(2) POI 读word表格

第一段

Java code

    
    
    
    

     
     
     
     try
     
     
     
      { FileInputStream in 
     
     
     
     =
     
     
     
      
     
     
     
     new
     
     
     
      FileInputStream(
     
     
     
     "
     
     
     
     D:\\XBRL.doc
     
     
     
     "
     
     
     
     );
     
     
     
     //
     
     
     
      载入文档
     
     
     
      
     
     
     
      POIFSFileSystem pfs 
     
     
     
     =
     
     
     
      
     
     
     
     new
     
     
     
      POIFSFileSystem(in); HWPFDocument hwpf 
     
     
     
     =
     
     
     
      
     
     
     
     new
     
     
     
      HWPFDocument(pfs); Range range 
     
     
     
     =
     
     
     
      hwpf.getRange();
     
     
     
     //
     
     
     
      得到文档的读取范围
     
     
     
      
     
     
     
      
     
     
     
     int
     
     
     
      paraNum 
     
     
     
     =
     
     
     
      range.numParagraphs(); 
     
     
     
     int
     
     
     
      m 
     
     
     
     =
     
     
     
      
     
     
     
     0
     
     
     
     ; 
     
     
     
     //
     
     
     
      数组下标
     
     
     
      
     
     
     
      String[] ret 
     
     
     
     =
     
     
     
      
     
     
     
     new
     
     
     
      String[paraNum]; 
     
     
     
     for
     
     
     
      (
     
     
     
     int
     
     
     
      i 
     
     
     
     =
     
     
     
      
     
     
     
     0
     
     
     
     ; i 
     
     
     
     <
     
     
     
      paraNum; 
     
     
     
     ++
     
     
     
     i) { 
     
     
     
     //
     
     
     
      从每一段落中获取文字,每一段是一个回车
     
     
     
      
     
     
     
      Paragraph p 
     
     
     
     =
     
     
     
      range.getParagraph(i); 
     
     
     
     boolean
     
     
     
      test 
     
     
     
     =
     
     
     
      p.isInTable(); 
     
     
     
     //
     
     
     
      判断该Paragraph是否在word的表格中
     
     
     
      
     
     
     
      
     
     
     
     if
     
     
     
      (test 
     
     
     
     ==
     
     
     
      
     
     
     
     true
     
     
     
     ) { Table table 
     
     
     
     =
     
     
     
      range.getTable(p); 
     
     
     
     //
     
     
     
      通过第一个在table中的Paragraph来获取整个table
     
     
     
      
     
     
     
      
     
     
     
     int
     
     
     
      numRow 
     
     
     
     =
     
     
     
      table.numRows(); 
     
     
     
     //
     
     
     
      获取table中的行数
     
     
     
      
     
     
     
      
     
     
     
     for
     
     
     
      (
     
     
     
     int
     
     
     
      j 
     
     
     
     =
     
     
     
      
     
     
     
     0
     
     
     
     ; j 
     
     
     
     <
     
     
     
      numRow; j
     
     
     
     ++
     
     
     
     ) { TableRow tablerow 
     
     
     
     =
     
     
     
      table.getRow(j); 
     
     
     
     //
     
     
     
      获得一行
     
     
     
      
     
     
     
      
     
     
     
     int
     
     
     
      numbercell 
     
     
     
     =
     
     
     
      tablerow.numCells(); 
     
     
     
     //
     
     
     
      通过tablerow获取单元格个数
     
     
     
      
     
     
     
      
     
     
     
     for
     
     
     
      (
     
     
     
     int
     
     
     
      k 
     
     
     
     =
     
     
     
      
     
     
     
     0
     
     
     
     ; k 
     
     
     
     <
     
     
     
      numbercell; k
     
     
     
     ++
     
     
     
     ) { TableCell tablexell 
     
     
     
     =
     
     
     
      tablerow.getCell(k); 
     
     
     
     //
     
     
     
      获得单元格
     
     
     
      
     
     
     
      
     
     
     
     int
     
     
     
      p1 
     
     
     
     =
     
     
     
      tablexell.numParagraphs(); 
     
     
     
     //
     
     
     
      获取单元格中的Paragraph的个数
     
     
     
      
     
     
     
      String str 
     
     
     
     =
     
     
     
      
     
     
     
     ""
     
     
     
     ; 
     
     
     
     for
     
     
     
      (
     
     
     
     int
     
     
     
      l 
     
     
     
     =
     
     
     
      
     
     
     
     0
     
     
     
     ; l 
     
     
     
     <
     
     
     
      p1; l
     
     
     
     ++
     
     
     
     ) { Paragraph para 
     
     
     
     =
     
     
     
      tablexell.getParagraph(l); str 
     
     
     
     =
     
     
     
      str 
     
     
     
     +
     
     
     
      para.text().trim(); 
     
     
     
     //
     
     
     
