[Swftools-common] Different textSnapshot behavior in pdf2swf-generated files

[Marc Hämmerle]

hello,

we need text selection functionality for a project and stumbled upon  
some very strange things in swfs generated by pdf2swf. our application  
is written in flex & actionscript 3 so we need the files to be avm2- 
compatible; localconnection is not an option. pdf2swf outputs them  
fine since the last 2 (?) releases. but if you select text via  
textSnapshot and setSelected on a swf generated by pdf2swf, you get a  
(visual) selection with each character selected on it's own and spaces  
*not* selected instead of one continous line of selection (as it  
should be). if make myself a flash-file in the flash9 ide, put in some  
static textfields and a dynamic textfield for character metrics and  
publish it as avm2-file, the selection works as expected with our  
application. as i remembered that old examples for flash 7 or 8 *did*  
indeed work i tried that out and voila: they still work if you write  
your application for the avm1 and use a swf generated by pdf2swf for  
flash 6 (-T 6 parameter).
and to make things even more weird: you can use classes in  
actionscript 3 that write a new header in the bytearray of loaded avm1- 
files to dupe them as avm2-files for the actionscript virtual machine  
*if* you use flash-ide generated avm1 swfs, but *not* when you use  
pdf2swf-generated avm1-files. it seems to me that there's a failure in  
the swf header outputted by swftools.

(it's probably nothing, but flash ide-generated files and pdf2swf  
generated files differ on the framerate when you swfdump them: ide  
states it as 25.00 fps and pdf2swf states them as 0.25 fps)

has anybody else stumbled upon this problems and maybe has got a  
solution?

marc