Annotate

tests/test_utf8.lua @ 6594:fa6eb6fb4a80

tests: Add UTF-8 validity tests
author Matthew Wild <mwild1@gmail.com>
date Mon, 23 Mar 2015 17:16:54 +0000
child 7515:9e48299799d8
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
6594
fa6eb6fb4a80 tests: Add UTF-8 validity tests
Matthew Wild <mwild1@gmail.com>
parents:
diff changeset
1 package.cpath = "../?.so"
fa6eb6fb4a80 tests: Add UTF-8 validity tests
Matthew Wild <mwild1@gmail.com>
parents:
diff changeset
2 package.path = "../?.lua";
fa6eb6fb4a80 tests: Add UTF-8 validity tests
Matthew Wild <mwild1@gmail.com>
parents:
diff changeset
3
fa6eb6fb4a80 tests: Add UTF-8 validity tests
Matthew Wild <mwild1@gmail.com>
parents:
diff changeset
4 function valid()
fa6eb6fb4a80 tests: Add UTF-8 validity tests
Matthew Wild <mwild1@gmail.com>
parents:
diff changeset
5 local encodings = require "util.encodings";
fa6eb6fb4a80 tests: Add UTF-8 validity tests
Matthew Wild <mwild1@gmail.com>
parents:
diff changeset
6 local utf8 = assert(encodings.utf8, "no encodings.utf8 module");
fa6eb6fb4a80 tests: Add UTF-8 validity tests
Matthew Wild <mwild1@gmail.com>
parents:
diff changeset
7
fa6eb6fb4a80 tests: Add UTF-8 validity tests
Matthew Wild <mwild1@gmail.com>
parents:
diff changeset
8 for line in io.lines("utf8_sequences.txt") do
fa6eb6fb4a80 tests: Add UTF-8 validity tests
Matthew Wild <mwild1@gmail.com>
parents:
diff changeset
9 local data = line:match(":%s*([^#]+)"):gsub("%s+", ""):gsub("..", function (c) return string.char(tonumber(c, 16)); end)
fa6eb6fb4a80 tests: Add UTF-8 validity tests
Matthew Wild <mwild1@gmail.com>
parents:
diff changeset
10 local expect = line:match("(%S+):");
fa6eb6fb4a80 tests: Add UTF-8 validity tests
Matthew Wild <mwild1@gmail.com>
parents:
diff changeset
11 if expect ~= "pass" and expect ~= "fail" then
fa6eb6fb4a80 tests: Add UTF-8 validity tests
Matthew Wild <mwild1@gmail.com>
parents:
diff changeset
12 error("unknown expectation: "..line:match("^[^:]+"));
fa6eb6fb4a80 tests: Add UTF-8 validity tests
Matthew Wild <mwild1@gmail.com>
parents:
diff changeset
13 end
fa6eb6fb4a80 tests: Add UTF-8 validity tests
Matthew Wild <mwild1@gmail.com>
parents:
diff changeset
14 local prefix, style = " ", valid_style;
fa6eb6fb4a80 tests: Add UTF-8 validity tests
Matthew Wild <mwild1@gmail.com>
parents:
diff changeset
15 local valid = utf8.valid(data);
fa6eb6fb4a80 tests: Add UTF-8 validity tests
Matthew Wild <mwild1@gmail.com>
parents:
diff changeset
16 assert_equal(valid, utf8.valid(data.." "));
fa6eb6fb4a80 tests: Add UTF-8 validity tests
Matthew Wild <mwild1@gmail.com>
parents:
diff changeset
17 assert_equal(valid, expect == "pass", line);
fa6eb6fb4a80 tests: Add UTF-8 validity tests
Matthew Wild <mwild1@gmail.com>
parents:
diff changeset
18 end
fa6eb6fb4a80 tests: Add UTF-8 validity tests
Matthew Wild <mwild1@gmail.com>
parents:
diff changeset
19 end