新しい質問への回答
最終的な説明が正しければ、このコードは必要以上に複雑になります。あなたがそれを投稿する前に、私はあなたが必要と思っているよりもはるかに多様なテーブルを処理できるルーチンを作成しました。「実際の」ファイルを見たことがないので、考えられる完全な複雑さを処理するためのコードを削除していません。
次のようなテストワークシートを作成します。
このワークシートには、私が考える可能性のあるすべての厄介な問題が含まれているため、複製することをお勧めします。このワークシートでこのコードを試してください。コードが何をしているのか、そしてその理由を理解するようにしてください。そうすれば、実際のテーブルがあなたに投げかけるものすべてに備える必要があります。
一部のコードは複雑で、ユーザー定義のデータ型を定義する必要がありました。「vbaUser-DefinedDataType」をグーグルで検索してみましたが、見つけたチュートリアルに非常に失望したので、自分で試してみます。
私のマクロが多くの人々の名前と年齢を保持する必要があると仮定します。私は明らかにいくつかの配列が必要になります:
Dim NameFamily() As String
Dim NameGiven() As String
Dim Age() As Long
ReDim NameFamily(1 to 20)
ReDim NameGiven(1 to 3, 1 to 20)
ReDim Age(1 to 20)
NameFamily(5) = "Dallimore"
NameGiven(1, 5) = "Anthony"
NameGiven(2, 5) = "John"
NameGiven(3, 5) = ""
Age(5) = 65
非常に簡単に、保守が困難なコードが大量に発生する可能性があります。特に一人当たりの変数の数が増えるにつれて。
別の方法は、ほとんどの言語が構造と呼び、VBAがユーザー定義のデータ型と呼ぶものを使用することです。
Type Person
NameFamily As String
NameGiven() As String
NumGivenNames as Long
Age As Long
End Type
Personは新しいデータ型であり、次の型を使用して変数を宣言できます。
Dim Boss As Person
Dim OtherStaff() As Person
ReDim OtherStaff(1 to 20)
OtherStaff(5).NameFamily = "Dallimore"
OtherStaff(5).NumGivenNames = 2
Redim OtherStaff(5).NameGiven(1 To OtherStaff(5).NumGivenNames)
OtherStaff(5).NameGiven(1) = "Anthony"
OtherStaff(5).NameGiven(2) = "John"
OtherStaff(5).Age = 65
これはおそらく簡単には見えません。人々に関する別の情報項目を追加したい場合、利点はより明白になります。おそらく子供の数。通常のアレイでは、最初に新しいアレイを追加する必要があります。次に、person配列のサイズを変更し、新しい配列のReDimステートメントを追加するコード内のすべてのポイントを見つける必要があります。ReDimを見逃すと、奇妙なエラーが発生します。ユーザー定義のデータ型を使用して、型定義に1行追加します。
Type Person
NameFamily As String
NameGiven() As String
NumGivenNames as Long
Age As Long
NumChildren As Long
End Type
これで、既存のすべてのコードがこの新しい変数に対して完全に更新されます。
上記は非常に簡単な紹介ですが、コードで使用したユーザー定義のデータ型のすべての機能を網羅していると思います。
私のコードを理解できるように、十分なコメントを含めたことを願っています。ゆっくりと作業し、必要に応じて質問してください。
以下のコードは、以前のバージョンの質問に対処するために更新された3番目のバージョンです。
変数の命名規則
名前の形式はAaaaBbbbCcccで、各名前部分によって名前の範囲が縮小されます。したがって、「Col」は列の略です。列番号として使用される変数はすべて「Col」で始まります。「Dest」はdestinationの略で、「Src」は「Source」の略です。したがって、「ColSrc」で始まる変数は、ソースワークシートの列番号です。
配列AaaaBbbbCcccがある場合、結果の名前が長すぎる場合を除いて、その配列のインデックスはInxAaaaBbbbCcccで始まります。長すぎる場合は、Aaaa、Bbbb、およびCcccが省略または破棄されます。したがって、「InxNameDtl」が長すぎると思うので、「NameDtl()」のすべてのインデックスは「InxName」で始まります。
「Crnt」は「Current」の略で、通常、forループ変数またはforループの1回の反復で配列から抽出された値を示します。
Option Explicit
Type typNameDtl
InxPredCrntMax As Long
Name As String
Output As Boolean
Predecessor() As String
End Type
Sub ExtractValue3()
Dim ColDestCrnt As Long ' Current column of destination worksheet
Dim ColSrcCrnt As Long ' Current column of source worksheet
Dim ColSrcSheetLast As Long ' Last column of worksheet
Dim InxNISCrnt As Long ' Current index into NameInSeq array
Dim InxNISCrntMax As Long ' Index of last used entry in NameInSeq array
Dim InxNISFirstThisPass As Long ' Index of first entry in NameInSeq array
' used this pass
