Add 'effective utf8' field to ncinput struct. This field will contain the utf8 emitted by a key with all modifiers taken into account.

doc/man/man3/notcurses_input.3.md

@@ -37,6 +37,7 @@ typedef struct ncinput {
   ncintype_e evtype;
   unsigned modifiers;// bitmask over NCKEY_MOD_*
   int ypx, xpx;      // pixel offsets within cell, -1 for undefined
+  char utf8_eff[5];  // Effective utf8 representation, taking modifier keys into account
 } ncinput;
 
 
@@ -242,6 +243,12 @@ issues are resolved. You can determine whether the protocol is in use
 by examining the output of **notcurses-info(1)**. If the **kbd** property
 is indicated, you're using the Kitty protocol.
 
+Note that utf_eff will always return the effective text value of the key,
+taking into account any meta keys pressed. (So shift-a will always return
+'A' in this field, regardless of if the Kitty keyboard disambiguation
+protocol is used.) This field generally only can be trusted on
+**NCTYPE_PRESS** and **NCTYPE_REPEAT** events, however.
+
 Mouse events in the left margins will never be delivered to the
 application (as is intended), but mouse events in the bottom and right margins
 sometimes can be if the event occurs prior to a window resize.

include/notcurses/notcurses.h

@@ -1214,6 +1214,7 @@ typedef struct ncinput {
   ncintype_e evtype;
   unsigned modifiers;// bitmask over NCKEY_MOD_*
   int ypx, xpx;      // pixel offsets within cell, -1 for undefined
+  char utf8_eff[5];  // Effective utf8 representation, taking modifier keys into account
 } ncinput;
 
 static inline bool

src/lib/in.c

@@ -797,7 +797,7 @@ kitty_functional(uint32_t val){
 }
 
 static void
-kitty_kbd(inputctx* ictx, int val, int mods, int evtype){
+kitty_kbd_txt(inputctx* ictx, int val, int mods, unsigned *txt, int evtype){
   assert(evtype >= 0);
   assert(mods >= 0);
   assert(val > 0);
@@ -830,9 +830,33 @@ kitty_kbd(inputctx* ictx, int val, int mods, int evtype){
       tni.evtype = NCTYPE_UNKNOWN;
       break;
   }
+  // Validate the txt array. It needs to be a non-zero-length set of up to 4 bytes.
+  int txt_valid = 0;
+  if(txt){
+    for (int i = 0 ; i<4 ; i++){
+      if(txt[i]==0) break;
+      if(txt[i]>255){
+        txt_valid = 0;
+        break;
+      }
+      txt_valid = 1;
+    }
+  }
+  //note: if we don't populate .utf8_eff here, it will be set to what becomes .utf8 in
+  //internal_get().
+  if(txt_valid){
+    for(int i=0 ; i<4 ; i++){
+      tni.utf8_eff[i] = (char)txt[i];
+    }
+  }
   load_ncinput(ictx, &tni);
 }
 
+static void
+kitty_kbd(inputctx* ictx, int val, int mods, int evtype){
+  kitty_kbd_txt(ictx, val, mods, NULL, evtype);
+}
+
 static int
 kitty_cb_simple(inputctx* ictx){
   unsigned val = amata_next_numeric(&ictx->amata, "\x1b[", 'u');
@@ -849,6 +873,66 @@ kitty_cb(inputctx* ictx){
   return 2;
 }
 
