This is a follow-up patch.
* src/truetype/ttinterp.c, src/truetype/ttinterp.h
[TT_CONFIG_OPTION_STATIC_INTERPRETER,
TT_CONFIG_OPTION_STATIC_RASTER]: Remove macros and related code.
`API functions' are functions tagged with `FT_EXPORT_DEF'.
Besides trivial fixes, the following changes are included, too.
* src/base/ftbdf.c (FT_Get_BDF_Charset_ID, FT_Get_BDF_Property): Set
error code if no service is available.
* src/base/ftinit.c (FT_Done_FreeType): Change return value for
invalid `library' parameter to `Invalid_Library_Handle'.
* src/base/ftobjs.c (FT_New_Size): Change return value for invalid
`asize' parameter to `Invalid_Argument'.
* src/base/ftoutln.c (FT_Outline_Copy): Change return value for
invalid `source' and `target' parameters to `Invalid_Outline'.
(FT_Outline_Done_Internal): Change return value for invalid
`outline' parameter to `Invalid_Outline'.
Before this patch, it was impossible to ever call DELTAP[123] in
subpixel hinting mode as described in the ClearType whitepaper; it
only worked if in `compatibility mode'. However, compatibility mode
essentially disables SHPIX, completely ruining hinting of
ttfautohint output, for example.
We now follow the whitepaper more closely so that DELTAP[123]
instructions for touched points in the non-subpixel direction are
executed.
Zero distance does not have to be treated specially if you follow
specifications and check the sign as the very last step of rounding.
* src/truetype/ttinterp.c (Round_None, Round_To_Grid,
Round_Down_To_Grid, Round_Up_To_Grid, Round_To_Double_Grid): Use
macros when available, do not check for non-zero distance.
(Round_To_Half_Grid, Round_Super, Round_Super_45): Ditto, return phase
if sign changed.
The legal range for delta shift is zero through six. Negative values
are illegal according to
https://developer.apple.com/fonts/TrueType-Reference-Manual/RM04/Chap4.html#delta%20shift
* src/truetype/ttobjs.h (delta_shift, delta_base): Make unsigned.
* src/truetype/ttinterp.h (DO_SDS): Throw an error if delta_shift
out of range.
(Ins_DELTAP, Ins_DELTAC): Optimize for legal delta_shift.
* src/truetype/ttinterp.h (TT_ExecContextRec): New field
`cur_ppem_func' with a function pointer.
* src/truetype/ttinterp.c (TT_RunIns): Initialize `cur_ppem_func'
depending on the pixel geometry to either...
(Current_Ppem_Stretched): ... this for stretched pixels.
(Current_Ppem): ... or this for square pixels.
(DO_MPPEM, DO_MPS, Ins_DELTAP, Ins_DELTAC): Use `cur_ppem_func'.
In case of an error in the `prep' table, no longer try to execute it
again and again. This makes FreeType handle endless loops in buggy
fonts much faster.
* src/truetype/ttobjs.h (TT_SizeRec): The fields `bytecode_ready'
and `cvt_ready' are now negative if not initialized yet, otherwise
they indicate the error code of the last run.
* src/truetype/ttobjs.c (tt_size_run_fpgm, tt_size_run_prep,
tt_size_done_bytecode, tt_size_init_bytecode,
tt_size_ready_bytecode, tt_size_init, tt_size_done, tt_size_reset):
Updated.
* src/truetype/ttgload.c (tt_loader_init): Updated.
* src/truetype/ttinterp.c (TT_RunIns): Force reexecution of `fpgm'
and `prep' only if we are in the `glyf' table.
This patch unifies the subpixel and non-subpixel cases.
* src/truetype/ttinterp.h (TT_ExecContextRec): Remove
`grayscale_hinting'; all code should refer to `grayscale' instead.
Remove unused `native_hinting' member.
Rename `subpixel_hinting' member to `subpixel.
* src/truetype/ttgload.c (TT_LOADER_SET_PP): Updated.
(tt_loader_init): Updated.
* src/truetype/ttinterp.c (Ins_GETINFO): Simplify.
Updated.