Dim InxNameCrnt As Long ' Current index into NameDtl array
Dim InxNameCrntMax As Long ' Index of last used entry in NameDtl array
Dim InxPredCrnt As Long ' Current index into NameDtl(N).Predecessor
' array
Dim InxPredCrntMaxCrnt As Long ' Temporary copy of
' NameDtl(N).InxPredecessorCrntMax
Dim InxTableCrnt As Long ' Current index into RowSrcTableTitle and
' RowSrcTableEnd arrays
Dim InxTableCrntMax As Long ' Last used entry in RowSrcTableTitle and
' RowSrcTableEnd arrays
Dim Found As Boolean ' Set to True if a loop finds what is
' being sought
Dim NameCrnt As String ' Current index into NameDtl array
Dim NameInSeq() As String ' Array of names in output sequence
Dim NameLenMax As Long ' Maximum length of a name. Only used to
' align columns in diagnostic output.
Dim NameDtl() As typNameDtl ' Array of names found and their predecessors
Dim PredNameCrnt As String ' Current predecessor name. Used when
' searching NameDtl(N).Predecessor
Dim RowDestCrnt As Long ' Current row of destination worksheet
Dim RowSrcCrnt1 As Long ' \ Indices into source worksheet allowing
Dim RowSrcCrnt2 As Long ' / nested searches
Dim RowSrcTableEnd() As Long ' Array holding last row of each table within
' source worksheet
Dim RowSrcTableEndCrnt As Long ' The last row of the current table
Dim RowSrcSheetLast As Long ' Last row of source worksheet
Dim RowSrcTableTitle() As Long ' Array holding title row of each table within
' source worksheet
Dim RowSrcTableTitleCrnt As Long ' Title row of current table
Dim SheetValue() As Variant ' Copy of source worksheet.
' Column A of source worksheet used to test this code:
' Start
' row Values in starting and following rows
' 2 block1 name c d e f
' 9 block2 name b c d e
' 16 block3 name a c d
' 22 block4 name a d e
' 29 block5 name a d f
' 36 block6 name d e f
' Note that a and b never appear together in a table; it is impossible
' to deduce their preferred sequence from this data.
' Stage 1: Load entire source worksheet into array.
' =================================================
With Worksheets("Jia Source")
' Detrmine dimensions of worksheet
RowSrcSheetLast = .Cells.Find("*", .Range("A1"), xlFormulas, , _
xlByRows, xlPrevious).Row
ColSrcSheetLast = .Cells.Find("*", .Range("A1"), xlFormulas, , _
xlByColumns, xlPrevious).Column
SheetValue = .Range(.Cells(1, 1), _
.Cells(RowSrcSheetLast, ColSrcSheetLast)).Value
' SheetValue is a one-based array with rows as the first dimension and
' columns as the second. An array loaded from a worksheet is always one-based
' even if the range does not start at Cells(1,1). Because this range starts
' at Cells(1,1), indices into SheetValue match row and column numbers within
' the worksheet. This match is convenient for diagnostic output but is not
' used by the macro which does not reference the worksheet, RowSrcSheetLast or
' ColSrcSheet again.
