'commit'

src/main/java/com/dsideal/base/Test/ReadWordTOC.java

@@ -9,6 +9,7 @@ import org.dom4j.Element;
 import org.dom4j.io.SAXReader;
 
 import java.io.*;
+import java.util.Arrays;
 import java.util.List;
 import java.util.zip.ZipEntry;
 import java.util.zip.ZipInputStream;
@@ -59,7 +60,7 @@ public class ReadWordTOC {
         Document document = reader.read(new File(xmlPath));
         Element root = document.getRootElement();// 获取根元素
         List<Element> children = root.element("body").elements("p");//工作区
-
+        boolean out = false;
         for (Element child : children) {
             if (child.getName().equals("p")) {
 //                if (child.element("pPr") == null) continue;
@@ -84,17 +85,46 @@ public class ReadWordTOC {
                             for (Element t : pChild.elements("t")) {
                                 //此元素的字体与字号,同受它同级上面<rPr>的<rFonts>属性ascii控制
                                 String font = pChild.element("rPr").element("rFonts").attribute("ascii").getValue();
-                                if (font.equals("方正仿宋简体")) continue;
                                 if (font.equals("华文楷体")) continue;
                                 int fontSize = Integer.parseInt(pChild.element("rPr").element("sz").attribute("val").getValue());
                                 if (fontSize != 32) continue;
+                                if (font.equals("方正仿宋简体")) {
+                                    //加粗，有/b标签的保留，其它的不保留continue掉
+                                    if (pChild.element("rPr").element("b") == null) continue;
+                                }
+
                                 content = content + t.getText();
                             }
                         }
                     }
                 }
-                if (!StrKit.isBlank(content))
-                    System.out.println(content);
+
+                if (!StrKit.isBlank(content)) {
+                    //如果content是 "图"+数字形式的，不输出
+                    if (!content.contains("(图") && !content.contains("（图")) {
+                        //输出全部内容
+//                        if(content.startsWith("（")) System.out.print("\t");
+//                        //如果content是以 数字+.开头的，那么多输出两个tab
+//                        if (content.matches("^[0-9]+\\..*")) System.out.print("\t\t");
+//                        System.out.println(content);
+                        //只输出四和五
+                        String[] printDx = {"一", "二", "三", "六", "七", "八", "九", "十"};
+                        //转为 List<String>
+                        List<String> printDxList = Arrays.asList(printDx);
+
+                        if (content.startsWith("四") || content.startsWith("五")) {
+                            out = true;
+                        }
+                        for (String s : printDxList) {
+                            if (content.startsWith(s)) out = false;
+                        }
+
+                        if (out) {
+                            if(content.startsWith("（")) System.out.print("\t");
+                            System.out.println(content);
+                        }
+                    }
+                }
             }
         }
     }