[HarfBuzz] Simplifying the hb-shape.cc example
Deron Kazmaier
deron at pagestream.org
Fri Jan 6 09:00:02 PST 2012
I've been where you are! I have a set of hb unicode functions I wrote up
because I did not want to depend on glib or icu either for my test
program. It is a handful of simple functions and a rather large file of
data, and might not work for you, but if you want I am happy to send
them to you (about 210k uncompressed). I also wrote a small (rather
ugly) program that parses the unicode data files to generate the tables
it uses.
Also, it is possible to get simple strings of right-to-left to work, if
you use the functions to determine the primary script and direction of
the run. No mixed rtl/ltr, but at least arabic/hebrew would work (It
does for me at least). Others could probably help you tweak it to be
more complete. (get_script and get_bidi are calls from my unicode
functions above).
run_script = HB_SCRIPT_UNKNOWN;
run_direction = HB_DIRECTION_UNKNOWN;
Roughly for each character:
if (run_script == HB_SCRIPT_UNKNOWN)
{
script = get_script(character);
if ((script != HB_SCRIPT_COMMON) && (script !=
HB_SCRIPT_INHERITED) && (script != HB_SCRIPT_UNKNOWN)) run_script = script;
}
if (run_direction == HB_DIRECTION_UNKNOWN)
{
direction = get_bidi(character);
if (direction == UNICODE_BIDI_STRONG_LTR) run_direction
= HB_DIRECTION_LTR;
else if (direction == UNICODE_BIDI_STRONG_RTL)
run_direction = HB_DIRECTION_RTL;
}
and then
if (run_direction != HB_DIRECTION_UNKNOWN) hb_buffer_set_direction
(buffer, run_direction);
if (run_script != HB_SCRIPT_UNKNOWN) hb_buffer_set_script (buffer,
run_script);
Deron
On 1/6/12 3:33 AM, Nicolas Rougier wrote:
>
>
> Thanks for the explanation but now I'm a bit lost...
> I thought that harfbuzz could work alone for shaping but from what you
> suggest (I need to try), I would need either glib or ICU, is that
> right ? But if I have to depend on ICU, I could as well use it
> directly for shaping and layout, no ?
>
>
> Nicolas
>
>
> On Jan 6, 2012, at 10:46 , Kenichi Ishibashi wrote:
>
>> I took a look at your code and noticed that you don't call
>> hb_buffer_set_unicode_funcs(). Since both HAVE_GLIB and HAVE_ICU are
>> not defined, _hb_unicode_funcs_nil is used as default. It won't work
>> for many cases(including ligatures). You need to implement your
>> custom hb_unicode_funcs_t or use existing
>> functions(hb_{glib,icu}_unicode_funcs).
>>
>> I think using ICU is a handy way if you can depend on ICU. hb-icu.cc
>> contains two 'gint' variables, but you can just remove 'g' from them
>> (Someone already sent a patch, IIRC).
>>
>> Thanks,
>>
>> On Fri, Jan 6, 2012 at 5:09 PM, Nicolas Rougier
>> <Nicolas.Rougier at inria.fr <mailto:Nicolas.Rougier at inria.fr>> wrote:
>>
>>
>> Thanks, this was certainly an error but it did not fix the problem.
>>
>> Nicolas
>>
>>
>> On Jan 6, 2012, at 9:00 , Kenichi Ishibashi wrote:
>>
>>> You may want to define "HAVE_OT" if you use your custom makefile
>>> (see configure.ac <http://configure.ac/>).
>>>
>>> On Fri, Jan 6, 2012 at 4:40 PM, Nicolas Rougier
>>> <Nicolas.Rougier at inria.fr <mailto:Nicolas.Rougier at inria.fr>> wrote:
>>>
>>>
>>> Thanks for the quick answer. I will try to investigate a bit
>>> further to see if there are any errors.
>>> If anyone on the list can test the code and report results,
>>> that might be helpful as well.
>>>
>>> I'm attaching the Makefile I used to compile (removed all
>>> glib stuff, maybe this where the error is):
>>>
>>> Nicolas
>>>
>>> Makefile
>>> ======
>>>
>>> PLATFORM = $(shell uname)
>>> CC = g++
>>> CFLAGS = -Wall `freetype-config --cflags`
>>> LIBS = `freetype-config --libs`
>>>
>>> SOURCES = main.cc \
>>> hb-blob.cc hb-buffer.cc hb-common.cc
>>> hb-fallback-shape.cc hb-font.cc \
>>> hb-ft.cc hb-ot-layout.cc hb-ot-map.cc
>>> hb-ot-shape-complex-arabic.cc \
>>> hb-ot-shape-complex-indic.cc hb-ot-shape-complex-misc.cc
>>> hb-shape.cc \
>>> hb-ot-shape-normalize.cc hb-ot-shape.cc hb-ot-tag.cc
>>> hb-unicode.cc
>>>
>>> HEADERS = \
>>> hb-buffer-private.hh hb-fallback-shape-private.hh
>>> hb-font-private.hh \
>>> hb-ot-shape-complex-arabic-table.hh
>>> hb-ot-shape-complex-indic-table.hh \
>>> hb-ot-shape-complex-private.hh hb-mutex-private.hh
>>> hb-object-private.hh \
>>> hb-open-file-private.hh hb-open-type-private.hh
>>> hb-ot-head-table.hh \
>>> hb-ot-hhea-table.hh hb-ot-hmtx-table.hh
>>> hb-ot-layout-common-private.hh \
>>> hb-ot-layout-gpos-table.hh hb-ot-layout-gsub-table.hh
>>> hb-unicode.h \
>>> hb-ot-layout-gsubgpos-private.hh
>>> hb-ot-layout-private.hh hb-shape.h \
>>> hb-ot-map-private.hh hb-ot-maxp-table.hh
>>> hb-ot-name-table.hh hb-blob.h \
>>> hb-private.hh hb-unicode-private.hh hb.h hb-buffer.h
>>> hb-version.h hb-ot.h \
>>> hb-common.h hb-font.h hb-ft.h hb-ot-layout.h
>>> hb-ot-shape.h hb-ot-tag.h
>>>
>>> OBJECTS := $(SOURCES:.cc=.o)
>>>
>>> all: main
>>>
>>> main: $(OBJECTS)
>>> $(CC) $(OBJECTS) $(LIBS) -o $@
>>>
>>> %.o : %.cc
>>> @echo "$@..."
