Profiling VBA code for microsoft word - vba

I have some legacy code that uses VBA to parse a word document and build some XML output;
Needless to say it runs like a dog but I was interested in profiling it to see where it's breaking down and maybe if there are some options to make it faster.
I don't want to try anything until I can start measuring my results so profiling is a must - I've done a little searching around but can't find anything that would do this job easily. There was one tool by brentwood? that requires modifying your code but it didn't work and I ran outa time.
Anyone know anything simple that works?
Update: The code base is about 20 or so files, each with at least 100 methods - manually adding in start/end calls for each method just isn't appropriate - especially removing them all afterwards - I was actually thinking about doing some form of REGEX to solve this issue and another to remove them all after but its just a little too intrusive but may be the only solution. I've found some nice timing code on here earlier so the timing part of it isn't an issue.

Using a class and #if would make that "adding code to each method" a little easier...
Profiler Class Module::
#If PROFILE = 1 Then
Private m_locationName As String
Private Sub Class_Initialize()
m_locationName = "unknown"
End Sub
Public Sub Start(locationName As String)
m_locationName = locationName
MsgBox m_locationName
End Sub
Private Sub Class_Terminate()
MsgBox m_locationName & " end"
End Sub
#Else
Public Sub Start(locationName As String)
'no op
End Sub
#End If
some other code module:
' helper "factory" since VBA classes don't have ctor params (or do they?)
Private Function start_profile(location As String) As Profiler
Set start_profile = New Profiler
start_profile.Start location
End Function
Private Sub test()
Set p = start_profile("test")
MsgBox "do work"
subroutine
End Sub
Private Sub subroutine()
Set p = start_profile("subroutine")
End Sub
In Project Properties set Conditional Compilation Arguments to:
PROFILE = 1
Remove the line for normal, non-profiled versions.
Adding the lines is a pain, I don't know of any way to automatically get the current method name which would make adding the profiling line to each function easy. You could use the VBE object model to inject the code for you - but I wonder is doing this manually would be ultimately faster.

It may be possible to use a template to add a line to each procedure:
http://msdn.microsoft.com/en-us/library/aa191135(office.10).aspx
Error handler templates usually include an ExitHere label of some description.. The first line after the label could be the timer print.
It is also possible to modify code through code: "Example: Add some lines required for DAO" is an Access example, but something similar could be done with Word.
This would, hopefully, narrow down the area to search for problems. The line could then be commented out, or you could revert to back-ups.

Insert a bunch of
Debug.Print "before/after foo", Now
before and after snippets that you think might run for long terms, then just compare them and voila there you are.

My suggestion would be to divide and conquer, by inserting some timing lines in a few key places to try to isolate the problem, and then drill down on that area.
If the problem is more diffused and not obvious, I'd suggest simplifying by progressively disabling whole chunks of code one at a time, as far as is possible without breaking the process. This is the analogy of finding speed bumps in an Excel workbook by progressively hard coding sheets or parts of sheets until the speed problem disappears.

About that "Now" function (above, svinto) ...
I've used the "Timer" function (in Excel VBA), which returns a Single.
It seems to work just fine. Larry

Related

Splitting large code into Modules to Call

My end goal was to generate text into a TextBox (let's call it TextBox06) based on the values inserted into a series of Textboxes -- let's name them TextBox01, TextBox02, TextBox03, TextBox04, TextBox05. These TextBoxes are in a Userform (let's call it Userform01).
These TextBoxes can have a lot of different values and combinations (over 50 values each, and depending on which order the values are placed in the textboxes there will be a different output). Moral of the story: code probably works, but gives "procedure too large" error with no real way around it (pretty sure I have shrunk the code as much as possible, but there just aren't any more generic statements that repeat, and most of it must be a bruteforce-style method of inserting most of the combinations individually).
So I went looking on the internet for a way to split the code. I tried using the "call procedure" feature. What I did was:
I created two new Modules (Module01 and Module02) in the same "project".
I split the big code half in Module01, and half in Module02. The Modules' content is something like this:
Sub Module01()
[half the code]
End Sub
I wrote the following lines of code in each sub of the textboxes involved (this is in Userform01).
Private Sub TextBox01_Change()
Application.Run ("Module01")
Application.Run ("Module02")
End Sub
[Then five more copies for Private Sub TextBox02_Change () until Private Sub TextBox06_Change () included]
Assuming this is even the correct way of doing it, I then hit a roadblock. Apparently, copy-pasting the big code in Module01 and Module02 wasn't enough. From what I understood, the program doesn't seem to understand what the TextBox01.Text (and the like) values in the code are. So I tried looking for a solution to THAT, and found several possible codes, none of which seems to have worked so far. The one that I currently have is this:
Sub Module01()
Dim Text01 As String
Set Text01 = TextBox01.Text
[repeat for Text02 to Text06 included]
[half the code]
End Sub
[repeat for Module02 with the other half of the code]
Assuming that I'm even on the right track when it comes to properly using the "call" function, what is the proper way to make this work (i.e. how do I correctly reference the TextBox01.Text content in the Module, so I can then call the Modules in the Userform01) ?

