unicode width updates

- 0xFE0F (VS16) needs to move the cursor and clear the overlapping cell
  when it's combined with a previous cell, which adds some complexity if
  the next print was going to need to wrap
- regional indicators (0x1F1E6..0x1F1FF) are technically width 1, but
  nearly all temrinal emulators/fonts and tmux consider them wide, so
  let's force these to width=2
- unassigned codepoints should default to a width of 1 instead of not
  being printed at all, which allows unknown codepoints to still
  possibly be rendered

Author: jdrouhard

src/terminal/terminal.cc

@@ -134,8 +134,26 @@ void Emulator::print( const Parser::Print* act )
       }
       if ( !combining_cell->full() ) {
         combining_cell->append( ch );
-        if ( ch == 0xFE0F ) {
+        // VS16 causes the previous codepoint to be rendered as its emoji representation
+        // which could cause it to change from 1 to 2 characters wide
+        if ( ch == 0xFE0F && !combining_cell->get_wide() ) {
+          // have to move this emoji to the next line
+          if ( fb.ds.auto_wrap_mode && fb.ds.next_print_will_wrap ) {
+            fb.get_mutable_row( -1 )->set_wrap( false );
+            fb.ds.move_col( 0 );
+            fb.move_rows_autoscroll( 1 );
+            *fb.get_mutable_cell() = *combining_cell;
+            fb.reset_cell( combining_cell );
+            fb.ds.move_col( 1, true, true );
+            combining_cell = fb.get_combining_cell();
+          }
           combining_cell->set_wide( true );
+          if ( fb.ds.insert_mode ) {
+            fb.insert_cell( fb.ds.get_cursor_row(), fb.ds.get_cursor_col() );
+          } else if ( fb.ds.get_cursor_col() < fb.ds.get_width() ) {
+            fb.reset_cell( fb.get_mutable_cell() );
+          }
+          fb.ds.move_col( 1, true, true );
         }
       }
     } break;

src/util/char_utils.h

@@ -2,9 +2,20 @@
 #define CHAR_UTILS_HPP
 
 #include "widechar_width.h"
+#include <algorithm>
+#include <array>
 
 static int mosh_wcwidth( uint32_t c )
 {
+  static constexpr auto force_wide
+    = std::array { 0x1F1E6, 0x1F1E7, 0x1F1E8, 0x1F1E9, 0x1F1EA, 0x1F1EB, 0x1F1EC, 0x1F1ED, 0x1F1EE,
+                   0x1F1EF, 0x1F1F0, 0x1F1F1, 0x1F1F2, 0x1F1F3, 0x1F1F4, 0x1F1F5, 0x1F1F6, 0x1F1F7,
+                   0x1F1F8, 0x1F1F9, 0x1F1FA, 0x1F1FB, 0x1F1FC, 0x1F1FD, 0x1F1FE, 0x1F1FF };
+
+  if ( std::binary_search( std::begin( force_wide ), std::end( force_wide ), c ) ) {
+    return 2;
+  }
+
   int width = widechar_wcwidth( c );
   if ( width >= 0 ) {
     return width;
@@ -21,7 +32,7 @@ static int mosh_wcwidth( uint32_t c )
     case widechar_private_use:
       return 1;
     case widechar_unassigned:
-      return -1;
+      return 1;
     case widechar_non_character:
       return -1;
     case widechar_widened_in_9: