neovim

commit 5e64d92411737a8fe0ea44faebf986944bfea8ea
parent ba1cc9e10c67e9cf8873ba67b2c0d6b8d5ecd6dd
Author: monkoose <pythonproof@gmail.com>
Date:   Thu, 22 May 2025 13:00:22 +0300

perf(runtime): vim.trim for long only whitespace strings

Diffstat:
M
runtime/lua/vim/shared.lua
 | 
4
+++-
A
test/benchmark/trim_spec.lua
 | 
70
++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
M
test/functional/lua/vim_spec.lua
 | 
4
+++-

3 files changed, 76 insertions(+), 2 deletions(-)
diff --git a/runtime/lua/vim/shared.lua b/runtime/lua/vim/shared.lua
@@ -801,7 +801,9 @@ end
 ---@return string String with whitespace removed from its beginning and end
 function vim.trim(s)
   vim.validate('s', s, 'string')
-  return s:match('^%s*(.*%S)') or ''
+  -- `s:match('^%s*(.*%S)')` is slow for long whitespace strings,
+  -- so we are forced to split it into two parts to prevent this
+  return s:gsub('^%s+', ''):match('^.*%S') or ''
 end
 
 --- Escapes magic chars in |lua-pattern|s.
diff --git a/test/benchmark/trim_spec.lua b/test/benchmark/trim_spec.lua
@@ -0,0 +1,70 @@
+describe('vim.trim()', function()
+  --- @param t number[]
+  local function mean(t)
+    assert(#t > 0)
+    local sum = 0
+    for _, v in ipairs(t) do
+      sum = sum + v
+    end
+    return sum / #t
+  end
+
+  --- @param t number[]
+  local function median(t)
+    local len = #t
+    if len % 2 == 0 then
+      return t[len / 2]
+    end
+    return t[(len + 1) / 2]
+  end
+
+  --- @param f fun(t: number[]): table<number, number|string|table>
+  local function measure(f, input, N)
+    local stats = {} ---@type number[]
+    for _ = 1, N do
+      local tic = vim.uv.hrtime()
+      f(input)
+      local toc = vim.uv.hrtime()
+      stats[#stats + 1] = (toc - tic) / 1000000
+    end
+    table.sort(stats)
+    print(
+      string.format(
+        '\nN: %d, Min: %0.6f ms, Max: %0.6f ms, Median: %0.6f ms, Mean: %0.6f ms',
+        N,
+        math.min(unpack(stats)),
+        math.max(unpack(stats)),
+        median(stats),
+        mean(stats)
+      )
+    )
+  end
+
+  local strings = {
+    ['10000 whitespace characters'] = string.rep(' ', 10000),
+    ['10000 whitespace characters and one non-whitespace at the end'] = string.rep(' ', 10000)
+      .. '0',
+    ['10000 whitespace characters and one non-whitespace at the start'] = '0'
+      .. string.rep(' ', 10000),
+    ['10000 non-whitespace characters'] = string.rep('0', 10000),
+    ['10000 whitespace and one non-whitespace in the middle'] = string.rep(' ', 5000)
+      .. 'a'
+      .. string.rep(' ', 5000),
+    ['10000 whitespace characters surrounded by non-whitespace'] = '0'
+      .. string.rep(' ', 10000)
+      .. '0',
+    ['10000 non-whitespace characters surrounded by whitespace'] = ' '
+      .. string.rep('0', 10000)
+      .. ' ',
+  }
+
+  --- @type string[]
+  local string_names = vim.tbl_keys(strings)
+  table.sort(string_names)
+
+  for _, name in ipairs(string_names) do
+    it(name, function()
+      measure(vim.trim, strings[name], 100)
+    end)
+  end
+end)
diff --git a/test/functional/lua/vim_spec.lua b/test/functional/lua/vim_spec.lua
@@ -719,10 +719,12 @@ describe('lua stdlib', function()
       { ' b  ', 'b' },
       { '\tc', 'c' },
       { 'r\n', 'r' },
+      { '', '' },
+      { ' \t \n', '' },
     }
 
     for _, q in ipairs(trims) do
-      assert(q[2], trim(q[1]))
+      eq(q[2], trim(q[1]))
     end
 
     -- Validates args.