refactor(treesitter): add Range type aliase for Range4|Range6

runtime/doc/news.txt

@@ -50,6 +50,11 @@ The following changes may require adaptations in user config or plugins.
 • Unsaved changes are now preserved rather than discarded when |channel-stdio|
   is closed.
 
+• Changes to |vim.treesitter.get_node_text()|:
+  - It now returns `string`, as opposed to `string|string[]|nil`.
+  - The `concat` option has been removed as it was not consistently applied.
+  - Invalid ranges now cause an error instead of returning `nil`.
+
 ==============================================================================
 NEW FEATURES                                                    *news-features*
 

runtime/doc/treesitter.txt

@@ -898,7 +898,8 @@ Query:iter_captures({self}, {node}, {source}, {start}, {stop})
       • {self}
 
     Return: ~
-        (fun(): integer, TSNode, TSMetadata ): capture id, capture node, metadata
+        (fun(): integer, TSNode, TSMetadata): capture id, capture node,
+        metadata
 
                                                         *Query:iter_matches()*
 Query:iter_matches({self}, {node}, {source}, {start}, {stop})

runtime/lua/vim/treesitter/_range.lua

@@ -16,6 +16,8 @@ local M = {}
 ---@field [5] integer end column
 ---@field [6] integer end bytes
 
+---@alias Range Range4|Range6
+
 ---@private
 ---@param a_row integer
 ---@param a_col integer
@@ -85,8 +87,8 @@ function M.validate(r)
 end
 
 ---@private
----@param r1 Range4|Range6
+---@param r1 Range
----@param r2 Range4|Range6
+---@param r2 Range
 ---@return boolean
 function M.intercepts(r1, r2)
   local srow_1, scol_1, erow_1, ecol_1 = M.unpack4(r1)
@@ -106,7 +108,7 @@ function M.intercepts(r1, r2)
 end
 
 ---@private
----@param r Range4|Range6
+---@param r Range
 ---@return integer, integer, integer, integer
 function M.unpack4(r)
   local off_1 = #r == 6 and 1 or 0
@@ -121,8 +123,8 @@ function M.unpack6(r)
 end
 
 ---@private
----@param r1 Range4|Range6
+---@param r1 Range
----@param r2 Range4|Range6
+---@param r2 Range
 ---@return boolean whether r1 contains r2
 function M.contains(r1, r2)
   local srow_1, scol_1, erow_1, ecol_1 = M.unpack4(r1)
@@ -143,7 +145,7 @@ end
 
 ---@private
 ---@param source integer|string
----@param range Range4|Range6
+---@param range Range
 ---@return Range6
 function M.add_bytes(source, range)
   if type(range) == 'table' and #range == 6 then

runtime/lua/vim/treesitter/languagetree.lua

@@ -484,7 +484,6 @@ function LanguageTree:included_regions()
 end
 
 ---@private
---- TODO(lewis6991): cleanup of the node_range interface
 ---@param node TSNode
 ---@param source string|integer
 ---@param metadata TSMetadata
@@ -530,7 +529,7 @@ end
 ---@param pattern integer
 ---@param lang string
 ---@param combined boolean
----@param ranges Range4[]
+---@param ranges Range6[]
 local function add_injection(t, tree_index, pattern, lang, combined, ranges)
   assert(type(lang) == 'string')
 
@@ -558,13 +557,11 @@ end
 --- https://tree-sitter.github.io/tree-sitter/syntax-highlighting#language-injection
 ---@param match table<integer,TSNode>
 ---@param metadata TSMetadata
----@return string, boolean, Range4[]
+---@return string?, boolean, Range6[]
 function LanguageTree:_get_injection(match, metadata)
-  local ranges = {} ---@type Range4[]
+  local ranges = {} ---@type Range6[]
   local combined = metadata['injection.combined'] ~= nil
-  local lang = metadata['injection.language']
+  local lang = metadata['injection.language'] --[[@as string?]]
-  assert(type(lang) == 'string')
-
   local include_children = metadata['injection.include-children'] ~= nil
 
   for id, node in pairs(match) do
@@ -584,7 +581,7 @@ end
 ---@private
 ---@param match table<integer,TSNode>
 ---@param metadata TSMetadata
----@return string, boolean, Range4[]
+---@return string, boolean, Range6[]
 function LanguageTree:_get_injection_deprecated(match, metadata)
   local lang = nil ---@type string
   local ranges = {} ---@type Range6[]
@@ -910,7 +907,7 @@ end
 
 ---@private
 ---@param tree TSTree
----@param range Range4
+---@param range Range
 ---@return boolean
 local function tree_contains(tree, range)
   return Range.contains({ tree:root():range() }, range)

runtime/lua/vim/treesitter/query.lua

@@ -59,10 +59,18 @@ end
 
 ---@private
 ---@param buf integer
----@param range Range6
+---@param range Range
 ---@returns string
 local function buf_range_get_text(buf, range)
   local start_row, start_col, end_row, end_col = Range.unpack4(range)
+  if end_col == 0 then
+    if start_row == end_row then
+      start_col = -1
+      start_row = start_row - 1
+    end
+    end_col = -1
+    end_row = end_row - 1
+  end
   local lines = a.nvim_buf_get_text(buf, start_row, start_col, end_row, end_col, {})
   return table.concat(lines, '\n')
 end
@@ -396,7 +404,7 @@ local predicate_handlers = {
 predicate_handlers['vim-match?'] = predicate_handlers['match?']
 
 ---@class TSMetadata
----@field range Range4|Range6
+---@field range Range
 ---@field [integer] TSMetadata
 ---@field [string] integer|string
 

scripts/lua2dox.lua

@@ -302,7 +302,7 @@ local types = { 'integer', 'number', 'string', 'table', 'list', 'boolean', 'func
 local tagged_types = { 'TSNode', 'LanguageTree' }
 
 -- Document these as 'table'
-local alias_types = { 'Range4', 'Range6' }
+local alias_types = { 'Range', 'Range4', 'Range6', 'TSMetadata' }
 
 -- Processes the file and writes filtered output to stdout.
 function TLua2DoX_filter.filter(this, AppStamp, Filename)

test/functional/treesitter/parser_spec.lua

@@ -196,7 +196,7 @@ void ui_refresh(void)
     local manyruns = q(100)
 
     -- First run should be at least 400x slower than an 100 subsequent runs.
-    local factor = is_os('win') and 300 or 400
+    local factor = is_os('win') and 200 or 400
     assert(factor * manyruns < firstrun, ('firstrun: %f ms, manyruns: %f ms'):format(firstrun / 1e6, manyruns / 1e6))
   end)
 
@@ -277,13 +277,13 @@ void ui_refresh(void)
     eq('void', res2)
   end)
 
-  it('support getting text where start of node is past EOF', function()
+  it('support getting text where start of node is one past EOF', function()
     local text = [[
 def run
   a = <<~E
 end]]
     insert(text)
-    local result = exec_lua([[
+    eq('', exec_lua[[
       local fake_node = {}
       function fake_node:start()
         return 3, 0, 23
@@ -291,12 +291,14 @@ end]]
       function fake_node:end_()
         return 3, 0, 23
       end
-      function fake_node:range()
+      function fake_node:range(bytes)
+        if bytes then
+          return 3, 0, 23, 3, 0, 23
+        end
         return 3, 0, 3, 0
       end
-      return vim.treesitter.get_node_text(fake_node, 0) == nil
+      return vim.treesitter.get_node_text(fake_node, 0)
     ]])
-    eq(true, result)
   end)
 
   it('support getting empty text if node range is zero width', function()