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, 49 insertions, 0 deletions

diff --git a/src/ugc.hh b/src/ugc.hh
new file mode 100644
index 0000000..c49d50f
--- /dev/null
+++ b/src/ugc.hh
@@ -0,0 +1,49 @@
+#ifndef UGC_HH
+#define UGC_HH
+
+#include <cstdint>
+
+namespace u {
+
+enum class GeneralCategory : uint8_t {
+  LETTER_UPPERCASE,
+  LETTER_LOWERCASE,
+  LETTER_TITLECASE,
+  LETTER_MODIFIER,
+  LETTER_OTHER,
+
+  MARK_NONSPACING,
+  MARK_SPACING_COMBINDING,
+  MARK_SPACING_ENCLOSING,
+
+  NUMBER_DIGIT,
+  NUMBER_LETTER,
+  NUMBER_OTHER,
+
+  PUNCTUATION_CONNECTOR,
+  PUNCTUATION_DASH,
+  PUNCTUATION_OPEN,
+  PUNCTUATION_CLOSE,
+  PUNCTUATION_INITIAL_QUOTE,
+  PUNCTUATION_FINAL_QUOTE,
+  PUNCTUATION_OTHER,
+
+  SYMBOL_MATH,
+  SYMBOL_CURRENCY,
+  SYMBOL_MODIFIER,
+  SYMBOL_OTHER,
+
+  SEPARATOR_SPACE,
+  SEPARATOR_LINE,
+  SEPARATOR_PARAGRAPH,
+
+  OTHER_CONTROL,
+  OTHER_FORMAT,
+  OTHER_SURROGATE,
+  OTHER_PRIVATE_USE,
+  OTHER_UNASSIGNED,
+};
+
+}  // namespace u
+
+#endif  // UGC_HH