Calculation of numberOfTilesX and numberOfTilesY subtracts xOsiz/yOsiz instead of xTOsiz/yTOsiz

[bitsgalore]

As described here by @comstock and @aous72:

aous72/OpenJPH#113 (comment)

Some of the files (both JPH and codestream) fail validation with:

    <tests>
        <foundExpectedNumberOfTiles>False</foundExpectedNumberOfTiles>
        <foundExpectedNumberOfTileParts>False</foundExpectedNumberOfTileParts>
    </tests>

TODO: figure out whether this is a Jpylyzer bug or a real fault of the images.

[bitsgalore]

Looks like a fault of the images, not Jpylyzer, see:

aous72/OpenJPH#113 (comment)

[aous72]

Hi Johan,

Thank you for looking into this -- it always useful to confirm against other peoples' code.
I apologize for not having the time to have a more detailed report when I first looked at it.

With that said, -- please correct me if I am wrong -- the image simple_enc_irv97_tall_narrow1.j2c has TWO tiles.
Please refer to equation B-5 of the standard. This also explains the tpsot and tnsot values.

Thank you.

Kind regards,
Aous.

[bitsgalore]

I think you're right - looking at the code again:

jpylyzer/jpylyzer/boxvalidator.py

Line 1924 in b5d4d97

# Number of tiles

Here I see:

        numberOfTilesX = math.ceil((xsiz - xOsiz) / xTsiz)
        numberOfTilesY = math.ceil((ysiz - yOsiz) / yTsiz)

But of course this should be:

          numberOfTilesX = math.ceil((xsiz - xTOsiz) / xTsiz)
          numberOfTilesY = math.ceil((ysiz - yTOsiz) / yTsiz)

Will correct this in the stable release.

Thanks again for reporting this, and apologies for any confusion caused by my initial response!

[aous72]

Thank you Johan.

[bitsgalore]

Fixed: 1d47452
Also added file that triggered this to test corpus + added test: fc1a2d0