Nice answer from Manhattan: that's exactly how I use Excel's built-in functionality to select a folder and fetch a set of file names.
However, there's an interesting side-question in there:
Are those single-sheet Excel files workbooks, of just .csv text files?
If they have a .csv extension, you don't need to open them in Excel to count the rows!
Here's the code to do it:
Fast VBA for Counting Rows in a CSV file
Public Function FileRowCount(FilePath As String, Optional RowDelimiter As String = vbCr) As Long
' Returns the row count of a text file, including the header row
' Returns - 1 on error
' Unicode-compliant, works on UTF-8, UTF-16, ASCII, with or without a Byte order Marker.
' Reads a typical 30Mb file over the network in 200-300ms. Hint: always copy to a local folder.
' If you're scanning files for use with a SQL driver, use basSQL.TableRowCount: it's 20x slower,
' but it returns a proper test of the file's usability as a SQL 'table'
' Nigel Heffernan Excellerando.Blogspot.com 2015
' Unit test:
' s=Timer : for i = 0 to 99 : n=FileRowCount("C:\Temp\MyFile.csv") : Next i : Print Format(n,"#,##0") & " rows in " & FORMAT((Timer-s)/i,"0.000") & " sec"
' Network performance on a good day: reads ~ 150 MB/second, plus an overhead of 70 ms for each file
' Local-drive performance: ~ 4.5 GB/second, plus an overhead of 4 ms for each file
On Error Resume Next
Dim hndFile As Long
Dim lngRowCount As Long
Dim lngOffset As Long
Dim lngFileLen As Long
Const CHUNK_SIZE As Long = 8192
Dim strChunk As String * CHUNK_SIZE
If Len(Dir(FilePath, vbNormal)) < 1 Then
FileRowCount = -1
Exit Function
End If
' trap the error of a folder path without a filename:
If FileName(FilePath) = "" Then
FileRowCount = -1
Exit Function
End If
hndFile = FreeFile
Open FilePath For Binary Access Read Shared As #hndFile
lngFileLen = LOF(hndFile)
lngOffset = 1
Do Until EOF(hndFile)
Get #hndFile, , strChunk
FileRowCount = FileRowCount + UBound(Split(strChunk, RowDelimiter))
Loop
Close #hndFile
Erase arrBytes
End Function
Public Function FileName(Path As String) As String
' Strip the folder and path from a file's path string, leaving only the file name
' This does not check for the existence or accessibility of the file:
' all we're doing here is string-handling
' Nigel Heffernan Excellerando.Blogspot.com 2011
Dim strPath As String
Dim arrPath() As String
Const BACKSLASH As String * 1 = "\"
strPath = Trim(Path)
If strPath = "" Then Exit Function
If Right$(strPath, 1) = BACKSLASH Then Exit Function
arrPath = Split(strPath, BACKSLASH)
If UBound(arrPath) = 0 Then ' does not contain "\"
FileName = Path
Else
FileName = arrPath(UBound(arrPath))
End If
Erase arrPath
End Function
Note the use of the Split
function to count the row separators: VBA's string-handling is generally slow, especially when you concatenate strings, but there are a couple of places where VBA can perform a string manipulation without internal allocation and deallocation; if you know where they are, you'll find that parts of your code run as fast as a 'C' developer's best work.
Warning: Horrible Hack
Strictly speaking, I should declare Dim arrBytes(CHUNK_SIZE) As Byte
and use this Byte array instead of strChunk
to receive the Get
from a file opened for binary read.
There are two reasons for not doing it the 'right' way:
- The last
Get
, which will set end-of-file TRUE, will extract less data from the file than the full 'chunk'. What happens next is that these last few bytes of the file are written into the array without clearing out the data from the previous 'Get'. So you have to do additional plumbing, counting bytes off against LOF(#hwndFile)
to detect the 'Last Get' and branching into a statement that clears the buffer, or allocates a smaller byte array and uses that instead; - The code will only cope with UTF-8 2-byte encoded character sets, or with single-byte encoded ASCII 'Latin' text if you do a bit of byte-array substitution around your row delimiters.
The VBA
String
type is a byte array with a wrapper that allows your code (or rather, the compiler) to handle all that complexity in the background.
However, it's much faster to go back into the primordial C, using old-school Get
statements, than using later libraries like Scripting.FileSystemObject
. Also, you have some ability to examine the incoming data at the byte level, to debug issues where you're getting '???????' characters instead of the text you were expecting.
Anyway: this is late to the game, as StackOverflow answers go, and it's an answer to the less-interesting part of your question. But it's going to be interesting to people who need a quick rowcount in their data files, and your question comes at the top of the list when they search for that.