FR: [show CJK font full name instead font family name when preview fonts]

[windirt]

Does your feature request involve difficulty completing a task? Please describe.
not difficult

Describe the solution you'd like
when preview font name, if detect CJK font, showing the full font name which has CJK characters, instead showing the font family name which only contain english characters.

[poire-z]

No alternative name seems to be accessible from the font library we use:
https://www.freetype.org/freetype2/docs/reference/ft2-base_interface.html#ft_facerec

[windirt]

in the doc, I found this. maybe useful.

family_name The face's family name. This is an ASCII string, usually in English, that describes the typeface's family (like ‘Times New Roman’, ‘Bodoni’, ‘Garamond’, etc). This is a least common denominator used to list fonts. Some formats (TrueType & OpenType) provide localized and Unicode versions of this string. Applications should use the format-specific interface to access them. Can be NULL (e.g., in fonts embedded in a PDF file). In case the font doesn't provide a specific family name entry, FreeType tries to synthesize one, deriving it from other name entries.

[windirt]

for example, here is the same ttf in kindle's custom font menu, it's CJK name can be recognized by kindle.

[poire-z]

Some formats (TrueType & OpenType) provide localized and Unicode versions of this string. Applications should use the format-specific interface to access them

I read that :) and what I meant is that our current library does not have the interface to access them. And adding other libraries just to get them feel a bit heavy :/

[NiLuJe]

Does HB have methods to read those?

(Also: that kind of metadata is often quirky and/or bogus and/or wrong and/or outdated, so, here be dragons anyway).

[poire-z]

Does HB have methods to read those?

Possibly: https://harfbuzz.github.io/harfbuzz-hb-ot-name.html

/**
 * hb_ot_name_id_t:
 * @HB_OT_NAME_ID_INVALID: Value to represent a nonexistent name ID.
 *
 * An integral type representing an OpenType 'name' table name identifier.
 * There are predefined name IDs, as well as name IDs return from other
 * API.  These can be used to fetch name strings from a font face.
 *
 * Since: 2.0.0
 **/
enum
{
  HB_OT_NAME_ID_COPYRIGHT               = 0,
  HB_OT_NAME_ID_FONT_FAMILY             = 1,
  HB_OT_NAME_ID_FONT_SUBFAMILY          = 2,
  HB_OT_NAME_ID_UNIQUE_ID               = 3,
  HB_OT_NAME_ID_FULL_NAME               = 4,
  HB_OT_NAME_ID_VERSION_STRING          = 5,
  HB_OT_NAME_ID_POSTSCRIPT_NAME         = 6,
  HB_OT_NAME_ID_TRADEMARK               = 7,
  HB_OT_NAME_ID_MANUFACTURER            = 8,
  HB_OT_NAME_ID_DESIGNER                = 9,
  HB_OT_NAME_ID_DESCRIPTION             = 10,
  HB_OT_NAME_ID_VENDOR_URL              = 11,
  HB_OT_NAME_ID_DESIGNER_URL            = 12,
  HB_OT_NAME_ID_LICENSE                 = 13,
  HB_OT_NAME_ID_LICENSE_URL             = 14,
/*HB_OT_NAME_ID_RESERVED                = 15,*/
  HB_OT_NAME_ID_TYPOGRAPHIC_FAMILY      = 16,
  HB_OT_NAME_ID_TYPOGRAPHIC_SUBFAMILY   = 17,
  HB_OT_NAME_ID_MAC_FULL_NAME           = 18,
  HB_OT_NAME_ID_SAMPLE_TEXT             = 19,
  HB_OT_NAME_ID_CID_FINDFONT_NAME       = 20,
  HB_OT_NAME_ID_WWS_FAMILY              = 21,
  HB_OT_NAME_ID_WWS_SUBFAMILY           = 22,
  HB_OT_NAME_ID_LIGHT_BACKGROUND        = 23,
  HB_OT_NAME_ID_DARK_BACKGROUND         = 24,
  HB_OT_NAME_ID_VARIATIONS_PS_PREFIX    = 25,

  HB_OT_NAME_ID_INVALID                 = 0xFFFF
};

typedef unsigned int hb_ot_name_id_t;

I don't think crengine should be the proxy for that, and neither xtext.
So, may be a small ffi/cdef wrapper to harfbuzz, just to see what kind of data is in there ? (If you feel like playing with it :)

[ezdiy]

No need to bother HB with this. It would just go back to freetype and ask it anyway :)

https://www.freetype.org/freetype2/docs/reference/ft2-sfnt_names.html
https://www.freetype.org/freetype2/docs/reference/ft2-truetype_tables.html#tt_name_id_xxx

That said, interpreting this stuff is pretty annoying. Add to this that a lot of chinese fonts just plaster those strings in big5/gb, with software having to do manual hacks to display stuff properly.

[poire-z]

Had a try with adding a function to xtext to use the above Harfbuzz facilities (and not FreeType, as HB can use the current UI language tag and do the table inspection/lookup for us, and give us proper UTF8).

With my fonts, I only get something for zh_CN (Simplified chinese), nothing for zh_TW (Traditional Chinese, nothing for Japanese, nothing for any other languages (except for "Windings 2" ... which translate to... "Windings 2" in many languages :/).

I don't read chinese, but is the chinese stuff displayed there what you expect to see instead of the original font name (in parenthesis) ?

static int xtext_getLocalizedFontName(lua_State *L) {
    // argument should be our Lua font object (face_obj), with
    // a getFallbackFont() callback.
    luaL_checktype(L, 1, LUA_TTABLE);
    lua_getfield(L, 1, "ftface");
    // We expect it to be a luajit ffi cdata, but the C API does not have a #define for
    // that type. But it looks like its value is higher than the greatest LUA_T* type.
    if ( lua_type(L, -1) <= LUA_TTHREAD ) {// Higher plain Lua datatype (lua.h)
        luaL_typerror(L, -1, "cdata");
    }
    // Get the usable (for Harfbuzz) FT_Face object
    FT_Face * face = (FT_Face *)lua_topointer(L, -1);
    lua_pop(L, -1); // remove ftface object
    hb_face_t * hb_face = hb_ft_face_create_referenced(*face);;
    char name[256];
    unsigned int len = 256;
    unsigned int res = hb_ot_name_get_utf8 (hb_face, HB_OT_NAME_ID_FULL_NAME, default_lang_hb_language, &len, name);
    hb_face_destroy(hb_face);
    if (res > 0) {
        lua_pushlstring(L, name, len);
        return 1;
    }
    return 0;
}

(For most people, it would be useless and expensive for nothing - I don't know if it should be an option, that people might enable and see nothing, and keep enabling and wasting cpu time for nothing - or just be enabled when lang= zh_CN...)

[windirt]

I don't read chinese, but is the chinese stuff displayed there what you expect to see instead of the original font name (in parenthesis) ?

The Chinese characters outside the parenthesis are indeed the font name I want to display.