[RESOLVED] Search engine speed

[GARY MICHAEL]

Hello,

Anybody know of the fastest search engine in VB?

I loaded the whole kjv Bible into a rtf box and it took me 37 seconds to fine the word "day."

Private Function WordCount(strLine As String, strWord As String) As Long
   WordCount = (Len(strLine) - Len(Replace(strLine, strWord, ""))) / Len(strWord)

End Function

Private Sub cmdFindNext_Click()
 Dim intIdx As Long
 Dim CountLeft As Integer
 Dim response As String
 Dim strtext As String
 Dim NumLoop As Integer

 FoundLabel.Visible = True

 strtext = RTFText1.Text

 RTFText1.SetFocus

 If txtSearch.Text = "" Then
   COUNTED = 0
   Exit Sub
 End If     
 
 COUNTED = COUNTED + 1
 CountLeft = WordCount(strtext, txtSearch.Text) - COUNTED

 NumLoop = NumLoop + 1
 FoundLabel.Caption = " Hit #" & COUNTED & " of " & WordCount(strtext, txtSearch.Text) & ", there are "  & CountLeft & " Left"

 RTFText1.Find txtSearch, RTFText1.SelStart + 1, Len(RTFText1.TextRTF)

 If NumLoop = 2 Then
   FoundLabel.Caption = " Hit #" & COUNTED - 1 & " of " & WordCount(strtext, txtSearch.Text)
   Beep
   COUNTED = 0
   Exit Sub
 End If
    POS = Form1a.RTFText1.SelStart
    POS = InStr(POS + 1, Form1a.RTFText1.Text, txtSearch.Text)
         
  If POS = 0 Then
             
   response = MsgBox("your search for ~ " & txtSearch & " ~ ... was found a total of " & COUNTED - 1 & " time(s). ...  Do you want to start from the beginning?", vbYesNo)
   COUNTED = 0
   FoundLabel.Caption = " Hit #" & COUNTED & " of " & WordCount(strtext, txtSearch.Text)

           If response = vbNo Then
               Exit Sub
           Else

               RTFText1.Find txtSearch, RTFText1.SelStart + 1, Len(RTFText1.TextRTF)
         
               Form1a.RTFText1.SelStart = POS 'GOES TO THE TOP
           End If
  End If

If OptVar1 = True Then
   Exit Sub
End If

RTFText1.SelColor = vbRed
RTFText1.SetFocus
SendKeys "{down 20}"

If SearchBoldVar = True Then
   RTFText1.SelBold = Not RTFText1.SelBold
End If

If SearchItalicVar = True Then
   RTFText1.SelItalic = Not RTFText1.SelItalic
End If

If SearchUnderlineVar = True Then
   RTFText1.SelUnderline = Not RTFText1.SelUnderline
End If

If SearchRegularVar = True Then
   SearchItalicVar = False
   SearchBoldVar = False
   SearchUnderlineVar = False
End If

Exit Sub
 
End Sub

[si_the_geek]

You haven't shown your WordCount function, but it is being called twice (with exactly the same parameters) before you even begin searching. In some cases you also run it again (with exactly the same parameters again) after the search.

Rather than re-running it multiple times, run it once and store the result.


There are several other things about your code that are less than ideal, such as:

• Using a String to store the result of a MsgBox... when the data type it returns is numeric (so use either Long, or vbMsgBoxResult).
• Joining two hard-coded strings (" " & "Hit #"), when you could just put the result (" Hit #").
• Terrible indenting, which makes it extremely hard to read - and thus check what it is doing.

Due to the last point, I'm not even going to read most of your code.

[GARY MICHAEL]

Thanks for your comments. Maybe somebody else will have the patience to go through it for me. I appreciate what you DID say. Will the things you mentioned slow my search down?

[GARY MICHAEL]

You know the

" " & "Hit #",

code was pretty stupid. I know better. Don't know what I was thinking about. Part of this code was done a long time ago and I never noticed the need to change it.

[si_the_geek]

Thanks for your comments. Maybe somebody else will have the patience to go through it for me.

That will be significantly more likely if you stop intentionally making it hard to read.

Will the things you mentioned slow my search down?

Three of the four things I mentioned definitely will.

We cannot tell for sure how much each would be by (especially as we can't see all of the code), but the first could be very significant.

[GARY MICHAEL]

I don't know what would make you think that I would intentionally make it hard to read. That is a pretty strong inditement on my character. You are mistaken and you don't know me well enough to make such a statement. I need to learn how to code in such a fashion that will be more readable. I'll try to do better.

