Merge #32082 refactor(treesitter): use coroutines for resuming _parse()

runtime/doc/news.txt

@@ -379,6 +379,8 @@ TREESITTER
   activated by passing the `on_parse` callback parameter.
 • |vim.treesitter.query.set()| can now inherit and/or extend runtime file
   queries in addition to overriding.
+• |LanguageTree:is_valid()| now accepts a range parameter to narrow the scope
+  of the validity check.
 
 TUI
 

runtime/doc/treesitter.txt

@@ -1581,7 +1581,8 @@ LanguageTree:invalidate({reload})                  *LanguageTree:invalidate()*
     Parameters: ~
       • {reload}  (`boolean?`)
 
-LanguageTree:is_valid({exclude_children})            *LanguageTree:is_valid()*
+                                                     *LanguageTree:is_valid()*
+LanguageTree:is_valid({exclude_children}, {range})
     Returns whether this LanguageTree is valid, i.e., |LanguageTree:trees()|
     reflects the latest state of the source. If invalid, user should call
     |LanguageTree:parse()|.
@@ -1589,6 +1590,7 @@ LanguageTree:is_valid({exclude_children})            *LanguageTree:is_valid()*
     Parameters: ~
       • {exclude_children}  (`boolean?`) whether to ignore the validity of
                             children (default `false`)
+      • {range}             (`Range?`) range to check for validity
 
     Return: ~
         (`boolean`)

runtime/lua/vim/treesitter/languagetree.lua

@@ -60,6 +60,8 @@ local default_parse_timeout_ms = 3
 ---| 'on_child_added'
 ---| 'on_child_removed'
 
+---@alias ParserThreadState { timeout: integer? }
+
 --- @type table<TSCallbackNameOn,TSCallbackName>
 local TSCallbackNames = {
   on_changedtree = 'changedtree',
@@ -83,6 +85,8 @@ local TSCallbackNames = {
 ---Table of callback queues, keyed by each region for which the callbacks should be run
 ---@field private _cb_queues table<string, fun(err?: string, trees?: table<integer, TSTree>)[]>
 ---@field private _regions table<integer, Range6[]>?
+---The total number of regions. Since _regions can have holes, we cannot simply read this value from #_regions.
+---@field private _num_regions integer
 ---List of regions this tree should manage and parse. If nil then regions are
 ---taken from _trees. This is mostly a short-lived cache for included_regions()
 ---@field private _lang string Language name
@@ -90,7 +94,8 @@ local TSCallbackNames = {
 ---@field private _source (integer|string) Buffer or string to parse
 ---@field private _trees table<integer, TSTree> Reference to parsed tree (one for each language).
 ---Each key is the index of region, which is synced with _regions and _valid.
----@field private _valid boolean|table<integer,boolean> If the parsed tree is valid
+---@field private _valid_regions table<integer,true> Set of valid region IDs.
+---@field private _is_entirely_valid boolean Whether the entire tree (excluding children) is valid.
 ---@field private _logger? fun(logtype: string, msg: string)
 ---@field private _logfile? file*
 local LanguageTree = {}
@@ -132,7 +137,9 @@ function LanguageTree.new(source, lang, opts)
     _injection_query = injections[lang] and query.parse(lang, injections[lang])
       or query.get(lang, 'injections'),
     _injections_processed = false,
-    _valid = false,
+    _valid_regions = {},
+    _num_regions = 1,
+    _is_entirely_valid = false,
     _parser = vim._create_ts_parser(lang),
     _ranges_being_parsed = {},
     _cb_queues = {},
@@ -238,7 +245,8 @@ end
 --- tree in treesitter. Doesn't clear filesystem cache. Called often, so needs to be fast.
 ---@param reload boolean|nil
 function LanguageTree:invalidate(reload)
-  self._valid = false
+  self._valid_regions = {}
+  self._is_entirely_valid = false
   self._parser:reset()
 
   -- buffer was reloaded, reparse all trees
@@ -271,53 +279,6 @@ function LanguageTree:lang()
   return self._lang
 end
 
---- Returns whether this LanguageTree is valid, i.e., |LanguageTree:trees()| reflects the latest
---- state of the source. If invalid, user should call |LanguageTree:parse()|.
----@param exclude_children boolean|nil whether to ignore the validity of children (default `false`)
----@return boolean
-function LanguageTree:is_valid(exclude_children)
-  local valid = self._valid
-
-  if type(valid) == 'table' then
-    for i, _ in pairs(self:included_regions()) do
-      if not valid[i] then
-        return false
-      end
-    end
-  end
-
-  if not exclude_children then
-    if not self._injections_processed then
-      return false
-    end
-
-    for _, child in pairs(self._children) do
-      if not child:is_valid(exclude_children) then
-        return false
-      end
-    end
-  end
-
-  if type(valid) == 'boolean' then
-    return valid
-  end
-
-  self._valid = true
-  return true
-end
-
---- Returns a map of language to child tree.
---- @return table<string,vim.treesitter.LanguageTree>
-function LanguageTree:children()
-  return self._children
-end
-
---- Returns the source content of the language tree (bufnr or string).
---- @return integer|string
-function LanguageTree:source()
-  return self._source
-end
-
 --- @param region Range6[]
 --- @param range? boolean|Range
 --- @return boolean
@@ -343,39 +304,88 @@ local function intercepts_region(region, range)
   return false
 end
 
