src/host/genminilua.lua

   1 ----------------------------------------------------------------------------
   2 -- Lua script to generate a customized, minified version of Lua.
   3 -- The resulting 'minilua' is used for the build process of LuaJIT.
   4 ----------------------------------------------------------------------------
   5 -- Copyright (C) 2005-2014 Mike Pall. All rights reserved.
   6 -- Released under the MIT license. See Copyright Notice in luajit.h
   7 ----------------------------------------------------------------------------
   8
   9 local sub, match, gsub = string.sub, string.match, string.gsub
  10
  11 local LUA_VERSION = "5.1.5"
  12 local LUA_SOURCE
  13
  14 local function usage()
  15   io.stderr:write("Usage: ", arg and arg[0] or "genminilua",
  16                   " lua-", LUA_VERSION, "-source-dir\n")
  17   os.exit(1)
  18 end
  19
  20 local function find_sources()
  21   LUA_SOURCE = arg and arg[1]
  22   if not LUA_SOURCE then usage() end
  23   if sub(LUA_SOURCE, -1) ~= "/" then LUA_SOURCE = LUA_SOURCE.."/" end
  24   local fp = io.open(LUA_SOURCE .. "lua.h")
  25   if not fp then
  26     LUA_SOURCE = LUA_SOURCE.."src/"
  27     fp = io.open(LUA_SOURCE .. "lua.h")
  28     if not fp then usage() end
  29   end
  30   local all = fp:read("*a")
  31   fp:close()
  32   if not match(all, 'LUA_RELEASE%s*"Lua '..LUA_VERSION..'"') then
  33     io.stderr:write("Error: version mismatch\n")
  34     usage()
  35   end
  36 end
  37
  38 local LUA_FILES = {
  39 "lmem.c", "lobject.c", "ltm.c", "lfunc.c", "ldo.c", "lstring.c", "ltable.c",
  40 "lgc.c", "lstate.c", "ldebug.c", "lzio.c", "lopcodes.c",
  41 "llex.c", "lcode.c", "lparser.c", "lvm.c", "lapi.c", "lauxlib.c",
  42 "lbaselib.c", "ltablib.c", "liolib.c", "loslib.c", "lstrlib.c", "linit.c",
  43 }
  44
  45 local REMOVE_LIB = {}
  46 gsub([[
  47 collectgarbage dofile gcinfo getfenv getmetatable load print rawequal rawset
  48 select tostring xpcall
  49 foreach foreachi getn maxn setn
  50 popen tmpfile seek setvbuf __tostring
  51 clock date difftime execute getenv rename setlocale time tmpname
  52 dump gfind len reverse
  53 LUA_LOADLIBNAME LUA_MATHLIBNAME LUA_DBLIBNAME
  54 ]], "%S+", function(name)
  55   REMOVE_LIB[name] = true
  56 end)
  57
  58 local REMOVE_EXTINC = { ["<assert.h>"] = true, ["<locale.h>"] = true, }
  59
  60 local CUSTOM_MAIN = [[
  61 typedef unsigned int UB;
  62 static UB barg(lua_State *L,int idx){
  63 union{lua_Number n;U64 b;}bn;
  64 bn.n=lua_tonumber(L,idx)+6755399441055744.0;
  65 if (bn.n==0.0&&!lua_isnumber(L,idx))luaL_typerror(L,idx,"number");
  66 return(UB)bn.b;
  67 }
  68 #define BRET(b) lua_pushnumber(L,(lua_Number)(int)(b));return 1;
  69 static int tobit(lua_State *L){
  70 BRET(barg(L,1))}
  71 static int bnot(lua_State *L){
  72 BRET(~barg(L,1))}
  73 static int band(lua_State *L){
  74 int i;UB b=barg(L,1);for(i=lua_gettop(L);i>1;i--)b&=barg(L,i);BRET(b)}
  75 static int bor(lua_State *L){
  76 int i;UB b=barg(L,1);for(i=lua_gettop(L);i>1;i--)b|=barg(L,i);BRET(b)}
  77 static int bxor(lua_State *L){
  78 int i;UB b=barg(L,1);for(i=lua_gettop(L);i>1;i--)b^=barg(L,i);BRET(b)}
  79 static int lshift(lua_State *L){
  80 UB b=barg(L,1),n=barg(L,2)&31;BRET(b<<n)}
  81 static int rshift(lua_State *L){
