diff options
| author | Matthias Clasen <mclasen@redhat.com> | 2023-05-02 03:41:42 +0000 |
|---|---|---|
| committer | Matthias Clasen <mclasen@redhat.com> | 2023-05-02 03:41:42 +0000 |
| commit | 12c14a188f0e880bde88e18fa7fe845458e8770a (patch) | |
| tree | 93849a594a8e7e73d7b4b2547e6a02b32895889d | |
| parent | b446637f4a4cbc8018151d1a7186644cdcad8455 (diff) | |
| parent | d26a0dcf3071ee428b3b4c436a0d156d3efcfe0d (diff) | |
| download | pango-12c14a188f0e880bde88e18fa7fe845458e8770a.tar.gz | |
Merge branch 'script-itemize' into 'main'
itemize: Improve script itemization
Closes #739
See merge request GNOME/pango!685
| -rw-r--r-- | pango/pango-script.c | 23 | ||||
| -rw-r--r-- | tests/itemize/three.expected | 9 | ||||
| -rw-r--r-- | tests/itemize/three.items | 1 |
3 files changed, 30 insertions, 3 deletions
diff --git a/pango/pango-script.c b/pango/pango-script.c index 71c519fd..fa9ab21d 100644 --- a/pango/pango-script.c +++ b/pango/pango-script.c @@ -286,12 +286,29 @@ get_pair_index (gunichar ch) return -1; } +static gboolean +is_cluster_extender (gunichar ch) +{ + GUnicodeType type = g_unichar_type (ch); + return (type >= G_UNICODE_SPACING_MARK && type <= G_UNICODE_NON_SPACING_MARK) || + (ch >= 0x200C && ch <= 0x200D) || /* ZWJ, ZWNJ */ + (ch >= 0xFF9E && ch <= 0xFF9F) || /* katakana sound marks */ + (ch >= 0x1F3FB && ch <= 0x1F3FF) || /* fitzpatrick skin tone modifiers */ + (ch >= 0xE0020 && ch <= 0xE007F); /* emoji (flag) tag characters */ +} + /* duplicated in pango-language.c */ #define REAL_SCRIPT(script) \ ((script) > PANGO_SCRIPT_INHERITED && (script) != PANGO_SCRIPT_UNKNOWN) -#define SAME_SCRIPT(script1, script2) \ - (!REAL_SCRIPT (script1) || !REAL_SCRIPT (script2) || (script1) == (script2)) +#define IS_CLUSTER_EXTENDER(ch) \ + g_unichar_type (ch) + +/* TODO: Use Unicode ScriptExtensions */ +#define SAME_SCRIPT(script1, script2, ch) \ + (!REAL_SCRIPT (script1) || !REAL_SCRIPT (script2) || \ + (script1) == (script2) || \ + is_cluster_extender (ch)) #define IS_OPEN(pair_index) (((pair_index) & 1) == 0) @@ -372,7 +389,7 @@ pango_script_iter_next (PangoScriptIter *iter) } } - if (SAME_SCRIPT (iter->script_code, sc)) + if (SAME_SCRIPT (iter->script_code, sc, ch)) { if (!REAL_SCRIPT (iter->script_code) && REAL_SCRIPT (sc)) { diff --git a/tests/itemize/three.expected b/tests/itemize/three.expected new file mode 100644 index 00000000..eb45fbd3 --- /dev/null +++ b/tests/itemize/three.expected @@ -0,0 +1,9 @@ +<span font="Cantarell 11" fallback="false">عַرַبִ</span> + +Items: عַرַبִ +Chars: 6(0) +Font: Cantarell 11 +Script: arabic +Lang: ar +Bidi: 1 +Attrs: 0 12 fallback false diff --git a/tests/itemize/three.items b/tests/itemize/three.items new file mode 100644 index 00000000..394767e5 --- /dev/null +++ b/tests/itemize/three.items @@ -0,0 +1 @@ +<span font="Cantarell 11" fallback="false">عַرַبִ</span> |