+static int 
+kitty_cb_atxtn(inputctx* ictx, int n, int with_event){
+  unsigned txt[5]={0};
+  unsigned val = amata_next_numeric(&ictx->amata, "\x1b[", ';');
+  unsigned ev = 0;
+  unsigned mods = 0;
+  if (with_event) {
+    mods = amata_next_numeric(&ictx->amata, "", ':');
+    ev = amata_next_numeric(&ictx->amata, "", ';');
+  } else {
+    mods = amata_next_numeric(&ictx->amata, "", ';');
+  }
+  for (int i = 0; i<n; i++) {
+    txt[i] = amata_next_numeric(&ictx->amata, "", (i==n-1)?'u':';');
+  }
+  kitty_kbd_txt(ictx, val, mods, txt, ev);
+  return 2;
+}
+
+static int
+kitty_cb_atxt1(inputctx* ictx){
+  return kitty_cb_atxtn(ictx, 1, 0);
+}
+
+static int
+kitty_cb_atxt2(inputctx* ictx){
+  return kitty_cb_atxtn(ictx, 2, 0);
+}
+
+static int
+kitty_cb_atxt3(inputctx* ictx){
+  return kitty_cb_atxtn(ictx, 3, 0);
+}
+
+static int
+kitty_cb_atxt4(inputctx* ictx){
+  return kitty_cb_atxtn(ictx, 4, 0);
+}
+
+
+static int
+kitty_cb_complex_atxt1(inputctx* ictx){
+  return kitty_cb_atxtn(ictx, 1, 1);
+}
+
+static int
+kitty_cb_complex_atxt2(inputctx* ictx){
+  return kitty_cb_atxtn(ictx, 2, 1);
+}
+
+static int
+kitty_cb_complex_atxt3(inputctx* ictx){
+  return kitty_cb_atxtn(ictx, 3, 1);
+}
+
+static int
+kitty_cb_complex_atxt4(inputctx* ictx){
+  return kitty_cb_atxtn(ictx, 4, 1);
+}
+
 static uint32_t
 legacy_functional(uint32_t id){
   switch(id){
@@ -1702,7 +1786,15 @@ build_cflow_automaton(inputctx* ictx){
     { "[\\Nu", kitty_cb_simple, },
     { "[\\N;\\N~", wezterm_cb, },
     { "[\\N;\\Nu", kitty_cb, },
+    { "[\\N;\\N;\\Nu", kitty_cb_atxt1, },
+    { "[\\N;\\N;\\N;\\Nu", kitty_cb_atxt2, },
+    { "[\\N;\\N;\\N;\\N;\\Nu", kitty_cb_atxt3, },
+    { "[\\N;\\N;\\N;\\N;\\N;\\Nu", kitty_cb_atxt4, },
     { "[\\N;\\N:\\Nu", kitty_cb_complex, },
+    { "[\\N;\\N:\\N;\\Nu", kitty_cb_complex_atxt1, },
+    { "[\\N;\\N:\\N;\\N;\\Nu", kitty_cb_complex_atxt2, },
+    { "[\\N;\\N:\\N;\\N;\\N;\\Nu", kitty_cb_complex_atxt3, },
+    { "[\\N;\\N:\\N;\\N;\\N;\\N;\\Nu", kitty_cb_complex_atxt4, },
     { "[\\N;\\N;\\N~", xtmodkey_cb, },
     { "[\\N;\\N:\\N~", kitty_cb_functional, },
     { "[1;\\NP", legacy_cb_f1, },
@@ -2617,6 +2709,9 @@ internal_get(inputctx* ictx, const struct timespec* ts, ncinput* ni){
     if(notcurses_ucs32_to_utf8(&ni->id, 1, (unsigned char*)ni->utf8, sizeof(ni->utf8)) < 0){
       ni->utf8[0] = 0;
     }
+    if(ni->utf8_eff[0] == 0){
+      memcpy(ni->utf8_eff, ni->utf8, sizeof(ni->utf8_eff));
+    }
   }
   if(++ictx->iread == ictx->isize){
     ictx->iread = 0;

src/lib/reader.c

@@ -400,12 +400,8 @@ bool ncreader_offer_input(ncreader* n, const ncinput* ni){
   }else if(nckey_synthesized_p(ni->id)){
     return false;
   }
-  // FIXME need to collect full EGCs
-  char wbuf[WCHAR_MAX_UTF8BYTES + 1];
-  // FIXME breaks for wint_t < 32bits
-  if(snprintf(wbuf, sizeof(wbuf), "%lc", (wint_t)ni->id) < (int)sizeof(wbuf)){
-    ncreader_write_egc(n, wbuf);
-  }
+
+  ncreader_write_egc(n, ni->utf8_eff);
   return true;
 }
 

src/lib/termdesc.c

@@ -351,9 +351,9 @@ init_terminfo_esc(tinfo* ti, const char* name, escape_e idx,
 #define KITTYQUERY
 #endif
 
-// request kitty keyboard protocol features 1, 2, and 8, first pushing current.
+// request kitty keyboard protocol features 1, 2, 8 and 16, first pushing current.
 // see https://sw.kovidgoyal.net/kitty/keyboard-protocol/#progressive-enhancement
-#define KKBDSUPPORT "\x1b[=11u"
+#define KKBDSUPPORT "\x1b[=27u"
 
 // the kitty keyboard protocol allows unambiguous, complete identification of
 // input events. this queries for the level of support. we want to do this