[Image-SIG] Patch for BmpImagePlugin

[Rob Hooft]

Here is a patch that makes PIL read Bmp images made by Windows, where the
number of colors is smaller than the maximum for the number of bits per
pixel (I have an 8bit image here that uses a 32 color palette).

*** /usr/people/hooft/p/basis4/Imaging-0.3b2/PIL/BmpImagePlugin.py	Sun Jul 19 16:25:14 1998
--- ./BmpImagePlugin.py	Thu Sep 24 11:19:04 1998
***************
*** 70,75 ****
--- 70,76 ----
  	    bits = i16(s[10:])
  	    self.size = i16(s[4:]), i16(s[6:])
  	    lutsize = 3
+             colors=1 << bits
  
  	elif len(s) in [40, 64]:
  
***************
*** 77,82 ****
--- 78,86 ----
  	    bits = i16(s[14:])
  	    self.size = i32(s[4:]), i32(s[8:])
  	    self.info["compression"] = i32(s[16:])
+             colors=i32(s[32:])
+             if colors==0:
+                 colors=1 << bits
  	    lutsize = 4
  
  	else:
***************
*** 90,96 ****
  
  	# LUT
  	if self.mode == "P":
- 	    colors = 1 << bits
              palette = []
              greyscale = 1
  	    if colors == 2:
--- 94,99 ----