From 3c1889fbd0d7b96aae16f3479ac8aae70a7e15f7 Mon Sep 17 00:00:00 2001
From: Florent Le Coz <louiz@louiz.org>
Date: Wed, 28 Oct 2015 19:13:53 +0100
Subject: Use Catch for our test suite

`make check` is also added to compile and run the tests
Catch is fetched with cmake automatically into the build directory when needed
---
 tests/encoding.cpp | 56 ++++++++++++++++++++++++++++++++++++++++++++++++++++++
 1 file changed, 56 insertions(+)
 create mode 100644 tests/encoding.cpp

(limited to 'tests/encoding.cpp')

diff --git a/tests/encoding.cpp b/tests/encoding.cpp
new file mode 100644
index 0000000..389cf23
--- /dev/null
+++ b/tests/encoding.cpp
@@ -0,0 +1,56 @@
+#include "catch.hpp"
+
+#include <utils/encoding.hpp>
+
+
+TEST_CASE("UTF-8 validation")
+{
+  const char* valid = "C̡͔͕̩͙̽ͫ̈́ͥ̿̆ͧ̚r̸̩̘͍̻͖̆͆͛͊̉̕͡o͇͈̳̤̱̊̈͢q̻͍̦̮͕ͥͬͬ̽ͭ͌̾ͅǔ͉͕͇͚̙͉̭͉̇̽ȇ͈̮̼͍͔ͣ͊͞͝ͅ ͫ̾ͪ̓ͥ̆̋̔҉̢̦̠͈͔̖̲̯̦ụ̶̯͐̃̋ͮ͆͝n̬̱̭͇̻̱̰̖̤̏͛̏̿̑͟ë́͐҉̸̥̪͕̹̻̙͉̰ ̹̼̱̦̥ͩ͑̈́͑͝ͅt͍̥͈̹̝ͣ̃̔̈̔ͧ̕͝ḙ̸̖̟̙͙ͪ͢ų̯̞̼̲͓̻̞͛̃̀́b̮̰̗̩̰̊̆͗̾̎̆ͯ͌͝.̗̙͎̦ͫ̈́ͥ͌̈̓ͬ";
+  CHECK(utils::is_valid_utf8(valid));
+  CHECK_FALSE(utils::is_valid_utf8("\xF0\x0F"));
+  CHECK_FALSE(utils::is_valid_utf8("\xFE\xFE\xFF\xFF"));
+
+  std::string in = "Biboumi ╯°□°）╯︵ ┻━┻";
+  INFO(in);
+  CHECK(utils::is_valid_utf8(in.data()));
+}
+
+TEST_CASE("UTF-8 conversion")
+{
+  std::string in = "Biboumi ╯°□°）╯︵ ┻━┻";
+  REQUIRE(utils::is_valid_utf8(in.data()));
+
+  SECTION("Converting UTF-8 to UTF-8 should return the same string")
+    {
+      std::string res = utils::convert_to_utf8(in, "UTF-8");
+      CHECK(utils::is_valid_utf8(res.c_str()) == true);
+      CHECK(res == in);
+    }
+
+  SECTION("Testing latin-1 conversion")
+    {
+      std::string original_utf8("couc¥ou");
+      std::string original_latin1("couc\xa5ou");
+
+      SECTION("Convert proper latin-1 to UTF-8")
+        {
+          std::string from_latin1 = utils::convert_to_utf8(original_latin1.c_str(), "ISO-8859-1");
+          CHECK(from_latin1 == original_utf8);
+        }
+      SECTION("Check the behaviour when the decoding fails (here because we provide a wrong charset)")
+        {
+          std::string from_ascii = utils::convert_to_utf8(original_latin1, "US-ASCII");
+          CHECK(from_ascii == "couc�ou");
+        }
+    }
+}
+
+TEST_CASE("Remove invalid XML chars")
+{
+  std::string without_ctrl_char("𤭢€¢$");
+  std::string in = "Biboumi ╯°□°）╯︵ ┻━┻";
+  INFO(in);
+  CHECK(utils::remove_invalid_xml_chars(without_ctrl_char) == without_ctrl_char);
+  CHECK(utils::remove_invalid_xml_chars(in) == in);
+  CHECK(utils::remove_invalid_xml_chars("\acouco\u0008u\uFFFEt\uFFFFe\r\n♥") == "coucoute\r\n♥");
+}
-- 
cgit v1.2.3