()
            {
                @Override
                public void warning(TransformerException exception) throws TransformerException
                {
                }

                @Override
                public void fatalError(TransformerException exception) throws TransformerException
                {
                }

                @Override
                public void error(TransformerException exception) throws TransformerException
                {
                }
            });
        }

        try
        {
            transformer.transform(new SAXSource(parser, new InputSource(new StringReader(logMsg))), new StreamResult(out));
        }
        catch (TransformerException e)
        {
            if(e.getCause() != null && e.getCause() instanceof SAXParseException)
            {
                if(!allowBadlyFormedFragments || !"XML document structures must start and end within the same entity.".equals(e.getCause().getMessage()))
                {
                    throw e;
                }
            }
            else
            {
                throw e;
            }
        }
        out.flush();
        return out.toString();
    }

我总是使用下面的函数:

public static String prettyPrintXml(String xmlStringToBeFormatted) {
    String formattedXmlString = null;
    try {
        DocumentBuilderFactory documentBuilderFactory = DocumentBuilderFactory.newInstance();
        documentBuilderFactory.setValidating(true);
        DocumentBuilder documentBuilder = documentBuilderFactory.newDocumentBuilder();
        InputSource inputSource = new InputSource(new StringReader(xmlStringToBeFormatted));
        Document document = documentBuilder.parse(inputSource);

        Transformer transformer = TransformerFactory.newInstance().newTransformer();
        transformer.setOutputProperty(OutputKeys.INDENT, "yes");
        transformer.setOutputProperty("{http://xml.apache.org/xslt}indent-amount", "2");

        StreamResult streamResult = new StreamResult(new StringWriter());
        DOMSource dOMSource = new DOMSource(document);
        transformer.transform(dOMSource, streamResult);
        formattedXmlString = streamResult.getWriter().toString().trim();
    } catch (Exception ex) {
        StringWriter sw = new StringWriter();
        ex.printStackTrace(new PrintWriter(sw));
        System.err.println(sw.toString());
    }
    return formattedXmlString;
}

凯文·哈肯森说: 但是，如果您知道您的XML字符串是有效的，并且您不想引起将字符串解析为DOM的内存开销，然后在DOM上运行转换以获得字符串—您可以通过字符解析进行一些老式的字符。在每个字符后插入换行符和空格，保持和缩进计数器(以确定空格的数量)，为每个<…>和递减你看到的每一个。”

同意了。这种方法要快得多，依赖关系也少得多。

示例解决方案:

/**
 * XML utils, including formatting.
 */
public class XmlUtils
{
  private static XmlFormatter formatter = new XmlFormatter(2, 80);

  public static String formatXml(String s)
  {
    return formatter.format(s, 0);
  }

  public static String formatXml(String s, int initialIndent)
  {
    return formatter.format(s, initialIndent);
  }

  private static class XmlFormatter
  {
    private int indentNumChars;
    private int lineLength;
    private boolean singleLine;

    public XmlFormatter(int indentNumChars, int lineLength)
    {
      this.indentNumChars = indentNumChars;
      this.lineLength = lineLength;
    }

    public synchronized String format(String s, int initialIndent)
    {
      int indent = initialIndent;
      StringBuilder sb = new StringBuilder();
      for (int i = 0; i < s.length(); i++)
      {
        char currentChar = s.charAt(i);
        if (currentChar == '<')
        {
          char nextChar = s.charAt(i + 1);
          if (nextChar == '/')
            indent -= indentNumChars;
          if (!singleLine)   // Don't indent before closing element if we're creating opening and closing elements on a single line.
            sb.append(buildWhitespace(indent));
          if (nextChar != '?' && nextChar != '!' && nextChar != '/')
            indent += indentNumChars;
          singleLine = false;  // Reset flag.
        }
        sb.append(currentChar);
        if (currentChar == '>')
        {
          if (s.charAt(i - 1) == '/')
          {
            indent -= indentNumChars;
            sb.append("\n");
          }
          else
          {
            int nextStartElementPos = s.indexOf('<', i);
            if (nextStartElementPos > i + 1)
            {
              String textBetweenElements = s.substring(i + 1, nextStartElementPos);

              // If the space between elements is solely newlines, let them through to preserve additional newlines in source document.
              if (textBetweenElements.replaceAll("\n", "").length() == 0)
              {
                sb.append(textBetweenElements + "\n");
              }
              // Put tags and text on a single line if the text is short.
              else if (textBetweenElements.length() <= lineLength * 0.5)
              {
                sb.append(textBetweenElements);
                singleLine = true;
              }
              // For larger amounts of text, wrap lines to a maximum line length.
              else
              {
                sb.append("\n" + lineWrap(textBetweenElements, lineLength, indent, null) + "\n");
              }
              i = nextStartElementPos - 1;
            }
            else
            {
              sb.append("\n");
            }
          }
        }
      }
      return sb.toString();
    }
  }

  private static String buildWhitespace(int numChars)
  {
    StringBuilder sb = new StringBuilder();
    for (int i = 0; i < numChars; i++)
      sb.append(" ");
    return sb.toString();
  }

