refactor: delete duplicate utf8-functionality

Also remove British National Replacement Character Set.

We keep the DEC Special Graphics and ASCII despite it not being unicode
as some old software such as calcurse still rely on this functionality.

References:
- https://github.com/neovim/neovim/pull/31934#discussion_r1911046426
- https://en.wikipedia.org/wiki/DEC_Special_Graphics
- https://vt100.net/docs/vt220-rm/chapter2.html#S2.4.3

src/nvim/mbyte_defs.h

@@ -74,3 +74,5 @@ typedef struct {
 } CharBoundsOff;
 
 typedef utf8proc_int32_t GraphemeState;
+
+enum { UNICODE_INVALID = 0xFFFD, };

src/nvim/tui/termkey/termkey.c

@@ -634,40 +634,13 @@ static void eat_bytes(TermKey *tk, size_t count)
   tk->buffcount -= count;
 }
 
-// TODO(dundargoc): we should be able to replace this with utf_char2bytes from mbyte.c
 int fill_utf8(int codepoint, char *str)
 {
-  int nbytes = utf_char2len(codepoint);
-
+  int nbytes = utf_char2bytes(codepoint, str);
   str[nbytes] = 0;
-
-  // This is easier done backwards
-  int b = nbytes;
-  while (b > 1) {
-    b--;
-    str[b] = (char)0x80 | (codepoint & 0x3f);
-    codepoint >>= 6;
-  }
-
-  switch (nbytes) {
-  case 1:
-    str[0] = (codepoint & 0x7f); break;
-  case 2:
-    str[0] = (char)0xc0 | (codepoint & 0x1f); break;
-  case 3:
-    str[0] = (char)0xe0 | (codepoint & 0x0f); break;
-  case 4:
-    str[0] = (char)0xf0 | (codepoint & 0x07); break;
-  case 5:
-    str[0] = (char)0xf8 | (codepoint & 0x03); break;
-  case 6:
-    str[0] = (char)0xfc | (codepoint & 0x01); break;
-  }
-
   return nbytes;
 }
 
-#define UTF8_INVALID 0xFFFD
 static TermKeyResult parse_utf8(const unsigned char *bytes, size_t len, int *cp, size_t *nbytep)
 {
   unsigned nbytes;
@@ -681,7 +654,7 @@ static TermKeyResult parse_utf8(const unsigned char *bytes, size_t len, int *cp,
     return TERMKEY_RES_KEY;
   } else if (b0 < 0xc0) {
     // Starts with a continuation byte - that's not right
-    *cp = UTF8_INVALID;
+    *cp = UNICODE_INVALID;
     *nbytep = 1;
     return TERMKEY_RES_KEY;
   } else if (b0 < 0xe0) {
@@ -700,7 +673,7 @@ static TermKeyResult parse_utf8(const unsigned char *bytes, size_t len, int *cp,
     nbytes = 6;
     *cp = b0 & 0x01;
   } else {
-    *cp = UTF8_INVALID;
+    *cp = UNICODE_INVALID;
     *nbytep = 1;
     return TERMKEY_RES_KEY;
   }
@@ -714,7 +687,7 @@ static TermKeyResult parse_utf8(const unsigned char *bytes, size_t len, int *cp,
 
     cb = bytes[b];
     if (cb < 0x80 || cb >= 0xc0) {
-      *cp = UTF8_INVALID;
+      *cp = UNICODE_INVALID;
       *nbytep = b;
       return TERMKEY_RES_KEY;
     }
@@ -725,14 +698,14 @@ static TermKeyResult parse_utf8(const unsigned char *bytes, size_t len, int *cp,
 
   // Check for overlong sequences
   if ((int)nbytes > utf_char2len(*cp)) {
-    *cp = UTF8_INVALID;
+    *cp = UNICODE_INVALID;
   }
 
   // Check for UTF-16 surrogates or invalid *cps
   if ((*cp >= 0xD800 && *cp <= 0xDFFF)
       || *cp == 0xFFFE
       || *cp == 0xFFFF) {
-    *cp = UTF8_INVALID;
+    *cp = UNICODE_INVALID;
   }
 
   *nbytep = nbytes;
@@ -962,9 +935,9 @@ static TermKeyResult peekkey_simple(TermKey *tk, TermKeyKey *key, int force, siz
     if (res == TERMKEY_RES_AGAIN && force) {
       // There weren't enough bytes for a complete UTF-8 sequence but caller
       // demands an answer. About the best thing we can do here is eat as many
-      // bytes as we have, and emit a UTF8_INVALID. If the remaining bytes
+      // bytes as we have, and emit a UNICODE_INVALID. If the remaining bytes
       // arrive later, they'll be invalid too.
-      codepoint = UTF8_INVALID;
+      codepoint = UNICODE_INVALID;
       *nbytep = tk->buffcount;
       res = TERMKEY_RES_KEY;
     }

src/nvim/vterm/encoding.c

@@ -210,6 +210,7 @@ static void decode_table(VTermEncoding *enc, void *data, uint32_t cp[], int *cpi
   }
 }
 
+// https://en.wikipedia.org/wiki/DEC_Special_Graphics
 static const struct StaticTableEncoding encoding_DECdrawing = {
   { .decode = &decode_table },
   {
@@ -247,13 +248,6 @@ static const struct StaticTableEncoding encoding_DECdrawing = {
   }
 };
 
-static const struct StaticTableEncoding encoding_uk = {
-  { .decode = &decode_table },
-  {
-    [0x23] = 0x00a3,  // £
-  }
-};
-
 static struct {
   VTermEncodingType type;
   char designation;
@@ -262,7 +256,6 @@ static struct {
 encodings[] = {
   { ENC_UTF8,      'u', &encoding_utf8 },
   { ENC_SINGLE_94, '0', (VTermEncoding *)&encoding_DECdrawing },
-  { ENC_SINGLE_94, 'A', (VTermEncoding *)&encoding_uk },
   { ENC_SINGLE_94, 'B', &encoding_usascii },
   { 0 },
 };

test/unit/vterm_spec.lua

@@ -28,6 +28,7 @@ local bit = require('bit')
 --- @field parser_sos function
 --- @field parser_text function
 --- @field print_color function
+--- @field schar_get fun(any, any):integer
 --- @field screen_sb_clear function
 --- @field screen_sb_popline function
 --- @field screen_sb_pushline function
@@ -43,6 +44,8 @@ local bit = require('bit')
 --- @field state_setpenattr function
 --- @field state_settermprop function
 --- @field term_output function
+--- @field utf_ptr2char fun(any):integer
+--- @field utf_ptr2len fun(any):integer
 --- @field vterm_input_write function
 --- @field vterm_keyboard_end_paste function
 --- @field vterm_keyboard_key function
@@ -360,7 +363,7 @@ local function screen_cell(row, col, expected, screen)
   pos['row'] = row
   pos['col'] = col
 
-  local cell = t.ffi.new('VTermScreenCell')
+  local cell = t.ffi.new('VTermScreenCell') ---@type any
   vterm.vterm_screen_get_cell(screen, pos, cell)
 
   local buf = t.ffi.new('unsigned char[32]')
@@ -1705,12 +1708,6 @@ putglyph 1f3f4,200d,2620,fe0f 2 0,4]])
     push('#', vt)
     expect('putglyph 23 1 0,0')
 
-    -- Designate G0=UK
-    reset(state, nil)
-    push('\x1b(A', vt)
-    push('#', vt)
-    expect('putglyph a3 1 0,0')
-
     -- Designate G0=DEC drawing
     reset(state, nil)
     push('\x1b(0', vt)