[reportlab-users] problem with PIL 1.1.7 and ReportLab 2.2

[Robin Becker]

On 10/12/2009 16:58, Bill Janssen wrote:

> Folks, I've recently upgraded to PIL 1.1.7, and I'm suddenly seeing an

> issue with ReportLab that I hadn't before.  Here's my code:

>

>              c = canvas.Canvas(outputfile,

>                                pagesize=(5 * inch, 3 * inch))

>              border = 0.3 * inch

>              x = border

>              y = (3 * inch) - (2 * border)

>              iconwidth = 0.5 * inch

>              iconheight = iconwidth / self._CARDIMAGEASPECT

>              c.drawImage(_CARDIMAGEFILEPATH,

>                          (5 * inch - border - iconwidth),

>                          (3 * inch - border - iconheight),

>                          iconwidth, iconheight)

>

> The path specified by _CARDIMAGEFILEPATH does exist and is readable and

> is a valid PNG (this code has been working fine for over a year).

>




>

> And here's the stack trace:

>

> [... my frames omitted ...]

> thr008:   File "/opt/GoodStuff/UpLib/lib/python2.5/site-packages/reportlab/pdfgen/canvas.py", line 623, in drawImage

> thr008:     imgObj = pdfdoc.PDFImageXObject(name, image, mask=mask)

> thr008:   File "/opt/GoodStuff/UpLib/lib/python2.5/site-packages/reportlab/pdfbase/pdfdoc.py", line 1994, in __init__

> thr008:     self.loadImageFromA85(src)

> thr008:   File "/opt/GoodStuff/UpLib/lib/python2.5/site-packages/reportlab/pdfbase/pdfdoc.py", line 1998, in loadImageFromA85

> thr008:     imagedata = map(string.strip,pdfutils.makeA85Image(source,IMG=IMG))

> thr008:   File "/opt/GoodStuff/UpLib/lib/python2.5/site-packages/reportlab/pdfbase/pdfutils.py", line 35, in makeA85Image

> thr008:     raw = img.getRGBData()

> thr008:   File "/opt/GoodStuff/UpLib/lib/python2.5/site-packages/reportlab/lib/utils.py", line 650, in getRGBData

> thr008:     self._dataA = ImageReader(im.split()[3])

> thr008:   File "/opt/GoodStuff/UpLib/lib/python2.5/site-packages/PIL/Image.py", line 1497, in split

> thr008:     if self.im.bands == 1:

> thr008: AttributeError: 'NoneType' object has no attribute 'bands'

>

> Anyone see what's going wrong here?  What do I need to change?

>

> If you'd like to, you can try it yourself by downloading UpLib from

> http://uplib.parc.com and building it with various combos of ReportLab

> and PIL.

>

> Bill

........

well since I assume it worked with < 1.1.7 it's likely the problem is with some 
change in PIL.

I haven't tried PIL 1.1.7 the latest I see is 1.1.6; is this code failing with 
all images? Looking at the beta stuff I see some changes were made for the PNG 
format so perhaps they affected this attempt at getting information about the 
alpha channel. Not having messed about I'm guessing that im.split() isn't 
returning what it used to. Looking at the code near the failure

	im = self._image
	mode = self.mode = im.mode
	if mode=='RGBA':

it seems as though the image has been identified as RGBA and then we're trying 
to use im.split() to create a new image representing the alpha channel (so it 
can be used as a mask later on).
-- 
Robin Becker