Microsoft Project VBA to update Custom field on task change

I have been wracking my brain trying to work out how to write a small piece of code that will activate only when particular fields at a task level have been modified.
I tried to make this code work at the project change level with a for each loop and select cases but that lags the whole program and still doesn't give me the result I need. I also tried to make it work when run manually with a for each loop and select cases or a bunch of If statements, but again, it can't tell me which field changed, but it can highlight a discrepancy between two fields.
The goal is to have a change log field (Text10) that auto updates based on the field that is modified and the date of the change. I only care about 4 fields changing (Date1, Date2, Date3, Date4).
e.g. If [Date1] is modified, Text10 = "Date1 modified 10/11/21"
Note: If 2 fields are modified, I would be happy enough with just listing the last one.
I was hoping there was some sort of "On Change, If Target = xxx" but I have not been able to find anything like that.
I also tried implementing the code as defined here >> Microsoft Documents: Project.Change Event but I am unclear what this is supposed to do and couldn't actually see it doing anything / I never got the message box I believe was supposed to appear.
I am using Microsoft Project Standard 2019.
After much research and trial and error, I ended up solving this.
To get it working, I added a Class Module and ran a piece of code on open to initialize it. This essentially tells Project to start watching for events. I then use the "Field" variant to fill the field name amongst the text string and "NewVal" variant to fill the result. This was an easy solution in the end. The code I found that worked is below:
In Class Module "cm_Events"
Public WithEvents MyMSPApp As MSProject.Application
Private Sub Class_Initialize()
Set MyMSPApp = Application
End Sub
Private Sub MyMSPApp_ProjectBeforeTaskChange(ByVal tsk As Task, ByVal Field As PjField, ByVal NewVal As Variant, Cancel As Boolean)
'What you want the code to do
End Sub
In Module "m_Events"
Public oMSPEvents As New cm_Events
Sub StartEvents()
Set oMSPEvents.MyMSPApp = MSProject.Application
End Sub
In ThisProject code
Private Sub Project_Open(ByVal pj As Project)
Call m_Events.StartEvents
End Sub

What role do modules play in writing code that is less memory-intensive?

