Painting of some pattern was broken because the paint operation was
implemented as a fill to a rect containing the clip, but this rect was
not transformed as appropriate (using the cairo_to_ps matrix).
PDF simply fills the whole surface rect, so we implement the same
behavior in PS.
Fixes clip-group-shapes-circles, clip-stroke, linear-gradient-extend,
linear-gradient-one-stop, radial-gradient-one-stop.
There were some difference between how the FT and Win32
load_truetype_table font backend functions worked due to the
difference between FT_Load_Sfnt_Table() and GetFontData(). eg FT
returns an error if less than the requested number of bytes could be
read while Win32 returns success and sets the length to the number of
bytes read.
Make the default case be an assert to catch a silly programming error
and silence the compiler:
cairo-pdf-operators.c: In function ‘_word_wrap_stream_write’:
cairo-pdf-operators.c:300: warning: ‘count’ may be used uninitialized in
this function
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
If the font has been synthesized we can't use the native subsetters as
the outlines won't be the same. Instead force the use of the fallback
subsetters so the synthesized outlines will used to generate the font.
One of the problems identified in
https://bugzilla.mozilla.org/show_bug.cgi?id=454532 is that there are
some older printer drivers that do not work with ExtTextOut and the
ETO_GLYPH_INDEX option.
Fix this by where possible mapping glyph indices back to unicode and
calling ExtTextOut without ETO_GLYPH_INDEX. Glyphs that can not be
mapped back to unicode are printed with ETO_GLYPH_INDEX.
Similar to the freetype and toy font backends, use a hash table
to map logfont,hfont to font faces.
This fixes the multiple embedding of the same font in PDF.
https://bugs.freedesktop.org/show_bug.cgi?id=24849
The issue is that we store our data flipped based on whether we're
rendering to an FBO or to a window. By not flipping our gl_FragCoord
usage based on that (either with math or ARB_frag_coord_conventions),
this caused linear gradients to be flipped when rendering either to a
window or to an FBO. To avoid this, pass in appropriate texcoords.
And, if we're passing in texcoords, just do the projection to the
linear gradient factor on the CPU side per vertex instead of providing
a bunch of uniforms to do the math per fragment.
Fixes 18 testcases.
The use of ActualText in a marked content sequence is a PDF 1.5
feature.
A 'use_actual_text' flag linked to the PDF version has already been
implemented in pdf-operators but for some reason this flag had not
been used to control the use of ActualText.
Commit c0008242b0 made cairo use libm's lround instead of its own _cairo_lround
by default. However, since commit ce58f874 from 2006, _cairo_lround does
arithmetic rounding instead of away-from-zero rounding (before said commit, it
was using baker's rounding).
So to make the rounding of _cairo_lround be independent from
DISABLE_SOME_FLOATING_POINT, we have to use another function. Turns out that
_cairo_round already does the same thing that _cairo_lround does. Their only
difference is the return type.
Signed-off-by: Uli Schlachter <psychon@znc.in>
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
_cairo_xcb_picture_set_matrix() checked if the matrix that it should
set is an identity matrix. In this case this function simply didn't do
anything at all. The assumption here seems to be that a picture's
matrix is the identity matrix by default.
The problem here is that we might first set a picture's matrix to
something else and then later need an identity transform again. Fix
this by still setting the new matrix if it is an identify matrix. We
just skip some unneeded checks and optimizations in this case.
This fixes the "finer-grained-fallbacks" test in the test suite.
Signed-off-by: Uli Schlachter <psychon@znc.in>
Similar to ps_standard_encoding array, use a string pool and lookup
indices for the winansi glyph names to reduce .data size. As
ps_encoding and winansi share mostly the same names, the perl script
has been modified to merge the symbols into the one string pool and
generate separate lookup indices for ps_encoding and winansi.
The latin subsets feature requires these names. As
cairo-type1-subsets.c depends on FreeType, move these names out to a
separate file to allow compilation without FT.
In 9b9952ab4f
_cairo_memory_stream_destroy was changed to take an unsigned long
instead of unsigned int, and the two callsites in cairo-gl-shaders.c
weren't updated.
In 06e9caf861 the type of the variables
was changed, but the type used to compute the allocation size was not.
Fixes a crash in user-font-mask (test-fallback backend).
Use accessors instead of directly accessing path optimization flags.
Change the conditions for outputting tolerance (was 'when
path->is_rectilinear is FALSE', now is 'whenever the path includes a
curve').
Always output tolerance for strokes, because pen depends on tolerance
(for round caps/joins and for cusps).
FALSE and TRUE are defined in cairoint.h, but cairoint.h depends on
cairo-path-fixed-private.h, so just use 0/1 to avoid the depencency
loop.
Fixes a number of errors reported by 'make check'.
This fixes compilation of the xcb backend by porting commit e9c1fc31887c5bfbb's
changes from the image backend.
Signed-off-by: Uli Schlachter <psychon@znc.in>
The additional time spent in the computation of tight extents for the
curve_to operation doesn't seem to be significant, but it makes the
extents computations faster and the approximations more accurate.
current_point and last_move_to were previously left in their old
position (which could lead to incorrect flag computation if other
operations were added to the path) and flags were not updated.
Only fill_maybe_region can change its value because the transformation
preserves vertical and horizontal lines, but can move the points and
make them integer if they were not or non-integer if they were.
Recomputing it is just as easy as checking if all the points are
integer and the path is fill_is_rectilinear.
