This code originally just searched for `eexec`. This was later modified
to check that the `eexec` found is valid (not in a string or comment).
This was done by searching for `eexec` as before and then, for each
`eexec` found, searching from the beginning using the correct parsing to
see if the `eexec` was still found. If the private dictionary is large
and contains many copies of `eexec` which are not valid, the initial
part of the private dictionary is scanned once for each, potentially
leading to n^2 parsing time.
Instead of finding an initial `eexec` and then re-parsing to discover if
it is valid, drop the initial search for `eexec` and just parse to find
a valid `eexec`. This is strictly faster since the validation must
happen anyway and avoids restarting from the beginning each time an
`eexec` is found in the data.
* src/type1/t1parse.c (T1_Get_Private_Dict): avoid n^2 parsing
Bug: https://bugs.chromium.org/p/chromium/issues/detail?id=1328883
We no longer have to take care of the 8.3 file name limit; this
allows us (a) to introduce longer, meaningful file names, and (b) to
avoid macro names in `#include' lines altogether since some
compilers (most notably Visual C++) doesn't support this properly.
*/*: Replace
#include FOO_H
with
#include <freetype/foo.h>
or something similar. Also update the documentation.
This monster commit was created by applying Nikhil's scripts
`docconverter.py' and `markify.py' to all C header and source files,
followed up by minor manual clean-up.
No change in functionality, of course.
I used commit f7419907bc6044b9b7057f9789866426c804ba82 from
https://github.com/nikramakrishnan/freetype-docs.git.
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.
* src/type1/t1parse.c (T1_New_Parser), src/type42/t42parse.c
(t42_parser_init): modifying functions to check the font header before
allocating anything on the heap.
* internal/freetype/ftmemory.h: introducing the new macros FT_ARRAY_MAX
and FT_ARRAY_CHECK
* src/pcf/pcfread.c, src/pcf/pcfutil.c: minor fixes and simplifications.
try to protect the PCF driver from doing stupid things with broken fonts.
accent offset.
Update code similarly to the seac support for Type 1 fonts.
(cff_decoder_parse_charstrings) <cff_op_endchar>: Fix magnitude
of accent offset.
Don't hint glyphs twice if seac is emulated.
<cff_op_flex>: Assign correct point tags.
* docs/CHANGES: Updated.
* src/type1/t1parse.c (T1_Get_Private_Dict): Use FT_MEM_MOVE, not
FT_MEM_COPY, for copying the private dict.
* src/type1/t1load.c (parse_subrs): Assign number of subrs only
in first run.
(parse_charstrings): Parse /CharStrings in second run without
assigning values.
(parse_dict): Skip all /CharStrings arrays but the first. We need
this for non-standard fonts like `Optima' which have different
outlines depending on the resolution. Note that there is no
guarantee that we get fitting /Subrs and /CharStrings arrays; this
can only be done by a real PS interpreter.
correctly. This doesn't slow down the loading of PS fonts
significantly since charstrings aren't affected.
* include/freetype/config/ftstdlib.h (ft_xdigit): Renamed to...
(ft_isxdigit): This. Updated all callers.
(ft_isdigit): New alias to `isdigit'.
* include/freetype/internal/psaux.h (PS_Parser_FuncsRec): Renamed
`skip_alpha' to `skip_PS_token'.
Add parameter to `to_bytes' and change some argument types.
* src/psaux/psauxmod.c (ps_parser_funcs): Updated.
* src/psaux/psobjs.c (ft_char_table): New array to map character
codes (ASCII and EBCDIC) of digits to numbers.
(OP): New auxiliary macro holding either `>=' or `<' depending on
the character encoding.
(skip_comment): New function.
(skip_spaces): Use it.
(skip_alpha): Removed.
(skip_literal_string, skip_string): New functions.
(ps_parser_skip_PS_token): New function. This is a better
replacement of...
(ps_parser_skip_alpha): Removed.
(ps_parser_to_token, ps+parser_to_token_array): Updated.
(T1Radix): Rewritten, using `ft_char_table'.
(t1_toint): Renamed to...
(ps_toint): This. Update all callers.
Use `ft_char_table'.
(ps_tobytes): Add parameter to handle delimiters and change some
argument types.
Use `ft_char_table'.
(t1_tofixed): Renamed to...
(ps_tofixed): This. Update all callers.
Use `ft_char_table'.
(t1_tocoordarray): Renamed and updated to...
(ps_tocoordarray): This. Update all callers.
(t1_tofixedarray): Renamed and updated to...
(ps_tofixedarray): This. Update all callers.
(t1_tobool): Renamed to...
(ps_tobool): This. Update all callers.
(ps_parser_load_field): Updated.
