refactor(grid): change schar_T representation to be more compact

Previously, a screen cell would occupy 28+4=32 bytes per cell
as we always made space for up to MAX_MCO+1 codepoints in a cell.

As an example, even a pretty modest 50*80 screen would consume

50*80*2*32 = 256000, i e a quarter megabyte

With the factor of two due to the TUI side buffer, and even more when
using msg_grid and/or ext_multigrid.

This instead stores a 4-byte union of either:
- a valid UTF-8 sequence up to 4 bytes
- an escape char which is invalid UTF-8 (0xFF) plus a 24-bit index to a
  glyph cache

This avoids allocating space for huge composed glyphs _upfront_, while
still keeping rendering such glyphs reasonably fast (1 hash table lookup
+ one plain index lookup). If the same large glyphs are using repeatedly
on the screen, this is still a net reduction of memory/cache
consumption. The only case which really gets worse is if you blast
the screen full with crazy emojis and zalgo text and even this case
only leads to 4 extra bytes per char.

When only <= 4-byte glyphs are used, plus the 4-byte attribute code,
i e 8 bytes in total there is a factor of four reduction of memory use.
Memory which will be quite hot in cache as the screen buffer is scanned
over in win_line() buffer text drawing

A slight complication is that the representation depends on host byte
order. I've tested this manually by compling and running this
in qemu-s390x and it works fine. We might add a qemu based solution
to CI at some point.

3 files changed, 21 insertions, 11 deletions

diff --git a/src/nvim/api/ui.c b/src/nvim/api/ui.c
index 70c97be984..0ea2310042 100644
--- a/src/nvim/api/ui.c
+++ b/src/nvim/api/ui.c
@@ -833,8 +833,7 @@ void remote_ui_raw_line(UI *ui, Integer grid, Integer row, Integer startcol, Int
     bool was_space = false;
     for (size_t i = 0; i < ncells; i++) {
       repeat++;
-      if (i == ncells - 1 || attrs[i] != attrs[i + 1]
-          || strcmp(chunk[i], chunk[i + 1]) != 0) {
+      if (i == ncells - 1 || attrs[i] != attrs[i + 1] || chunk[i] != chunk[i + 1]) {
         if (UI_BUF_SIZE - BUF_POS(data) < 2 * (1 + 2 + sizeof(schar_T) + 5 + 5) + 1) {
           // close to overflowing the redraw buffer. finish this event,
           // flush, and start a new "grid_line" event at the current position.
@@ -859,7 +858,9 @@ void remote_ui_raw_line(UI *ui, Integer grid, Integer row, Integer startcol, Int
         uint32_t csize = (repeat > 1) ? 3 : ((attrs[i] != last_hl) ? 2 : 1);
         nelem++;
         mpack_array(buf, csize);
-        mpack_str(buf, chunk[i]);
+        char sc_buf[MAX_SCHAR_SIZE];
+        schar_get(sc_buf, chunk[i]);
+        mpack_str(buf, sc_buf);
         if (csize >= 2) {
           mpack_uint(buf, (uint32_t)attrs[i]);
           if (csize >= 3) {
@@ -869,7 +870,7 @@ void remote_ui_raw_line(UI *ui, Integer grid, Integer row, Integer startcol, Int
         data->ncells_pending += MIN(repeat, 2);
         last_hl = attrs[i];
         repeat = 0;
-        was_space = strequal(chunk[i], " ");
+        was_space = chunk[i] == schar_from_ascii(' ');
       }
     }
     // If the last chunk was all spaces, add a clearing chunk even if there are
@@ -893,8 +894,10 @@ void remote_ui_raw_line(UI *ui, Integer grid, Integer row, Integer startcol, Int
     for (int i = 0; i < endcol - startcol; i++) {
       remote_ui_cursor_goto(ui, row, startcol + i);
       remote_ui_highlight_set(ui, attrs[i]);
-      remote_ui_put(ui, chunk[i]);
-      if (utf_ambiguous_width(utf_ptr2char((char *)chunk[i]))) {
+      char sc_buf[MAX_SCHAR_SIZE];
+      schar_get(sc_buf, chunk[i]);
+      remote_ui_put(ui, sc_buf);
+      if (utf_ambiguous_width(utf_ptr2char(sc_buf))) {
         data->client_col = -1;  // force cursor update
       }
     }
diff --git a/src/nvim/api/vim.c b/src/nvim/api/vim.c
index 916409b973..0a94b8aafc 100644
--- a/src/nvim/api/vim.c
+++ b/src/nvim/api/vim.c
@@ -1947,7 +1947,9 @@ Array nvim__inspect_cell(Integer grid, Integer row, Integer col, Arena *arena, E
   }
   ret = arena_array(arena, 3);
   size_t off = g->line_offset[(size_t)row] + (size_t)col;
-  ADD_C(ret, CSTR_AS_OBJ((char *)g->chars[off]));
+  char *sc_buf = arena_alloc(arena, MAX_SCHAR_SIZE, false);
+  schar_get(sc_buf, g->chars[off]);
+  ADD_C(ret, CSTR_AS_OBJ(sc_buf));
   int attr = g->attrs[off];
   ADD_C(ret, DICTIONARY_OBJ(hl_get_attr_by_id(attr, true, arena, err)));
   // will not work first time
@@ -1963,6 +1965,11 @@ void nvim__screenshot(String path)
   ui_call_screenshot(path);
 }
 
+void nvim__invalidate_glyph_cache(void)
+{
+  schar_cache_clear_force();
+}
+
 Object nvim__unpack(String str, Error *err)
   FUNC_API_FAST
 {
diff --git a/src/nvim/api/win_config.c b/src/nvim/api/win_config.c
index 63cf3bb701..a579b0dde5 100644
--- a/src/nvim/api/win_config.c
+++ b/src/nvim/api/win_config.c
@@ -16,7 +16,7 @@
 #include "nvim/drawscreen.h"
 #include "nvim/extmark_defs.h"
 #include "nvim/globals.h"
-#include "nvim/grid_defs.h"
+#include "nvim/grid.h"
 #include "nvim/highlight_group.h"
 #include "nvim/macros.h"
 #include "nvim/mbyte.h"
@@ -348,7 +348,7 @@ Dictionary nvim_win_get_config(Window window, Error *err)
       for (size_t i = 0; i < 8; i++) {
         Array tuple = ARRAY_DICT_INIT;
 
-        String s = cstrn_to_string(config->border_chars[i], sizeof(schar_T));
+        String s = cstrn_to_string(config->border_chars[i], MAX_SCHAR_SIZE);
 
         int hi_id = config->border_hl_ids[i];
         char *hi_name = syn_id2name(hi_id);
@@ -520,7 +520,7 @@ static void parse_border_style(Object style,  FloatConfig *fconfig, Error *err)
 {
   struct {
     const char *name;
-    schar_T chars[8];
+    char chars[8][MAX_SCHAR_SIZE];
     bool shadow_color;
   } defaults[] = {
     { "double", { "╔", "═", "╗", "║", "╝", "═", "╚", "║" }, false },
@@ -531,7 +531,7 @@ static void parse_border_style(Object style,  FloatConfig *fconfig, Error *err)
     { NULL, { { NUL } }, false },
   };
 
-  schar_T *chars = fconfig->border_chars;
+  char (*chars)[MAX_SCHAR_SIZE] = fconfig->border_chars;
   int *hl_ids = fconfig->border_hl_ids;
 
   fconfig->border = true;