使用POI读取Word207和Excel2007实例

1. /**
2. * 需要的jar包：
3. * poi-3.0.2-FINAL-20080204.jar
4. * poi-contrib-3.0.2-FINAL-20080204.jar
5. * poi-scratchpad-3.0.2-FINAL-20080204.jar
6. * poi-3.5-beta6-20090622.jar
7. * geronimo-stax-api_1.0_spec-1.0.jar
8. * ooxml-schemas-1.0.jar
9. * openxml4j-bin-beta.jar
10. * poi-ooxml-3.5-beta6-20090622.jar
11. * xmlbeans-2.3.0.jar
12. * dom4j-1.6.1.jar
13. */  
14.   
15. import java.io.FileInputStream;   
16. import java.io.IOException;   
17. import java.io.InputStream;   
18.   
19. import org.apache.poi.POIXMLDocument;   
20. import org.apache.poi.POIXMLTextExtractor;   
21. import org.apache.poi.hssf.usermodel.HSSFCell;   
22. import org.apache.poi.hssf.usermodel.HSSFRow;   
23. import org.apache.poi.hssf.usermodel.HSSFSheet;   
24. import org.apache.poi.hssf.usermodel.HSSFWorkbook;   
25. import org.apache.poi.hwpf.extractor.WordExtractor;   
26. import org.apache.poi.openxml4j.exceptions.OpenXML4JException;   
27. import org.apache.poi.openxml4j.opc.OPCPackage;   
28. import org.apache.poi.xssf.usermodel.XSSFCell;   
29. import org.apache.poi.xssf.usermodel.XSSFRow;   
30. import org.apache.poi.xssf.usermodel.XSSFSheet;   
31. import org.apache.poi.xssf.usermodel.XSSFWorkbook;   
32. import org.apache.poi.xwpf.extractor.XWPFWordExtractor;   
33. import org.apache.xmlbeans.XmlException;   
34.   
35. public class WordAndExcelExtractor {   
36. public static void main(String[] args){   
37.   try{   
38.     String wordFile = "D:/松山血战.docx";   
39.     String wordText2007 = WordAndExcelExtractor.extractTextFromDOC2007(wordFile);   
40.     System.out.println("wordText2007======="+wordText2007);   
41.       
42.     InputStream is = new FileInputStream("D:/XXX研发中心技术岗位职位需求.xls");      
43.     String excelText = WordAndExcelExtractor.extractTextFromXLS(is);      
44.     System.out.println("text2003==========" + excelText);   
45.       
46.     String excelFile = "D:/Hello2007.xlsx";      
47.     String excelText2007 = WordAndExcelExtractor.extractTextFromXLS2007(excelFile);   
48.     System.out.println("excelText2007==========" + excelText2007);   
49.   
50.       
51.    }catch(Exception e ){   
52.     e.printStackTrace();   
53.    }   
54. }   
55.   
56. /**
57.    * @Method: extractTextFromDOCX
58.    * @Description: 从word 2003文档中提取纯文本
59.    *
60.    * @param
61.    * @return String
62.    * @throws
63.    */  
64.     public static String extractTextFromDOC(InputStream is) throws IOException {   
65.          WordExtractor ex = new WordExtractor(is); //is是WORD文件的InputStream   
66.   
67.         return ex.getText();   
68.      }   
69.   
70. /**
71.    * @Method: extractTextFromDOCX
72.    * @Description: 从word 2007文档中提取纯文本
73.    *
74.    * @param
75.    * @return String
76.    * @throws
77.    */  
78.     public static String extractTextFromDOC2007(String fileName) throws IOException, OpenXML4JException, XmlException {   
79.       OPCPackage opcPackage = POIXMLDocument.openPackage(fileName);   
80.       POIXMLTextExtractor ex = new XWPFWordExtractor(opcPackage);      
81.   
82.         return ex.getText();   
83.      }   
84.   
