解析word中的表格

由于word表格的特殊性,其本身中的数据本来就不够完善,不能够很好的知道其具体的合并、跨行的相关属性,表格的单位可能是PT或者是百分比,并且是共存的,为处理带来了一定的负担,本代码实现了一个将Word表格解析为XMLTable。

 由于没找到上传附件的位置,仅提供部分代码,有需要的可email我(提供联系方式)或进群:490571636,我会提供全部代码。

/// <summary>         /// 分析表格通过位置跨度         /// </summary>         /// <param name="table">word表格对象</param>         /// <returns>自定义的表格对象</returns>         public static WordTable ParserTableByPositionSpan(Word.Table table)         {             List<double> positionList = new List<double>();             WordTable wordTable = new WordTable();

wordTable.RowCount = table.Rows.Count;             wordTable.ColCount = table.Columns.Count;             positionList.Add(0);             double tableWidth = 0;             //获取所有行中的单元格及位置列表 如果有的列没有获取,说明没有或被上边的单元格合并(只有上边被合并的列才会)             for (var row = 1; row <= table.Rows.Count; row++)             {                 WTRow currRow = wordTable.AddRow();                 currRow.RowIndex = row;                 double leftPosition = 0;                 for (var col = 1; col <= table.Columns.Count; col++)                 {                     Word.Cell cell = null;                     try                     {                         //获取指定位置的单元格,如果没有会触发异常                         cell = table.Cell(row, col);                     }                     catch (Exception e)                     {                         //System.Diagnostics.Trace.                     }                                         //指定的位置有单元格则直接添加到当前行                     if (cell != null)                     {                         //第一行的宽度肯定会有值不会了现9999999的情况                         if (row == 1)                         {                             tableWidth += cell.Width;                         }                                                                         WTCol wTCol = currRow.AddCol();                         wTCol.Value = GetRangeParagraphs(cell.Range);

double width = cell.Width;                                                 //cell的宽度有三种形式                         //按内容展示(磅值)Word.WdPreferredWidthType.wdPreferredWidthAuto 属性width有值 PreferredWidth为0                         //固定宽度(磅值) Word.WdPreferredWidthType.wdPreferredWidthPoints 属性width有值 PreferredWidth有时为9999999                         //匹配窗口(百分比)Word.WdPreferredWidthType.wdPreferredWidthPercent 属性width有时为9999999 PreferredWidth为百分比                         switch (cell.PreferredWidthType)                         {                             case Word.WdPreferredWidthType.wdPreferredWidthAuto:                                 width = cell.Width;                                 break;                             case Word.WdPreferredWidthType.wdPreferredWidthPoints:                                 ///9999999为百分比或磅值时  取PreferredWidth的宽度                                 if (width == 9999999)                                 {                                     width = cell.PreferredWidth;                                 }                                 break;                             case Word.WdPreferredWidthType.wdPreferredWidthPercent:                                 ///9999999为百分比或磅值时  width和PreferredWidth所对应的宽度                                 if (width == 9999999)                                 {                                     width = GetPercentWidth(tableWidth, cell.PreferredWidth);                                 }                                 break;                         }                                                 wTCol.Width = width;                         wTCol.RealCol = col;                         wTCol.Left = leftPosition;                         leftPosition += width;                         AddNewPosition(positionList, leftPosition);                         continue;                     }

//WTCol prevRowWTCol1 = wordTable.GetPreviousRowRefCol(row, col);                     //WTCol prevRowWTCol2 = wordTable.GetPreviousRowRefColByLeft(row, leftPosition);                     WTCol prevRowWTCol1 = wordTable.GetPreviousRowRefColByLeft(row, leftPosition);                                         //第一列 如果没有单元格说明被上边的行给合并了                     if (prevRowWTCol1 != null)                     {                         //if (prevRowWTCol1.Left + prevRowWTCol1.Width > leftPosition)                         {                             prevRowWTCol1.RowSpan += 1;                         }                         leftPosition += prevRowWTCol1.Width;                         continue;                     }                                        //if (prevRowWTCol2 != null)                     //{

//}                     //else                 }             }

wordTable.ColCount = positionList.Count - 1;

//矫正列合并处理 由于列合并不会产生空的单元格,以列宽判断列合并情况(默认均为合并1列)             foreach (var row in wordTable.Rows)             {                 //int realColCount = 1;                 foreach (var col in row.Cols)                 {                     int colSpan = GetPositionSpan(positionList, col.Left, col.Left + col.Width);

col.ColSpan = colSpan;                     col.RealCol = GetRealColByPositionList(positionList, col.Left); //realColCount;                     //realColCount += colSpan;                     //realColCount                 }             }

return wordTable;         }

上一篇:ruby 操作数据库语句


下一篇:lyGrid表格插件