使用Interop

Question

我有很多包含数据的excel文件，它包含空行和空列。 如下所示

我正在尝试使用interop从excel中删除Empty行和列。 我创建一个简单的winform应用程序并使用以下代码，它工作正常。

Dim lstFiles As New List(Of String)
lstFiles.AddRange(IO.Directory.GetFiles(m_strFolderPath, "*.xls", IO.SearchOption.AllDirectories))

Dim m_XlApp = New Excel.Application
Dim m_xlWrkbs As Excel.Workbooks = m_XlApp.Workbooks
Dim m_xlWrkb As Excel.Workbook

For Each strFile As String In lstFiles
    m_xlWrkb = m_xlWrkbs.Open(strFile)
    Dim m_XlWrkSheet As Excel.Worksheet = m_xlWrkb.Worksheets(1)
    Dim intRow As Integer = 1

    While intRow <= m_XlWrkSheet.UsedRange.Rows.Count
        If m_XlApp.WorksheetFunction.CountA(m_XlWrkSheet.Cells(intRow, 1).EntireRow) = 0 Then
            m_XlWrkSheet.Cells(intRow, 1).EntireRow.Delete(Excel.XlDeleteShiftDirection.xlShiftUp)
        Else
            intRow += 1
        End If
    End While

    Dim intCol As Integer = 1
    While intCol <= m_XlWrkSheet.UsedRange.Columns.Count
        If m_XlApp.WorksheetFunction.CountA(m_XlWrkSheet.Cells(1, intCol).EntireColumn) = 0 Then
            m_XlWrkSheet.Cells(1, intCol).EntireColumn.Delete(Excel.XlDeleteShiftDirection.xlShiftToLeft)
        Else
            intCol += 1
        End If
    End While
Next

m_xlWrkb.Save()
m_xlWrkb.Close(SaveChanges:=True)

Marshal.ReleaseComObject(m_xlWrkb)
Marshal.ReleaseComObject(m_xlWrkbs)
m_XlApp.Quit()
Marshal.ReleaseComObject(m_XlApp)

但是在清理大型Excel文件时需要花费很多时间。 有关优化此代码的任何建议吗？或者其他方式更快地清理这个excel文件？是否有一个函数可以一次删除空行？

如果答案使用C＃

，我没有问题

修改

我上传了一个示例文件Sample File。但并非所有文件都具有相同的结构。

Answer 1

我发现如果工作表很大，循环遍历excel工作表可能需要一些时间。所以我的解决方案试图避免工作表中的任何循环。为了避免在工作表中循环，我从usedRange返回的单元格中创建了一个二维对象数组：

Excel.Range targetCells = worksheet.UsedRange;
object[,] allValues = (object[,])targetCells.Cells.Value;

这是我循环访问的数组，以获取空行和列的索引。我制作2个int列表，一个保留行索引以删除另一个保持列索引删除。

List<int> emptyRows = GetEmptyRows(allValues, totalRows, totalCols);
List<int> emptyCols = GetEmptyCols(allValues, totalRows, totalCols);

这些列表将从高到低排序，以简化从下到上删除行并从右到左删除列的过程。然后只需循环遍历每个列表并删除相应的行/列。

DeleteRows(emptyRows, worksheet);
DeleteCols(emptyCols, worksheet);

最后删除所有空行和列后，我将文件另存为新文件名。

希望这有帮助。

修改

解决了UsedRange问​​题，如果工作表顶部有空行，那么现在将删除这些行。此外，这将删除起始数据左侧的所有空列。即使在数据启动之前存在空行或列，这也允许索引正常工作。 这是通过获取UsedRange中第一个单元格的地址来完成的，这将是“$ A $ 1：$ D $ 4”形式的地址。如果要保留并且不删除左侧的空行和左侧的空列，这将允许使用偏移量。在这种情况下，我只是删除它们。要从顶部删除要删除的行数，可以通过第一个“$ A $ 4”地址计算，其中“4”是第一个数据出现的行。所以我们需要删除前3行。列地址的格式为“A”，“AB”或甚至“AAD”，这需要一些翻译，并且感谢How to convert a column number (eg. 127) into an excel column (eg. AA)我能够确定左侧需要删除多少列。

class Program {
  static void Main(string[] args) {
    Excel.Application excel = new Excel.Application();
    string originalPath = @"H:\ExcelTestFolder\Book1_Test.xls";
    Excel.Workbook workbook = excel.Workbooks.Open(originalPath);
    Excel.Worksheet worksheet = workbook.Worksheets["Sheet1"];
    Excel.Range usedRange = worksheet.UsedRange;