End With
' Stage 2: Locate each table and store number of
' title row and last data row in arrays.
' ==============================================
' 100 entries may be enough. The arrays are enlarged if necessary.
ReDim RowSrcTableEnd(1 To 100)
ReDim RowSrcTableTitle(1 To 100)
InxTableCrntMax = 0 ' Arrays currently empty
RowSrcCrnt1 = 1
' Loop identifying dimensions of tables
Do While RowSrcCrnt1 <= RowSrcSheetLast
' Search down for the first row of a table
Found = False
Do While RowSrcCrnt1 <= RowSrcSheetLast
If SheetValue(RowSrcCrnt1, 1) <> "" Then
RowSrcTableTitleCrnt = RowSrcCrnt1
Found = True
Exit Do
End If
RowSrcCrnt1 = RowSrcCrnt1 + 1
Loop
If Not Found Then
' All tables located
Exit Do
End If
' Search down for the last row of a table
Found = False
Do While RowSrcCrnt1 <= RowSrcSheetLast
If SheetValue(RowSrcCrnt1, 1) = "" Then
RowSrcTableEndCrnt = RowSrcCrnt1 - 1
Found = True
Exit Do
End If
RowSrcCrnt1 = RowSrcCrnt1 + 1
Loop
If Not Found Then
' Last table extends down to bottom of worksheet
RowSrcTableEndCrnt = RowSrcSheetLast
End If
' Store details of this table.
InxTableCrntMax = InxTableCrntMax + 1
' Enlarge arrays if they are full
If InxTableCrntMax > UBound(RowSrcTableTitle) Then
' Redim Preserve requires the interpreter find a block of memory
' of the new size, copy values across from the old array and
' release the old array for garbage collection. I always allocate
' extra memory in large chunks and use an index like
' InxTableCrntMax to record how much of the array has been used.
ReDim Preserve RowSrcTableTitle(UBound(RowSrcTableTitle) + 100)
ReDim Preserve RowSrcTableEnd(UBound(RowSrcTableTitle) + 100)
End If
RowSrcTableTitle(InxTableCrntMax) = RowSrcTableTitleCrnt
RowSrcTableEnd(InxTableCrntMax) = RowSrcTableEndCrnt
Loop
' Output the arrays to the Immediate window to demonstrate they are correct.
' For my test data, the output is:
' Elements: 1 2 3 4 5 6
' Title: 2 9 16 22 29 36
' Last data: 7 14 20 26 33 40
Debug.Print "Location of each table"
Debug.Print " Elements:";
For InxTableCrnt = 1 To InxTableCrntMax
Debug.Print Right(" " & InxTableCrnt, 3);
Next
Debug.Print
Debug.Print " Title:";
For InxTableCrnt = 1 To InxTableCrntMax
Debug.Print Right(" " & RowSrcTableTitle(InxTableCrnt), 3);
Next
Debug.Print
Debug.Print "Last data:";
For InxTableCrnt = 1 To InxTableCrntMax
Debug.Print Right(" " & RowSrcTableEnd(InxTableCrnt), 3);
Next
Debug.Print
' Stage 3. Build arrays listing predecessors of each name
' ========================================================
' The names within the tables are all in the same sequence but no table
' contains more than a few names so that sequence is not obvious. This
' stage accumulates data from the tables so that Stage 4 can deduce the full
' sequence. More correctly, Stage 4 deduces a sequence that does not
' contradict the tables because the sequence of a and b and the sequence
' of f and g is not defined by these tables.
' For Stage 4, I need a list of every name used in the tables and, for each
' name, a list of its predecessors. Consider first the list of names.