Two benefits: The allocated FDEF (and IDEF) array gets slightly
smaller, and the `ttdebug' demo program has access to function
numbers without additional costs.
Fortunately, no changes to FontForge are necessary – this is the
only external TrueType debugger I know of, but others may exist and
should check the code accordingly.
* src/truetype/ttinterp.h (TT_CallRec): Replace `Cur_Restart' and
`Cur_End' with a pointer to the corresponding `TT_DefRecord'
structure.
* src/truetype/ttinterp.c (DO_JROT, DO_JMPR, DO_JROF, Ins_ENDF,
Ins_CALL, Ins_LOOPCALL, Ins_UNKNOWN, TT_RunIns <Invalid_Opcode>):
Updated.
When using `ADD' with an immediate operand, the instruction is
actually `ADD Rd, Rn, #<imm12>', that is, the maximum of the
immediate operand cannot exceed 4095. It will fail to compile with
LLVM.
However, in GCC, due to some legacy compatibility considerations,
`ADD.W' will be automatically emitted when the immediate operand is
larger than 4095.
* builds/unix/ftconfig.in, include/freetype/config/ftconfig.h
(FT_MulFix_arm) [__GNUC__]: Support clang compiler.
* src/truetype/ttinterp.c (TT_MulFix14_arm) [__GNUC__]: Ditto.
* src/truetype/ttinterp.c (TT_MulFix14_long_long,
TT_DotFix14_long_long): `#pragma gcc diagnostic {push,pop}' has been
introduced with gcc version 4.6.
This patch provides slightly optimized versions for ARM, x86, and
x86_64 CPUs if built with GCC.
Also remove some dead code.
* src/truetype/ttinterp.c (TT_MulFix14_arm, TT_MulFix14_long_long,
TT_DotFix14_long_long): New functions.
Some small enhancements have allowed the removal of many macros and
the simplification of existing rules in `ttsubpix.c'.
* src/truetype/ttsubpix.h (SPH_TWEAK_ALLOW_X_DMOVEX,
SPH_TWEAK_ALLOW_X_MOVE_ZP2,
SPH_TWEAK_DELTAP_SKIP_EXAGGERATED_VALUES,
SPH_TWEAK_SKIP_INLINE_DELTAS, SPH_TWEAK_MIRP_CVT_ZERO): Removed.
(SPH_TWEAK_SKIP_NONPIXEL_Y_MOVES_DELTAP): New rule macro.
* src/truetype/ttsubpix.c: Updated affected rules.
* src/truetype/ttinterp.c (Direct_Move_X): Updated.
(INS_FDEF): Add additional function detection.
(INS_ENDF): Set runtime flag.
(Ins_CALL): Skip the call under certain conditions.
Remove bad code.
(Ins_LOOPCALL): Skip the call under certain conditions.
Remove bad code.
(Move_Zp2_Point): Updated.
(Ins_SHPIX): Updated.
Skip the move under some situations.
(Ins_MIAP): Improve conditions.
(Ins_MIRP): Updated.
(Ins_DELTAP): Skip move under certain conditions.
Simplify conditions.
(TT_RunIns): Updated.
Add code to handle new function detection.
Trace messages.
This makes the option TT_CONFIG_OPTION_SUBPIXEL_HINTING controllable
at runtime.
* src/truetype/ttdriver.c: Include FT_TRUETYPE_DRIVER_H.
(tt_property_set, tt_property_get): Fill templates.
* src/truetype/ttobjs.h (TT_DriverRec): Add `interpreter_version'
member.
Remove unused `extension_component' member.
* src/truetype/ttgload.c: Include FT_TRUETYPE_DRIVER_H.
(tt_get_metrics, TT_Hint_Glyph, TT_Process_Simple_Glyph,
compute_glyph_metrics, tt_loader_init): Use `interpreter_version'.
* src/truetype/ttinterp.c: Include FT_TRUETYPE_DRIVER_H.
(SUBPIXEL_HINTING): New macro to check `interpreter_version' flag.
Update all affected functions to use it.
Use TT_INTERPRETER_VERSION_XXX where appropriate.
* src/truetype/ttobjs.c: Include FT_TRUETYPE_DRIVER_H.
