feat: ffi for char classing

lua/precognition/ffi.lua

@@ -0,0 +1,29 @@
+local M = {}
+
+---@return ffi.namespace*
+function M.load()
+    if not _G.precog_C then
+        local ffi = require("ffi")
+        local ok, err = pcall(
+            ffi.cdef,
+            [[
+                  int utf_class(const int c);
+            ]]
+        )
+        ---@diagnostic disable-next-line: need-check-nil
+        if not ok then
+            error(err)
+        end
+        _G.precog_C = ffi.C
+    end
+    return _G.precog_C
+end
+
+return setmetatable(M, {
+    __index = function(_, key)
+        return M.load()[key]
+    end,
+    __newindex = function(_, k, v)
+        M.load()[k] = v
+    end,
+})

lua/precognition/utils.lua

@@ -5,6 +5,7 @@ M.char_classes = {
     whitespace = 0,
     other = 1,
     word = 2,
+    emoji = 3,
 }
 
 ---@param char string
@@ -14,18 +15,19 @@ function M.char_class(char, big_word)
     assert(type(big_word) == "boolean", "big_word must be a boolean")
     local cc = M.char_classes
     local byte = string.byte(char)
+    if byte == nil then
+        return cc.other
+    end
+    if char == " " or char == "\t" or char == "\0" then
+        return cc.whitespace
+    end
 
-    if byte and byte < 0x100 then
-        if char == " " or char == "\t" or char == "\0" then
-            return cc.whitespace
-        end
-        if char == "_" or char:match("%w") then
-            return big_word and cc.other or cc.word
-        end
+    local c_class = require("precognition.ffi").utf_class(byte)
+    if big_word and c_class ~= 0 then
         return cc.other
     end
 
-    return cc.other -- scary unicode edge cases go here
+    return c_class
 end
 
 ---@param bufnr? integer

tests/precognition/char_spec.lua

@@ -53,13 +53,14 @@ describe("big_word classing", function()
     end)
 
     it("can class emoji characters", function()
-        eq(utils.char_class("🐱", false), 1)
-        eq(utils.char_class("😸", false), 1)
+        eq(utils.char_class("🐱", false), 2)
+        eq(utils.char_class("😸", false), 2)
+        eq(utils.char_class("💩", false), 2)
     end)
 
     it("can class nerdfont characters", function()
-        eq(utils.char_class("", false), 1)
-        eq(utils.char_class("", false), 1)
+        eq(utils.char_class("", false), 2)
+        eq(utils.char_class("", false), 2)
     end)
 end)