' NameDtl is initialised to NameDtl(1 to 50) and InxNameCrntMax is initialised
' to 0 to record the array is empty. In table 1, the code below finds c, d,
' e and f. NameDtl and InxNameCrntMax are updated as these names are found:
'
' Initial state: InxNameCrntMax = 0 NameDtl empty
' Name c found : InxNameCrntMax = 1 NameDtl(1).Name = "c"
' Name d found : InxNameCrntMax = 2 NameDtl(2).Name = "d"
' Name e found : InxNameCrntMax = 3 NameDtl(3).Name = "e"
' Name f found : InxNameCrntMax = 4 NameDtl(4).Name = "f"
' In table 2, the code finds; b, c, d and e. b is new but c, d and e are
' already recorded and they must not be added again. For each name found,
' the code checks entries 1 to InxNameCrntMax. Only if the new name is not
' found, is it added.
' For each name, Stage 4 needs to know its predecessors. From table 1 it
' records that:
' d is preceeded by c
' e is preceeded by c and d
' f is preceeded by c, d and e
' The same technique is used for build the list of predecessors. The
' differences are:
' 1) Names are accumulated in NameDtl().Name while the predecessors of
' the fifth name are accumulated in NameDtl(5).Predecessor.
' 2) InxNameCrntMax is replaced, for the fifth name, by
' NameDtl(5).InxPredCrntMax.
' Start with space for 50 names. Enlarge if necessary.
ReDim NameDtl(1 To 50)
InxNameCrntMax = 0 ' Array is empty
' For each table
For InxTableCrnt = 1 To InxTableCrntMax
RowSrcTableTitleCrnt = RowSrcTableTitle(InxTableCrnt)
RowSrcTableEndCrnt = RowSrcTableEnd(InxTableCrnt)
' For each data row in the current table
For RowSrcCrnt1 = RowSrcTableTitleCrnt + 2 To RowSrcTableEndCrnt
' Look in NameDtl for name from current data row
NameCrnt = SheetValue(RowSrcCrnt1, 1)
Found = False
For InxNameCrnt = 1 To InxNameCrntMax
' Not this comparison is case sensitive "John" and "john" would not
' match. Use LCase if case insensitive comparison required.
If NameCrnt = NameDtl(InxNameCrnt).Name Then
Found = True
Exit For
End If
Next
If Not Found Then
' This is a new name. Create entry in NameDtl for it.
InxNameCrntMax = InxNameCrntMax + 1
If InxNameCrntMax > UBound(NameDtl) Then
ReDim Preserve NameDtl(UBound(NameDtl) + 50)
End If
InxNameCrnt = InxNameCrntMax
NameDtl(InxNameCrnt).Output = False
NameDtl(InxNameCrnt).Name = NameCrnt
' Allow for up to 20 predecessors
ReDim NameDtl(InxNameCrnt).Predecessor(1 To 20)
NameDtl(InxNameCrnt).InxPredCrntMax = 0
End If
' Check that each predecessor for the current name within the
' current table is recorded against the current name
For RowSrcCrnt2 = RowSrcTableTitleCrnt + 2 To RowSrcCrnt1 - 1
Found = False
PredNameCrnt = SheetValue(RowSrcCrnt2, 1)
' Move current number of predecessors from array to variable
' to make code more compact and easier to read
InxPredCrntMaxCrnt = NameDtl(InxNameCrnt).InxPredCrntMax
For InxPredCrnt = 1 To InxPredCrntMaxCrnt
If PredNameCrnt = _
NameDtl(InxNameCrnt).Predecessor(InxPredCrnt) Then
Found = True
Exit For
End If
Next
If Not Found Then
' This predecessor has not been recorded against the current name
InxPredCrntMaxCrnt = InxPredCrntMaxCrnt + 1
If InxPredCrntMaxCrnt > _
UBound(NameDtl(InxNameCrnt).Predecessor) Then
ReDim Preserve NameDtl(UBound(NameDtl) + 20)
End If
NameDtl(InxNameCrnt).Predecessor(InxPredCrntMaxCrnt) = PredNameCrnt
' Place new value for number of predecessors in its permenent store.
