Merge #22736 from justinmk/gsplit

feat(vim.gsplit): gain features of vim.split

2 files changed, 116 insertions, 64 deletions

diff --git a/runtime/lua/vim/shared.lua b/runtime/lua/vim/shared.lua
index 1c8defc93a..884929e33a 100644
--- a/runtime/lua/vim/shared.lua
+++ b/runtime/lua/vim/shared.lua
@@ -59,21 +59,52 @@ end)()
 
 --- Splits a string at each instance of a separator.
 ---
----@see |vim.split()|
----@see |luaref-patterns|
----@see https://www.lua.org/pil/20.2.html
----@see http://lua-users.org/wiki/StringLibraryTutorial
----
----@param s string String to split
----@param sep string Separator or pattern
----@param plain (boolean|nil) If `true` use `sep` literally (passed to string.find)
----@return fun():string (function) Iterator over the split components
-function vim.gsplit(s, sep, plain)
-  vim.validate({ s = { s, 's' }, sep = { sep, 's' }, plain = { plain, 'b', true } })
+--- Example:
+---   <pre>lua
+---   for s in vim.gsplit(':aa::b:', ':', {plain=true}) do
+---     print(s)
+---   end
+---   </pre>
+---
+--- If you want to also inspect the separator itself (instead of discarding it), use
+--- |string.gmatch()|. Example:
+---   <pre>lua
+---   for word, num in ('foo111bar222'):gmatch('([^0-9]*)(%d*)') do
+---     print(('word: %s num: %s'):format(word, num))
+---   end
+---   </pre>
+---
+--- @see |string.gmatch()|
+--- @see |vim.split()|
+--- @see |luaref-patterns|
+--- @see https://www.lua.org/pil/20.2.html
+--- @see http://lua-users.org/wiki/StringLibraryTutorial
+---
+--- @param s string String to split
+--- @param sep string Separator or pattern
+--- @param opts (table|nil) Keyword arguments |kwargs|:
+---       - plain: (boolean) Use `sep` literally (as in string.find).
+---       - trimempty: (boolean) Discard empty segments at start and end of the sequence.
+---@return fun():string|nil (function) Iterator over the split components
+function vim.gsplit(s, sep, opts)
+  local plain
+  local trimempty = false
+  if type(opts) == 'boolean' then
+    plain = opts -- For backwards compatibility.
+  else
+    vim.validate({ s = { s, 's' }, sep = { sep, 's' }, opts = { opts, 't', true } })
+    opts = opts or {}
+    plain, trimempty = opts.plain, opts.trimempty
+  end
 
   local start = 1
   local done = false
 
+  -- For `trimempty`:
+  local empty_start = true -- Only empty segments seen so far.
+  local empty_segs = 0 -- Empty segments found between non-empty segments.
+  local nonemptyseg = nil
+
   local function _pass(i, j, ...)
     if i then
       assert(j + 1 > start, 'Infinite loop detected')
@@ -87,16 +118,44 @@ function vim.gsplit(s, sep, plain)
   end
 
   return function()
-    if done or (s == '' and sep == '') then
-      return
-    end
-    if sep == '' then
+    if trimempty and empty_segs > 0 then
+      -- trimempty: Pop the collected empty segments.
+      empty_segs = empty_segs - 1
+      return ''
+    elseif trimempty and nonemptyseg then
+      local seg = nonemptyseg
+      nonemptyseg = nil
+      return seg
+    elseif done or (s == '' and sep == '') then
+      return nil
+    elseif sep == '' then
       if start == #s then
         done = true
       end
       return _pass(start + 1, start)
     end
-    return _pass(s:find(sep, start, plain))
+
+    local seg = _pass(s:find(sep, start, plain))
+
+    -- Trim empty segments from start/end.
+    if trimempty and seg == '' then
+      while not done and seg == '' do
+        empty_segs = empty_segs + 1
+        seg = _pass(s:find(sep, start, plain))
+      end
+      if done and seg == '' then
+        return nil
+      elseif empty_start then
+        empty_start = false
+        empty_segs = 0
+        return seg
+      end
+      nonemptyseg = seg ~= '' and seg or nil
+      seg = ''
+      empty_segs = empty_segs - 1
+    end
+
+    return seg
   end
 end
 
