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, 34 insertions, 0 deletions

diff --git a/src/str.cc b/src/str.cc
new file mode 100644
index 0000000..f81617d
--- /dev/null
+++ b/src/str.cc
@@ -0,0 +1,34 @@
+#include "str.hh"
+
+#include <cstddef>
+#include <string_view>
+#include <vector>
+
+namespace str {
+
+void split(std::string_view str, std::vector<std::string_view>& out,
+           char separator, bool keep_empty) {
+  out.clear();
+
+  size_t offset = 0;
+  while (true) {
+    auto next = str.find(separator, offset);
+    if (next == std::string_view::npos) {
+      if (keep_empty || offset < str.size())
+        out.push_back(str.substr(offset));
+      break;
+    }
+    if (keep_empty || offset < next)
+      out.push_back(str.substr(offset, next - offset));
+    offset = next + 1;
+  }
+}
+
+std::vector<std::string_view> split(std::string_view str,
+                                    char separator, bool keep_empty) {
+  std::vector<std::string_view> vec;
+  split(str, vec, separator, keep_empty);
+  return vec;
+}
+
+}  // namespace str