At work, I'm venturing into the world of VBA to try to make a spreadsheet template, which will run a report when a command button is clicked.
On several occasions I have encountered an "Out of Memory" run-time error, which has been easily fixed by jigging with the code using combinations of suggestions by other users on this site!
However, I am now curious about how to make codes less memory-intensive. In particular, do modules help make code less memory-intensive and, if so, how should I use them effectively in this regard? For example, should I assign a Sub to each module or would that be overkill?
I'm new to VBA so any/all help and criticism is very welcome!
Short answer: None whatsoever.
An "out of memory" error has many reasons to occur, none of which relate to how many modules your code is organized in. Because that's all modules are: an organizational tool at your disposal.
It is completely impossible to diagnose an "out of memory" error without knowing anything about what the code is doing: <meta> you would have to narrow down the problem to a particular specific piece of code before you can ask about it on this site (dumping a whole module and asking "what's wrong with this code?" isn't going to fly) </meta>
Look for very, very large arrays, clipboard operations perhaps. Maybe you're running a loop that's attempting to create and store a bajillion New large objects - could be pretty much anything... just not the number of modules.
should I assign a Sub to each module or would that be overkill?
As with many things, the answer is "it depends". Once you've familiarized yourself with the language, and start learning about Object-Oriented Programming (OOP), you'll come across design guidelines and principles that strongly advocate towards keeping public interfaces as simple as possible: adhering to the Interface Segregation Principle (the "I" of "SOLID") means that this is a perfectly complete and acceptable class module to have:
'#Interface IComparable
'#ModuleDescription("Defines a generalized type-specific comparison method that a class implements to order or sort its instances.")
Option Explicit
'#Description("Compares the current instance with another object of the same type and returns an integer that indicates whether the current instance precedes (-1), follows (1), or occurs in the same position in the sort order (0) as the other object.")
Public Function CompareTo(ByVal other As Object) As Integer
End Function
But that's well beyond beginner-level procedural programming concepts.
[...] which will run a report when a command button is clicked.
That command button could be an ActiveX button, or a Shape object that's attached to a macro. Assuming it's ActiveX, the worksheet's code-behind might look like this:
Option Explicit
Private Sub RunSalesReportButton_Click()
SalesReportMacro.CreateWeeklyReport
End Sub
And then you can have a SalesReportMacro standard module that might start out like this:
'#ModuleDescription("A macro that generates the weekly sales report.")
Option Explicit
Option Private Module
'#Descrition("Generates the weekly sales report.")
Public Sub CreateWeeklyReport()
If Not RefreshSalesData Then Exit Sub
CreatePivotReport
End Sub
The Public procedure at the top has a very high abstraction level: it's easy to tell at a glance what the procedure is going to be doing, because the lower-level procedures have meaningful names that tell us exactly what's going on. These lower-level procedures would be right underneath:
Private Function RefreshSalesData() As Boolean
Dim reportWeek As String
reportWeek = PromptForReportWeek
If Not IsNumeric(reportWeek) Then Exit Function
On Error GoTo CleanFail
AdjustDataConnectionCommand reportWeek
RefreshSalesData = True
CleanExit:
Exit Function
CleanFail:
MsgBox "Could not refresh the data for week " & reportWeek & ".", vbExclamation
Resume CleanExit
End Function
And the deeper you go, the lower the abstraction level becomes:
Private Function PromptForReportWeek() As String
Dim currentWeek As Integer
currentWeek = GetCurrentWeekNumber
PromptForReportWeek = InputBox("Please specify week#", "Generate Report", currentWeek)
End Function
Private Function GetCurrentWeekNumber()
GetCurrentWeekNumber = 42 'todo
End Function
Private Sub AdjustDataConnectionCommand(ByVal weekNumber As String)
With ThisWorkbook.Connections(1).OLEDBConnection
.CommandText = "dbo.WeeklySalesReport " & weekNumber
.Refresh
End With
End Sub
Private Sub CreatePivotReport()
'todo
End Sub
Procedures (Sub, Function, and others you'll discover in due time) should have one purpose, and be kept as short and focused on that one task as possible. By doing that, you make it easier to later refactor your code - say in 2 months you need to add another ActiveX button for a new CreateWeeklyInventoryReport macro: there's a good chance that this new report will also need to PromptForReportWeek - if that functionality is already well-abstracted into its own scope, you can more easily just remove/cut it from that module, add a new CalendarParameterPrompt module (which... might end up only having that procedure for a while.. until you need a good place to put a similar PromptForReportMonth function), add/paste it in there, make it Public, and invoke it from any other macro/module.
Doing this instead of massive omnipotent procedures that know everything and do everything (and eventually stop compiling because they got so large VBA refuses to compile them), will spare you lots of headaches in the near future.

How do I effectively create controls dynamically in Excel's VBA or How do I use Application.OnTime()?

I am working on a very large VBA project in Excel at my job. We are about 1500 lines of code for just one feature and have about a dozen more features to add. Because of this, I've been trying to break everything down so that I can keep code for each feature in separate places. OOP sucks in VBA... The problem being that these controls MUST have events fired. Of course, some events (like the TextBox_AfterUpdate event) are not available when you dynamically create controls. It's a bit convoluted because of everything that is going on, so I'll break it down the best I can:
I have a class module that represents a tab for a multipage control. When a user clicks on a tab, the Userform calls this class module and THERE I have the controls created dynamically. This way I can keep the code in that class module. I have a sub that I deemed as the "AfterUpdate" sub and put code that I needed to run there. Now the problem is to get that sub to be called at the appropriate time.
So what I did is to set up a Timer of sorts to check and see if the "ActiveControl" is said textbox. If it is not, we can assume that focus has left and we can raise that event. Here's the code I'm using:
An abbreviated version of the tab creation...
Private WithEvents cmbMarketplace As MSForms.ComboBox
Public Sub LoadTab(ByVal oPageTab As Object)
If TabLoaded Then Exit Sub
Set PageTab = oPageTab
Dim tmp As Object
Set tmp = PageTab.Add("Forms.Label.1")
tmp.Top = 6: tmp.Left = 6: tmp.Width = 48
tmp.Caption = "Marketplace:"
Set cmbMarketplace = PageTab.Add("Forms.ComboBox.1", "cmbMarketplace")
' LOAD OTHER CONTROLS '
TabLoaded = True
Start_Timer
End Sub
Then Start_Timer:
Public Sub Start_Timer()
TimerActive = True
Application.OnTime Now() + TimeValue("00:00:01"), "Timer"
End Sub
And the sub that is to be fired:
Public Sub Timer()
If TimerActive Then
' DO SOME RANDOM THINGS '
Application.OnTime Now() + TimeValue("00:00:01"), "Timer"
End If
End Sub
Does this seem like a reasonable approach to solving the problem I'm facing? I'm open to suggestions...
That's the first problem. This seems like a lot of work to accomplish this. (I'm working on getting visual studio, but I don't know if that's going to happen)
The above code will work but the "Timer" sub will not get raised at all. I get no errors if I just run the code. Everything is created, everything works as I would hope. However, if I step through the code, I eventually will get the following error:
Cannot run the macro "...xlsm!Timer". The macro may not be available in this workbook or all macros may be disabled.
Obviously neither of those suggestions are valid. Macros ARE enabled and the sub is in the same darn class module. I tried making it public, same problem. Tried "ClassModule1!Timer" to no avail. I'm at my wits end trying to figure this out. Thinking of having people write ALL this in the Userform or just giving up.
Does anybody have any suggestions on how to effectively break up large chunks of code? And does anybody have a clue why this sub will not run and seemingly cannot be found?
I understand that this is a confusing situation, so if you need more info or code examples or want to know why I have something set up the way I do, let me know.
Thanks!
Obviously neither of those suggestions are valid. Macros ARE enabled and the sub is in the same darn class module.
There's the problem: a macro cannot be in a class module. The message is entirely correct: VBA cannot see the Timer procedure, because it's not accessible.
A class module is a blueprint for an object, VBA (or any OOP language for that matter) can't do anything with a class module, without an instance of that class - i.e. an object.
Your timer callback needs to be a Public Sub in a standard module, so that it can be called directly as a macro. Public procedures of a class modules are methods, not macros.
Depending on what ' DO SOME RANDOM THINGS ' actually stands for, this may or may not require some restructuring.
1500-liner spaghetti code can be written in any language BTW.

