Console View
|
|
|||
Sune Vuorela
sune @vuorela.dk |
|
||
Build Qt frontends with QT_NO_CAST_FROM.. | |||
Sune Vuorela
sune @vuorela.dk |
|
||
Test creation and verification of GnuPG pgp signatures | |||
Juraj Šarinay
juraj @sarinay.com |
|
||
Do not add a second copy of signer's certificate to CMS SignedData. Signer's certificate is included in the chain we add as a result of NSS_CMSSignerInfo_IncludeCerts(). There is no need to call NSS_CMSSignedData_AddCertificate(). |
|||
Albert Astals Cid
aacid @kde.org |
|
||
Revert "CI: Debian has gnupg 2.4 now" This reverts commit 2228a610c149f696afa8bfdd9e18a81cecfba0c4. It seems they are doing too much distro patching to the level it breaks our tests |
|||
Albert Astals Cid
aacid @kde.org |
|
||
DCTStream::reset: Fix crash | |||
Albert Astals Cid
aacid @kde.org |
|
||
Update (C) | |||
Juraj Šarinay
juraj @sarinay.com |
|
||
Clean up generation and update of signature appearance streams. Remove unnecessary calls to FormWidgetSignature::updateWidgetAppearance() and/or AnnotWidget::updateAppearanceStream(). The single call from FormWidgetSignature::signDocument() is sufficient. Lift updatedAppearanceStream from AnnotWidget/AnnotStamp to Annot to prevent AnnotWidget::updateAppearanceStream() from copying the content stream added earlier within AnnotWidget::setNewAppearance() called by AnnotWidget::generateFieldAppearance(). |
|||
Albert Astals Cid
aacid @kde.org |
|
||
Increase version so people that track master can add ifdefs on version | |||
Albert Astals Cid
aacid @kde.org |
|
||
poppler 25.05.0 | |||
Albert Astals Cid
aacid @kde.org |
|
||
Update (C) | |||
Albert Astals Cid
aacid @kde.org |
|
||
Update (C) | |||
Marco Trevisan
mail @3v1n0.net |
|
||
glib/poppler-document: Copy the signing data during verification Signature verification data is passed to a thread and not stolen from the caller (as it's transfer-none and const), so we're supposed to copy it when passed to the thread and free'd at thread return time. |
|||
Albert Astals Cid
aacid @kde.org |
|
||
Update (C) | |||
Marco Trevisan
mail @3v1n0.net |
|
||
glib/poppler-document: Do not crash if required signature data parameters are unset and support text-less signatures It's perfectly legal to sign without a visible text signature as `pdfsig` does, but the GLib bindings didn't allow it or they crashed when some optional parameter were provided. So: * Do not crash if signature text left data is unset * Do not require a signature text at all * Throw an error if the partial name is unset (instead of crashing) * Throw an error if the destination filename is unset (instead of crashing) |
|||
Sune Vuorela
sune @vuorela.dk |
|
||
Object takes ownership of stream. Document it in code | |||
Nelson Benítez León
nbenitezl @gmail.com |
|
||
Fix drawing checkbox fields without a DA entry Spec says DA entry is required "to all fields containing variable text" but it seems checkbox fields are de-facto not considered as such. Handle that in AnnotAppearanceBuilder::drawText() Fixes #1055 (when used together with the fix in #642) |
|||
Nelson Benítez León
nbenitezl @gmail.com |
|
||
Handle when /AP /N is a Ref to Dict of named streams When the Normal Appearance Stream (/AP -> /N) is a Ref, we previously assumed it could only be a Ref to Stream, but it can also be a Ref to Dict of named streams, so we need to add that logic in AnnotAppearance::getAppearanceStream() Fixes checkbox field in below referenced issue. Fixes #1558 |
|||
Nelson Benítez León
nbenitezl @gmail.com |
|
||
Fix setting the AnnotWidget of standalone fields When loading standalone Formfields, we were wrongly creating new AnnotWidgets to fill its FormWidget member, whereas we just needed to set it to the own AnnotWidget that contains the standalone field. That meant that when clicking on standalone checkbox fields we were really clicking on AnnotWidgets that were not the ones used for drawing the page. With this fix those checkboxes now work ok. Fixes #642 Improves #1055 |
