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, 46 insertions, 0 deletions

diff --git a/src/u.cc b/src/u.cc
new file mode 100644
index 0000000..3c06ba8
--- /dev/null
+++ b/src/u.cc
@@ -0,0 +1,46 @@
+#include "u.hh"
+
+#include <cstdint>
+#include <utility>
+
+namespace u {
+
+// These are generated by gen_ugc
+GeneralCategory u6_2_0_lookup_gc(uint32_t code);
+GeneralCategory u8_0_0_lookup_gc(uint32_t code);
+GeneralCategory u10_0_0_lookup_gc(uint32_t code);
+GeneralCategory u11_0_0_lookup_gc(uint32_t code);
+GeneralCategory u12_1_0_lookup_gc(uint32_t code);
+GeneralCategory u13_0_0_lookup_gc(uint32_t code);
+GeneralCategory u14_0_0_lookup_gc(uint32_t code);
+GeneralCategory u15_0_0_lookup_gc(uint32_t code);
+GeneralCategory u15_1_0_lookup_gc(uint32_t code);
+GeneralCategory u16_0_0_lookup_gc(uint32_t code);
+
+GeneralCategory lookup_gc(uint32_t code, Version version) {
+  switch (version) {
+    case Version::u6_2_0:
+      return u6_2_0_lookup_gc(code);
+    case Version::u8_0_0:
+      return u8_0_0_lookup_gc(code);
+    case Version::u10_0_0:
+      return u10_0_0_lookup_gc(code);
+    case Version::u11_0_0:
+      return u11_0_0_lookup_gc(code);
+    case Version::u12_1_0:
+      return u12_1_0_lookup_gc(code);
+    case Version::u13_0_0:
+      return u13_0_0_lookup_gc(code);
+    case Version::u14_0_0:
+      return u14_0_0_lookup_gc(code);
+    case Version::u15_0_0:
+      return u15_0_0_lookup_gc(code);
+    case Version::u15_1_0:
+      return u15_1_0_lookup_gc(code);
+    case Version::u16_0_0:
+      return u16_0_0_lookup_gc(code);
+  }
+  std::unreachable();
+}
+
+}  // namespace u