fix(terminal): don't pass incomplete UTF-8 sequence to libvterm (#27922)

src/nvim/channel.c

@@ -29,6 +29,7 @@
 #include "nvim/log.h"
 #include "nvim/lua/executor.h"
 #include "nvim/main.h"
+#include "nvim/mbyte.h"
 #include "nvim/memory.h"
 #include "nvim/message.h"
 #include "nvim/msgpack_rpc/channel.h"
@@ -646,35 +647,52 @@ static inline list_T *buffer_to_tv_list(const char *const buf, const size_t len)
 void on_channel_data(Stream *stream, RBuffer *buf, size_t count, void *data, bool eof)
 {
   Channel *chan = data;
-  on_channel_output(stream, chan, buf, count, eof, &chan->on_data);
+  on_channel_output(stream, chan, buf, eof, &chan->on_data);
 }
 
 void on_job_stderr(Stream *stream, RBuffer *buf, size_t count, void *data, bool eof)
 {
   Channel *chan = data;
-  on_channel_output(stream, chan, buf, count, eof, &chan->on_stderr);
+  on_channel_output(stream, chan, buf, eof, &chan->on_stderr);
 }
 
-static void on_channel_output(Stream *stream, Channel *chan, RBuffer *buf, size_t count, bool eof,
+static void on_channel_output(Stream *stream, Channel *chan, RBuffer *buf, bool eof,
                               CallbackReader *reader)
 {
-  // stub variable, to keep reading consistent with the order of events, only
+  size_t count;
-  // consider the count parameter.
+  char *output = rbuffer_read_ptr(buf, &count);
-  size_t r;
+
-  char *ptr = rbuffer_read_ptr(buf, &r);
+  if (chan->term) {
+    if (!eof) {
+      char *p = output;
+      char *end = output + count;
+      while (p < end) {
+        // Don't pass incomplete UTF-8 sequences to libvterm. #16245
+        // Composing chars can be passed separately, so utf_ptr2len_len() is enough.
+        int clen = utf_ptr2len_len(p, (int)(end - p));
+        if (clen > end - p) {
+          count = (size_t)(p - output);
+          break;
+        }
+        p += clen;
+      }
+    }
+
+    terminal_receive(chan->term, output, count);
+  }
+
+  if (count) {
+    rbuffer_consumed(buf, count);
+  }
+  // Move remaining data to start of buffer, so the buffer can never wrap around.
+  rbuffer_reset(buf);
+
+  if (callback_reader_set(*reader)) {
+    ga_concat_len(&reader->buffer, output, count);
+  }
 
   if (eof) {
     reader->eof = true;
-  } else {
-    if (chan->term) {
-      terminal_receive(chan->term, ptr, count);
-    }
-
-    rbuffer_consumed(buf, count);
-
-    if (callback_reader_set(*reader)) {
-      ga_concat_len(&reader->buffer, ptr, count);
-    }
   }
 
   if (callback_reader_set(*reader)) {

test/functional/fixtures/shell-test.c

@@ -49,6 +49,10 @@ int main(int argc, char **argv)
     help();
   }
 
+#ifdef _MSC_VER
+  SetConsoleOutputCP(CP_UTF8);
+#endif
+
   if (argc >= 2) {
     if (strcmp(argv[1], "-t") == 0) {
       if (argc < 3) {

test/functional/terminal/buffer_spec.lua

@@ -10,6 +10,7 @@ local pcall_err = helpers.pcall_err
 local eq, neq = helpers.eq, helpers.neq
 local api = helpers.api
 local retry = helpers.retry
+local testprg = helpers.testprg
 local write_file = helpers.write_file
 local command = helpers.command
 local exc_exec = helpers.exc_exec
@@ -395,6 +396,20 @@ describe(':terminal buffer', function()
       eq('a' .. composing:rep(5), api.nvim_get_current_line())
     end)
   end)
+
+  it('handles split UTF-8 sequences #16245', function()
+    local screen = Screen.new(50, 7)
+    screen:attach()
+    fn.termopen({ testprg('shell-test'), 'UTF-8' })
+    screen:expect([[
+      ^å                                                 |
+      ref: å̲                                            |
+      1: å̲                                              |
+      2: å̲                                              |
+      3: å̲                                              |
+                                                        |*2
+    ]])
+  end)
 end)
 
 describe('on_lines does not emit out-of-bounds line indexes when', function()