Add unicode general category lookup

Generate the lookup tables from UnicodeData.txt, do to that,
add gen_ugc, which uses csv, buffers, line, io and other modules
to do the job.

1 files changed, 38 insertions, 0 deletions

diff --git a/test/str.cc b/test/str.cc
new file mode 100644
index 0000000..35d70d7
--- /dev/null
+++ b/test/str.cc
@@ -0,0 +1,38 @@
+#include <gtest/gtest.h>
+
+#include "str.hh"
+
+TEST(str, split) {
+  auto ret = str::split("");
+  EXPECT_EQ(0, ret.size());
+
+  ret = str::split("", ' ', true);
+  ASSERT_EQ(1, ret.size());
+  EXPECT_EQ("", ret[0]);
+
+  ret = str::split(" ");
+  EXPECT_EQ(0, ret.size());
+
+  ret = str::split(" ", ' ', true);
+  ASSERT_EQ(2, ret.size());
+  EXPECT_EQ("", ret[0]);
+  EXPECT_EQ("", ret[1]);
+
+  ret = str::split(" a b ");
+  ASSERT_EQ(2, ret.size());
+  EXPECT_EQ("a", ret[0]);
+  EXPECT_EQ("b", ret[1]);
+
+  ret = str::split(" a b ", ' ', true);
+  ASSERT_EQ(4, ret.size());
+  EXPECT_EQ("", ret[0]);
+  EXPECT_EQ("a", ret[1]);
+  EXPECT_EQ("b", ret[2]);
+  EXPECT_EQ("", ret[3]);
+
+  ret = str::split(" a b", ' ', true);
+  ASSERT_EQ(3, ret.size());
+  EXPECT_EQ("", ret[0]);
+  EXPECT_EQ("a", ret[1]);
+  EXPECT_EQ("b", ret[2]);
+}