Filter out duplicates with a map

[netmaestro]

One of the useful attributes of a map is that it won't hold duplicate values. If you try to put a key into it that is identical to a key it already contains, the existing key will be replaced. Only one element of a given key will ever exist in the map. So it follows that if you have a dataset that contains duplicates and you want to filter the dupes out, you can just dump the whole thing in a map and voila - all the dupes are gone. All you're left with are the unique values. Forget about sorting, stepping through while checking for a new value, copying uniques to another array, all that fiddly stuff. This is like throwing the whole set at a wall and only the unique values will stick - the rest slide off. One practical example is counting the unique colors in a bitmap:

Code: Select all

Procedure CountColorsUsed(pBitmap)
  GetObject_(pBitmap, SizeOf(BITMAP), @bmp.BITMAP) 
  *bmi.BITMAPINFO = AllocateMemory(SizeOf(BITMAPINFO)+SizeOf(RGBQUAD)*255)
  With *bmi\bmiHeader
    \biSize         = SizeOf(BITMAPINFOHEADER) 
    \biWidth        = bmp\bmWidth 
    \biHeight       = bmp\bmHeight 
    \biPlanes       = 1 
    \biBitCount     = 32 
  EndWith 
  hDC = GetWindowDC_(#Null) 
  GetDIBits_(hDC, pBitmap, 0, bmp\bmHeight, #Null, *bmi, #DIB_RGB_COLORS)
  Global *pPixels = AllocateMemory(*bmi\bmiHeader\biSizeImage)
  iRes = GetDIBits_(hDC, pBitmap, 0, bmp\bmHeight, *pPixels, *bmi, #DIB_RGB_COLORS) 
  ReleaseDC_(#Null, hDC) 

  Global NewMap Colors(MemorySize(*pPixels)/2)
  *p.long = *pPixels
  For i=1 To MemorySize(*pPixels)/SizeOf(RGBQUAD)
    Colors(Hex(*p\l)) = *p\l
    *p+SizeOf(RGBQUAD)
  Next
  
  result = MapSize(colors())
  ClearMap(colors())
  FreeMemory(*pPixels)
  FreeMemory(*bmi)
  
  ProcedureReturn result
EndProcedure

[idle]

5/5 for style and creative use
but now I'm wondering if we should be calling you crazy frog!

[Fred]

Yes, good usage of fast map lookup and uniqueness.

[Xombie]

! I did literally the exact same thing a while back while messing around with character recognition. To filter duplicates and count unique values. That's humorous to me