@@ -111,48 +170,17 @@ end
 --- </pre>
 ---
 ---@see |vim.gsplit()|
+---@see |string.gmatch()|
 ---
 ---@param s string String to split
 ---@param sep string Separator or pattern
----@param kwargs (table|nil) Keyword arguments:
----       - plain: (boolean) If `true` use `sep` literally (passed to string.find)
----       - trimempty: (boolean) If `true` remove empty items from the front
----         and back of the list
+---@param opts (table|nil) Keyword arguments |kwargs| accepted by |vim.gsplit()|
 ---@return string[] List of split components
-function vim.split(s, sep, kwargs)
-  local plain
-  local trimempty = false
-  if type(kwargs) == 'boolean' then
-    -- Support old signature for backward compatibility
-    plain = kwargs
-  else
-    vim.validate({ kwargs = { kwargs, 't', true } })
-    kwargs = kwargs or {}
-    plain = kwargs.plain
-    trimempty = kwargs.trimempty
-  end
-
+function vim.split(s, sep, opts)
   local t = {}
-  local skip = trimempty
-  for c in vim.gsplit(s, sep, plain) do
-    if c ~= '' then
-      skip = false
-    end
-
-    if not skip then
-      table.insert(t, c)
-    end
+  for c in vim.gsplit(s, sep, opts) do
+    table.insert(t, c)
   end
-
-  if trimempty then
-    for i = #t, 1, -1 do
-      if t[i] ~= '' then
-        break
-      end
-      table.remove(t, i)
-    end
-  end
-
   return t
 end
 
diff --git a/runtime/lua/vim/version.lua b/runtime/lua/vim/version.lua
index 8d8b0d6da7..3aacf3d4e0 100644
--- a/runtime/lua/vim/version.lua
+++ b/runtime/lua/vim/version.lua
@@ -65,6 +65,35 @@ local M = {}
 local Version = {}
 Version.__index = Version
 
+--- @private
+---
+--- Compares prerelease strings: per semver, number parts must be must be treated as numbers:
+--- "pre1.10" is greater than "pre1.2". https://semver.org/#spec-item-11
+local function cmp_prerel(prerel1, prerel2)
+  if not prerel1 or not prerel2 then
+    return prerel1 and -1 or (prerel2 and 1 or 0)
+  end
+  -- TODO(justinmk): not fully spec-compliant; this treats non-dot-delimited digit sequences as
+  -- numbers. Maybe better: "(.-)(%.%d*)".
+  local iter1 = prerel1:gmatch('([^0-9]*)(%d*)')
+  local iter2 = prerel2:gmatch('([^0-9]*)(%d*)')
+  while true do
+    local word1, n1 = iter1()
+    local word2, n2 = iter2()
+    if word1 == nil and word2 == nil then -- Done iterating.
+      return 0
+    end
+    word1, n1, word2, n2 =
+      word1 or '', n1 and tonumber(n1) or 0, word2 or '', n2 and tonumber(n2) or 0
+    if word1 ~= word2 then
+      return word1 < word2 and -1 or 1
+    end
+    if n1 ~= n2 then
+      return n1 < n2 and -1 or 1
+    end
+  end
+end
+
 function Version:__index(key)
   return type(key) == 'number' and ({ self.major, self.minor, self.patch })[key] or Version[key]
 end
@@ -88,7 +117,7 @@ function Version:__eq(other)
       return false
     end
   end
-  return self.prerelease == other.prerelease
+  return 0 == cmp_prerel(self.prerelease, other.prerelease)
 end
 
 function Version:__tostring()
@@ -111,13 +140,7 @@ function Version:__lt(other)
       return true
     end
   end
-  if self.prerelease and not other.prerelease then
-    return true
-  end
-  if other.prerelease and not self.prerelease then
-    return false
-  end
-  return (self.prerelease or '') < (other.prerelease or '')
+  return -1 == cmp_prerel(self.prerelease, other.prerelease)
 end
 
 ---@param other Version
@@ -127,7 +150,7 @@ end
 
 --- @private
 ---
---- Creates a new Version object. Not public currently.
+--- Creates a new Version object, or returns `nil` if `version` is invalid.
 ---
 --- @param version string|number[]|Version
 --- @param strict? boolean Reject "1.0", "0-x", "3.2a" or other non-conforming version strings
@@ -173,6 +196,7 @@ function M._version(version, strict) -- Adapted from https://github.com/folke/la
       build = build ~= '' and build or nil,
     }, Version)
   end
+  return nil -- Invalid version string.
 end
 
 ---TODO: generalize this, move to func.lua
@@ -341,7 +365,7 @@ function M.cmp(v1, v2)
   return -1
 end
 
----Returns `true` if the given versions are equal.
+---Returns `true` if the given versions are equal. See |vim.version.cmp()| for usage.
 ---@param v1 Version|number[]
 ---@param v2 Version|number[]
 ---@return boolean
@@ -349,7 +373,7 @@ function M.eq(v1, v2)
   return M.cmp(v1, v2) == 0
 end
 
----Returns `true` if `v1 < v2`.
+---Returns `true` if `v1 < v2`. See |vim.version.cmp()| for usage.
 ---@param v1 Version|number[]
 ---@param v2 Version|number[]
 ---@return boolean
@@ -357,7 +381,7 @@ function M.lt(v1, v2)
   return M.cmp(v1, v2) == -1
 end
 
----Returns `true` if `v1 > v2`.
+---Returns `true` if `v1 > v2`. See |vim.version.cmp()| for usage.
 ---@param v1 Version|number[]
 ---@param v2 Version|number[]
 ---@return boolean