diff options
| author | Karen Arutyunov <karen@codesynthesis.com> | 2020-02-26 17:16:45 +0300 |
|---|---|---|
| committer | Karen Arutyunov <karen@codesynthesis.com> | 2020-02-26 17:17:49 +0300 |
| commit | 5ae9686adac1508873f2d980e84becd3496244c2 (patch) | |
| tree | d7c88e678b29ed6bb7ae30b74bd01aa2b5d2e9a8 /tests/manifest-serializer | |
| parent | afb726d2d59b3715960a8647738860f40e37cf4f (diff) | |
Add notion of validator to char_scanner and make sure manifest is UTF-8
This involves implementing utf8_validator and UTF-8 utility functions and
using them during the manifest parsing, serialization, and rewriting.
Diffstat (limited to 'tests/manifest-serializer')
| -rw-r--r-- | tests/manifest-serializer/driver.cxx | 18 |
1 files changed, 18 insertions, 0 deletions
diff --git a/tests/manifest-serializer/driver.cxx b/tests/manifest-serializer/driver.cxx index 148a281..c818b4a 100644 --- a/tests/manifest-serializer/driver.cxx +++ b/tests/manifest-serializer/driver.cxx @@ -46,6 +46,7 @@ main () assert (test ({{"#", "x"}}, "# x\n")); assert (test ({{"#", "x"},{"#", "y"},{"#", ""}}, "# x\n# y\n#\n")); assert (fail ({{"",""},{"#", "x"}})); // serialization after eos + assert (fail ({{"#", "\xB0"}})); // invalid UTF-8 sequence // Empty manifest stream. // @@ -89,6 +90,12 @@ main () assert (fail ({{"","1"},{"a b",""}})); assert (fail ({{"","1"},{"a\tb",""}})); assert (fail ({{"","1"},{"a\n",""}})); + assert (fail ({{"","1"},{"a\xB0",""}})); // invalid UTF-8 sequence + + // Invalid value. + // + assert (fail ({{"","1"},{"a","\xB0"}})); // invalid UTF-8 sequence + assert (fail ({{"","1"},{"a","\xD0"}})); // incomplete UTF-8 sequence // Simple value. // @@ -172,11 +179,22 @@ main () "xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx\\Y\\\n" "yyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyy"); + // Hard break after the UTF-8/delayed hard break. + // + string l6 ("xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx" + "xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx\xF0\x90\x8C\x82" + "\xF0\x90\x8C\x82yyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyy"); + + string e6 ("xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx" + "xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx\xF0\x90\x8C\x82\\\n" + "\xF0\x90\x8C\x82yyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyy"); + assert (test ({{"","1"},{"a",l1},{"",""},{"",""}}, ": 1\na: " + e1 + "\n")); assert (test ({{"","1"},{"a",l2},{"",""},{"",""}}, ": 1\na: " + e2 + "\n")); assert (test ({{"","1"},{"a",l3},{"",""},{"",""}}, ": 1\na: " + e3 + "\n")); assert (test ({{"","1"},{"a",l4},{"",""},{"",""}}, ": 1\na: " + e4 + "\n")); assert (test ({{"","1"},{"a",l5},{"",""},{"",""}}, ": 1\na: " + e5 + "\n")); + assert (test ({{"","1"},{"a",l6},{"",""},{"",""}}, ": 1\na: " + e6 + "\n")); // Multi-line value. // |