|||
Nelson Benítez León
nbenitezl @gmail.com |
|
||
CairoFontEngine: remove unused code Spotted by Anton Thomasson |
|||
Nelson Benítez León
nbenitezl @gmail.com |
|
||
CairoFontEngine: invalidate broken embedded fonts When unable to create a font face from data read from gfxFont->readEmbFontFile() it means such data is broken (errors are also printed to console) so mark such gfxFont as containing INVALID embedded data, otherwise it will end up being use to draw text and failed at that, like shown in issue #1453 SplashOutputDev (which does not exhibit this bug) does a similar thing in SplashOutputDev::doUpdateFont() Fixes #1453 |
|||
Albert Astals Cid
aacid @kde.org |
|
||
Update (C) | |||
Sune Vuorela
sune @vuorela.dk |
|
||
Use named constructors for empty objects null, eof and error objects are the only ones where we actually need to create them completely empty. Do them with named constructors and move the empty-object-with-type to private section to avoid creating them by accident |
|||
Albert Astals Cid
aacid @kde.org |
|
||
CI: Debian has gnupg 2.4 now | |||
Albert Astals Cid
aacid @kde.org |
|
||
CI: Fedora 42 | |||
Albert Astals Cid
aacid @kde.org |
|
||
Update (C) | |||
Stefan Brüns
stefan.bruens @rwth-aachen.de |
|
||
[SplashXPath] Reuse auxiliary curve point arrays Allocate the data on first use, reuse it for all curved segments of a path, and discard it afterwards. CCBUG: #1555 |
|||
Stefan Brüns
stefan.bruens @rwth-aachen.de |
|
||
[SplashXPath] Reserve space for segments | |||
Stefan Brüns
stefan.bruens @rwth-aachen.de |
|
||
[SplashPath] Reserve segments before flattening/stroking For longer paths, reallocation can take up a significant amount of time. A fairly good estimate can be made from the input path length. For flattening, twice the input length is used to accomodate for added curve segments. For stroking, each segment create two new segments, (at least) two segments for the path caps, and some segments for each rounded line joins. CCBUG: #1555 |
|||
Stefan Brüns
stefan.bruens @rwth-aachen.de |
|
||
[SplashXPathScanner] Some more variable scope cleanup | |||
Stefan Brüns
stefan.bruens @rwth-aachen.de |
|
||
[SplashXPathScanner] Remove unused return value in addIntersection CCBUG: #1555 |
|||
Stefan Brüns
stefan.bruens @rwth-aachen.de |
|
||
[Splash] Reduce required work for path pruning For long path, quite some amount of work is spent if the bounding box of all points intersects with the clipping rectangle, and if there is some intersection the path is processed further. For this check, it is often sufficient to only check one, or two points. Only if both points are outside (which is often false) the remaining points should be checked. CCBUG: #1555 |
|||
Albert Astals Cid
aacid @kde.org |
|
||
CI: Also build with non default options | |||
Albert Astals Cid
aacid @kde.org |
|
||
Fix signature of FlateStream::getPSFilter | |||
Albert Astals Cid
aacid @kde.org |
|
||
Fix re-fetching after xref reconstruction Fixes #1584 |
|||
Albert Astals Cid
aacid @kde.org |
|
||
Increase version so people that track master can add ifdefs on version | |||
Albert Astals Cid
aacid @kde.org |
|
||
poppler 25.04.0 | |||
Juraj Šarinay
juraj @sarinay.com |
|
||
Properly verify adbe.pkcs7.sha1 signatures. For signatures with non-empty encapsulated content (typically adbe.pkcs7.sha1), we only compared hash values and never actually checked SignatureValue within SignerInfo. The bug introduced by c7c0207b1cfe49a4353d6cda93dbebef4508138f made trivial signature forgeries possible. Fix this by calling NSS_CMSSignerInfo_Verify() after the hash values compare equal. |
|||
Albert Astals Cid
aacid @kde.org |
|
||
Update (C) | |||
Albert Astals Cid
aacid @kde.org |
|
||
Move isOk check to inside JBIG2Bitmap::combine | |||
Albert Astals Cid
aacid @kde.org |
|
||
Fix comment | |||