refactor(char_classing): use inbuilt vim functions

lua/precognition/horizontal_motions.lua

@@ -69,7 +69,8 @@ function M.end_of_word(str, cursorcol, linelen, big_word)
     local rev_offset
 
     if
-        (c_class == cc.other and next_char_class ~= cc.other) or (next_char_class == cc.other and c_class ~= cc.other)
+        (c_class == cc.punctuation and next_char_class ~= cc.punctuation)
+        or (next_char_class == cc.punctuation and c_class ~= cc.punctuation)
     then
         offset = offset + 1
         char = vim.fn.strcharpart(str, offset - 1, 1)
@@ -132,7 +133,7 @@ function M.prev_word_boundary(str, cursorcol, linelen, big_word)
     while utils.char_class(char, big_word) == c_class and offset >= 0 do
         offset = offset - 1
         char = vim.fn.strcharpart(str, offset - 1, 1)
-        --if remaining string is whitespace, return nil_wrap
+        --if remaining string is whitespace, return 0
         local remaining = string.sub(str, offset)
         if remaining:match("^%s*$") and #remaining > 0 then
             return 0

lua/precognition/utils.lua

@@ -3,28 +3,32 @@ local M = {}
 ---@enum cc
 M.char_classes = {
     whitespace = 0,
-    other = 1,
+    punctuation = 1,
     word = 2,
     emoji = 3,
+    other = "other",
+    UNKNOWN = -1,
 }
 
 ---@param char string
 ---@param big_word boolean
----@return integer
+---@return cc
 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
+
+    if char == "" then
+        return cc.UNKNOWN
     end
-    if char == " " or char == "\t" or char == "\0" then
+
+    if char == "\0" then
         return cc.whitespace
     end
 
-    local c_class = require("precognition.ffi").utf_class(byte)
+    local c_class = vim.fn.charclass(char)
+
     if big_word and c_class ~= 0 then
-        return cc.other
+        return cc.punctuation
     end
 
     return c_class

tests/precognition/char_spec.lua

@@ -6,8 +6,11 @@ local eq = assert.are.same
 describe("static classes", function()
     it("are set correctly", function()
         eq(cc.whitespace, 0)
-        eq(cc.other, 1)
+        eq(cc.punctuation, 1)
         eq(cc.word, 2)
+        eq(cc.emoji, 3)
+        eq(cc.other, "other")
+        eq(cc.UNKNOWN, -1)
     end)
 end)
 
@@ -30,6 +33,17 @@ describe("char classing", function()
         eq(utils.char_class("@", false), 1)
         eq(utils.char_class(".", false), 1)
     end)
+
+    it("can class emoji characters", function()
+        eq(utils.char_class("🐱", false), 3)
+        eq(utils.char_class("😸", false), 3)
+        eq(utils.char_class("💩", false), 3)
+    end)
+
+    it("can class nerdfont characters", function()
+        eq(utils.char_class("", false), 2)
+        eq(utils.char_class("", false), 2)
+    end)
 end)
 
 describe("big_word classing", function()
@@ -51,17 +65,6 @@ describe("big_word classing", function()
         eq(utils.char_class("@", true), 1)
         eq(utils.char_class(".", true), 1)
     end)
-
-    it("can class emoji characters", function()
-        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), 2)
-        eq(utils.char_class("", false), 2)
-    end)
 end)
 
 describe("pad arrays", function()

tests/precognition/horizontal_motions_spec.lua

@@ -109,6 +109,9 @@ describe("boundaries", function()
             local str = "a big.word string"
             eq(3, hm.prev_word_boundary(str, 10, #str, true))
             eq(3, hm.prev_word_boundary(str, 10, #str, true))
+
+            str = "big.word"
+            eq(1, hm.prev_word_boundary(str, 5, #str, true))
         end)
 
         it("can walk string with b", function()