    RemoveEmptyTopRowsAndLeftCols(worksheet, usedRange);

    DeleteEmptyRowsCols(worksheet);

    string newPath = @"H:\ExcelTestFolder\Book1_Test_Removed.xls";
    workbook.SaveAs(newPath, Excel.XlSaveAsAccessMode.xlNoChange);

    workbook.Close();
    excel.Quit();
    System.Runtime.InteropServices.Marshal.ReleaseComObject(workbook);
    System.Runtime.InteropServices.Marshal.ReleaseComObject(excel);
    Console.WriteLine("Finished removing empty rows and columns - Press any key to exit");
    Console.ReadKey();
  }

  private static void DeleteEmptyRowsCols(Excel.Worksheet worksheet) {
    Excel.Range targetCells = worksheet.UsedRange;
    object[,] allValues = (object[,])targetCells.Cells.Value;
    int totalRows = targetCells.Rows.Count;
    int totalCols = targetCells.Columns.Count;

    List<int> emptyRows = GetEmptyRows(allValues, totalRows, totalCols);
    List<int> emptyCols = GetEmptyCols(allValues, totalRows, totalCols);

    // now we have a list of the empty rows and columns we need to delete
    DeleteRows(emptyRows, worksheet);
    DeleteCols(emptyCols, worksheet);
  }

  private static void DeleteRows(List<int> rowsToDelete, Excel.Worksheet worksheet) {
    // the rows are sorted high to low - so index's wont shift
    foreach (int rowIndex in rowsToDelete) {
      worksheet.Rows[rowIndex].Delete();
    }
  }

  private static void DeleteCols(List<int> colsToDelete, Excel.Worksheet worksheet) {
    // the cols are sorted high to low - so index's wont shift
    foreach (int colIndex in colsToDelete) {
      worksheet.Columns[colIndex].Delete();
    }
  }

  private static List<int> GetEmptyRows(object[,] allValues, int totalRows, int totalCols) {
    List<int> emptyRows = new List<int>();

    for (int i = 1; i < totalRows; i++) {
      if (IsRowEmpty(allValues, i, totalCols)) {
        emptyRows.Add(i);
      }
    }
    // sort the list from high to low
    return emptyRows.OrderByDescending(x => x).ToList();
  }

  private static List<int> GetEmptyCols(object[,] allValues, int totalRows, int totalCols) {
    List<int> emptyCols = new List<int>();

    for (int i = 1; i < totalCols; i++) {
      if (IsColumnEmpty(allValues, i, totalRows)) {
        emptyCols.Add(i);
      }
    }
    // sort the list from high to low
    return emptyCols.OrderByDescending(x => x).ToList();
  }

  private static bool IsColumnEmpty(object[,] allValues, int colIndex, int totalRows) {
    for (int i = 1; i < totalRows; i++) {
      if (allValues[i, colIndex] != null) {
        return false;
      }
    }
    return true;
  }

  private static bool IsRowEmpty(object[,] allValues, int rowIndex, int totalCols) {
    for (int i = 1; i < totalCols; i++) {
      if (allValues[rowIndex, i] != null) {
        return false;
      }
    }
    return true;
  }

  private static void RemoveEmptyTopRowsAndLeftCols(Excel.Worksheet worksheet, Excel.Range usedRange) {
    string addressString = usedRange.Address.ToString();
    int rowsToDelete = GetNumberOfTopRowsToDelete(addressString);
    DeleteTopEmptyRows(worksheet, rowsToDelete);
    int colsToDelete = GetNumberOfLeftColsToDelte(addressString);
    DeleteLeftEmptyColumns(worksheet, colsToDelete);
  }

  private static void DeleteTopEmptyRows(Excel.Worksheet worksheet, int startRow) {
    for (int i = 0; i < startRow - 1; i++) {
      worksheet.Rows[1].Delete();
    }
  }

  private static void DeleteLeftEmptyColumns(Excel.Worksheet worksheet, int colCount) {
    for (int i = 0; i < colCount - 1; i++) {
      worksheet.Columns[1].Delete();
    }
  }

  private static int GetNumberOfTopRowsToDelete(string address) {
    string[] splitArray = address.Split(':');
    string firstIndex = splitArray[0];
    splitArray = firstIndex.Split('$');
    string value = splitArray[2];
    int returnValue = -1;
    if ((int.TryParse(value, out returnValue)) && (returnValue >= 0))
      return returnValue;
    return returnValue;
  }