[si_the_geek]

Your indenting is completely random, which makes it even harder to read than if there was no indenting at all.

You did that indenting, which means that to some degree it was intentional - even if you didn't consciously set out to make it hard to read.

If you were writing a letter, would you start each line a random amount across the page? I'm assuming not, but that is essentially what you have done.


The only proper indenting you have got is on these three lines:

  End If
End If
If OptVar1 = True Then

..but due to the rest of the code, we have no idea which If each of those End If's refers to - and the only way to find out is to read each of the previous lines of code to work out where it is.

With proper indenting there would be no reading needed, and you could see almost instantly.


As a simple guideline, each time you have a code structure like If-Else-End If (or For-Next / Do-Loop / etc), you should have the same number of spaces at the start of each of those lines, and more spaces in front of all of the lines between them, eg:

  If response = vbNo Then
    Exit Sub
  Else
    RTFText1.Find txtSearch, RTFText1.SelStart + 1, Len(RTFText1.TextRTF)
    Form1a.RTFText1.SelStart = POS 'GOES TO THE TOP
  End If

Now you can see at a glance which If the End If (and Else) refer to.

When you have one code structure inside another, you increase the indenting as apt, eg:

If Number > 5 Then
  If Number > 10 Then
    MsgBox "Number is more than 10"
  Else
    MsgBox "Number is more than 5, but not more than 10"
  End If
End If

..and again, you can tell at a glance which refers to which.


Most people use the Tab key for indenting, and VB allows you to easily indent a whole group of lines at the same time, by selecting the lines and then pressing Tab. You can set the amount of spaces a Tab causes in VB's options (the default is 4, I personally prefer 2).

[GARY MICHAEL]

Oh! Now I get it. I think. I will change my post above and see how it goes. Thanks SI.

[GARY MICHAEL]

Alright Si, check me out. Is my post better as changed above?
I've included this code too.

Private Function WordCount(strLine As String, strWord As String) As Long
   WordCount = (Len(strLine) - Len(Replace(strLine, strWord, ""))) / Len(strWord)

End Function

[si_the_geek]

That is much better.

It isn't quite perfect, because there is still some randomness in there, mainly in this section:

 End If
    POS = Form1a.RTFText1.SelStart
    POS = InStr(POS + 1, Form1a.RTFText1.Text, txtSearch.Text)
         
  If POS = 0 Then
             
   response = MsgBox("your search for ~ " & txtSearch & " ~ ... was found a total of " & COUNTED - 1 & " time(s). ...  Do you want to start from the beginning?", vbYesNo)
   COUNTED = 0
   FoundLabel.Caption = " Hit #" & COUNTED & " of " & WordCount(strtext, txtSearch.Text)

           If response = vbNo Then
               Exit Sub
           Else

               RTFText1.Find txtSearch, RTFText1.SelStart + 1, Len(RTFText1.TextRTF)
         
               Form1a.RTFText1.SelStart = POS 'GOES TO THE TOP
           End If
  End If

..which should be like this:

 End If
 POS = Form1a.RTFText1.SelStart
 POS = InStr(POS + 1, Form1a.RTFText1.Text, txtSearch.Text)
         
 If POS = 0 Then
             
   response = MsgBox("your search for ~ " & txtSearch & " ~ ... was found a total of " & COUNTED - 1 & " time(s). ...  Do you want to start from the beginning?", vbYesNo)
   COUNTED = 0
   FoundLabel.Caption = " Hit #" & COUNTED & " of " & WordCount(strtext, txtSearch.Text)

   If response = vbNo Then
     Exit Sub
   Else

     RTFText1.Find txtSearch, RTFText1.SelStart + 1, Len(RTFText1.TextRTF)
         
     Form1a.RTFText1.SelStart = POS 'GOES TO THE TOP
   End If
 End If

Note that the amount of spaces you should use for indenting is a matter of opinion, just use whatever you feel happiest with (but always use the same amount, and set the VB option for spaces per Tab to match it).



The next thing to do is act on the rest of post #2 (the first half, and the first bullet point).

[GARY MICHAEL]

Thanks again.

What would you change in my code to speed it up to the max?

[si_the_geek]

The changes I've already recommended are likely to produce a noticeable improvement - it wouldn't be surprising if the 37 seconds is reduced to something less than 30 (perhaps even under 20).

