[COMMIT] Use non-ASCII Lisp characters as keysyms under X11 too, thank you Ben

APPROVE COMMIT NOTE: This patch has been committed # HG changeset patch # User Aidan Kehoe <kehoea(a)parhasard.net&gt; # Date 1513728833 0 # Wed Dec 20 00:13:53 2017 +0000 # Node ID a69c6e871607eee0c429c827c5e0be49ede0bb17 # Parent 268f55c89b5c4bd6e5fe72eef1be8a2d6856a844 Use non-ASCII Lisp characters as keysyms under X11 too, thank you Ben src/ChangeLog addition: 2017-12-20 Aidan Kehoe <kehoea(a)parhasard.net&gt; Use Lisp characters as the keysym for event objects reflecting keys with character mappings on X11, as we do on every other platform, and as Ben suggests. * event-Xt.c: * event-Xt.c (USE_UNICODE_MAP): Don't use the current buffer for the Unicode conversion precedence list, use the language-specific default in Vdefault_unicode_precedence_array. * event-Xt.c (x_keysym_to_character): Give useful values for ASCII values too. * event-Xt.c (maybe_define_x_key_as_self_inserting_character): Don't mess around with the character-of-keysym property here, now we no longer use symbol keysyms for X11 keys with character mappings. * event-Xt.c (x_has_keysym): Call m_d_x_k_a_s_i_c() with its new simpler calling convention. * event-Xt.c (x_keysym_to_emacs_keysym): Call x_keysym_to_character() early. * event-Xt.c (x_to_emacs_keysym): Drop the SIMPLE_P argument, not used. * event-Xt.c (x_event_to_emacs_event): Update calls to x_to_emacs_keysym() correspondingly in this function. diff -r 268f55c89b5c -r a69c6e871607 src/ChangeLog --- a/src/ChangeLog Wed Dec 20 00:04:36 2017 +0000 +++ b/src/ChangeLog Wed Dec 20 00:13:53 2017 +0000 ＠＠ -1,3 +1,28 ＠＠ +2017-12-20 Aidan Kehoe <kehoea(a)parhasard.net&gt; + + Use Lisp characters as the keysym for event objects reflecting + keys with character mappings on X11, as we do on every other + platform, and as Ben suggests. + * event-Xt.c: + * event-Xt.c (USE_UNICODE_MAP): + Don't use the current buffer for the Unicode conversion precedence + list, use the language-specific default in + Vdefault_unicode_precedence_array. + * event-Xt.c (x_keysym_to_character): + Give useful values for ASCII values too. + * event-Xt.c (maybe_define_x_key_as_self_inserting_character): + Don't mess around with the character-of-keysym property here, now + we no longer use symbol keysyms for X11 keys with character + mappings. + * event-Xt.c (x_has_keysym): + Call m_d_x_k_a_s_i_c() with its new simpler calling convention. + * event-Xt.c (x_keysym_to_emacs_keysym): + Call x_keysym_to_character() early. + * event-Xt.c (x_to_emacs_keysym): + Drop the SIMPLE_P argument, not used. + * event-Xt.c (x_event_to_emacs_event): + Update calls to x_to_emacs_keysym() correspondingly in this function. + 2017-12-20 Aidan Kehoe <kehoea(a)parhasard.net&gt; * redisplay-gtk.c (XLIKE_bevel_area): diff -r 268f55c89b5c -r a69c6e871607 src/event-Xt.c --- a/src/event-Xt.c Wed Dec 20 00:04:36 2017 +0000 +++ b/src/event-Xt.c Wed Dec 20 00:13:53 2017 +0000 ＠＠ -125,7 +125,7 ＠＠ 0 }; -static Lisp_Object x_keysym_to_emacs_keysym (KeySym keysym, int simple_p); +static Lisp_Object x_keysym_to_emacs_keysym (KeySym keysym); void emacs_Xt_mapping_action (Widget w, XEvent *event); void debug_process_finalization (Lisp_Process *p); void emacs_Xt_event_handler (Widget wid, XtPointer closure, XEvent *event, ＠＠ -147,13 +147,12 ＠＠ && map[keysym - FIRST_KNOWN_##map ]) do \ { \ keysym -= FIRST_KNOWN_##map ; \ - return make_char (buffer_unicode_to_ichar ((int) map[keysym], \ - /* ＠＠#### need to get some sort \ - of buffer to compute this off; \ - only applies in the old-Mule \ - world */ \ - current_buffer, \ - CONVERR_SUCCEED)); \ + /* Use the language-environment precedence list, not that of */ \ + /* the current buffer. */ \ + return make_char (filtered_unicode_to_ichar \ + ((int) map[keysym], \ + Vdefault_unicode_precedence_array, NULL, \ + CONVERR_SUCCEED)); \ } while (0) /* Maps to Unicode