shell: handle split-up UTF-8 sequences

src/nvim/os/shell.c

@@ -422,7 +422,7 @@ static void out_data_ring(char *output, size_t size)
   }
 
   if (output == NULL && size == SIZE_MAX) {   // Print mode
-    out_data_append_to_screen(last_skipped, last_skipped_len, true);
+    out_data_append_to_screen(last_skipped, &last_skipped_len, true);
     return;
   }
 
@@ -450,30 +450,40 @@ static void out_data_ring(char *output, size_t size)
 /// @param output       Data to append to screen lines.
 /// @param remaining    Size of data.
 /// @param new_line     If true, next data output will be on a new line.
-static void out_data_append_to_screen(char *output, size_t remaining,
-                                      bool new_line)
+static void out_data_append_to_screen(char *output, size_t *count,
+                                      bool eof)
 {
-  char *p = output, *end = output + remaining;
+  char *p = output, *end = output + *count;
   while (p < end) {
     if (*p == '\n' || *p == '\r' || *p == TAB || *p == BELL) {
       msg_putchar_attr((uint8_t)(*p), 0);
       p++;
     } else {
+      // Note: this is not 100% precise:
+      // 1. we don't check if received continuation bytes are already invalid
+      //    and we thus do some buffering that could be avoided
+      // 2. we don't compose chars over buffer boundaries, even if we see an
+      //    incomplete UTF-8 sequence that could be composing with the last
+      //    complete sequence.
+      // This will be corrected when we switch to vterm based implementation
       int i = *p ? mb_ptr2len_len((char_u *)p, (int)(end-p)) : 1;
+      if (!eof && i == 1 && utf8len_tab_zero[*(uint8_t *)p] > (end-p) && *p) {
+        *count = (size_t)(p - output);
+        goto end;
+      }
 
       (void)msg_outtrans_len_attr((char_u *)p, i, 0);
       p += i;
     }
   }
 
+end:
   ui_flush();
 }
 
 static void out_data_cb(Stream *stream, RBuffer *buf, size_t count, void *data,
     bool eof)
 {
-  // We always output the whole buffer, so the buffer can never
-  // wrap around.
   size_t cnt;
   char *ptr = rbuffer_read_ptr(buf, &cnt);
 
@@ -482,12 +492,16 @@ static void out_data_cb(Stream *stream, RBuffer *buf, size_t count, void *data,
     // Save the skipped output. If it is the final chunk, we display it later.
     out_data_ring(ptr, cnt);
   } else {
-    out_data_append_to_screen(ptr, cnt, eof);
+    out_data_append_to_screen(ptr, &cnt, eof);
   }
 
   if (cnt) {
     rbuffer_consumed(buf, cnt);
   }
+
+  // Move remaining data to start of buffer, so the buffer can never
+  // wrap around.
+  rbuffer_reset(buf);
 }
 
 /// Parses a command string into a sequence of words, taking quotes into

test/functional/fixtures/shell-test.c

@@ -4,6 +4,13 @@
 #include <stdio.h>
 #include <string.h>
 #include <stdint.h>
+#include <unistd.h>
+
+static void wait(void)
+{
+  fflush(stdout);
+  usleep(10*1000);
+}
 
 static void help(void)
 {
@@ -61,6 +68,22 @@ int main(int argc, char **argv)
       for (uint8_t i = 0; i < number; i++) {
         printf("%d: %s\n", (int) i, argv[3]);
       }
+    } else if (strcmp(argv[1], "UTF-8") == 0) {
+      // test split-up UTF-8 sequence
+      printf("\xc3"); wait();
+      printf("\xa5\n"); wait();
+
+      // split up a 2+2 grapheme clusters all possible ways
+      printf("ref: \xc3\xa5\xcc\xb2\n"); wait();
+
+      printf("1: \xc3"); wait();
+      printf("\xa5\xcc\xb2\n"); wait();
+
+      printf("2: \xc3\xa5"); wait();
+      printf("\xcc\xb2\n"); wait();
+
+      printf("3: \xc3\xa5\xcc"); wait();
+      printf("\xb2\n"); wait();
     } else {
       fprintf(stderr, "Unknown first argument\n");
       return 3;

test/functional/ui/output_spec.lua

@@ -9,6 +9,7 @@ local feed_command = helpers.feed_command
 local iswin = helpers.iswin
 local clear = helpers.clear
 local command = helpers.command
+local nvim_dir = helpers.nvim_dir
 
 describe("shell command :!", function()
   if helpers.pending_win32(pending) then return end
@@ -196,5 +197,29 @@ describe("shell command :!", function()
     end)
     end)
 
+    it('handles multibyte sequences split over buffer boundaries', function()
+      command('cd '..nvim_dir)
+      local cmd
+      if iswin() then
+        cmd = '!shell-test UTF-8  '
+      else
+        cmd = '!./shell-test UTF-8'
+      end
+      feed_command(cmd)
+      -- Note: only the first example of split composed char works
+      screen:expect([[
+        {1:~                                                    }|
+        {1:~                                                    }|
+        :]]..cmd..[[                                 |
+        å                                                    |
+        ref: å̲                                               |
+        1: å̲                                                 |
+        2: å ̲                                               |
+        3: å ̲                                               |
+                                                             |
+        {3:Press ENTER or type command to continue}^              |
+      ]])
+    end)
+
   end)
 end)