FindString to support whole words

[BarryG]

Can FindString maybe be updated with a flag to only find whole words? Like this:

Code: Select all

Debug FindString("text1$ text2$ text3$","t2$") ; Want 0 returned, not 11.

[AZJIO]

Takes into account the word found at the beginning and at the end of the line.

Code: Select all

EnableExplicit

Procedure FindString2(Source.s, SearchString.s)
	Protected pos = 1
	Protected length = Len(SearchString)
	Protected Result, charStart, charEnd
	Repeat
		pos = FindString(Source, SearchString, pos)
		If pos
			charStart = Asc(Mid(Source, pos - 1, 1))
			charEnd = Asc(Mid(Source, pos + length, 1))
; 			Debug Chr(charStart)
; 			Debug Chr(charEnd)
; 			Debug pos
; 			If Not ((charStart >= 65 And charStart <= 90) Or (charStart >= 97 And charStart <= 122) Or (charStart >= 48 And charStart <= 57 Or charStart = 36) Or (charEnd >= 65 And charEnd <= 90) Or (charEnd >= 97 And charEnd <= 122) Or (charEnd >= 48 And charEnd <= 57) Or charEnd = 36)
			If (Not ((charStart >= 65 And charStart <= 90) Or (charStart >= 97 And charStart <= 122) Or (charStart >= 48 And charStart <= 57 Or charStart = 36)) Or pos = 1) And Not ((charEnd >= 65 And charEnd <= 90) Or (charEnd >= 97 And charEnd <= 122) Or (charEnd >= 48 And charEnd <= 57) Or charEnd = 36)
				Result = pos
				Break
			EndIf
			pos + 1
		Else
			Break
		EndIf
	ForEver
	
    ProcedureReturn Result
EndProcedure

Debug FindString2("text1$ text2$ ext text3$","ext")
Debug FindString2("t2$ text1$ text2$ text3$ t2$","t2$")

Takes into account the word found at the beginning and at the end of the line.

Code: Select all

EnableExplicit

Procedure FindString2(Source.s, SearchString.s, separator.s = #TAB$ + " .,:<>()[]{}!")
	Protected pos = 1
	Protected length = Len(SearchString)
	Protected Result, charStart.s, charEnd.s
	Repeat
		pos = FindString(Source, SearchString, pos)
		If pos
			charStart = Mid(Source, pos - 1, 1)
			charEnd = Mid(Source, pos + length, 1)
; 			Debug charStart
; 			Debug charEnd
; 			Debug pos
; 			If FindString(separator, charStart) And FindString(separator, charEnd)
			If (pos = 1 Or FindString(separator, charStart)) And (FindString(separator, charEnd) Or charEnd = "")
				Result = pos
				Break
			EndIf
			pos + 1
		Else
			Break
		EndIf
	ForEver
	
    ProcedureReturn Result
EndProcedure

Debug FindString2("text1$ text2$ text3$","t2$")
Debug FindString2("text1$ text2$ text3$ t2$","t2$")

Code: Select all

EnableExplicit

Procedure FindString2(Source.s, SearchString.s)
	Protected pos = 1
	Protected length = Len(SearchString)
	Protected Result, charStart, charEnd
	Repeat
		pos = FindString(Source, SearchString, pos)
		If pos
			charStart = Asc(Mid(Source, pos - 1, 1))
			charEnd = Asc(Mid(Source, pos + length, 1))
			If pos = 1
				charStart = 1
			Else
				Select charStart
					Case 65 To 90, 97 To 122, 48 To 57, 36
						pos + 1
						Continue
				EndSelect
			EndIf
			Select charEnd
				Case 65 To 90, 97 To 122, 48 To 57, 36
					pos + 1
					Continue
			EndSelect
			Result = pos
			Break
		Else
			Break
		EndIf
	ForEver
	
	ProcedureReturn Result
EndProcedure

Debug FindString2("text1$ text2$ ext text3$","ext")
Debug FindString2("t2$ text1$ text2$ text3$ t2$","t2$")

[ebs]

Not official, but how about

Code: Select all

Debug FindString("text1$ text2$ text3$"," t2$") ; returns 0

[NicTheQuick]

Just use Regular Expressions. It's way easier with them. With a `\b` you can match word boundaries.

