Update version of the patch, without modifications to PNGWriter, is attached to the bug https://bugs.freedesktop.org/show_bug.cgi?id=32340 It works similar to Splash: all rows are buffered in the memory, and then if needed - flipped. I have also implemented the x flip, though I have never seen PDFs needing it. But it was a nice exercise. Additionally, since I can flip only PNGs, if I see a JPEG which needs flipping (and PNG support is on) I convert the JPEGs to PNGs and flip them too.
The patch in bug applies on top of my patch from https://bugs.freedesktop.org/show_bug.cgi?id=47186 On 3/11/12, Ihar `Philips` Filipau <[email protected]> wrote: > Hi All! > > I had some spare time and source code at hand and tried to solve > another of my problems with the pdftohtml: vertically flipped images. > (Also a long standing bug: > https://bugs.freedesktop.org/show_bug.cgi?id=32340 ) > > An hour (w)hacking at the code produced that piece of frankenstein > code you might see below. > > Short summary: implement very crude yflip in PNGWriter by buffering > rows and transpose/write them before closing the PNG image; disable > JPEGs and write always PNGs. (The piece of code detecting flip was > lifted from pdf2xml project.) > > > Is there any official ideas on how to implement the image flip in the > pdftohtml? > > Are other flip possible too? rotations? To date I have seen only PDFs > with upside down images. > > Have Fun. > _______________________________________________ poppler mailing list [email protected] http://lists.freedesktop.org/mailman/listinfo/poppler
