refactor(shada): rework msgpack decoding without msgpack-c

This also makes shada reading slightly faster due to avoiding
some copying and allocation.

Use keysets to drive decoding of msgpack maps for shada entries.

3 files changed, 31 insertions, 24 deletions

diff --git a/src/nvim/generators/c_grammar.lua b/src/nvim/generators/c_grammar.lua
index 9ce9f3d7a6..4e7de7adb4 100644
--- a/src/nvim/generators/c_grammar.lua
+++ b/src/nvim/generators/c_grammar.lua
@@ -91,7 +91,9 @@ local c_proto = Ct(
     * (P(';') + (P('{') * nl + (impl_line ^ 0) * P('}')))
 )
 
-local c_field = Ct(Cg(c_id, 'type') * ws * Cg(c_id, 'name') * fill * P(';') * fill)
+local dict_key = P('DictKey(') * Cg(rep1(any - P(')')), 'dict_key') * P(')')
+local keyset_field =
+  Ct(Cg(c_id, 'type') * ws * Cg(c_id, 'name') * fill * (dict_key ^ -1) * fill * P(';') * fill)
 local c_keyset = Ct(
   P('typedef')
     * ws
@@ -99,7 +101,7 @@ local c_keyset = Ct(
     * fill
     * P('{')
     * fill
-    * Cg(Ct(c_field ^ 1), 'fields')
+    * Cg(Ct(keyset_field ^ 1), 'fields')
     * P('}')
     * fill
     * P('Dict')
diff --git a/src/nvim/generators/gen_api_dispatch.lua b/src/nvim/generators/gen_api_dispatch.lua
index 61c80a3d2e..3567831c41 100644
--- a/src/nvim/generators/gen_api_dispatch.lua
+++ b/src/nvim/generators/gen_api_dispatch.lua
@@ -72,14 +72,19 @@ local keysets = {}
 local function add_keyset(val)
   local keys = {}
   local types = {}
+  local c_names = {}
   local is_set_name = 'is_set__' .. val.keyset_name .. '_'
   local has_optional = false
   for i, field in ipairs(val.fields) do
+    local dict_key = field.dict_key or field.name
     if field.type ~= 'Object' then
-      types[field.name] = field.type
+      types[dict_key] = field.type
     end
     if field.name ~= is_set_name and field.type ~= 'OptionalKeys' then
-      table.insert(keys, field.name)
+      table.insert(keys, dict_key)
+      if dict_key ~= field.name then
+        c_names[dict_key] = field.name
+      end
     else
       if i > 1 then
         error("'is_set__{type}_' must be first if present")
@@ -94,6 +99,7 @@ local function add_keyset(val)
   table.insert(keysets, {
     name = val.keyset_name,
     keys = keys,
+    c_names = c_names,
     types = types,
     has_optional = has_optional,
   })
@@ -332,19 +338,6 @@ output:write([[
 keysets_defs:write('// IWYU pragma: private, include "nvim/api/private/dispatch.h"\n\n')
 
 for _, k in ipairs(keysets) do
-  local c_name = {}
-
-  for i = 1, #k.keys do
-    -- some keys, like "register" are c keywords and get
-    -- escaped with a trailing _ in the struct.
-    if vim.endswith(k.keys[i], '_') then
-      local orig = k.keys[i]
-      k.keys[i] = string.sub(k.keys[i], 1, #k.keys[i] - 1)
-      c_name[k.keys[i]] = orig
-      k.types[k.keys[i]] = k.types[orig]
-    end
-  end
-
   local neworder, hashfun = hashy.hashy_hash(k.name, k.keys, function(idx)
     return k.name .. '_table[' .. idx .. '].str'
   end)
@@ -354,6 +347,8 @@ for _, k in ipairs(keysets) do
   local function typename(type)
     if type == 'HLGroupID' then
       return 'kObjectTypeInteger'
+    elseif type == 'StringArray' then
+      return 'kUnpackTypeStringArray'
     elseif type ~= nil then
       return 'kObjectType' .. type
     else
@@ -374,7 +369,7 @@ for _, k in ipairs(keysets) do
         .. '", offsetof(KeyDict_'
         .. k.name
         .. ', '
-        .. (c_name[key] or key)
+        .. (k.c_names[key] or key)
         .. '), '
         .. typename(k.types[key])
         .. ', '
diff --git a/src/nvim/generators/hashy.lua b/src/nvim/generators/hashy.lua
index 711e695742..ea35064962 100644
--- a/src/nvim/generators/hashy.lua
+++ b/src/nvim/generators/hashy.lua
@@ -73,11 +73,15 @@ function M.switcher(put, tab, maxlen, worst_buck_size)
           vim.list_extend(neworder, buck)
           local endidx = #neworder
           put("      case '" .. c .. "': ")
-          put('low = ' .. startidx .. '; ')
-          if bucky then
-            put('high = ' .. endidx .. '; ')
+          if len == 1 then
+            put('return ' .. startidx .. ';\n')
+          else
+            put('low = ' .. startidx .. '; ')
+            if bucky then
+              put('high = ' .. endidx .. '; ')
+            end
+            put 'break;\n'
           end
-          put 'break;\n'
         end
         put '      default: break;\n'
         put '    }\n    '
@@ -105,13 +109,19 @@ function M.hashy_hash(name, strings, access)
   end
   local len_pos_buckets, maxlen, worst_buck_size = M.build_pos_hash(strings)
   put('int ' .. name .. '_hash(const char *str, size_t len)\n{\n')
-  if worst_buck_size > 1 then
+  if maxlen == 1 then
+    put('\n') -- nothing
+  elseif worst_buck_size > 1 then
     put('  int low = 0, high = 0;\n')
   else
     put('  int low = -1;\n')
   end
   local neworder = M.switcher(put, len_pos_buckets, maxlen, worst_buck_size)
-  if worst_buck_size > 1 then
+  if maxlen == 1 then
+    put([[
+  return -1;
+]])
+  elseif worst_buck_size > 1 then
     put([[
   for (int i = low; i < high; i++) {
     if (!memcmp(str, ]] .. access('i') .. [[, len)) {