Extract UTF-8 sequence parser into its own function, fix bugs, apply unit tests

1 files changed, 92 insertions, 1 deletions

diff --git a/t/03utf8.c b/t/03utf8.c
index d4d8da3..34fce6f 100644
--- a/t/03utf8.c
+++ b/t/03utf8.c
@@ -7,7 +7,7 @@ int main(int argc, char *argv[])
   TermKey   *tk;
   TermKeyKey key;
 
-  plan_tests(21);
+  plan_tests(57);
 
   pipe(fd);
 
@@ -72,6 +72,97 @@ int main(int argc, char *argv[])
   is_int(key.type,        TERMKEY_TYPE_UNICODE, "key.type UTF-8 4 high");
   is_int(key.code.number, 0x10FFFF,             "key.code.number UTF-8 4 high");
 
+  /* Invalid continuations */
+
+  write(fd[1], "\xC2!", 2);
+
+  termkey_advisereadable(tk);
+  is_int(termkey_getkey(tk, &key), TERMKEY_RES_KEY, "getkey yields RES_KEY UTF-8 2 invalid cont");
+  is_int(key.code.number, 0xFFFD, "key.code.number UTF-8 2 invalid cont");
+  is_int(termkey_getkey(tk, &key), TERMKEY_RES_KEY, "getkey yields RES_KEY UTF-8 2 invalid after");
+  is_int(key.code.number, '!', "key.code.number UTF-8 2 invalid after");
+
+  write(fd[1], "\xE0!", 2);
+
+  termkey_advisereadable(tk);
+  is_int(termkey_getkey(tk, &key), TERMKEY_RES_KEY, "getkey yields RES_KEY UTF-8 3 invalid cont");
+  is_int(key.code.number, 0xFFFD, "key.code.number UTF-8 3 invalid cont");
+  is_int(termkey_getkey(tk, &key), TERMKEY_RES_KEY, "getkey yields RES_KEY UTF-8 3 invalid after");
+  is_int(key.code.number, '!', "key.code.number UTF-8 3 invalid after");
+
+  write(fd[1], "\xE0\xA0!", 3);
+
+  termkey_advisereadable(tk);
+  is_int(termkey_getkey(tk, &key), TERMKEY_RES_KEY, "getkey yields RES_KEY UTF-8 3 invalid cont 2");
+  is_int(key.code.number, 0xFFFD, "key.code.number UTF-8 3 invalid cont 2");
+  is_int(termkey_getkey(tk, &key), TERMKEY_RES_KEY, "getkey yields RES_KEY UTF-8 3 invalid after");
+  is_int(key.code.number, '!', "key.code.number UTF-8 3 invalid after");
+
+  write(fd[1], "\xF0!", 2);
+
+  termkey_advisereadable(tk);
+  is_int(termkey_getkey(tk, &key), TERMKEY_RES_KEY, "getkey yields RES_KEY UTF-8 4 invalid cont");
+  is_int(key.code.number, 0xFFFD, "key.code.number UTF-8 4 invalid cont");
+  is_int(termkey_getkey(tk, &key), TERMKEY_RES_KEY, "getkey yields RES_KEY UTF-8 4 invalid after");
+  is_int(key.code.number, '!', "key.code.number UTF-8 4 invalid after");
+
+  write(fd[1], "\xF0\x90!", 3);
+
+  termkey_advisereadable(tk);
+  is_int(termkey_getkey(tk, &key), TERMKEY_RES_KEY, "getkey yields RES_KEY UTF-8 4 invalid cont 2");
+  is_int(key.code.number, 0xFFFD, "key.code.number UTF-8 4 invalid cont 2");
+  is_int(termkey_getkey(tk, &key), TERMKEY_RES_KEY, "getkey yields RES_KEY UTF-8 4 invalid after");
+  is_int(key.code.number, '!', "key.code.number UTF-8 4 invalid after");
+
+  write(fd[1], "\xF0\x90\x80!", 4);
+
+  termkey_advisereadable(tk);
+  is_int(termkey_getkey(tk, &key), TERMKEY_RES_KEY, "getkey yields RES_KEY UTF-8 4 invalid cont 3");
+  is_int(key.code.number, 0xFFFD, "key.code.number UTF-8 4 invalid cont 3");
+  is_int(termkey_getkey(tk, &key), TERMKEY_RES_KEY, "getkey yields RES_KEY UTF-8 4 invalid after");
+  is_int(key.code.number, '!', "key.code.number UTF-8 4 invalid after");
+
+  /* Partials */
+
+  write(fd[1], "\xC2", 1);
+  termkey_advisereadable(tk);
+  is_int(termkey_getkey(tk, &key), TERMKEY_RES_AGAIN, "getkey yields RES_AGAIN UTF-8 2 partial");
+
+  write(fd[1], "\xA0", 1);
+  termkey_advisereadable(tk);
+  is_int(termkey_getkey(tk, &key), TERMKEY_RES_KEY, "getkey yields RES_KEY UTF-8 2 partial");
+  is_int(key.code.number, 0x00A0, "key.code.number UTF-8 2 partial");
+
+  write(fd[1], "\xE0", 1);
+  termkey_advisereadable(tk);
+  is_int(termkey_getkey(tk, &key), TERMKEY_RES_AGAIN, "getkey yields RES_AGAIN UTF-8 3 partial");
+
+  write(fd[1], "\xA0", 1);
+  termkey_advisereadable(tk);
+  is_int(termkey_getkey(tk, &key), TERMKEY_RES_AGAIN, "getkey yields RES_AGAIN UTF-8 3 partial");
+
+  write(fd[1], "\x80", 1);
+  termkey_advisereadable(tk);
+  is_int(termkey_getkey(tk, &key), TERMKEY_RES_KEY, "getkey yields RES_KEY UTF-8 3 partial");
+  is_int(key.code.number, 0x0800, "key.code.number UTF-8 3 partial");
+
+  write(fd[1], "\xF0", 1);
+  termkey_advisereadable(tk);
+  is_int(termkey_getkey(tk, &key), TERMKEY_RES_AGAIN, "getkey yields RES_AGAIN UTF-8 4 partial");
+
+  write(fd[1], "\x90", 1);
+  termkey_advisereadable(tk);
+  is_int(termkey_getkey(tk, &key), TERMKEY_RES_AGAIN, "getkey yields RES_AGAIN UTF-8 4 partial");
+
+  write(fd[1], "\x80", 1);
+  termkey_advisereadable(tk);
+  is_int(termkey_getkey(tk, &key), TERMKEY_RES_AGAIN, "getkey yields RES_AGAIN UTF-8 4 partial");
+
+  write(fd[1], "\x80", 1);
+  termkey_advisereadable(tk);
+  is_int(termkey_getkey(tk, &key), TERMKEY_RES_KEY, "getkey yields RES_KEY UTF-8 4 partial");
+  is_int(key.code.number, 0x10000, "key.code.number UTF-8 4 partial");
+
   termkey_destroy(tk);
 
   return exit_status();