for X11 KeySyms, where we don't have a direct internal ＠＠ -1004,8 +1003,12 ＠＠ int code = 0; #endif /* MULE */ + if (keysym < 0x100) + { + return make_char ((Ichar) keysym); + } + /* ＠＠#### Add support for 0xFE?? and 0xFF?? keysyms - Add support for KOI8-U extensions in the 0x06?? range See http://www.cl.cam.ac.uk/~mgk25/ucs/keysyms.txt */ ＠＠ -1021,11 +1024,12 ＠＠ return make_char (keysym & 0xFFFFFF); #else if (keysym >= 0x01000000 && keysym <= 0x0110FFFF) - return make_char (buffer_unicode_to_ichar - ((int) (keysym & 0xFFFFFF), - /* ＠＠#### need to get some sort of buffer to compute - this off; only applies in the old-Mule world */ - current_buffer, CONVERR_SUCCEED)); + /* Use the language-environment precedence list, not that of + the current buffer. */ + return make_char (filtered_unicode_to_ichar + ((int)(keysym & 0xFFFFFF), + Vdefault_unicode_precedence_array, NULL, + CONVERR_SUCCEED)); #endif /* not MULE */ if ((keysym & 0xff) < 0xa0) ＠＠ -1131,12 +1135,10 ＠＠ case 32: /* Currency. The lower sixteen bits of these keysyms happily correspond exactly to the Unicode code points of the associated characters */ - return make_char (buffer_unicode_to_ichar - ((int) (keysym & 0xffff), - /* ＠＠#### need to get some sort of buffer to - compute this off; only applies in the old-Mule - world */ - current_buffer, CONVERR_SUCCEED)); + return make_char (filtered_unicode_to_ichar + ((int) (keysym & 0xffff), + Vdefault_unicode_precedence_array, NULL, + CONVERR_SUCCEED)); /* ＠＠#### Support me! ＠＠ -1204,11 +1206,16 ＠＠ break; } - if (code == 0) - return Qnil; - - /* #### Is this check on !NILP (charset) needed? Maybe should be assert? */ - if (!NILP (charset)) + assert (!NILP (charset)); + + if (code == 0 || EQ (charset, Qzero)) + { + /* CHARSET being zero is unlikely, but we shouldn't crash, even with + error-checking turned on, if X11 gives us a keysym we don't + understand. Don't make it an assertion. */ + return Qnil; + } + else { /* First try to generate a unified character by converting through Unicode, then try converting directly to an Ichar (only matters ＠＠ -1216,16 +1223,21 ＠＠ int ucs = charset_codepoint_to_unicode (charset, 0, code, CONVERR_FAIL); if (ucs >= 0) { - /* ＠＠#### current_buffer dependency */ - Ichar ich = buffer_unicode_to_ichar (ucs, current_buffer, - CONVERR_FAIL); + Ichar ich + /* Use the language-environment precedence list, not that of the + current buffer. */ + = filtered_unicode_to_ichar (ucs, + Vdefault_unicode_precedence_array, + NULL, CONVERR_SUCCEED); if (ich >= 0) - return make_char (ich); + { + return make_char (ich); + } } else { - Ichar ich = - charset_codepoint_to_ichar (charset, 0, code, CONVERR_FAIL); + Ichar ich + = charset_codepoint_to_ichar (charset, 0, code, CONVERR_FAIL); if (ich >= 0) return make_char (ich); } ＠＠ -1242,29 +1254,18 ＠＠ /* keymap handling */ /************************************************************************/ -/* ＠＠#### This is the wrong approach, I think. We should not be exposing the - name of the keysym anywhere, or forcing the user to use this name. - At the Lisp level, the user should simply see the character itself, and - should be able to bind the actual character. Furthermore, introducing - the symbol introduces an X-specific dependency; I can't expect to set - a binding for a particular Unicode character and have it work on both - Windows and X. --ben */ +extern Lisp_Object Vcurrent_global_map; /* See comment near character_to_event(). */ static void -maybe_define_x_key_as_self_inserting_character (KeySym keysym, - Lisp_Object symbol) +maybe_define_x_key_as_self_inserting_character (Lisp_Object emacs_keysym) { - Lisp_Object character = x_keysym_to_character (keysym); - - if (CHARP (character)) + if (CHARP (emacs_keysym)) { - extern Lisp_Object Vcurrent_global_map; - extern Lisp_Object Qcharacter_of_keysym; - if (NILP (Flookup_key (Vcurrent_global_map, symbol, Qnil))) + if (NILP (Flookup_key (Vcurrent_global_map, emacs_keysym, Qnil))) { - Fput (symbol, Qcharacter_of_keysym, character); - Fdefine_key (Vcurrent_global_map, symbol, Qself_insert_command); + Fdefine_key (Vcurrent_global_map, emacs_keysym, + Qself_insert_command); } } } ＠＠ -1275,7 +1276,7 ＠＠ KeySym upper_lower[2]; int j; - if (keysym < 0x80) /* Optimize for ASCII keysyms */ + if (keysym < 0x80) /* ASCII keysyms have default bindings anyway. */ return; /* If you execute: ＠＠ -1292,7 +1293,7 ＠＠ for (j = 0; j < (upper_lower[0] == upper_lower[1] ? 