      得到单元格中的内容
     
     
     
      
     
     
     
      } 
     
     
     
     for
     
     
     
      (
     
     
     
     int
     
     
     
      n 
     
     
     
     =
     
     
     
      m; n 
     
     
     
     <
     
     
     
      paraNum;) { ret[n] 
     
     
     
     =
     
     
     
      str; 
     
     
     
     //
     
     
     
      将单元格中的内容方入数组元素中
     
     
     
      
     
     
     
      
     
     
     
     break
     
     
     
     ; } m
     
     
     
     ++
     
     
     
     ; } } } } 
     
     
     
     for
     
     
     
      (
     
     
     
     int
     
     
     
      i 
     
     
     
     =
     
     
     
      
     
     
     
     0
     
     
     
     ; i 
     
     
     
     <
     
     
     
      paraNum; i
     
     
     
     ++
     
     
     
     ) { System.out.println(ret[i]); } } 
     
     
     
     catch
     
     
     
      (Exception e) { e.printStackTrace(); }
    
    
    
    

第二段

Java code

    
    
    
    

     
     
     
     try
     
     
     
      { POIFSFileSystem fs 
     
     
     
     =
     
     
     
      
     
     
     
     new
     
     
     
      POIFSFileSystem(
     
     
     
     new
     
     
     
      FileInputStream( 
     
     
     
     "
     
     
     
     D:\\XBRL.doc
     
     
     
     "
     
     
     
     )); HWPFDocument doc 
     
     
     
     =
     
     
     
      
     
     
     
     new
     
     
     
      HWPFDocument(fs); Range range 
     
     
     
     =
     
     
     
      doc.getOverallRange(); TableIterator ti 
     
     
     
     =
     
     
     
      
     
     
     
     new
     
     
     
      TableIterator(range); 
     
     
     
     //
     
     
     
      Table table = null;
     
     
     
      
     
     
     
      
     
     
     
     while
     
     
     
      (ti.hasNext()) { System.out.println(
     
     
     
     "
     
     
     
     Getting table!
     
     
     
     "
     
     
     
     ); Table table 
     
     
     
     =
     
     
     
      (Table)ti.next(); System.out.println(
     
     
     
     "
     
     
     
     Number of rows: 
     
     
     
     "
     
     
     
      
     
     
     
     +
     
     
     
      table.numRows()); 
     
     
     
     for
     
     
     
      (
     
     
     
     int
     
     
     
      a 
     
     
     
     =
     
     
     
      
     
     
     
     0
     
     
     
     ; a 
     
     
     
     <
     
     
     
      table.numRows(); a
     
     
     
     ++
     
     
     
     ) { TableRow row 
     
     
     
     =
     
     
     
      table.getRow(a); System.out.println(
     
     
     
     "
     
     
     
     \tTable row number: 
     
     
     
     "
     
     
     
      
     
     
     
     +
     
     
     
      a); 
     
     
     
     for
     
     
     
      (
     
     
     
     int
     
     
     
      b 
     
     
     
     =
     
     
     
      
     
     
     
     0
     
     
     
     ; b 
     
     
     
     <
     
     
     
      row.numCells(); b
     
     
     
     ++
     
     
     
     ) { System.out.println(
     
     
     
     "
     
     
     
     \t\tTable cell number: 
     
     
     
     "
     
     
     
      
     
     
     
     +
     
     
     
      b); TableCell tablecell 
     
     
     
     =
     
     
     
      row.getCell(b); 
     
     
     
     for
     
     
     
      (
     
     
     
     int
     
     
     
      c 
     
     
     
     =
     
     
     
      
     
     
     
     0
     
     
     
     ; c 
     
     
     
     <
     
     
     
      tablecell.numParagraphs(); c
     
     
     
     ++
     
     
     
     ) { Paragraph tablepara 
     
     
     
     =
     
     
     
      tablecell.getParagraph(c); 
     
     
     
     for
     
     
     
      (
     
     
     
     int
     
     
     
      d 
     
     
     
     =
     
     
     
      
     
     
     
     0
     
     
     
     ; d 
     
     
     
     <
     
     
     
      tablepara.numCharacterRuns(); d
     
     
     
     ++
     
     
     
     ) { CharacterRun run 
     
     
     
     =
     
     
     
      tablepara.getCharacterRun(d); System.out.println(
     
     
     
     "
     
     
     
     \t\tText: \"
     
     
     
     "
     
     
     
      
     
     
     
     +
     
     
     
      run.text() 
     
     
     
     +
     
     
     
      
     
     
     
     "
     
     
     
     \"
     
     
     
     "
     
     
     
     ); } } } } System.out.println(
     
     
     
     "
     
     
     
     \n
     
     
     
     "
     
     
     
     ); } } 
     
     
     
     catch
     
     
     
      (Exception e) { e.printStackTrace(); }