(tt_driver_init): Initialize `interpreter_version'.
* src/truetype/ttsubpix.c: Include FT_TRUETYPE_DRIVER_H.
Use TT_INTERPRETER_VERSION_XXX where appropriate.
* src/base/ftcalc.c (FT_DivFix): Use unsigned values for
computations which use the left shift operator and convert to signed
as the last step.
* src/base/fttrigon.c (ft_trig_prenorm, FT_Vector_Rotate,
FT_Vector_Length, FT_Vector_Polarize): Ditto.
* src/cff/cffgload.c (cff_decoder_parse_charstrings): Simplify.
* src/cff/cffload.c (cff_subfont_load): Fix constant.
* src/cff/cffparse.c (cff_parse_integer, cff_parse_real, do_fixed,
cff_parse_fixed_dynamic): Use unsigned values for computations which
use the left shift operator and convert to signed as the last step.
* src/cid/cidload.c (cid_get_offset): Ditto.
* src/psaux/psconv.c (PS_Conv_ToFixed): Ditto.
* src/psaux/t1decode.c (t1_decoder_parse_charstrings): Ditto.
* src/truetype/ttinterp.c (TT_MulFix14, TT_DotFix14): Ditto.
This is essentially a mechanical conversion, adding inclusion of
`FT_INTERNAL_DEBUG_H' where necessary, and providing the macros for
stand-alone compiling modes of the rasterizer modules.
To convert the remaining occurrences of FT_Err_XXX and friends it is
necessary to rewrite the code. Note, however, that it doesn't harm
if some cases are not handled since FT_THROW is a no-op.
* include/freetype/fttrigon.h (FT_Hypot): Declare it.
* src/base/fttrigon.c (FT_Hypot): Define it.
* src/truetype/ttgload.c (TT_Process_Composite_Component): Use it
instead of explicit expressions.
* src/truetype/ttinterp.c (Current_Ratio, Normalize): Use it instead
of TT_VecLen.
(TT_VecLen): Removed.
Unit vector components are stored as 2.14 fixed-point numbers. In
order to calculate all 14 bits accurately, a short vector to be
normalized has to be upscaled to at least 14 bits before its length
is calculated. This has been safe since accurate CORDIC algorithms
were adopted.
* src/truetype/ttinterp.c (Normalize): Scale short vectors by 0x4000.
Back in the days, vector length calculations were not very accurate
and the vector normalization function, Normalize, had to meticulously
correct the errors for long vectors [commit b7ef2b0968]. It was no
longer necessary after accurate CORDIC algorithms were adopted, but
the code remained. It is time to kill it.
* src/truetype/ttinterp.c (Normalize): Remove error compensation.
(TT_VecLen): Remove any mention of old less accurate implementation.
The dot product between freeVector and projVector or cosine of
the angle between these FT_F2Dot14 unit vectors used to be scaled up
by 4 and routinely occupied 32 bits in an FT_Long field F_dot_P.
This patch scales the value down by 2^14 instead, which simplifies
its use throughout the bytecode interpreter.
This does not lead to the loss of precision because the lower bits
are unreliable anyway. Consider two unit vectors (1,0) and (.6,.8)
for which the true value of F_dot_P is .6 * 0x40000000 = 0x26666666.
These vectors are stored as (0x4000,0) and (0x2666,0x3333) after
rounding and F_dot_P is assigned 0x26660000. The lower bits were
already lost while rounding the unit vector components.
Besides code simplification, this change can lead to better
performance when FT_MulDiv with the scaled-down F_dot_P is less
likely to use the costly 64-bit path. We are not changing the type
of F_dot_P to FT_F2Dot14 at this point.
* src/truetype/ttinterp.c (Compute_Funcs): Scale F_dot_P down by 14
bits and modify its use accordingly.
(Direct_Move, Direct_Move_Orig, Compute_Point_Displacement): Modify
the use of F_dot_P field.
* src/truetype/ttobjs.c (tt_size_run_fpgm): Change arbitrary
assignment of F_dot_P to its theoretical maximum in case we decide
to scale back its type later.
Werner Lemberg