NameDtl(InxNameCrnt).InxPredCrntMax = InxPredCrntMaxCrnt
End If
Next
Next
Next
' Output NameDtl to the Immediate window to demonstrate it is correct.
' Find length of longest name so columns can be justified
NameLenMax = 4 ' Minimum length is that of title
For InxNameCrnt = 1 To InxNameCrntMax
If Len(NameDtl(InxNameCrnt).Name) > NameLenMax Then
NameLenMax = Len(NameDtl(InxNameCrnt).Name)
End If
Next
' Output headings
Debug.Print vbLf & "Contents of NameDtl table"
Debug.Print Space(NameLenMax + 10) & "Max"
Debug.Print Left("Name" & Space(NameLenMax), NameLenMax + 2) & _
"Output inx Predecessors"
' Output table contents
For InxNameCrnt = 1 To InxNameCrntMax
Debug.Print Left(NameDtl(InxNameCrnt).Name & Space(NameLenMax), _
NameLenMax + 4) & _
IIf(NameDtl(InxNameCrnt).Output, " True ", " False") & _
" " & Right(" " & _
NameDtl(InxNameCrnt).InxPredCrntMax, 3) & " ";
For InxPredCrnt = 1 To NameDtl(InxNameCrnt).InxPredCrntMax
Debug.Print " " & _
NameDtl(InxNameCrnt).Predecessor(InxPredCrnt);
Next
Debug.Print
Next
' Stage 4: Sequence names for list.
' =================================
' The output from the above routine for the test data is:
' Max
' Name Output inx Predecessors
' c False 2 b a
' d False 3 c b a
' e False 4 c d b a
' g False 3 c d e
' b False 0
' a False 0
' f False 3 a d e
' Note 1: All this information is in the sequence found.
' Note 2: We do not know the "true" sequence of b and a or of g and f.
' The loop below has three steps:
' 1) Transfer any names to NamesInSeq() that have not already been
' transferred and have a value of 0 for Max inx.
' 2) If no names are transferred, the loop has completed its task.
' 3) Remove any names transferred during this pass from the predecessor
' lists and mark the name as output.
' Before the loop NameInSeq() is empty, InxNISCrntMax = 0 and
' InxNISFirstThisPass = InxNISCrntMax+1 = 1.
' After step 1 of pass 1:
' NameInSeq(1) = "b" and NameInSeq(2) = "a"
' InxNISCrntMax = 2
' Entries InxNISFirstThisPass (1) to InxNISCrntMax (2) of NamesInSeq have
' been transferred during this pass so names a and b are removed from the
' lists by copying the last entry in each list over the name to be removed
' and reducing Max inx. For pass 1, only the list for f is changed.
' At the end of pass 1, NameDtl is:
' Max
' Name Output inx Predecessors
' c False 0
' d False 1 c
' e False 2 c d
' g False 3 c d e
' b True 0
' a True 0
' f False 2 e d
' During pass 2, c is moved to NamesInSeq and removed form the lists to give:
' Max
' Name Output inx Predecessors
' c True 0
' d False 0
' e False 1 d
' g False 2 e d
' b True 0
' a True 0
' f False 2 e d
' This process continues until all names have been transferred.
' Size array for total number of names.
ReDim NameInSeq(1 To InxNameCrntMax)
InxNISCrntMax = 0 ' Array empty
' Loop until every name has been moved
' from ProdecessorDtl to NameInSeq.