85. /**
86.    * @Method: extractTextFromXLS
87.    * @Description: 从excel 2003文档中提取纯文本
88.    *
89.    * @param
90.    * @return String
91.    * @throws
92.    */  
93.     @SuppressWarnings("deprecation")   
94. private static String extractTextFromXLS(InputStream is)   
95.         throws IOException {   
96.          StringBuffer content   = new StringBuffer();   
97.          HSSFWorkbook workbook = new HSSFWorkbook(is); //创建对Excel工作簿文件的引用   
98.   
99.         for (int numSheets = 0; numSheets < workbook.getNumberOfSheets(); numSheets++) {   
100.             if (null != workbook.getSheetAt(numSheets)) {   
101.                  HSSFSheet aSheet = workbook.getSheetAt(numSheets); //获得一个sheet   
102.   
103.                 for (int rowNumOfSheet = 0; rowNumOfSheet <= aSheet.getLastRowNum(); rowNumOfSheet++) {   
104.                     if (null != aSheet.getRow(rowNumOfSheet)) {   
105.                          HSSFRow aRow = aSheet.getRow(rowNumOfSheet); //获得一行   
106.   
107.                         for (short cellNumOfRow = 0; cellNumOfRow <= aRow.getLastCellNum(); cellNumOfRow++) {   
108.                             if (null != aRow.getCell(cellNumOfRow)) {   
109.                                  HSSFCell aCell = aRow.getCell(cellNumOfRow); //获得列值   
110.                                                                    
111.                                 if(aCell.getCellType() == HSSFCell.CELL_TYPE_NUMERIC){   
112.                                   content.append(aCell.getNumericCellValue());   
113.                                  }else if(aCell.getCellType() == HSSFCell.CELL_TYPE_BOOLEAN){   
114.                                   content.append(aCell.getBooleanCellValue());   
115.                                  }else {   
116.                                   content.append(aCell.getStringCellValue());   
117.                                  }   
118.                              }   
119.                          }   
120.                      }   
121.                  }   
122.              }   
123.          }   
124.   
125.         return content.toString();   
126.      }   
127.        
128.     /**
129.       * @Method: extractTextFromXLS2007
130.       * @Description: 从excel 2007文档中提取纯文本
131.       *
132.       * @param
133.       * @return String
134.       * @throws
135.       */  
136.     private static String extractTextFromXLS2007(String fileName) throws Exception{   
137.       StringBuffer content = new StringBuffer();   
138.         
139.      //构造 XSSFWorkbook 对象，strPath 传入文件路径       
140.    XSSFWorkbook xwb = new XSSFWorkbook(fileName);   
141.      
142.   //循环工作表Sheet   
143.   for(int numSheet = 0; numSheet < xwb.getNumberOfSheets(); numSheet++){   
144.     XSSFSheet xSheet = xwb.getSheetAt(numSheet);   
145.    if(xSheet == null){   
146.     continue;   
147.     }   
148.       
149.    //循环行Row   
150.    for(int rowNum = 0; rowNum <= xSheet.getLastRowNum(); rowNum++){   
151.      XSSFRow xRow = xSheet.getRow(rowNum);   
152.     if(xRow == null){   
153.      continue;   
154.      }   
155.        
156.     //循环列Cell   
157.     for(int cellNum = 0; cellNum <= xRow.getLastCellNum(); cellNum++){   
158.       XSSFCell xCell = xRow.getCell(cellNum);   
159.      if(xCell == null){   
160.       continue;   
161.       }   
162.         
163.      if(xCell.getCellType() == XSSFCell.CELL_TYPE_BOOLEAN){   
164.        content.append(xCell.getBooleanCellValue());   
165.       }else if(xCell.getCellType() == XSSFCell.CELL_TYPE_NUMERIC){   
166.        content.append(xCell.getNumericCellValue());   
167.       }else{   
168.        content.append(xCell.getStringCellValue());   
169.       }   
170.      }   
171.     }   
172.    }   
173.      
174.   return content.toString();   
175.      }   
176.        
177. }