(ps_parser_load_field_table): Use `T1_MAX_TABLE_ELEMENTS'
everywhere.
(ps_parser_to_int, ps_parser_to_fixed, ps_parser_to_coord_array,
ps_parser_to_fixed_array): Skip spaces. Updated.
(ps_parser_to_bytes): Add parameter to handle delimiters and change
some argument types. Updated.
* src/psaux/psobjs.h: Updated.
* src/cid/cidload.c (cid_parse_dict): Updated.
* src/cid/cidparse.c (cid_parser_new): Check whether the `StartData'
token was really found.
* src/cid/cidparse.h (cid_parser_skip_alpha): Updated and renamed
to...
(cid_parser_skip_PS_token): This.
* src/type1/t1parse.h (T1_ParserRec): Use `FT_Bool' for boolean
fields.
(T1_Skip_Alpha): Replaced with...
(T1_Skip_PS_Token): This new macro.
* src/type1/t1parse.c (hexa_value): Removed.
(T1_Get_Private_Dict): Use `ft_isxdigit' and
`psaux->ps_parser_funcs_to_bytes' for handling ASCII hexadecimal
encoding.
After decrypting, replace the four random bytes at the beginning
with whitespace.
* src/type1/t1load.c (t1_allocate_blend): Use proper error values.
(parser_blend_design_positions, parse_blend_design_map,
parse_weight_vector): Updated.
(is_space): Handle `\f' also.
(is_name_char): Removed.
(read_binary_data): Updated.
(parse_encoding): Use `ft_isdigit'.
Updated.
(parse_subrs): Updated.
(TABLE_EXTEND): New macro.
(parse_charstrings): Updated.
Provide a workaround for buggy fonts which have more entries in the
/CharStrings dictionary then expected; the function now adds some
slots and skips entries which still exceed the new limit.
(parse_dict): Updated.
Terminate on the token `closefile'.
* src/type42/t42parse.c (T1_Skip_Alpha): Replaced with...
(T1_Skip_PS_Token): This new macro. Updated all callers.
(t42_parse_encoding): Use `ft_isdigit'.
* src/base/ftmm.c (ft_face_get_mm_service): Return FT_Err_OK if
success.
* src/type1/t1parse.h (T1_ParserRec): Change type of `base_len'
and `private_len' to FT_Long.
* src/type1/t1parse.c (T1_Get_Private_Dict): Remove cast for
`private_len'.
* src/type1/t1load.c: Use FT_Int cast for most calls of T1_ToInt.
Use FT_PtrDist where appropriate.
(parse_encoding): Use FT_Long for `count' and `n'.
(read_binary_data): Use FT_Long* for second parameter.
* src/type1/t1afm.c (afm_atoindex): Use FT_PtrDist.
* src/cache/ftcsbits.c (ftc_sbit_node_load): Remove unused label.
* src/pshinter/pshalgo3.c (psh3_hint_align): Remove unused variable.
warn about the GNU Make requirement at compile time..
* include/freetype/config/ftstdlib.h,
include/freetype/config/ftconfig.h,
include/freetype/config/ftheader.h,
include/freetype/internal/ftmemory.h,
include/freetype/internal/ftobjs.h,
src/autohint/ahoptim.c,
src/base/ftdbgmem.c, src/base/ftdebug.c,
src/base/ftmac.c, src/base/ftobjs.c,
src/base/ftsystem.c,
src/cache/ftcimage.c, src/cache/ftcsbits.c,
src/cff/cffdriver.c, src/cff/cffload.c, src/cff/cffobjs.c,
src/cid/cidload.c, src/cid/cidparse.c, src/cid/cidriver.c,
src/pcf/pcfdriver.c, src/pcf/pcfread.c,
src/psaux/t1cmap.c, src/psaux/t1decode.c,
src/pshinter/pshalgo1.c, src/pshinter/pshalgo2.c,
src/pshinter/pshrec.c,
src/psnames/psmodule.c,
src/raster/ftraster.c,
src/sfnt/sfdriver.c, src/sfnt/ttload.c, src/sfnt/ttpost.c,
src/smooth/ftgrays.c,
src/type1/t1afm.c, src/type1/t1driver.c, src/type1/t1gload.c,
src/type1/t1load.c, src/type1/t1objs.c, src/type1/t1parse.c:
added the new configuration file "ftstdlib.h" used to define
aliases for all ISO C library functions used by the engine
(e.g. strlen, qsort, setjmp, etc...)
this eases the porting of FreeType 2 to exotic environments like
XFree86 modules/extensions..
also removed many #include <string.h>, #include <stdlib.h>, etc...
from the engine's sources where they're not needed..
Werner Lemberg