Do While True
Found = False ' No name found to move during this pass
' Record index of first name, if any, to be added during this pass
InxNISFirstThisPass = InxNISCrntMax + 1
' Transfer names without predecessors to NameInSeq()
For InxNameCrnt = 1 To InxNameCrntMax
If Not NameDtl(InxNameCrnt).Output Then
' This name has not been output
If NameDtl(InxNameCrnt).InxPredCrntMax = 0 Then
' This name has no predecessors or no predecessors that
' have not already been transferred to NameInSeq()
InxNISCrntMax = InxNISCrntMax + 1
NameInSeq(InxNISCrntMax) = NameDtl(InxNameCrnt).Name
NameDtl(InxNameCrnt).Output = True
Found = True
End If
End If
Next
If Not Found Then
' All names already transferred to NameInSeq
Exit Do
End If
' Remove references to names transferred to NameinSeq()
' during this pass
For InxNISCrnt = InxNISFirstThisPass To InxNISCrntMax
NameCrnt = NameInSeq(InxNISCrnt)
For InxNameCrnt = 1 To InxNameCrntMax
If Not NameDtl(InxNameCrnt).Output Then
' This name has not been output
For InxPredCrnt = 1 To NameDtl(InxNameCrnt).InxPredCrntMax
If NameCrnt = _
NameDtl(InxNameCrnt).Predecessor(InxPredCrnt) Then
' Remove this name by overwriting it
' with the last name in the list
NameDtl(InxNameCrnt).Predecessor(InxPredCrnt) = _
NameDtl(InxNameCrnt).Predecessor _
(NameDtl(InxNameCrnt).InxPredCrntMax)
NameDtl(InxNameCrnt).InxPredCrntMax = _
NameDtl(InxNameCrnt).InxPredCrntMax - 1
Exit For
End If
Next
End If
Next
Next
Loop
Debug.Print vbLf & "Name list"
For InxNISCrnt = 1 To InxNISCrntMax
Debug.Print NameInSeq(InxNISCrnt)
Next
' Stage 5: Transfer data
' ======================
' We now have everything we need for the transfer:
' * NameInSeq() contains the names in the output sequence
' * SheetValue() contains all the data from the source worksheet
' * RowSrcTableTitle() and RowSrcTableEnd() identify the
' start and end row of each table
With Worksheets("Jia Destination")
.Cells.EntireRow.Delete ' Clear destination sheet
ColDestCrnt = 1
.Cells(1, ColDestCrnt).Value = "Name"
' Output names
RowDestCrnt = 2
For InxNISCrnt = 1 To InxNISCrntMax
.Cells(RowDestCrnt, ColDestCrnt).Value = NameInSeq(InxNISCrnt)
RowDestCrnt = RowDestCrnt + 1
Next
' Output values from each table
For InxTableCrnt = 1 To InxTableCrntMax
RowSrcTableTitleCrnt = RowSrcTableTitle(InxTableCrnt)
RowSrcTableEndCrnt = RowSrcTableEnd(InxTableCrnt)
' Find value column, if any
Found = False
ColSrcCrnt = 2
Do While SheetValue(RowSrcTableTitleCrnt + 1, ColSrcCrnt) <> ""
If LCase(SheetValue(RowSrcTableTitleCrnt + 1, ColSrcCrnt)) = _
"value" Then
Found = True
Exit Do
End If
ColSrcCrnt = ColSrcCrnt + 1
Loop
If Found Then
' Value column found for this table
ColDestCrnt = ColDestCrnt + 1
' Transfer table name
.Cells(1, ColDestCrnt).Value = SheetValue(RowSrcTableTitleCrnt, 1)
' Transfer values
RowDestCrnt = 2
RowSrcCrnt1 = RowSrcTableTitleCrnt + 2
For InxNISCrnt = 1 To InxNISCrntMax
If NameInSeq(InxNISCrnt) = SheetValue(RowSrcCrnt1, 1) Then
' Value for this name in this table
.Cells(RowDestCrnt, ColDestCrnt).Value = _
SheetValue(RowSrcCrnt1, ColSrcCrnt)
' Value transferred from this row. Step to next if any
RowSrcCrnt1 = RowSrcCrnt1 + 1
If RowSrcCrnt1 > RowSrcTableEndCrnt Then
' No more rows in this table
Exit For
End If
End If
RowDestCrnt = RowDestCrnt + 1
Next
Else
Call MsgBox("Table starting at row " & RowSrcTableTitleCrnt & _
" does not have a value column", vbOKOnly)
End If
Next
End With
End Sub