Try and play nicer with UTF-8 input sequences

embedded_cli.c

@@ -146,10 +146,13 @@ const char *embedded_cli_get_history(struct embedded_cli *cli,
         if (len == 0)
             return NULL;
         pos += len + 1;
-        if (pos == sizeof(cli->history))
+        if (pos >= sizeof(cli->history))
             return NULL;
     }
 
+    if (cli->history[pos] == '\0')
+        return NULL;
+
     return &cli->history[pos];
 #else
     (void)cli;
@@ -168,7 +171,7 @@ static void embedded_cli_extend_history(struct embedded_cli *cli)
         if (strcmp(cli->buffer, cli->history) == 0)
             return;
         memmove(&cli->history[len + 1], &cli->history[0],
-                sizeof(cli->history) - len + 1);
+                sizeof(cli->history) - (len + 1));
         memcpy(cli->history, cli->buffer, len + 1);
         // Make sure it's always nul terminated
         cli->history[sizeof(cli->history) - 1] = '\0';
@@ -399,7 +402,7 @@ bool embedded_cli_insert_char(struct embedded_cli *cli, char ch)
             cli_putchar(cli, '\n', true);
             break;
         default:
-            if (ch > 0)
+            if (ch < 0 || ch >= 32)
                 embedded_cli_insert_default_char(cli, ch);
         }
     }

embedded_cli.h

@@ -133,7 +133,7 @@ const char *embedded_cli_get_line(const struct embedded_cli *cli);
 
 /**
  * Parses the internal buffer and returns it as an argc/argc combo
- * @return number of values in argv (maximum of EMBEDDED_CLI_MAX_ARGC)
+ * @return number of values in argv (maximum of EMBEDDED_CLI_MAX_ARGC - 1)
  */
 int embedded_cli_argc(struct embedded_cli *cli, char ***argv);
 

examples/posix_demo.c

@@ -95,7 +95,8 @@ int main(void)
             cli_argc = embedded_cli_argc(&cli, &cli_argv);
             printf("Got %d args\n", cli_argc);
             for (int i = 0; i < cli_argc; i++) {
-                printf("Arg %d/%d: '%s'\n", i, cli_argc, cli_argv[i]);
+                printf("Arg %d/%d: [%lu bytes] '%s'\n", i, cli_argc,
+                       strlen(cli_argv[i]), cli_argv[i]);
             }
             done = cli_argc >= 1 && (strcmp(cli_argv[0], "quit") == 0);
 

tests/embedded_cli_test.c

@@ -18,6 +18,7 @@
 #define CTRL_L "\x0c"
 #define CTRL_R "\x12"
 #define CTRL_U "\x15"
+#define CTRL_X "\x18"
 
 static void cli_equals(const struct embedded_cli *cli, const char *line)
 {
@@ -214,6 +215,8 @@ static void test_multiple(void)
         {"abc" LEFT LEFT CTRL_L "\n", "abc"},
         {"abc" CTRL_U "\n", ""},
         {"abc" LEFT LEFT CTRL_U "\n", "bc"},
+        // The check below ensures we ignore unknown control sequences
+        {CTRL_X " " CTRL_X " " CTRL_X "\n", "  "},
         {NULL, NULL},
     };
 
@@ -300,22 +303,36 @@ static void test_max_chars(void)
     TEST_ASSERT(cli.buffer[sizeof(cli.buffer) - 2] == 'f');
 }
 
-TEST_LIST = {
-    {"simple", test_simple},
-    {"argc", test_argc},
-    {"delete", test_delete},
-    {"cursor_left", test_cursor_left},
-    {"cursor_right", test_cursor_right},
+static void test_utf8(void)
+{
+    struct embedded_cli cli;
+    char **argv;
+    embedded_cli_init(&cli, NULL, NULL, NULL);
+    test_insert_line(&cli, "normal ñ ü 中文 text\n");
+    int argc = embedded_cli_argc(&cli, &argv);
+    TEST_ASSERT(argc == 5);
+    TEST_ASSERT(strcmp(argv[0], "normal") == 0);
+    TEST_ASSERT(strcmp(argv[1], "ñ") == 0);
+    TEST_ASSERT(strcmp(argv[2], "ü") == 0);
+    TEST_ASSERT(strcmp(argv[3], "中文") == 0);
+    TEST_ASSERT(strcmp(argv[4], "text") == 0);
+}
+
+TEST_LIST = {{"simple", test_simple},
+             {"argc", test_argc},
+             {"delete", test_delete},
+             {"cursor_left", test_cursor_left},
+             {"cursor_right", test_cursor_right},
 #if EMBEDDED_CLI_HISTORY_LEN
-    {"history", test_history},
-    {"history_keys", test_history_keys},
-    {"search", test_search},
-    {"up_down", test_up_down},
+             {"history", test_history},
+             {"history_keys", test_history_keys},
+             {"search", test_search},
+             {"up_down", test_up_down},
 #endif
-    {"multiple", test_multiple},
-    {"echo", test_echo},
-    {"quotes", test_quotes},
-    {"too_many_args", test_too_many_args},
-    {"max_chars", test_max_chars},
-    {NULL, NULL},
-};
+             {"multiple", test_multiple},
+             {"echo", test_echo},
+             {"quotes", test_quotes},
+             {"too_many_args", test_too_many_args},
+             {"max_chars", test_max_chars},
+             {"utf8", test_utf8},
+             {NULL, NULL}};