  private static int GetNumberOfLeftColsToDelte(string address) {
    string[] splitArray = address.Split(':');
    string firstindex = splitArray[0];
    splitArray = firstindex.Split('$');
    string value = splitArray[1];
    return ParseColHeaderToIndex(value);
  }

  private static int ParseColHeaderToIndex(string colAdress) {
    int[] digits = new int[colAdress.Length];
    for (int i = 0; i < colAdress.Length; ++i) {
      digits[i] = Convert.ToInt32(colAdress[i]) - 64;
    }
    int mul = 1; int res = 0;
    for (int pos = digits.Length - 1; pos >= 0; --pos) {
      res += digits[pos] * mul;
      mul *= 26;
    }
    return res;
  }
}

编辑2：为了进行测试，我创建了一个循环通过工作表的方法，并将其与通过对象数组循环的代码进行比较。它显示出显着的差异。

循环通过工作表并删除空行和列的方法。

enum RowOrCol { Row, Column };
private static void ConventionalRemoveEmptyRowsCols(Excel.Worksheet worksheet) {
  Excel.Range usedRange = worksheet.UsedRange;
  int totalRows = usedRange.Rows.Count;
  int totalCols = usedRange.Columns.Count;

  RemoveEmpty(usedRange, RowOrCol.Row);
  RemoveEmpty(usedRange, RowOrCol.Column);
}

private static void RemoveEmpty(Excel.Range usedRange, RowOrCol rowOrCol) {
  int count;
  Excel.Range curRange;
  if (rowOrCol == RowOrCol.Column)
    count = usedRange.Columns.Count;
  else
    count = usedRange.Rows.Count;

  for (int i = count; i > 0; i--) {
    bool isEmpty = true;
    if (rowOrCol == RowOrCol.Column)
      curRange = usedRange.Columns[i];
    else
      curRange = usedRange.Rows[i];

    foreach (Excel.Range cell in curRange.Cells) {
      if (cell.Value != null) {
        isEmpty = false;
        break; // we can exit this loop since the range is not empty
      }
      else {
        // Cell value is null contiue checking
      }
    } // end loop thru each cell in this range (row or column)

    if (isEmpty) {
      curRange.Delete();
    }
  }
}

然后是Main用于测试/计时两种方法。

enum RowOrCol { Row, Column };

static void Main(string[] args)
{
  Excel.Application excel = new Excel.Application();
  string originalPath = @"H:\ExcelTestFolder\Book1_Test.xls";
  Excel.Workbook workbook = excel.Workbooks.Open(originalPath);
  Excel.Worksheet worksheet = workbook.Worksheets["Sheet1"];
  Excel.Range usedRange = worksheet.UsedRange;

  // Start test for looping thru each excel worksheet
  Stopwatch sw = new Stopwatch();
  Console.WriteLine("Start stopwatch to loop thru WORKSHEET...");
  sw.Start();
  ConventionalRemoveEmptyRowsCols(worksheet);
  sw.Stop();
  Console.WriteLine("It took a total of: " + sw.Elapsed.Milliseconds + " Miliseconds to remove empty rows and columns...");

  string newPath = @"H:\ExcelTestFolder\Book1_Test_RemovedLoopThruWorksheet.xls";
  workbook.SaveAs(newPath, Excel.XlSaveAsAccessMode.xlNoChange);
  workbook.Close();
  Console.WriteLine("");

  // Start test for looping thru object array
  workbook = excel.Workbooks.Open(originalPath);
  worksheet = workbook.Worksheets["Sheet1"];
  usedRange = worksheet.UsedRange;
  Console.WriteLine("Start stopwatch to loop thru object array...");
  sw = new Stopwatch();
  sw.Start();
  DeleteEmptyRowsCols(worksheet);
  sw.Stop();

  // display results from second test
  Console.WriteLine("It took a total of: " + sw.Elapsed.Milliseconds + " Miliseconds to remove empty rows and columns...");
  string newPath2 = @"H:\ExcelTestFolder\Book1_Test_RemovedLoopThruArray.xls";
  workbook.SaveAs(newPath2, Excel.XlSaveAsAccessMode.xlNoChange);
  workbook.Close();
  excel.Quit();
  System.Runtime.InteropServices.Marshal.ReleaseComObject(workbook);
  System.Runtime.InteropServices.Marshal.ReleaseComObject(excel);
  Console.WriteLine("");
  Console.WriteLine("Finished testing methods - Press any key to exit");
  Console.ReadKey();
}

