From fa4ef8d7ff12a38a5f1d07e2a0b64d7f92bd1142 Mon Sep 17 00:00:00 2001 From: Matthew Wild Date: Mon, 23 Mar 2015 17:16:54 +0000 Subject: tests: Add UTF-8 validity tests --- tests/test_utf8.lua | 19 +++++++++++++++++++ 1 file changed, 19 insertions(+) create mode 100644 tests/test_utf8.lua (limited to 'tests/test_utf8.lua') diff --git a/tests/test_utf8.lua b/tests/test_utf8.lua new file mode 100644 index 00000000..481eff5d --- /dev/null +++ b/tests/test_utf8.lua @@ -0,0 +1,19 @@ +package.cpath = "../?.so" +package.path = "../?.lua"; + +function valid() + local encodings = require "util.encodings"; + local utf8 = assert(encodings.utf8, "no encodings.utf8 module"); + + for line in io.lines("utf8_sequences.txt") do + local data = line:match(":%s*([^#]+)"):gsub("%s+", ""):gsub("..", function (c) return string.char(tonumber(c, 16)); end) + local expect = line:match("(%S+):"); + if expect ~= "pass" and expect ~= "fail" then + error("unknown expectation: "..line:match("^[^:]+")); + end + local prefix, style = " ", valid_style; + local valid = utf8.valid(data); + assert_equal(valid, utf8.valid(data.." ")); + assert_equal(valid, expect == "pass", line); + end +end -- cgit v1.2.3