| Age | Commit message (Collapse) | Author |
|---|
|
TIFF 5.0 uses a slightly laxer set of rules for TIFF decode.
Specifically, when we hit the maximum code, we are not required
to send a clear code immediately, but it can overrrun.
We don't bother storing codes > 12 bits, because they can never
be used. This avoids the need to extend the table.
|
|
|
|
|
|
If thirdparty/luratech is populated then this decoder will be preferred
over jbig2dec (even if both are present).
|
|
Normal glyphs and inital glyphs in ligatures have their start/stop (p
and q) set before determining whether to append to an existing span or
insert a space. For non-initial glyphs the start/stop were never set
which introduced uninitialized values into the span data structure.
Now, all glyphs have their start/stop set and then if it is a
non-initial glyph in a ligature the append and space detection is
ignored. This means that no values are uninitialized.
|
|
If verbose=yes is set then the full list of files will be printed.
|
|
|
|
Allow us to write a document to an fz_output as opposed to just a
filename.
We cannot write digital signatures in this method though. Will
ponder that in a later commit.
|
|
Slight tweak to the #ifdeffery required.
|
|
Include VS2012 fix.
|
|
Missing qualifier from prototype.
|
|
The recent js change requires a makefile tweak.
|
|
If a file cannot be saved incrementally, then don't accept that
as an option. In practise this means if someone asks to save
a file incrementally, and it was repaired, or it uses encryption
then throw an error.
Add a new function to ask if it's safe to save a file incrementally,
and use that in the appropriate places.
|
|
|
|
Uses the more modern/concise syntax that arrived in Xcode 4.4
|
|
|
|
|
|
We're a subclass of UIViewController, so must call one of UIViewController's
designated initialisers.
|
|
What is effectively our external API on iOS would be expected to use
NSString rather than char *.
|
|
We shouldn't output any debug in default release build.
|
|
This has knock on effects in the store.
fix
|
|
It seems that we can end up with a null item in MuPDFReaderView
onSingleTap. Add some simple checks to avoid this.
|
|
Calculations that involved non power of 2 bpps were going wrong.
|
|
When blanking pixmaps (or converting pixmap colors), watch out for
integer overflows.
|
|
|
|
|
|
Omitting the unlisted UTF-8 and UTF-32 CMaps saves ~1M.
Omitting the unlisted other CMaps saves ~200k.
Define CJK_CMAPS=0 to skip all CMaps.
Define EXTRA_CMAPS=1 to include the various other CMaps.
Define UTF8_CMAPS=1 and UTF32_CMAPS to include the UTF-8 and UTF-32 CMaps.
|
|
Compile with -Os, -ffunction-sections, -fdata-sections and link
with --gc-sections and -s (strip) to include only the functions and
data tables actually required.
|
|
I had changed some code to *3 when creating the routine from the
4 bpp variant, but this particular multiplication should still
have been *4, as it was looking up in an int table.
|
|
In C we can't have a label as the only thing in a block.
|
|
Need to specify that we are using alphas now.
|
|
clang only accepts the unified format - so use the unified syntax
everywhere, and add .syntax unified to tell the gnu assembler we're
using the unified syntax.
|
|
Lab colorspaces had been broken due to incorrect clipping of
color values introduced in an effort to fix Bug 696796.
|
|
Just a few places in the MSVC projects didn't have this enabled.
|
|
Update the test device so it can be used 'wrapping' another
device.
In particular, it can be used to wrap the display list device
so that we can evaluate 'color or not' while building the
display list rather than having to rerun the display list
afterwards.
Also, give improved control over whether we test every pixel
of images/shadings.
|
|
memory.h isn't part of the C standard, and Xcode was getting upset
when building mupdf as a module:
error: include of non-modular header inside framework module
'MuPDF.mupdf.memento'
[-Werror,-Wnon-modular-include-in-framework-module]
|
|
|
|
We store an hb_font in every font, and currently have fz_drop_font
know to call harfbuzz to destroy it. This causes harfbuzz to be
included even in builds that never use it.
We improve this situation by storing both an hb_font, and a
function pointer to destroy it within fz_font. This costs us
an extra pointer per fz_font, but solves the problem.
|
|
This enables linker to drop .o files that aren't required.
|
|
Don't try to load SVG images if SVG support is disabled.
|
|
|
|
|
|
|
|
Introduce FZ_PLOTTER defines to set which defines we required.
Add FZ_ENABLE define to set which document handlers are built
by default.
|
|
|
|
|
|
Use do {} while(--w) rather than while(w--) {} as this safes a
test each time around the loop.
|
|
|
|
Again, mainly so profiling works nicely.
|
|
For the common cases, avoid the loop.
|
