InsertString: #PB_String_InPlace would be useful

[netmaestro]

ReplaceString has this option and today I find a need for it with InsertString, but it isn't available. I wrote my own:

Code: Select all

Procedure.s UpdateString(target$, insert$, pos)
  *target.Character = @target$ + (pos*SizeOf(character))
  *insert.Character = @insert$
  While *insert\c
    *target\c = *insert\c
    *target+SizeOf(Character)
    *insert+SizeOf(Character)
  Wend
  ProcedureReturn target$
EndProcedure

a$ = "Hello World!"
b$ = "Girls"

a$ = UpdateString(a$, b$, 6)

Debug a$

But I think having the core functionality would be superior to this.

[edit] Made code unicode-friendly before somebody yells at me

[ts-soft]

[edit] Made code unicode-friendly before somebody yells at me

better is it

[STARGÅTE]

that is no insert, it is an overwrite.

and your code don't work like ReplaceString with #PB_String_InPlace without return:

Code: Select all

Define Text.s = "Hallo World!"
ReplaceString(Text, "World", "Girls", #PB_String_InPlace)
Debug Text

what you want would be something like this:

Code: Select all

Procedure OverwriteString(*String, *NewString, Position.i)
	If MemoryStringLength(*NewString) + Position - 1 <= MemoryStringLength(*String) And Position > 0
		CopyMemory(*NewString, *String+(Position-1)*SizeOf(Character), MemoryStringLength(*NewString)*SizeOf(Character))
	EndIf
EndProcedure

Define Text.s = "Hello World!"
Define NewString.s = "Girls"
OverwriteString(@Text, @NewString, 7)
Debug Text

[netmaestro]

Destructive insert = overwrite, that's just semantics. I prefer it to return the result, which is why my code works as it does. If the team implements the flag, I expect it will work in place as ReplaceString does.

[netmaestro]

what you want would be something like this:

Actually, what I'd want would be something like this:

http://www.purebasic.fr/english/viewtop ... 59#p362759

where the execution time is cut by two thirds.

[Little John]

Code: Select all

Procedure.s UpdateString(target$, insert$, pos)
  *target.Character = @target$ + (pos*SizeOf(character))
  *insert.Character = @insert$
  While *insert\c
    *target\c = *insert\c
    *target+SizeOf(Character)
    *insert+SizeOf(Character)
  Wend
  ProcedureReturn target$
EndProcedure

a$ = "Hello World!"
b$ = "Girls"

a$ = UpdateString(a$, b$, 6)

Debug a$

If insert$ is too long, or pos is too big, respectively, then accidentally the excess characters of insert$ will be written behind the end of target$. And additional check will prevent this:

Code: Select all

While *insert\c And *target\c

Regards, Little John

PS: I saw that you wrote an assembly procedure now, but since I'm not good in understanding assembly at 8 o'clock in the morning, I commented on this PB code.

[freak]

If you don't care about the length check on the target, you can just use PokeS()

[netmaestro]

If you don't care about the length check on the target, you can just use PokeS()

That was my first thought, but I rejected it because PokeS() will write a terminating zero, effectively ending the string at the end of my insert, is that right? If not I went to a lot of trouble for nothing.

[luis]

It does so (adding the zero).

[Demivec]

If you don't care about the length check on the target, you can just use PokeS()

That was my first thought, but I rejected it because PokeS() will write a terminating zero, effectively ending the string at the end of my insert, is that right? If not I went to a lot of trouble for nothing.

@netmaestro: PokeS() always writes a terminating zero. I favor an overwrite method also. Preferably one replaces characters without making the target string longer or shorter. Does your ideal solution include making the target string longer if placing the new characters near the end?

[netmaestro]

Does your ideal solution include making the target string longer if placing the new characters near the end?

No, because it's for updating an ISAM file with fixed-length records. The length of the target string never changes.

[Little John]

That was my first thought, but I rejected it because PokeS() will write a terminating zero, effectively ending the string at the end of my insert, is that right?

What about CopyMemory() or CopyMemoryString()?

Regards, Little John

[Demivec]

That was my first thought, but I rejected it because PokeS() will write a terminating zero, effectively ending the string at the end of my insert, is that right?

What about CopyMemory() or CopyMemoryString()?

@Little John: CopyMemoryString() will always place a terminating zero also. CopyMemory() would be the only viable choice.

[netmaestro]

CopyMemory is usable but the code you have to write to allow for unicode, while it works, looks ugly to me. CopyMemoryString also writes the terminating zero, so it's no use either. I just want the flag.

[wilbert]

It would make sense to have a flag for CopyMemoryString to not write the terminating zero.