Hi,

Before giving you the answers, I may have an idea: it seems that the problematic pdf is encrypted. I think that PDF::API2 doesn't work because it tries to copy some sort raw content from an encrypted pdf to a non encrypted file (which produces blank pages because it's incorrect data). CAM::PDF might work because it starts with the original PDF and then remove the unwanted pages leaving the file encrypted (I presume that sejda either do the same or first decipher the content before copying it).

PDF-API2 folder in my Strawberry Perl installation gives 2.033:

Here is the result of your script with juste a printf added on line 11 to check the encryption (printf "isEncrypted : %s\n",$pdf->isEncrypted();) :

PDF Version : 1.3
Pages       : 540
isEncrypted : 1
Page   1 Media  0.00  0.00 595.00 864.00
Page   2 Media  0.00  0.00 595.00 864.00
Page   3 Media  0.00  0.00 595.00 864.00
Page   4 Media  0.00  0.00 595.00 864.00
Page   5 Media  0.00  0.00 595.00 864.00
Page   6 Media  0.00  0.00 595.00 864.00
Page   7 Media  0.00  0.00 595.00 864.00
Page   8 Media  0.00  0.00 595.00 864.00
Page   9 Media  0.00  0.00 595.00 864.00
Page  10 Media  0.00  0.00 595.00 864.00
Page  11 Media  0.00  0.00 595.00 864.00
Page  12 Media  0.00  0.00 595.00 864.00
...an so on until page 540 (values are exactly the same)
[download]