+--- Returns whether this LanguageTree is valid, i.e., |LanguageTree:trees()| reflects the latest
+--- state of the source. If invalid, user should call |LanguageTree:parse()|.
+---@param exclude_children boolean? whether to ignore the validity of children (default `false`)
+---@param range Range? range to check for validity
+---@return boolean
+function LanguageTree:is_valid(exclude_children, range)
+  local valid_regions = self._valid_regions
+
+  if not self._is_entirely_valid then
+    if not range then
+      return false
+    end
+    -- TODO: Efficiently search for possibly intersecting regions using a binary search
+    for i, region in pairs(self:included_regions()) do
+      if not valid_regions[i] and intercepts_region(region, range) then
+        return false
+      end
+    end
+  end
+
+  if not exclude_children then
+    if not self._injections_processed then
+      return false
+    end
+
+    for _, child in pairs(self._children) do
+      if not child:is_valid(exclude_children, range) then
+        return false
+      end
+    end
+  end
+
+  return true
+end
+
+--- Returns a map of language to child tree.
+--- @return table<string,vim.treesitter.LanguageTree>
+function LanguageTree:children()
+  return self._children
+end
+
+--- Returns the source content of the language tree (bufnr or string).
+--- @return integer|string
+function LanguageTree:source()
+  return self._source
+end
+
 --- @private
 --- @param range boolean|Range?
---- @param timeout integer?
+--- @param thread_state ParserThreadState
 --- @return Range6[] changes
 --- @return integer no_regions_parsed
 --- @return number total_parse_time
 --- @return boolean finished whether async parsing still needs time
-function LanguageTree:_parse_regions(range, timeout)
+function LanguageTree:_parse_regions(range, thread_state)
   local changes = {}
   local no_regions_parsed = 0
   local total_parse_time = 0
 
-  if type(self._valid) ~= 'table' then
-    self._valid = {}
-  end
-
   -- If there are no ranges, set to an empty list
   -- so the included ranges in the parser are cleared.
   for i, ranges in pairs(self:included_regions()) do
     if
-      not self._valid[i]
+      not self._valid_regions[i]
       and (
         intercepts_region(ranges, range)
         or (self._trees[i] and intercepts_region(self._trees[i]:included_ranges(false), range))
       )
     then
       self._parser:set_included_ranges(ranges)
-      self._parser:set_timeout(timeout and timeout * 1000 or 0) -- ms -> micros
+      self._parser:set_timeout(thread_state.timeout and thread_state.timeout * 1000 or 0) -- ms -> micros
+
       local parse_time, tree, tree_changes =
         tcall(self._parser.parse, self._parser, self._trees[i], self._source, true)
+      while true do
+        if tree then
+          break
+        end
+        coroutine.yield(changes, no_regions_parsed, total_parse_time, false)
 
-      if not tree then
+        parse_time, tree, tree_changes =
-        return changes, no_regions_parsed, total_parse_time, false
+          tcall(self._parser.parse, self._parser, self._trees[i], self._source, true)
       end
 
       self:_do_callback('changedtree', tree_changes, tree)
@@ -384,7 +394,13 @@ function LanguageTree:_parse_regions(range, timeout)
 
       total_parse_time = total_parse_time + parse_time
       no_regions_parsed = no_regions_parsed + 1
-      self._valid[i] = true
+      self._valid_regions[i] = true
+
+      -- _valid_regions can have holes, but that is okay because this equality is only true when it
+      -- has no holes (meaning all regions are valid)
+      if #self._valid_regions == self._num_regions then
+        self._is_entirely_valid = true
+      end
     end
   end
 
@@ -476,7 +492,11 @@ function LanguageTree:_async_parse(range, on_parse)
   local ct = is_buffer_parser and buf.changedtick or nil
   local total_parse_time = 0
   local redrawtime = vim.o.redrawtime
-  local timeout = not vim.g._ts_force_sync_parsing and default_parse_timeout_ms or nil
+
+  local thread_state = {} ---@type ParserThreadState
+
+  ---@type fun(): table<integer, TSTree>, boolean
+  local parse = coroutine.wrap(self._parse)
 
   local function step()
     if is_buffer_parser then
@@ -490,10 +510,12 @@ function LanguageTree:_async_parse(range, on_parse)
       if buf.changedtick ~= ct then
         ct = buf.changedtick
         total_parse_time = 0
+        parse = coroutine.wrap(self._parse)
       end
     end
 