To make it faster than that would require replacing RTFText1.Find with something else, and/or tidying up the rest of the code (for example, NumLoop can only ever have the value 1, yet you have code which checks if it is 2).

[GARY MICHAEL]

If you could change my code to reflect what you are talking about, I could see it better and compare to two codes. Can you do that?

[GARY MICHAEL]

Here is a revision of my code.

Private Function WordCount(strLine As String, strWord As String) As Long
   WordCount = (Len(strLine) - Len(Replace(strLine, strWord, ""))) / Len(strWord)

End Function


Private Sub cmdFindNext_Click()

Dim intIdx As Long
Dim CountLeft As Integer
Dim response As Long
Dim strtext As String

FoundLabel.Visible = True
        
strtext = RTFText1.Text

RTFText1.SetFocus

If txtSearch.Text = "" Then
    COUNTED = 0
    Exit Sub
End If
     
      
COUNTED = COUNTED + 1
CountLeft = WordCount(strtext, txtSearch.Text) - COUNTED

FoundLabel.Caption = " Hit #" & COUNTED & " of " & WordCount(strtext, txtSearch.Text) & ", there are " & CountLeft & " Left"

RTFText1.Find txtSearch, RTFText1.SelStart + 1, Len(RTFText1.TextRTF)

    POS = Form1a.RTFText1.SelStart
    POS = InStr(POS + 1, Form1a.RTFText1.Text, txtSearch.Text)
         
If POS = 0 Then
             
    response = MsgBox("your search for ... " & txtSearch & " ... was found a total of " & COUNTED - 1 & " time(s). ...  Do you want to start from the beginning?", vbYesNo)
    COUNTED = 0
    FoundLabel.Caption = " " & "Hit #" & COUNTED & " of " & WordCount(strtext, txtSearch.Text)

    If response = vbNo Then
        Exit Sub
    Else

        RTFText1.Find txtSearch, RTFText1.SelStart + 1, Len(RTFText1.TextRTF)
       
        Form1a.RTFText1.SelStart = POS 'GOES TO THE TOP
    End If
End If


RTFText1.SelColor = vbRed
RTFText1.SetFocus
SendKeys "{down 20}"

 
End Sub

[GARY MICHAEL]

I don't think VB has the speed I'm looking for. I have a program and has a very, very large file that searches the whole document in a split second. I think that it was written in visual c++. That is the kind of speed I need. I will hunt for a visual C++ search engine.

How do I end this this thread, since it was not resolved? Do I unsubscribe to it?

[si_the_geek]

VB has the speed, but you aren't attempting making use of it... instead you are using one of the easiest methods available, doing very little to improve it even after apt advice, and are pretending to yourself that somehow the speed is the fault of the language.

Based on the level of coding skills and effort you have shown, there is no reason to believe that in a reasonable time frame you could get a C based program working at all (as it is very strict in terms of data types, etc), let alone get it to be faster than what you have above.

If you want more speed you have to put in extra effort... yet it took 11 replies for you to get around to this:

Here is a revision of my code.

That has finally corrected one of the simplest issues I mentioned back in post #2, but it was the smallest of the three speed based issues (its main effect is reducing bugs).


In terms of what I have advised you to do already, the most significant speed issue by far is to only call the WordCount function once, and store the result to a variable - then use the variable everywhere that you currently call the function.

As you currently have it, each time you run it you are using exactly the same parameters, so it is doing exactly the same work again, and returning exactly the same result. This is not specifically a programming issue, it applies in many other situations too... why do the same work repeatedly, when re-doing it has no extra effect?

If you could change my code to reflect what you are talking about, I could see it better and compare to two codes. Can you do that?

No, you do it yourself - and show us what you come up with.

If it doesn't work, we can help correct it, and explain why those corrections are needed. If it works but has room for improvement, we can tell you how to do that, and why. Whichever way it goes, you will learn far more than if we just gave you code.

Doing things for you takes us more time, and leads to more of the same... and as we are giving up our own free time to help a random stranger, does that sound like a reasonable request to make?

How do I end this this thread, since it was not resolved? Do I unsubscribe to it?

If you want to give up then just mark it as Resolved so that people don't keep trying to help, and reply (and/or edit the title) to say that you have given up so that people who search can tell.

[CVMichael]

GARY MICHAEL,

Take a look at this for indenting your code with just one click of a button:
http://www.vbforums.com/showthread.php?t=479449

[GARY MICHAEL]

Thanks CV Michael.