Nico Weber
Nico Weber
In particular, implements support for AES in LibPDF. Also fixes an issue where the linearization detection logic would crash on inputs where the first object is an encrypted compressed stream...
This works: Build/lagom/bin/pdf --render out.png ~/Downloads/0000/0000342.pdf --page 5 This doesn't: Build/lagom/bin/pdf --render out.png ~/Downloads/0000/0000342.pdf --page 5 --dump-contents It does work if I move the `--dump-contents` in front of the name...
I think the jpeg decoder might have channels off by a small amount, due to various things suggesting it: 1. If I make `green.png` filled with (0, 255, 0) and...
This is useful if you want an array with an explicit type but still want its size to be inferred. --- β¦and random minor stuff along the way, such as...
...and do string expansion at the call site. CID-keyed fonts treat the charset as CIDs instead of as SIDs, so having access to the SIDs in numberic form will be...
This reverts commit 52afa936c48f62df8a9b7fec0e78fa75d55407ec. No longer necessary after #23122 -- turns out things work better when you do them right. No behavior change.
1. Get 0000.zip from here: https://downloads.digitalcorpora.org/corpora/files/CC-MAIN-2021-31-PDF-UNTRUNCATED/zipfiles/0000-0999/ 2. Open 0000612.pdf 3. Go to page 27 4. Toggle Debug->Show Clipping Paths Expected: Clipping paths draw Actual: App hangs, things get super slow...
This is one of the two top dict entries we need for CID-keyed fonts. We don't send any CID-keyed font data into the CFF parser yet, so no behavior change.
No behavior change, except that we now dbgln() if we see a PrivDictOperator we don't know about. (I haven't seen this in practice, but I found this useful while debugging...