编辑3 根据OP请求... 我更新并更改了代码以匹配OP代码。有了这个我发现了一些有趣的结果。见下文。

我更改了代码以匹配您正在使用的函数，即... EntireRow和CountA。下面的代码我发现它的表现非常糟糕。运行一些测试我发现下面的代码执行时间超过800毫秒。然而，一个微妙的变化带来了巨大的变化。

在线：

while (rowIndex <= worksheet.UsedRange.Rows.Count)

这会减慢很多事情。如果你为UsedRang创建一个范围变量而不是保持regrabbibg它与while循环的每次迭代将产生巨大的差异。所以...当我将while循环更改为...时

Excel.Range usedRange = worksheet.UsedRange;
int rowIndex = 1;

while (rowIndex <= usedRange.Rows.Count)
and
while (colIndex <= usedRange.Columns.Count)

这与我的对象阵列解决方案非常接近。我没有发布结果，因为您可以使用下面的代码并更改while循环以在每次迭代时获取UsedRange或使用变量usedRange来测试它。

private static void RemoveEmptyRowsCols3(Excel.Worksheet worksheet) {
  //Excel.Range usedRange = worksheet.UsedRange;     // <- using this variable makes the while loop much faster 
  int rowIndex = 1;

  // delete empty rows
  //while (rowIndex <= usedRange.Rows.Count)     // <- changing this one line makes a huge difference - not grabbibg the UsedRange with each iteration...
  while (rowIndex <= worksheet.UsedRange.Rows.Count) {
    if (excel.WorksheetFunction.CountA(worksheet.Cells[rowIndex, 1].EntireRow) == 0) {
      worksheet.Cells[rowIndex, 1].EntireRow.Delete(Excel.XlDeleteShiftDirection.xlShiftUp);
    }
    else {
      rowIndex++;
    }
  }

  // delete empty columns
  int colIndex = 1;
  // while (colIndex <= usedRange.Columns.Count) // <- change here also

  while (colIndex <= worksheet.UsedRange.Columns.Count) {
    if (excel.WorksheetFunction.CountA(worksheet.Cells[1, colIndex].EntireColumn) == 0) {
      worksheet.Cells[1, colIndex].EntireColumn.Delete(Excel.XlDeleteShiftDirection.xlShiftToLeft);
    }
    else {
      colIndex++;
    }
  }
}

按@Hadi更新

如果excel在上次使用后包含额外的空白行和列，则可以更改DeleteCols和DeleteRows函数以获得更好的性能：

private static void DeleteRows(List<int> rowsToDelete, Microsoft.Office.Interop.Excel.Worksheet worksheet)
{
    // the rows are sorted high to low - so index's wont shift

    List<int> NonEmptyRows = Enumerable.Range(1, rowsToDelete.Max()).ToList().Except(rowsToDelete).ToList();

    if (NonEmptyRows.Max() < rowsToDelete.Max())
    {

        // there are empty rows after the last non empty row

        Microsoft.Office.Interop.Excel.Range cell1 = worksheet.Cells[NonEmptyRows.Max() + 1,1];
        Microsoft.Office.Interop.Excel.Range cell2 = worksheet.Cells[rowsToDelete.Max(), 1];

        //Delete all empty rows after the last used row
        worksheet.Range[cell1, cell2].EntireRow.Delete(Microsoft.Office.Interop.Excel.XlDeleteShiftDirection.xlShiftUp);


    }    //else last non empty row = worksheet.Rows.Count



    foreach (int rowIndex in rowsToDelete.Where(x => x < NonEmptyRows.Max()))
    {
        worksheet.Rows[rowIndex].Delete();
    }
}

private static void DeleteCols(List<int> colsToDelete, Microsoft.Office.Interop.Excel.Worksheet worksheet)
{
    // the cols are sorted high to low - so index's wont shift

    //Get non Empty Cols
    List<int> NonEmptyCols = Enumerable.Range(1, colsToDelete.Max()).ToList().Except(colsToDelete).ToList();

    if (NonEmptyCols.Max() < colsToDelete.Max())
    {

        // there are empty rows after the last non empty row

        Microsoft.Office.Interop.Excel.Range cell1 = worksheet.Cells[1,NonEmptyCols.Max() + 1];
        Microsoft.Office.Interop.Excel.Range cell2 = worksheet.Cells[1,NonEmptyCols.Max()];

        //Delete all empty rows after the last used row
        worksheet.Range[cell1, cell2].EntireColumn.Delete(Microsoft.Office.Interop.Excel.XlDeleteShiftDirection.xlShiftToLeft);