How to call another module without returning to the first one after completion?

This is probably the dumbest question I've ever asked here, but it's hard to find answers to things like this.
I have a program with a bunch of modules/subs that each calculate a different variable. They're pretty complex, so I like to keep them separate. Now I want an earlier module to skip to another module based on user input. I thought I could use the call (sub name) method for this, but then the program returns to where the call line was and continues on that module from where it left off.
Example:
Module 1:
Sub NewPracticeSub()
Call otherpracticesub
MsgBox ("We've gone back to this sub... :(")
End Sub
Module 2:
Sub otherpracticesub()
MsgBox ("We're in the other practice sub!")
End Sub
I don't want it to return to Module 1. What can I do to have it switch control to Module 2 without it then returning to complete Module 1 upon completion of Module 2?
I feel like I just used the most confusing language possible to explain all of this, but thank you for your help anyways!!
Edit: I know I used the words module and sub interchangeably, and I know they're different. I like to keep each sub (which are each very large in my program) in their own modules because it's easier to keep track of them, and easier to explain/demonstrate the application flow to other people.
I think all you're looking for is the command Exit Sub which will make the program leave the subroutine without continuing any further, But the way you usually want to do this is, rather than calling a Sub, rather call a Function that returns a boolean value.
So, for example:
Public Function MyFunc() as Boolean
....
If [good] MyFunc = True
Else MyFunc = False
End Function
Then you could do something along the lines of:
Sub MyCallingSub()
...
If MyFunc = True then Exit Sub
Else ...
End Sub
It just adds in A LOT more felxibility and ability to choose whether you want to continue further in your sub or not.
Hope that makes sense.
Other than using the ugly End statement which I will describe below (and strongly recommend you to avoid), I'm not aware of any way to circumvent the call stack. Even John's response necessarily returns to the calling procedure, and evaluates another statement to determine whether to proceed or end.
This may yield undesirable outcomes, which is why I hesitate to recommend it, in favor of properly structuring your code, loops, etc., with respect to the call stack.
In any case, here is how you can use the End statement within your child subroutines, without needing any sort of public/global variables. This still allows you the flexibility to decide when & where to invoke the End statement, so it need not always be invoked.
Sub NewPracticeSub()
Call otherpracticesub, True
MsgBox ("We've gone back to this sub... :(")
End Sub
Sub otherpracticesub(Optional endAll as Boolean=False)
MsgBox ("We're in the other practice sub!")
If endAll then End '## Only invoke End when True is passed to this subroutine
End Sub
Why I say this method should be avoided, via MSDN:
"Note The End statement stops code execution abruptly, without
invoking the Unload, QueryUnload, or Terminate event, or any other
Visual Basic code. Code you have placed in the Unload, QueryUnload,
and Terminate events of forms and class modules is not executed.
Objects created from class modules are destroyed, files opened using
the Open statement are closed, and memory used by your program is
freed. Object references held by other programs are invalidated.
The End statement provides a way to force your program to halt. For
normal termination of a Visual Basic program, you should unload all
forms. Your program closes as soon as there are no other programs
holding references to objects created from your public class modules
and no code executing."
It will always return but that doesn't mean its a problem. I suggest you use Exit Sub as follows:
Sub NewPracticeSub()
Call otherpracticesub
**Exit Sub**
'Nothing more can execute here so its no longer a worry
End Sub
Module 2:
Sub otherpracticesub()
MsgBox ("We're in the other practice sub!")
End Sub