[Caronte3D]

Code: Select all

Procedure FindStringWhole(textString$, toFind$)

  Protected count = Len(toFind$)
  Protected count2 = count + 1
  Protected pos = 0
  
  If textString$ = toFind$ Or Left(textString$, count2) = toFind$ + " "
    pos = 1
  ElseIf Right(textString$, count2) = " " + toFind$
    pos = Len(textString$) - count+1
  Else
    pos = FindString(textString$, " " + toFind$ + " ")
    If pos > 0
      pos + 1
    EndIf
  EndIf
  
  ProcedureReturn pos
EndProcedure

Debug FindStringWhole("text1$ text2$ text3$","t2$")

[Kiffi]

@Caronte3D:

Code: Select all

Debug FindStringWhole("Hello World!","World")

[NicTheQuick]

This is the simplest way for doing it with Regular Expressions:

Code: Select all

Procedure FindWholeWord(Source.s, SearchString.s, Mode.i = 0)
	Protected regExFlags.i = #PB_RegularExpression_MultiLine
	If Mode = #PB_String_NoCase
		regExFlags | #PB_RegularExpression_NoCase
	EndIf
	
	Protected hRegEx.i = CreateRegularExpression(#PB_Any, "\b" + SearchString + "\b", regExFlags)
	If Not hRegEx
		ProcedureReturn -1
	EndIf
	
	If Not ExamineRegularExpression(hRegEx, Source)
		ProcedureReturn -2
	EndIf
	Protected position.i
	If NextRegularExpressionMatch(hRegEx)
		position = RegularExpressionMatchPosition(hRegEx)
	EndIf
	FreeRegularExpression(hRegEx)
	ProcedureReturn position
EndProcedure

Debug FindWholeWord("text1$ text2$ text3$", "t2$")
Debug FindWholeWord("Hello World!", "world", #PB_String_NoCase)

But this is not a complete procedure. You also have to escape all characters that have a special meaning in a RegEx before searching for it.
Also this version creates a lot of overhead because the whole RegEx engine must be loaded, the pattern must be parsed and finally the matching has to be done.
You can improve that in case you always want to find the same pattern. Then just create one RegEx out of it and use `ExamineRegularExpression()` on multiple strings.

[AZJIO]

\b - works only for Latin letters and numbers and _, but is not suitable for $.
The string may contain meta characters and they must be escaped.
Import regex +150-200 kb to the file size.

[AZJIO]

Can FindString maybe be updated with a flag to only find whole words? Like this:

Adding parameters can slow down the function.
If only the compiler counted the number of function parameters and chose a more optimized one. For example, if there are 4 parameters, then the FindString4 function is embedded, and if there are 5 parameters, then the FindString5 function is embedded, that is, as the parameters increase, the compiler would embed the function with the largest number of parameters, while the function name remains FindString. Then increasing the number of parameters would not harm those who want a fast function with fewer parameters. If there are many function calls with different parameters, then the only function with the largest number of parameters is embedded.

[Quin]

Adding parameters can slow down the function.
If only the compiler counted the number of function parameters and chose a more optimized one. For example, if there are 4 parameters, then the FindString4 function is embedded, and if there are 5 parameters, then the FindString5 function is embedded, that is, as the parameters increase, the compiler would embed the function with the largest number of parameters, while the function name remains FindString. Then increasing the number of parameters would not harm those who want a fast function with fewer parameters. If there are many function calls with different parameters, then the only function with the largest number of parameters is embedded.

This sounds false to me.
Sure, the more parameters you add, the larger the paramsize and as a result the stack frame, and copying all the values from the parameters will take longer if you have more of them, but it shouldn't actually slow down the operation of the function.

[NicTheQuick]

\b - works only for Latin letters and numbers and _, but is not suitable for $.

Oh, then Purebasic's RegEx engine seems to be configured badly. Usually it works with UTF-8 characters and not just latin characters.

The string may contain meta characters and they must be escaped.

Exactly. That's what I said.

Import regex +150-200 kb to the file size.

I don't think a lot of people care about the executable size today.

[Quin]

Import regex +150-200 kb to the file size.

I don't think a lot of people care about the executable size today.

I'm a huge proponent of executable size, trying to make my binaries as small as they can reasonably be.
And that's why I use PureBasic. Statically linking SQLite doesn't even bring my binary to a meg!
So, even if you care about binary size, PB is the best choice. Ever tried using regular expressions in Python or .NET and then seeing all the stuff it has to add to your dist?

[miso]

Ever tried using regular expressions in Python or .NET and then seeing all the stuff it has to add to your dist?

I,ve seen a monster executable created by python over a gig... :S
Was not made by me.

[Quin]

Ever tried using regular expressions in Python or .NET and then seeing all the stuff it has to add to your dist?

I,ve seen a monster executable created by python over a gig... :S
Was not made by me.

Sheesh!
Python isn't a bad language, but it is noooot for desktop apps. Web scrapers? Sure. Random scripts? Yeah. But please don't write your desktop apps in Python...

[NicTheQuick]

So, even if you care about binary size, PB is the best choice. Ever tried using regular expressions in Python or .NET and then seeing all the stuff it has to add to your dist?

Well, you can not compare a language that gets compiled into an executable with a script language that needs a VM. Also regular expressions are one of the included libraries of Python. So if you use it or not, it has always the same size. And I don't know anybody who compiles Python into an executable. Maybe some Windows guys do weird things like that. On Linux I never saw that. It compiles itself into bytecode but that's all.