    }            //else last non empty column = worksheet.Columns.Count

    foreach (int colIndex in colsToDelete.Where(x => x < NonEmptyCols.Max()))
    {
        worksheet.Columns[colIndex].Delete();
    }
}

在Get Last non empty column and row index from excel using Interop

检查我的回答

Answer 2

也许需要考虑的事情：

Sub usedRangeDeleteRowsCols()
    Dim LastRow, LastCol, i As Long

    LastRow = Cells.Find(What:="*", SearchDirection:=xlPrevious, SearchOrder:=xlByRows).Row
    LastCol = Cells.Find(What:="*", SearchDirection:=xlPrevious, SearchOrder:=xlByColumns).Column

    For i = LastRow To 1 Step -1
        If WorksheetFunction.CountA(Range(Cells(i, 1), Cells(i, LastCol))) = 0 Then
            Cells(i, 1).EntireRow.Delete
        End If
    Next

    For i = LastCol To 1 Step -1
        If WorksheetFunction.CountA(Range(Cells(1, i), Cells(LastRow, i))) = 0 Then
            Cells(1, i).EntireColumn.Delete
        End If
    Next
End Sub

我认为与原始代码中的等效函数相比，有两种效率。首先，我们不是使用Excel的不可靠的UsedRange属性，而是找到最后一个值，只扫描真实使用范围内的行和列。

其次，工作表计数功能仅在真正使用的范围内起作用 - 例如，当搜索空白行时，我们只查看已使用列的范围（而不是.EntireRow）。

For循环向后工作，例如，每次删除行时，后续数据的行地址都会发生变化。向后工作意味着“要处理的数据”的行地址不会改变。

Answer 3

在我看来，最耗时的部分可能是枚举和查找空行和列。

怎么样： http://www.howtogeek.com/206696/how-to-quickly-and-easily-delete-blank-rows-and-columns-in-excel-2013/

修改

怎么样：

m_XlWrkSheet.Columns("A:A").SpecialCells(xlCellTypeBlanks).EntireRow.Delete
m_XlWrkSheet.Rows("1:1").SpecialCells(xlCellTypeBlanks).EntireColumn.Delete

测试样本数据结果看起来不错，性能更好（从VBA测试但差异很大）。

<强>更新

使用14k行（由样本数据制作）原始代码~30秒对样本Excel进行测试，此版本<1s

Answer 4

我所知道的最简单的方法是隐藏非空白单元格并删除可见的单元格：

#include <stdio.h>
#include <stdlib.h>
#include <pthread.h> //compile and link with -pthread
#include <semaphore.h>

#define BUFFER_SIZE 10

int buffer[BUFFER_SIZE];
int in, out;
int num;
semaphore s = 1, n =0, e = BUFFER_SIZE;

void *produce( void *ptr ) { 
    int item; 

    while(1){
        usleep(rand() % 1000); //sleep
        printf("Producer wants to produce.\n");
        semWait(e);
        semWait(s);
        //critical section below
        item = rand() % 10;
        num++;
        buffer[in] = item;
        in = (in + 1) % BUFFER_SIZE;
        //critical section above
        semSignal(s);
        semSignal(n);
        printf("Producer entered %d. Buffer Size = %d\n", item, num); 
    }
}

void *consume( void *ptr ) { 
    int item; 

    while(1){
        usleep(rand() % 1000); //sleep
        printf("Consumer wants to consume.\n");
        semWait(n);
        semWait(s);
        //critical section below
        num--;
        int item = buffer[out];
        out = (out + 1) % BUFFER_SIZE;
        //critical section above
        semSignal(s);
        semSignal(e);
        printf("Consumer consumed %d. Buffer Size = %d\n", item, num); 
    }
}

int main() { 
    pthread_t consumer, producer; 

    out = 0; //index of the item to be consumed next 
    in = 0; //index of the item to be produced next 
    num = 0; //number of items in the buffer 
    srand(time(NULL)); 
    pthread_create(&consumer, NULL, consume, NULL); 
    pthread_create(&producer, NULL, produce, NULL); 
    pthread_join(producer, NULL); 
    pthread_join(consumer, NULL); 
    pthread_exit(NULL);
}

更快的方法是不删除任何东西，而是移动（剪切+粘贴）非空白区域。

最快的Interop方式（没有打开文件的方法更复杂）是获取数组中的所有值，移动数组中的值，然后将值放回：

view(response)

Answer 5

您可以打开与工作表的ADO连接，获取字段列表，发出仅包含已知字段的SQL语句，还可以排除已知字段中没有值的记录。