Question

Excel VBA的查找与二进制搜索有多好/多快？我的平台是Office 11 | 2003，我将在三张值上搜索A列的字符串。行总数~140,000

如果值得哪个图书馆＆amp;我应该引用函数进行排序然后进行二进制搜索？据报道，二进制搜索字符串/文本存在潜在问题。

......一件事必须注意。使用二进制搜索具有sortedtextrequires的公式警告。 Aladin A., Excel MVP

Excel查找：

Worksheets(1).Range("A:A").Find("PN-String-K9", LookIn:=xlValues, LookAt:=xlWhole)

Answer 1

与我的直觉相反，VBA二进制搜索强烈优于Excel查找。至少在下面的场景中，120,000个6个字符串均匀分布在3个工作表上。

Excel查找需要1分58秒，
我的特定机器上的VBA二进制搜索需要36秒。

知道文本有序的优势显然超过了Excel的天然优势。注意Aladin A关于排序顺序的警告。

Option Explicit

' Call Search to look for a thousand random strings
' in 3 worksheets of a workbook

' requires a workbook with 3 sheets and
' column A populated with values between "00001" to "120000"
' split evenly 40,000 to a worksheet in ascending order.
' They must be text, not numbers.

Private Const NUM_ROWS As Long = 120000
Private Const SHEET_1 As String = "Sheet1"
Private Const SHEET_2 As String = "Sheet2"
Private Const SHEET_3 As String = "Sheet3"

' This uses VBA Binary Search
Public Sub Search()
    Worksheets(SHEET_1).Range("B:B").ClearContents
    Worksheets(SHEET_2).Range("B:B").ClearContents
    Worksheets(SHEET_3).Range("B:B").ClearContents
    DoSearch True       ' change to False to test Excel search
End Sub

' Searches for a thousand values using binary  or excel search depending on
' value of bBinarySearch
Public Sub DoSearch(ByVal bBinarySearch As Boolean)
    Debug.Print Now
    Dim ii As Long

    For ii = 1 To 1000
        Dim rr As Long
        rr = Int((NUM_ROWS) * Rnd + 1)
        If bBinarySearch Then
            Dim strSheetName As String
            Dim nRow As Long
            If BinarySearch(MakeSearchArg(rr), strSheetName, nRow) Then
                Worksheets(strSheetName).Activate
                Cells(nRow, 1).Activate
            End If
        Else
            If Not ExcelSearch(SHEET_1, MakeSearchArg(rr)) Then
                If Not ExcelSearch(SHEET_2, MakeSearchArg(rr)) Then
                    ExcelSearch SHEET_3, MakeSearchArg(rr)
                End If
            End If
        End If
        ActiveCell.Offset(0, 1).Value = "FOUND"
    Next
    Debug.Print Now

End Sub

' look for one cell value using Excel Find
Private Function ExcelSearch(ByVal strWorksheet As String _
  , ByVal strSearchArg As String) As Boolean
    On Error GoTo Err_Exit
    Worksheets(strWorksheet).Activate
    Worksheets(strWorksheet).Range("A:A").Find(What:=strSearchArg, LookIn:=xlValues, LookAt:= 
        xlWhole, SearchOrder:=xlByRows, SearchDirection:=xlNext, MatchCase:=True 
        , SearchFormat:=False).Activate
    ExcelSearch = True
    Exit Function
Err_Exit:
    ExcelSearch = False
End Function

' Look for value using a vba based binary search
' returns true if the search argument is found in the workbook
' strSheetName contains the name of the worksheet on exit and nRow gives the row
Private Function BinarySearch(ByVal strSearchArg As String _
  , ByRef strSheetName As String, ByRef nRow As Long) As Boolean
    Dim nFirst As Long, nLast As Long
    nFirst = 1
    nLast = NUM_ROWS
    Do While True
        Dim nMiddle As Long
        Dim strValue As String
        If nFirst > nLast Then
            Exit Do     ' Failed to find search arg
        End If
        nMiddle = Round((nLast - nFirst) / 2 + nFirst)
        SheetNameAndRowFromIdx nMiddle, strSheetName, nRow
        strValue = Worksheets(strSheetName).Cells(nRow, 1)
        If strSearchArg < strValue Then
            nLast = nMiddle - 1
        ElseIf strSearchArg > strValue Then
            nFirst = nMiddle + 1
        Else
            BinarySearch = True
            Exit Do
        End If
    Loop
End Function

' convert 1 -> "000001", 120000 -> "120000", etc
Private Function MakeSearchArg(ByVal nArg As Long) As String
    MakeSearchArg = Right(CStr(nArg + 1000000), 6)
End Function

' converts some number to a worksheet name and a row number
' This is depenent on the worksheets being named sheet1, sheet2, sheet3

' and containing an equal number of vlaues in each sheet where
' the total number of values is NUM_ROWS
Private Sub SheetNameAndRowFromIdx(ByVal nIdx As Long _
  , ByRef strSheetName As String, ByRef nRow As Long)
    If nIdx <= NUM_ROWS / 3 Then

        strSheetName = SHEET_1
        nRow = nIdx
    ElseIf nIdx > (NUM_ROWS / 3) * 2 Then
        strSheetName = SHEET_3
        nRow = nIdx - (NUM_ROWS / 3) * 2
    Else
        strSheetName = SHEET_2
        nRow = nIdx - (NUM_ROWS / 3)
    End If
End Sub

Answer 2

我发现使用AutoFilter比使用任何方法手动搜索记录要快得多。

我过滤，检查是否有任何结果，然后继续。如果发现任何（通过检查结果的数量），我可以搜索手动过滤的小部分或全部返回。

我在大约44,000条记录中使用它，搜索100多个部分的列表。

如果你不小心，二进制搜索很容易陷入无限循环。

Answer 3

如果你使用带有排序选项的vlookup，它可能会比你的vba更快。

Answer 4

我对此很感兴趣，因为我使用的是.Find函数，在一台PC上无法进行某些查找，但在另一台PC上可以运行！因此，我对计时进行了一些测试-我整理了一个表，其中按顺序排列了985个名称，并且编写了一个小的子例程来遍历它们，并使用不同的方法在同一列表中查找每个子例程（时间以毫秒为单位）：

蛮力：2000
。发现：750
Application.VLookup：265
二分搜索：234

VLookup的问题在于，除非将行号包括在表中，否则它无法返回行号。

这是我的二进制搜索代码，我假设工作表中有一个标题行，但是您可以轻松地修改标题和代码以传递该信息。可选的Col参数用于指示您是否需要行号或单元格的值。如果查找失败，该函数将返回0（零）。

Function Find(Sheet As Worksheet, What As String, Optional Col As Long = 0) As Variant
Dim Top As Long
Dim Mid As Long
Dim Bot As Long 'Bottom
Dim S As String
Dim T As String

   With Sheet
     Top = 2 'Sheet has a header row
     Bot = .UsedRange.Rows.Count
     S = LCase(What)
     Do
       Mid = (Top + Bot) / 2
       T = LCase(.Cells(Mid, 1))
       Select Case True
       Case T > S
         Bot = Mid - 1
       Case T < S
         Top = Mid + 1
       Case Else 'T = S
         If Col = 0 Then
           Find = Mid  'Return the row
         Else
           Find = .Cells(Mid, Col).Value2 'Return the cell's value
         End If
         Exit Function
       End Select
     Loop Until Bot < Top
   End With
   Find = 0
End Function

Excel查找速度与VBA二进制搜索？

4 个答案: