perf(blame): better cache invalidation

The blame cache is now maintained in the CacheEntry object
and invalidated incrementally on buffer updates.

In addition git-blame is bypassed if the cursor line is within a hunk.

lua/gitsigns/actions.lua

@@ -288,7 +288,7 @@ M.stage_hunk = mk_repeatable(async.void(function(range, opts)
 
   table.insert(bcache.staged_diffs, hunk)
 
-  bcache:invalidate()
+  bcache:invalidate(true)
   update(bufnr)
 end))
 
@@ -371,7 +371,7 @@ M.undo_stage_hunk = async.void(function()
   end
 
   bcache.git_obj:stage_hunks({ hunk }, true)
-  bcache:invalidate()
+  bcache:invalidate(true)
   update(bufnr)
 end)
 
@@ -389,7 +389,7 @@ M.stage_buffer = async.void(function()
 
   -- Only process files with existing hunks
   local hunks = bcache.hunks
-  if #hunks == 0 then
+  if not hunks or #hunks == 0 then
     print('No unstaged changes in file to stage')
     return
   end
@@ -405,7 +405,7 @@ M.stage_buffer = async.void(function()
     table.insert(bcache.staged_diffs, hunk)
   end
 
-  bcache:invalidate()
+  bcache:invalidate(true)
   update(bufnr)
 end)
 
@@ -432,7 +432,7 @@ M.reset_buffer_index = async.void(function()
 
   bcache.git_obj:unstage_file()
 
-  bcache:invalidate()
+  bcache:invalidate(true)
   update(bufnr)
 end)
 
@@ -893,17 +893,16 @@ M.blame_line = async.void(function(opts)
   end, 1000)
 
   async.scheduler_if_buf_valid()
-  local buftext = util.buf_lines(bufnr)
   local fileformat = vim.bo[bufnr].fileformat
   local lnum = api.nvim_win_get_cursor(0)[1]
-  local results = bcache.git_obj:run_blame(buftext, lnum, opts.ignore_whitespace)
+  local result = bcache:get_blame(lnum, opts)
   pcall(function()
     loading:close()
   end)
 
-  assert(results and results[lnum])
+  assert(result)
 
-  local result = util.convert_blame_info(results[lnum])
+  result = util.convert_blame_info(result)
 
   local is_committed = result.sha and tonumber('0x' .. result.sha) ~= 0
 
@@ -934,10 +933,12 @@ C.blame_line = function(args, _)
   M.blame_line(args)
 end
 
-local function update_buf_base(buf, bcache, base)
+---@param bcache Gitsigns.CacheEntry
+---@param base string?
+local function update_buf_base(bcache, base)
   bcache.base = base
-  bcache:invalidate()
+  bcache:invalidate(true)
-  update(buf)
+  update(bcache.bufnr)
 end
 
 --- Change the base revision to diff against. If {base} is not
@@ -978,8 +979,8 @@ M.change_base = async.void(function(base, global)
   if global then
     config.base = base
 
-    for bufnr, bcache in pairs(cache) do
+    for _, bcache in pairs(cache) do
-      update_buf_base(bufnr, bcache, base)
+      update_buf_base(bcache, base)
     end
   else
     local bufnr = current_buf()
@@ -988,7 +989,7 @@ M.change_base = async.void(function(base, global)
       return
     end
 
-    update_buf_base(bufnr, bcache, base)
+    update_buf_base(bcache, base)
   end
 end)
 
@@ -1317,7 +1318,7 @@ M.refresh = async.void(function()
   require('gitsigns.highlight').setup_highlights()
   require('gitsigns.current_line_blame').setup()
   for k, v in pairs(cache) do
-    v:invalidate()
+    v:invalidate(true)
     manager.update(k)
   end
 end)

lua/gitsigns/attach.lua

@@ -10,7 +10,6 @@ local hl = require('gitsigns.highlight')
 
 local gs_cache = require('gitsigns.cache')
 local cache = gs_cache.cache
-local CacheEntry = gs_cache.CacheEntry
 local Status = require('gitsigns.status')
 
 local gs_config = require('gitsigns.config')
@@ -111,6 +110,7 @@ end
 --- @param bufnr integer
 local function on_reload(_, bufnr)
   local __FUNC__ = 'on_reload'
+  cache[bufnr]:invalidate()
   dprint('Reload')
   manager.update_debounced(bufnr)
 end
@@ -336,7 +336,7 @@ local attach_throttled = throttle_by_id(function(cbuf, ctx, aucmd)
     return
   end
 
-  cache[cbuf] = CacheEntry.new({
+  cache[cbuf] = gs_cache.new({
     bufnr = cbuf,
     base = ctx and ctx.base or config.base,
     file = file,

lua/gitsigns/cache.lua

@@ -1,26 +1,27 @@
+local async = require('gitsigns.async')
 local config = require('gitsigns.config').config
+local util = require('gitsigns.util')
 
 local M = {
   CacheEntry = {},
 }
 
--- Timer object watching the gitdir
+--- @class (exact) Gitsigns.CacheEntry
-
---- @class Gitsigns.CacheEntry
 --- @field bufnr              integer
 --- @field file               string
 --- @field base?              string
 --- @field compare_text?      string[]
---- @field hunks              Gitsigns.Hunk.Hunk[]
+--- @field hunks?             Gitsigns.Hunk.Hunk[]
 --- @field force_next_update? boolean
 ---
 --- @field compare_text_head? string[]
 --- @field hunks_staged?      Gitsigns.Hunk.Hunk[]
 ---
---- @field staged_diffs       Gitsigns.Hunk.Hunk[]
+--- @field staged_diffs?      Gitsigns.Hunk.Hunk[]
 --- @field gitdir_watcher?    uv.uv_fs_event_t
 --- @field git_obj            Gitsigns.GitObj
 --- @field commit?            string
+--- @field blame?             table<integer,Gitsigns.BlameInfo?>
 local CacheEntry = M.CacheEntry
 
 function CacheEntry:get_compare_rev(base)
@@ -47,20 +48,95 @@ function CacheEntry:get_rev_bufname(rev)
   return string.format('gitsigns://%s/%s:%s', self.git_obj.repo.gitdir, rev, self.git_obj.relpath)
 end
 
-function CacheEntry:invalidate()
+--- Invalidate any state dependent on the buffer content.
-  self.compare_text = nil
+--- If 'all' is passed, then invalidate everything.
-  self.compare_text_head = nil
+--- @param all? boolean
+function CacheEntry:invalidate(all)
   self.hunks = nil
   self.hunks_staged = nil
+  self.blame = nil
+  if all then
+    -- The below doesn't need to be invalidated
+    -- if the buffer changes
+    self.compare_text = nil
+    self.compare_text_head = nil
+  end
 end
 
 --- @param o Gitsigns.CacheEntry
 --- @return Gitsigns.CacheEntry
-function CacheEntry.new(o)
+function M.new(o)
   o.staged_diffs = o.staged_diffs or {}
   return setmetatable(o, { __index = CacheEntry })
 end
 
+local sleep = async.wrap(function(duration, cb)
+  vim.defer_fn(cb, duration)
+end, 2)
+
+--- @private
+function CacheEntry:wait_for_hunks()
+  local loop_protect = 0
+  while not self.hunks and loop_protect < 10 do
+    loop_protect = loop_protect + 1
+    sleep(100)
+  end
+end
+
+--- @private
+--- @param opts Gitsigns.CurrentLineBlameOpts
+--- @return table<integer,Gitsigns.BlameInfo?>?
+function CacheEntry:run_blame(opts)
+  local blame_cache --- @type table<integer,Gitsigns.BlameInfo?>?
+  repeat
+    local buftext = util.buf_lines(self.bufnr)
+    local tick = vim.b[self.bufnr].changedtick
+    -- TODO(lewis6991): Cancel blame on changedtick
+    blame_cache = self.git_obj:run_blame(buftext, nil, opts.ignore_whitespace)
+    async.scheduler_if_buf_valid(self.bufnr)
+  until vim.b[self.bufnr].changedtick == tick
+  return blame_cache
+end
+
+--- @param file string
+--- @param lnum integer
+--- @return Gitsigns.BlameInfo
+local function get_blame_nc(file, lnum)
+  local Git = require('gitsigns.git')
+
+  return {
+    orig_lnum = 0,
+    final_lnum = lnum,
+    commit = Git.not_commited(file),
+    filename = file,
+  }
+end
+
+--- @param lnum integer
+--- @param opts Gitsigns.CurrentLineBlameOpts
+--- @return Gitsigns.BlameInfo?
+function CacheEntry:get_blame(lnum, opts)
+  local blame_cache = self.blame
+
+  if not blame_cache or not blame_cache[lnum] then
+    self:wait_for_hunks()
+    local Hunks = require('gitsigns.hunks')
+    if Hunks.find_hunk(lnum, self.hunks) then
+      --- Bypass running blame (which can be expensive) if we know lnum is in a hunk
+      blame_cache = blame_cache or {}
+      blame_cache[lnum] = get_blame_nc(self.git_obj.relpath, lnum)
+    else
+      -- Refresh cache
+      blame_cache = self:run_blame(opts)
+    end
+    self.blame = blame_cache
+  end
+
+  if blame_cache then
+    return blame_cache[lnum]
+  end
+end
+
 function CacheEntry:destroy()
   local w = self.gitdir_watcher
   if w and not w:is_closing() then

lua/gitsigns/config.lua

@@ -70,7 +70,6 @@
 --- @field trouble boolean
 --- -- Undocumented
 --- @field _refresh_staged_on_update boolean
---- @field _blame_cache boolean
 --- @field _threaded_diff boolean
 --- @field _inline2 boolean
 --- @field _extmark_signs boolean
@@ -760,14 +759,6 @@ M.schema = {
     ]],
   },
 
-  _blame_cache = {
-    type = 'boolean',
-    default = true,
-    description = [[
-      Cache blame results for current_line_blame
-    ]],
-  },
-
   _threaded_diff = {
     type = 'boolean',
     default = true,

lua/gitsigns/current_line_blame.lua

@@ -20,13 +20,6 @@ local function reset(bufnr)
   vim.b[bufnr].gitsigns_blame_line_dict = nil
 end
 
---- @class (exact) Gitsigns.BlameCache
---- @field cache Gitsigns.BlameInfo[]?
---- @field tick integer
-
---- @type table<integer,Gitsigns.BlameCache>
-local blame_cache = {}
-
 --- @param fmt string
 --- @param name string
 --- @param info Gitsigns.BlameInfoPublic
@@ -48,36 +41,6 @@ local function flatten_virt_text(virt_text)
   return table.concat(res)
 end
 
---- @param bufnr integer
---- @param lnum integer
---- @param opts Gitsigns.CurrentLineBlameOpts
---- @return Gitsigns.BlameInfo?
-local function run_blame(bufnr, lnum, opts)
-  -- init and invalidate
-  local tick = vim.b[bufnr].changedtick
-  if not blame_cache[bufnr] or blame_cache[bufnr].tick ~= tick then
-    blame_cache[bufnr] = { tick = tick }
-  end
-
-  local result = blame_cache[bufnr].cache
-
-  if result then
-    return result[lnum]
-  end
-
-  local buftext = util.buf_lines(bufnr)
-  local bcache = cache[bufnr]
-  result = bcache.git_obj:run_blame(buftext, nil, opts.ignore_whitespace)
-
-  if not result then
-    return
-  end
-
-  blame_cache[bufnr].cache = result
-
-  return result[lnum]
-end
-
 --- @param winid integer
 --- @return integer
 local function win_width(winid)
@@ -203,14 +166,12 @@ local function update0(bufnr)
 
   local opts = config.current_line_blame_opts
 
-  local blame_info = run_blame(bufnr, lnum, opts)
+  local blame_info = bcache:get_blame(lnum, opts)
 
   if not blame_info then
     return
   end
 
-  async.scheduler_if_buf_valid(bufnr)
-
   if lnum ~= get_lnum(winid) then
     -- Cursor has moved during events; abort and tr-trigger another update
     update0(bufnr)

lua/gitsigns/git.lua

@@ -8,8 +8,6 @@ local subprocess = require('gitsigns.subprocess')
 local gs_config = require('gitsigns.config')
 local config = gs_config.config
 
-local gs_hunks = require('gitsigns.hunks')
-
 local uv = vim.loop
 local startswith = vim.startswith
 
@@ -569,11 +567,30 @@ end
 
 local NOT_COMMITTED = {
   author = 'Not Committed Yet',
-  ['author_mail'] = '<not.committed.yet>',
+  author_mail = '<not.committed.yet>',
   committer = 'Not Committed Yet',
-  ['committer_mail'] = '<not.committed.yet>',
+  committer_mail = '<not.committed.yet>',
 }
 
+--- @param file string
+--- @return Gitsigns.CommitInfo
+function M.not_commited(file)
+  local time = os.time()
+  return {
+    sha = string.rep('0', 40),
+    abbrev_sha = string.rep('0', 8),
+    author = 'Not Committed Yet',
+    author_mail = '<not.committed.yet>',
+    author_tz = '+0000',
+    author_time = time,
+    committer = 'Not Committed Yet',
+    committer_time = time,
+    committer_mail = '<not.committed.yet>',
+    committer_tz = '+0000',
+    summary = 'Version of ' .. file,
+  }
+end
+
 ---@param x any
 ---@return integer
 local function asinteger(x)
@@ -585,11 +602,22 @@ end
 --- @param ignore_whitespace? boolean
 --- @return table<integer,Gitsigns.BlameInfo?>?
 function Obj:run_blame(lines, lnum, ignore_whitespace)
+  local ret = {} --- @type table<integer,Gitsigns.BlameInfo>
+
   if not self.object_name or self.repo.abbrev_head == '' then
     -- As we support attaching to untracked files we need to return something if
     -- the file isn't isn't tracked in git.
     -- If abbrev_head is empty, then assume the repo has no commits
-    return NOT_COMMITTED
+    local commit = M.not_commited(self.file)
+    for i in ipairs(lines) do
+      ret[i] = {
+        orig_lnum = 0,
+        final_lnum = i,
+        commit = commit,
+        filename = self.file,
+      }
+    end
+    return ret
   end
 
   local args = { 'blame', '--contents', '-', '--incremental' }
@@ -619,7 +647,6 @@ function Obj:run_blame(lines, lnum, ignore_whitespace)
     return
   end
 
