java - java中读取XML标签，代码优化

Question

我实际上做的是一个递归函数，它读取 xml 中的标签。下面是代码:

private void readTag(org.w3c.dom.Node item, String histoTags, String fileName, Hashtable<String, String> tagsInfos) {
    try {
        if (item.getNodeType() == Node.ELEMENT_NODE) {
            NodeList itemChilds = item.getChildNodes();

            for (int i=0; i < itemChilds.getLength(); i++) {
                org.w3c.dom.Node itemChild = itemChilds.item(i);
                readTag(itemChild, histoTags + "|" + item.getNodeName(), fileName, tagsInfos);
            }
      }
       else if (item.getNodeType() == Node.TEXT_NODE) {
           tagsInfosSoft.put(histoTags, item.getNodeValue());
      }
}

该函数需要一些时间来执行。该函数读取的 xml 格式如下:

<?xml version="1.0" encoding="UTF-8"?>
<Document>
     <Mouvement>
       <Com>
          <IdCom>32R01000000772669473</IdCom>
          <RefCde>32R</RefCde>
          <Edit>0</Edit>
       <Com>
     <Mouvement>
<Document>

有什么方法可以优化java中的这段代码吗？

Answer 1

两个优化，不知道有多大帮助:

• 请勿使用getChildNodes() 。使用getFirstChild()和getNextSibling() .
• 重复使用单个StringBuilder而不是为每个元素创建一个新元素(由 histoTags + "|" + item.getNodeName() 隐式完成)。

但是，您还应该注意，元素节点的文本内容可能会被视为多个 TEXT 和 CDATA 节点的组合。

如果您的代码适用于元素而不是节点，它也会更好地工作。

private static void readTag(Element elem, StringBuilder histoTags, String fileName, Hashtable<String, String> tagsInfos) {
    int histoLen = histoTags.length();
    CharSequence textContent = null;
    boolean hasChildElement = false;
    for (Node child = elem.getFirstChild(); child != null; child = child.getNextSibling()) {
        switch (child.getNodeType()) {
            case Node.ELEMENT_NODE:
                histoTags.append('|').append(child.getNodeName());
                readTag((Element)child, histoTags, fileName, tagsInfos);
                histoTags.setLength(histoLen);
                hasChildElement = true;
                break;
            case Node.TEXT_NODE:
            case Node.CDATA_SECTION_NODE:
                //uncomment to test: System.out.println(histoTags + ": \"" + child.getTextContent() + "\"");
                if (textContent == null)
                    // Optimization: Don't copy to a StringBuilder if only one text node will be found 
                    textContent = child.getTextContent();
                else if (textContent instanceof StringBuilder)
                    // Ok, now we need a StringBuilder to collect text from multiple nodes
                    ((StringBuilder)textContent).append(child.getTextContent());
                else
                    // And we keep collecting text from multiple nodes
                    textContent = new StringBuilder(textContent).append(child.getTextContent());
                break;
            default:
                // ignore all others
        }
    }
    if (textContent != null) {
        String text = textContent.toString();
        // Suppress pure whitespace content on elements with child elements, i.e. structural whitespace
        if (! hasChildElement || ! text.trim().isEmpty())
            tagsInfos.put(histoTags.toString(), text);
    }
}

测试

String xml = "<root>\n" +
             "  <tag>hello <![CDATA[world]]> Foo <!-- comment --> Bar</tag>\n" +
             "</root>\n";
Element docElem = DocumentBuilderFactory.newInstance()
                                        .newDocumentBuilder()
                                        .parse(new InputSource(new StringReader(xml)))
                                        .getDocumentElement();
Hashtable<String, String> tagsInfos = new Hashtable<>();
readTag(docElem, new StringBuilder(docElem.getNodeName()), "fileName", tagsInfos);
System.out.println(tagsInfos);

输出(打印内容未注释)

root: "
  "
root|tag: "hello "
root|tag: "world"
root|tag: " Foo "
root|tag: " Bar"
root: "
"
{root|tag=hello world Foo  Bar}

查看如何拆分 <tag> 内的文本使用 CDATA 和注释的节点导致 DOM 节点包含多个 TEXT/CDATA 子节点。