>>> @$(CC) -c $(CFLAGS) $< -o $@
>>>
>>> clean:
>>> @-rm -f glut-main
>>> @-rm -f $(OBJECTS)
>>>
>>> distclean: clean
>>> @-rm -f *~
>>>
>>>
>>>
>>> On Jan 5, 2012, at 23:07 , Behdad Esfahbod wrote:
>>>
>>> > On 01/05/2012 04:44 PM, Nicolas Rougier wrote:
>>> >>
>>> >> Hi,
>>> >>
>>> >> I've been trying to simplify the hb-shape.c example from
>>> the util directory in order to remove any glib dependencies
>>> and come up with source below ( ~50 lines) but it does not
>>> seem to work (no ligature for fi):
>>> >>
>>> >> $ hb-shape DejaVuSans.ttf "fi"
>>> >> <fi=0+1290>
>>> >>
>>> >> $ main DejaVuSans.ttf "fi"
>>> >> <f=0+721|i=2+569>
>>> >>
>>> >>
>>> >> Did I do something wrong ?
>>> >
>>> > The code looks fine. I'm out of ideas. Can you recompile
>>> harfbuzz with glib
>>> > and test the same code?
>>> >
>>> > behdad
>>> >
>>> >>
>>> >> Nicolas
>>> >>
>>> >>
>>> >> main.cc
>>> >> ======
>>> >>
>>> >> #include <stdio.h>
>>> >> #include <ft2build.h>
>>> >> #include FT_FREETYPE_H
>>> >> #include "hb-ft.h"
>>> >>
>>> >> int main( int argc, char **argv )
>>> >> {
>>> >> const char *text = "fi";
>>> >> char glyph_name[30];
>>> >> size_t i;
>>> >>
>>> >> FT_Library library;
>>> >> FT_Face face;
>>> >> FT_Init_FreeType( &library );
>>> >> FT_New_Face( library, "./DejaVuSans.ttf", 0, &face );
>>> >> FT_Set_Char_Size( face, (int)(32*64), 0, 72, 72 );
>>> >>
>>> >> hb_font_t * font = hb_ft_font_create( face, 0 );
>>> >> hb_ft_font_set_funcs( font );
>>> >>
>>> >> hb_buffer_t * buffer = hb_buffer_create( );
>>> >> hb_buffer_set_direction( buffer, HB_DIRECTION_LTR );
>>> >> hb_buffer_add_utf8( buffer, text, strlen(text), 0,
>>> strlen(text) );
>>> >> hb_shape( font, buffer, NULL, 0 );
>>> >> unsigned int num_glyphs = hb_buffer_get_length (buffer);
>>> >> hb_glyph_info_t* glyphs = hb_buffer_get_glyph_infos(
>>> buffer, 0 );
>>> >> hb_glyph_position_t *pos =
>>> hb_buffer_get_glyph_positions( buffer, 0 );
>>> >> hb_glyph_info_t *info = hb_buffer_get_glyph_infos
>>> (buffer, NULL);
>>> >>
>>> >> for( i=0; i<num_glyphs; ++i )
>>> >> {
>>> >> if( i ) printf( "|" );
>>> >> else printf( "<" );
>>> >> if (!FT_Get_Glyph_Name( face, glyphs[i].codepoint,
>>> glyph_name, sizeof (glyph_name) ))
>>> >> printf( "%s", glyph_name );
>>> >> else
>>> >> printf( "%u", info->codepoint );
>>> >> printf( "=%u", info->cluster );
>>> >> if (pos->x_offset || pos->y_offset) {
>>> >> printf("@");
>>> >> if (pos->x_offset) printf ("%d", pos->x_offset);
>>> >> if (pos->y_offset) printf (",%d", pos->y_offset);
>>> >> }
>>> >> if (pos->x_advance || pos->y_advance) {
>>> >> printf ("+");
>>> >> if (pos->x_advance) printf ("%d", pos->x_advance);
>>> >> if (pos->y_advance) printf (",%d",
>>> pos->y_advance);
>>> >> }
>>> >> ++pos; ++info;
>>> >> }
>>> >> printf( ">\n" );
>>> >>
>>> >> hb_font_destroy(font);
>>> >> FT_Done_Face( face );
>>> >> FT_Done_FreeType( library );
>>> >> }
>>> >>
>>>
--
Happy Trails,
Deron Kazmaier deron at dogstarkennel.com
KD0AYO Western South Dakota
Dog Star Kennel http://www.dogstarkennel.com
--- Racing AKC Sepp-Star Seppala Siberian Husky Sled Dogs
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.freedesktop.org/archives/harfbuzz/attachments/20120106/82f90d51/attachment.html>
More information about the HarfBuzz
mailing list