-  local ret = {} --- @type Gitsigns.BlameInfo[]
   local commits = {} --- @type table<string,Gitsigns.CommitInfo>
   local i = 1
 
@@ -660,6 +687,9 @@ function Obj:run_blame(lines, lnum, ignore_whitespace)
         local l = get()
         local key, value = l:match('^([^%s]+) (.*)')
         if key then
+          if vim.endswith(key, '_time') then
+            value = tonumber(value)
+          end
           key = key:gsub('%-', '_') --- @type string
           commit[key] = value
         else
@@ -749,6 +779,8 @@ end
 function Obj.stage_hunks(self, hunks, invert)
   ensure_file_in_index(self)
 
+  local gs_hunks = require('gitsigns.hunks')
+
   local patch = gs_hunks.create_patch(self.relpath, hunks, self.mode_bits, invert)
 
   if not self.i_crlf and self.w_crlf then

lua/gitsigns/manager.lua

@@ -75,6 +75,28 @@ local function apply_win_signs(bufnr, top, bot, clear)
   end
 end
 
+--- @param blame table<integer,Gitsigns.BlameInfo?>?
+--- @param first integer
+--- @param last_orig integer
+--- @param last_new integer
+local function on_lines_blame(blame, first, last_orig, last_new)
+  if not blame then
+    return
+  end
+
+  if last_new ~= last_orig then
+    if last_new < last_orig then
+      util.list_remove(blame, last_new, last_orig)
+    else
+      util.list_insert(blame, last_orig, last_new)
+    end
+  end
+
+  for i = math.min(first + 1, last_new), math.max(first + 1, last_new) do
+    blame[i] = nil
+  end
+end
+
 --- @param buf integer
 --- @param first integer
 --- @param last_orig integer
@@ -87,6 +109,8 @@ function M.on_lines(buf, first, last_orig, last_new)
     return true
   end
 
+  on_lines_blame(bcache.blame, first, last_orig, last_new)
+
   signs_normal:on_lines(buf, first, last_orig, last_new)
   if signs_staged then
     signs_staged:on_lines(buf, first, last_orig, last_new)

lua/gitsigns/util.lua

@@ -191,19 +191,6 @@ function M.get_relative_time(timestamp)
   end
 end
 
---- @generic T
---- @param x T[]
---- @return T[]
-function M.copy_array(x)
-  local r = {}
-  --- @diagnostic disable-next-line:no-unknown
-  for i, e in ipairs(x) do
-    --- @diagnostic disable-next-line:no-unknown
-    r[i] = e
-  end
-  return r
-end
-
 --- Strip '\r' from the EOL of each line only if all lines end with '\r'
 --- @param xs0 string[]
 --- @return string[]
@@ -304,4 +291,49 @@ function M.convert_blame_info(x)
   return ret
 end
 
+--- Efficiently remove items from middle of a list a list.
+---
+--- Calling table.remove() in a loop will re-index the tail of the table on
+--- every iteration, instead this function will re-index  the table exactly
+--- once.
+---
+--- Based on https://stackoverflow.com/questions/12394841/safely-remove-items-from-an-array-table-while-iterating/53038524#53038524
+---
+---@param t any[]
+---@param first integer
+---@param last integer
+function M.list_remove(t, first, last)
+  local n = #t
+  for i = 0, n - first do
+    t[first + i] = t[last + 1 + i]
+    t[last + 1 + i] = nil
+  end
+end
+
+--- Efficiently insert items into the middle of a list.
+---
+--- Calling table.insert() in a loop will re-index the tail of the table on
+--- every iteration, instead this function will re-index  the table exactly
+--- once.
+---
+--- Based on https://stackoverflow.com/questions/12394841/safely-remove-items-from-an-array-table-while-iterating/53038524#53038524
+---
+---@param t any[]
+---@param first integer
+---@param last integer
+---@param v any
+function M.list_insert(t, first, last, v)
+  local n = #t
+
+  -- Shift table forward
+  for i = n - first, 0, -1 do
+    t[last + 1 + i] = t[first + i]
+  end
+
+  -- Fill in new values
+  for i = first, last do
+    t[i] = v
+  end
+end
+
 return M

lua/gitsigns/watcher.lua

@@ -87,7 +87,7 @@ local handler = debounce_trailing(
       buf_check(bufnr)
     end
 
-    cache[bufnr]:invalidate()
+    cache[bufnr]:invalidate(true)
 
     require('gitsigns.manager').update(bufnr)
   end),