Allow `tokenizer` to pause and resume in the middle of a line (#1444)

data/core/doc/highlighter.lua

@@ -19,25 +19,34 @@ function Highlighter:start()
   if self.running then return end
   self.running = true
   core.add_thread(function()
-    while self.first_invalid_line < self.max_wanted_line do
+    while self.first_invalid_line <= self.max_wanted_line do
       local max = math.min(self.first_invalid_line + 40, self.max_wanted_line)
       local retokenized_from
       for i = self.first_invalid_line, max do
         local state = (i > 1) and self.lines[i - 1].state
         local line = self.lines[i]
-        if not (line and line.init_state == state and line.text == self.doc.lines[i]) then
+        if line and line.resume and (line.init_state ~= state or line.text ~= self.doc.lines[i]) then
+          -- Reset the progress if no longer valid
+          line.resume = nil
+        end
+        if not (line and line.init_state == state and line.text == self.doc.lines[i] and not line.resume) then
           retokenized_from = retokenized_from or i
-          self.lines[i] = self:tokenize_line(i, state)
+          self.lines[i] = self:tokenize_line(i, state, line and line.resume)
+          if self.lines[i].resume then
+            self.first_invalid_line = i
+            goto yield
+          end
         elseif retokenized_from then
           self:update_notify(retokenized_from, i - retokenized_from - 1)
           retokenized_from = nil
         end
       end
+
+      self.first_invalid_line = max + 1
+      ::yield::
       if retokenized_from then
         self:update_notify(retokenized_from, max - retokenized_from)
       end
-
-      self.first_invalid_line = max + 1
       core.redraw = true
       coroutine.yield()
     end
@@ -48,7 +57,7 @@ end
 
 local function set_max_wanted_lines(self, amount)
   self.max_wanted_line = amount
-  if self.first_invalid_line < self.max_wanted_line then
+  if self.first_invalid_line <= self.max_wanted_line then
     self:start()
   end
 end
@@ -91,11 +100,11 @@ function Highlighter:update_notify(line, n)
 end
 
 
-function Highlighter:tokenize_line(idx, state)
+function Highlighter:tokenize_line(idx, state, resume)
   local res = {}
   res.init_state = state
   res.text = self.doc.lines[idx]
-  res.tokens, res.state = tokenizer.tokenize(self.doc.syntax, res.text, state)
+  res.tokens, res.state, res.resume = tokenizer.tokenize(self.doc.syntax, res.text, state, resume)
   return res
 end
 

data/core/tokenizer.lua

@@ -1,5 +1,6 @@
 local core = require "core"
 local syntax = require "core.syntax"
+local config = require "core.config"
 
 local tokenizer = {}
 local bad_patterns = {}
@@ -8,7 +9,7 @@ local function push_token(t, type, text)
   type = type or "normal"
   local prev_type = t[#t-1]
   local prev_text = t[#t]
-  if prev_type and (prev_type == type or prev_text:ufind("^%s*$")) then
+  if prev_type and (prev_type == type or (prev_text:ufind("^%s*$") and type ~= "incomplete")) then
     t[#t-1] = type
     t[#t] = prev_text .. text
   else
@@ -128,8 +129,8 @@ end
 ---@param incoming_syntax table
 ---@param text string
 ---@param state string
-function tokenizer.tokenize(incoming_syntax, text, state)
-  local res = {}
+function tokenizer.tokenize(incoming_syntax, text, state, resume)
+  local res
   local i = 1
 
   if #incoming_syntax.patterns == 0 then
@@ -137,6 +138,20 @@ function tokenizer.tokenize(incoming_syntax, text, state)
   end
 
   state = state or string.char(0)
+
+  if resume then
+    res = resume.res
+    -- Remove "incomplete" tokens
+    while res[#res-1] == "incomplete" do
+      table.remove(res, #res)
+      table.remove(res, #res)
+    end
+    i = resume.i
+    state = resume.state
+  end
+
+  res = res or {}
+
   -- incoming_syntax    : the parent syntax of the file.
   -- state              : a string of bytes representing syntax state (see above)
 
@@ -245,7 +260,22 @@ function tokenizer.tokenize(incoming_syntax, text, state)
   end
 
   local text_len = text:ulen()
+  local start_time = system.get_time()
+  local starting_i = i
   while i <= text_len do
+    -- Every 200 chars, check if we're out of time
+    if i - starting_i > 200 then
+      starting_i = i
+      if system.get_time() - start_time > 0.5 / config.fps then
+        -- We're out of time
+        push_token(res, "incomplete", string.usub(text, i))
+        return res, string.char(0), {
+          res = res,
+          i = i,
+          state = state
+        }
+      end
+    end
     -- continue trying to match the end pattern of a pair if we have a state set
     if current_pattern_idx > 0 then
       local p = current_syntax.patterns[current_pattern_idx]