-    local parse_time, trees, finished = tcall(self._parse, self, range, timeout)
+    thread_state.timeout = not vim.g._ts_force_sync_parsing and default_parse_timeout_ms or nil
+    local parse_time, trees, finished = tcall(parse, self, range, thread_state)
     total_parse_time = total_parse_time + parse_time
 
     if finished then
@@ -535,17 +557,17 @@ function LanguageTree:parse(range, on_parse)
   if on_parse then
     return self:_async_parse(range, on_parse)
   end
-  local trees, _ = self:_parse(range)
+  local trees, _ = self:_parse(range, {})
   return trees
 end
 
 --- @private
 --- @param range boolean|Range|nil
---- @param timeout integer?
+--- @param thread_state ParserThreadState
 --- @return table<integer, TSTree> trees
 --- @return boolean finished
-function LanguageTree:_parse(range, timeout)
+function LanguageTree:_parse(range, thread_state)
-  if self:is_valid() then
+  if self:is_valid(nil, type(range) == 'table' and range or nil) then
     self:_log('valid')
     return self._trees, true
   end
@@ -558,12 +580,19 @@ function LanguageTree:_parse(range, timeout)
   local total_parse_time = 0
 
   -- At least 1 region is invalid
-  if not self:is_valid(true) then
+  if not self:is_valid(true, type(range) == 'table' and range or nil) then
+    ---@type fun(self: vim.treesitter.LanguageTree, range: boolean|Range?, thread_state: ParserThreadState): Range6[], integer, number, boolean
+    local parse_regions = coroutine.wrap(self._parse_regions)
+    while true do
       local is_finished
-    changes, no_regions_parsed, total_parse_time, is_finished = self:_parse_regions(range, timeout)
+      changes, no_regions_parsed, total_parse_time, is_finished =
-    timeout = timeout and math.max(timeout - total_parse_time, 0)
+        parse_regions(self, range, thread_state)
-    if not is_finished then
+      thread_state.timeout = thread_state.timeout
-      return self._trees, false
+        and math.max(thread_state.timeout - total_parse_time, 0)
+      if is_finished then
+        break
+      end
+      coroutine.yield(self._trees, false)
     end
     -- Need to run injections when we parsed something
     if no_regions_parsed > 0 then
@@ -585,13 +614,20 @@ function LanguageTree:_parse(range, timeout)
   })
 
   for _, child in pairs(self._children) do
-    if timeout == 0 then
+    if thread_state.timeout == 0 then
-      return self._trees, false
+      coroutine.yield(self._trees, false)
     end
-    local ctime, _, child_finished = tcall(child._parse, child, range, timeout)
+
-    timeout = timeout and math.max(timeout - ctime, 0)
+    ---@type fun(): table<integer, TSTree>, boolean
-    if not child_finished then
+    local parse = coroutine.wrap(child._parse)
-      return self._trees, child_finished
+
+    while true do
+      local ctime, _, child_finished = tcall(parse, child, range, thread_state)
+      if child_finished then
+        thread_state.timeout = thread_state.timeout and math.max(thread_state.timeout - ctime, 0)
+        break
+      end
+      coroutine.yield(self._trees, child_finished)
     end
   end
 
@@ -687,38 +723,34 @@ end
 ---region is valid or not.
 ---@param fn fun(index: integer, region: Range6[]): boolean
 function LanguageTree:_iter_regions(fn)
-  if not self._valid then
+  if vim.deep_equal(self._valid_regions, {}) then
     return
   end
 
-  local was_valid = type(self._valid) ~= 'table'
+  if self._is_entirely_valid then
-
+    self:_log('was valid')
-  if was_valid then
-    self:_log('was valid', self._valid)
-    self._valid = {}
   end
 
   local all_valid = true
 
   for i, region in pairs(self:included_regions()) do
-    if was_valid or self._valid[i] then
+    if self._valid_regions[i] then
-      self._valid[i] = fn(i, region)
+      -- Setting this to nil rather than false allows us to determine if all regions were parsed
-      if not self._valid[i] then
+      -- just by checking the length of _valid_regions.
+      self._valid_regions[i] = fn(i, region) and true or nil
+      if not self._valid_regions[i] then
         self:_log(function()
           return 'invalidating region', i, region_tostr(region)
         end)
       end
     end
 
-    if not self._valid[i] then
+    if not self._valid_regions[i] then
       all_valid = false
     end
   end
 
-  -- Compress the valid value to 'true' if there are no invalid regions
+  self._is_entirely_valid = all_valid
-  if all_valid then
-    self._valid = all_valid
-  end
 end
 
 --- Sets the included regions that should be parsed by this |LanguageTree|.
@@ -768,6 +800,7 @@ function LanguageTree:set_included_regions(new_regions)
   end
 
   self._regions = new_regions
+  self._num_regions = #new_regions
 end
 
 ---Gets the set of included regions managed by this LanguageTree. This can be different from the

test/functional/treesitter/parser_spec.lua

@@ -633,7 +633,7 @@ int x = INT_MAX;
         }, get_ranges())
 
         n.feed('7ggI//<esc>')
-        exec_lua([[parser:parse({6, 7})]])
+        exec_lua([[parser:parse({5, 6})]])
         eq('table', exec_lua('return type(parser:children().c)'))
         eq(2, exec_lua('return #parser:children().c:trees()'))
         eq({