MS Office writes the EmrFormat.Version field as 0 in the
EMR_COMMENT_MULTIFORMATS record carrying an embedded PDF. Per
[MS-EMF] 2.3.3.4.2 EmrFormat, Version MUST be 0x00000001 only when
Signature == EPS_SIGNATURE; for any other signature it MUST be
ignored. The reader rejected anything but 1, so the embedded PDF
was dropped on the floor.
Real-world files also use an empty OutputRect (0,0,0,0); per
[MS-EMF] 2.3.3.4.3 it is in logical coordinates, so fall back to
the EMF's logical bounds (mrclBounds) when it is degenerate, so
ResolveBitmapActions has somewhere to draw the PDF bitmap.
The new fixture is the existing pdf-in-emf.pptx EMF with the two
fields patched to reproduce the regression in 776 bytes.
Signed-off-by: Andras Timar <andras.timar@collabora.com>
Change-Id: I432f75a1cac11d84448db07c1a1815cd0be60744
Reviewed-on: https://gerrit.collaboraoffice.com/c/online/+/1904
Tested-by: Jenkins CPCI <releng@collaboraoffice.com>
Reviewed-by: Miklos Vajna <vmiklos@collabora.com>
(cherry picked from commit a5d7b2935718bf7f1f6cc0352c8d8681e6713f4b)
Cherry-pick note: the test fails on macOS in CI
Reviewed-on: https://gerrit.libreoffice.org/c/core/+/205084
Tested-by: Jenkins
Reviewed-by: Xisco Fauli <xiscofauli@libreoffice.org>
Andras Timar
c043f65f66