Re: [ft] Ideographic Variation Sequence (IVS)

[mpsuzuki]

Hi,

OpenType specification had defined cmap table format 14 that
was designed by Adobe to use Unicode VS, and current FreeType2
supports the format (the driver was written by George Williams).
It was included in 2007-Oct, and freetype-2.3.6 had included it.

There are APIs to access VS data in cmap format 14:
        FT_Face_GetCharsOfVariant()
        FT_Face_GetCharVariantIndex()
        FT_Face_GetCharVariantIsDefault()
        FT_Face_GetVariantSelectors()
        FT_Face_GetVariantsOfChar()
For detail, please check API reference:
http://freetype.sourceforge.net/freetype2/docs/reference/ft2-index.html

However, if you expect that FT2 eats an Unicode string including
UVS and generates an image with specified glyph shape, it is
misunderstanding. Decomposition of Unicode string into each
base character and UVS should be done before calling FT2
rasterizer. Also UVS availability should be checked explicitly.

Regards,
mpsuzuki


On Thu, 23 Jul 2009 11:17:42 +0200
Jeroen Hoek <address@hidden> wrote:
>In 2006 the Unicode Consortium introduced the concept of Ideographic Variation
>Sequences (IVS) and the Ideographic Variation Database (IVD). This system
>allows font designers to include variant glyphs for characters encoded at a
>single codepoint. This is needed for the languages used in CJKV cultures
>(China, Taiwan, the Koreas, Japan, and historically Vietnam), where regional
>names and historical usage sometimes necessitates a less common variant glyph.
>An example is given in UTS #37, appendix B.1:
>
>http://www.unicode.org/reports/tr37/#w3ac17b5b1
>
>The system appears fairly simple; any character can be followed by a variation
>selector, a Unicode character from the range U+E0100 to U+E01EF. If the font
>used supports IVS and a glyph corresponding to that variation selector is
>available, that glyph is used instead of the default glyph for that codepoint.
>
>I am not very familiar with the technical details, but I believe those variant
>glyphs are not directly encoded, but exclusively accessible through the
>combination of a character followed by a variation selector, using cmaps.
>
>My question is: is this something Freetype should eventually support? And if
>so, is this on the radar for the Freetype project?
>
>Kind regards,
>
>Jeroen Hoek
>
>
>_______________________________________________
>Freetype mailing list
>address@hidden
>http://lists.nongnu.org/mailman/listinfo/freetype