  /**
   * Wraps the supplied text to the specified line length.
   * @lineLength the maximum length of each line in the returned string (not including indent if specified).
   * @indent optional number of whitespace characters to prepend to each line before the text.
   * @linePrefix optional string to append to the indent (before the text).
   * @returns the supplied text wrapped so that no line exceeds the specified line length + indent, optionally with
   * indent and prefix applied to each line.
   */
  private static String lineWrap(String s, int lineLength, Integer indent, String linePrefix)
  {
    if (s == null)
      return null;

    StringBuilder sb = new StringBuilder();
    int lineStartPos = 0;
    int lineEndPos;
    boolean firstLine = true;
    while(lineStartPos < s.length())
    {
      if (!firstLine)
        sb.append("\n");
      else
        firstLine = false;

      if (lineStartPos + lineLength > s.length())
        lineEndPos = s.length() - 1;
      else
      {
        lineEndPos = lineStartPos + lineLength - 1;
        while (lineEndPos > lineStartPos && (s.charAt(lineEndPos) != ' ' && s.charAt(lineEndPos) != '\t'))
          lineEndPos--;
      }
      sb.append(buildWhitespace(indent));
      if (linePrefix != null)
        sb.append(linePrefix);

      sb.append(s.substring(lineStartPos, lineEndPos + 1));
      lineStartPos = lineEndPos + 1;
    }
    return sb.toString();
  }

  // other utils removed for brevity
}

我在这里找到的针对Java 1.6+的解决方案不会重新格式化已经格式化的代码。对我有效的方法(重新格式化已经格式化的代码)如下。

import org.apache.xml.security.c14n.CanonicalizationException;
import org.apache.xml.security.c14n.Canonicalizer;
import org.apache.xml.security.c14n.InvalidCanonicalizerException;
import org.w3c.dom.Element;
import org.w3c.dom.bootstrap.DOMImplementationRegistry;
import org.w3c.dom.ls.DOMImplementationLS;
import org.w3c.dom.ls.LSSerializer;
import org.xml.sax.InputSource;
import org.xml.sax.SAXException;

import javax.xml.parsers.DocumentBuilderFactory;
import javax.xml.parsers.ParserConfigurationException;
import javax.xml.transform.TransformerException;
import java.io.IOException;
import java.io.StringReader;

public class XmlUtils {
    public static String toCanonicalXml(String xml) throws InvalidCanonicalizerException, ParserConfigurationException, SAXException, CanonicalizationException, IOException {
        Canonicalizer canon = Canonicalizer.getInstance(Canonicalizer.ALGO_ID_C14N_OMIT_COMMENTS);
        byte canonXmlBytes[] = canon.canonicalize(xml.getBytes());
        return new String(canonXmlBytes);
    }

    public static String prettyFormat(String input) throws TransformerException, ParserConfigurationException, IOException, SAXException, InstantiationException, IllegalAccessException, ClassNotFoundException {
        InputSource src = new InputSource(new StringReader(input));
        Element document = DocumentBuilderFactory.newInstance().newDocumentBuilder().parse(src).getDocumentElement();
        Boolean keepDeclaration = input.startsWith("<?xml");
        DOMImplementationRegistry registry = DOMImplementationRegistry.newInstance();
        DOMImplementationLS impl = (DOMImplementationLS) registry.getDOMImplementation("LS");
        LSSerializer writer = impl.createLSSerializer();
        writer.getDomConfig().setParameter("format-pretty-print", Boolean.TRUE);
        writer.getDomConfig().setParameter("xml-declaration", keepDeclaration);
        return writer.writeToString(document);
    }
}

它是在单元测试中比较全字符串xml的好工具。

private void assertXMLEqual(String expected, String actual) throws ParserConfigurationException, IOException, SAXException, CanonicalizationException, InvalidCanonicalizerException, TransformerException, IllegalAccessException, ClassNotFoundException, InstantiationException {
    String canonicalExpected = prettyFormat(toCanonicalXml(expected));
    String canonicalActual = prettyFormat(toCanonicalXml(actual));
    assertEquals(canonicalExpected, canonicalActual);
}

如果您确信您有一个有效的XML，那么这个很简单，并且避免了XML DOM树。可能有一些错误，如果你看到任何错误，请评论

public String prettyPrint(String xml) {
            if (xml == null || xml.trim().length() == 0) return "";

            int stack = 0;
            StringBuilder pretty = new StringBuilder();
            String[] rows = xml.trim().replaceAll(">", ">\n").replaceAll("<", "\n<").split("\n");

            for (int i = 0; i < rows.length; i++) {
                    if (rows[i] == null || rows[i].trim().length() == 0) continue;

                    String row = rows[i].trim();
                    if (row.startsWith("<?")) {
                            // xml version tag
                            pretty.append(row + "\n");
                    } else if (row.startsWith("</")) {
                            // closing tag
                            String indent = repeatString("    ", --stack);
                            pretty.append(indent + row + "\n");
                    } else if (row.startsWith("<")) {
                            // starting tag
                            String indent = repeatString("    ", stack++);
                            pretty.append(indent + row + "\n");
                    } else {
                            // tag data
                            String indent = repeatString("    ", stack);
                            pretty.append(indent + row + "\n");
                    }
            }

            return pretty.toString().trim();
    }