1 : 2); j++) { KeySym ks = upper_lower[j]; - Lisp_Object sym = x_keysym_to_emacs_keysym (ks, 0); + Lisp_Object sym = x_keysym_to_emacs_keysym (ks); Lisp_Object new_value = with_modifiers ? Qt : Qsans_modifiers; Lisp_Object old_value = Fgethash (sym, hash_table, Qnil); ＠＠ -1300,7 +1301,7 ＠＠ && ! (EQ (old_value, Qsans_modifiers) && EQ (new_value, Qt))) { - maybe_define_x_key_as_self_inserting_character (ks, sym); + maybe_define_x_key_as_self_inserting_character (sym); Fputhash (sym, new_value, hash_table); } } ＠＠ -1846,32 +1847,19 ＠＠ /************************************************************************/ static Lisp_Object -x_keysym_to_emacs_keysym (KeySym keysym, int simple_p) +x_keysym_to_emacs_keysym (KeySym keysym) { Extbyte *name; DECLARE_EISTRING(einame); + Lisp_Object character; if (keysym >= XK_exclam && keysym <= XK_asciitilde) /* We must assume that the X keysym numbers for the ASCII graphic characters are the same as their ASCII codes. */ return make_char (keysym); - if (keysym >= 0x01000000 && keysym <= 0x0110FFFF) - { - /* These keysyms malloc with XKeysymToString(), *every time the - function is called.* Avoid leaking, construct the keysym string - ourselves. */ - Ascbyte buf [10]; - qxesprintf ((Ibyte *) buf, keysym & 0xff0000 ? "U%06X" : "U%04X", - (unsigned int) (keysym & 0xffffff)); - return KEYSYM (buf); - } - switch (keysym) { - /* These would be handled correctly by the default case, but by - special-casing them here we don't garbage a string or call - intern(). */ case XK_BackSpace: return QKbackspace; case XK_Tab: return QKtab; case XK_Linefeed: return QKlinefeed; ＠＠ -1881,8 +1869,15 ＠＠ case XK_Delete: return QKdelete; case 0: return Qnil; default: - if (simple_p) return Qnil; + character = x_keysym_to_character (keysym); + + if (CHARP (character)) + { + return character; + } + name = XKeysymToString (keysym); + if (!name || !name[0]) /* This happens if there is a mismatch between the Xlib of XEmacs and the Xlib of the X server... ＠＠ -1947,8 +1942,7 ＠＠ } static Lisp_Object -x_to_emacs_keysym (XKeyPressedEvent *event, int simple_p, KeySym *x_keysym_out) - /* simple_p means don't try too hard (ASCII only) */ +x_to_emacs_keysym (XKeyPressedEvent *event, KeySym *x_keysym_out) { KeySym keysym = NoSymbol; ＠＠ -1984,7 +1978,7 ＠＠ XLookupString (event, dummy, 200, &keysym, 0); *x_keysym_out = keysym; return (IsModifierKey (keysym) || keysym == XK_Mode_switch ) - ? Qnil : x_keysym_to_emacs_keysym (keysym, simple_p); + ? Qnil : x_keysym_to_emacs_keysym (keysym); } #endif /* ! XIM_MOTIF */ ＠＠ -2045,7 +2039,7 ＠＠ case XLookupBoth: *x_keysym_out = keysym; return (IsModifierKey (keysym) || keysym == XK_Mode_switch ) - ? Qnil : x_keysym_to_emacs_keysym (keysym, simple_p); + ? Qnil : x_keysym_to_emacs_keysym (keysym); case XLookupChars: { ＠＠ -2229,7 +2223,7 ＠＠ /* This used to compute the frame from the given X window and store it here, but we really don't care about the frame. */ SET_EVENT_CHANNEL (emacs_event, DEVICE_CONSOLE (d)); - keysym = x_to_emacs_keysym (&x_event->xkey, 0, &x_keysym); + keysym = x_to_emacs_keysym (&x_event->xkey, &x_keysym); /* If the emacs keysym is nil, then that means that the X keysym was either a Modifier or NoSymbol, which -- ‘As I sat looking up at the Guinness ad, I could never figure out / How your man stayed up on the surfboard after forty pints of stout’ (C. Moore)