perf(treesitter): incremental foldupdate

Problem:
While the fold level computation is incremental, the evaluation of the
foldexpr is done on the full buffer. Despite that the foldexpr reads
from the cache, it can take tens of milliseconds for moderately big (10K
lines) buffers.

Solution:
Track the range of lines on which the foldexpr should be evaluated.

1 files changed, 17 insertions, 6 deletions

diff --git a/src/nvim/lua/stdlib.c b/src/nvim/lua/stdlib.c
index 788185f2b4..032cb246bf 100644
--- a/src/nvim/lua/stdlib.c
+++ b/src/nvim/lua/stdlib.c
@@ -543,14 +543,25 @@ static int nlua_iconv(lua_State *lstate)
   return 1;
 }
 
-// Update foldlevels (e.g., by evaluating 'foldexpr') for all lines in the current window without
-// invoking other side effects. Unlike `zx`, it does not close manually opened folds and does not
-// open folds under the cursor.
+// Update foldlevels (e.g., by evaluating 'foldexpr') for the given line range in the given window,
+// without invoking other side effects. Unlike `zx`, it does not close manually opened folds and
+// does not open folds under the cursor.
 static int nlua_foldupdate(lua_State *lstate)
 {
-  curwin->w_foldinvalid = true;  // recompute folds
-  foldUpdate(curwin, 1, (linenr_T)MAXLNUM);
-  curwin->w_foldinvalid = false;
+  handle_T window = (handle_T)luaL_checkinteger(lstate, 1);
+  Error err = ERROR_INIT;
+  win_T *win = find_window_by_handle(window, &err);
+  if (ERROR_SET(&err)) {
+    nlua_push_errstr(lstate, err.msg);
+    api_clear_error(&err);
+    lua_error(lstate);
+    return 0;
+  }
+
+  linenr_T start = (linenr_T)luaL_checkinteger(lstate, 2);
+  linenr_T end = (linenr_T)luaL_checkinteger(lstate, 3);
+
+  foldUpdate(win, start + 1, end);
 
   return 0;
 }