  82 UB b=barg(L,1),n=barg(L,2)&31;BRET(b>>n)}
  83 static int arshift(lua_State *L){
  84 UB b=barg(L,1),n=barg(L,2)&31;BRET((int)b>>n)}
  85 static int rol(lua_State *L){
  86 UB b=barg(L,1),n=barg(L,2)&31;BRET((b<<n)|(b>>(32-n)))}
  87 static int ror(lua_State *L){
  88 UB b=barg(L,1),n=barg(L,2)&31;BRET((b>>n)|(b<<(32-n)))}
  89 static int bswap(lua_State *L){
  90 UB b=barg(L,1);b=(b>>24)|((b>>8)&0xff00)|((b&0xff00)<<8)|(b<<24);BRET(b)}
  91 static int tohex(lua_State *L){
  92 UB b=barg(L,1);
  93 int n=lua_isnone(L,2)?8:(int)barg(L,2);
  94 const char *hexdigits="0123456789abcdef";
  95 char buf[8];
  96 int i;
  97 if(n<0){n=-n;hexdigits="0123456789ABCDEF";}
  98 if(n>8)n=8;
  99 for(i=(int)n;--i>=0;){buf[i]=hexdigits[b&15];b>>=4;}
 100 lua_pushlstring(L,buf,(size_t)n);
 101 return 1;
 102 }
 103 static const struct luaL_Reg bitlib[] = {
 104 {"tobit",tobit},
 105 {"bnot",bnot},
 106 {"band",band},
 107 {"bor",bor},
 108 {"bxor",bxor},
 109 {"lshift",lshift},
 110 {"rshift",rshift},
 111 {"arshift",arshift},
 112 {"rol",rol},
 113 {"ror",ror},
 114 {"bswap",bswap},
 115 {"tohex",tohex},
 116 {NULL,NULL}
 117 };
 118 int main(int argc, char **argv){
 119   lua_State *L = luaL_newstate();
 120   int i;
 121   luaL_openlibs(L);
 122   luaL_register(L, "bit", bitlib);
 123   if (argc < 2) return sizeof(void *);
 124   lua_createtable(L, 0, 1);
 125   lua_pushstring(L, argv[1]);
 126   lua_rawseti(L, -2, 0);
 127   lua_setglobal(L, "arg");
 128   if (luaL_loadfile(L, argv[1]))
 129     goto err;
 130   for (i = 2; i < argc; i++)
 131     lua_pushstring(L, argv[i]);
 132   if (lua_pcall(L, argc - 2, 0, 0)) {
 133   err:
 134     fprintf(stderr, "Error: %s\n", lua_tostring(L, -1));
 135     return 1;
 136   }
 137   lua_close(L);
 138   return 0;
 139 }
 140 ]]
 141
 142 local function read_sources()
 143   local t = {}
 144   for i, name in ipairs(LUA_FILES) do
 145     local fp = assert(io.open(LUA_SOURCE..name, "r"))
 146     t[i] = fp:read("*a")
 147     assert(fp:close())
 148   end
 149   t[#t+1] = CUSTOM_MAIN
 150   return table.concat(t)
 151 end
 152
 153 local includes = {}
 154
 155 local function merge_includes(src)
 156   return gsub(src, '#include%s*"([^"]*)"%s*\n', function(name)
 157     if includes[name] then return "" end
 158     includes[name] = true
 159     local fp = assert(io.open(LUA_SOURCE..name, "r"))
 160     local src = fp:read("*a")
 161     assert(fp:close())
 162     src = gsub(src, "#ifndef%s+%w+_h\n#define%s+%w+_h\n", "")
 163     src = gsub(src, "#endif%s*$", "")
 164     return merge_includes(src)
 165   end)
 166 end
 167
 168 local function get_license(src)
 169   return match(src, "/%*+\n%* Copyright %(.-%*/\n")
 170 end
 171
 172 local function fold_lines(src)
 173   return gsub(src, "\\\n", " ")
 174 end
 175
 176 local strings = {}
 177
 178 local function save_str(str)
 179   local n = #strings+1
 180   strings[n] = str
 181   return "\1"..n.."\2"
 182 end
 183
 184 local function save_strings(src)
 185   src = gsub(src, '"[^"\n]*"', save_str)
 186   return gsub(src, "'[^'\n]*'", save_str)
 187 end
 188
 189 local function restore_strings(src)
 190   return gsub(src, "\1(%d+)\2", function(numstr)
 191     return strings[tonumber(numstr)]
 192   end)
 193 end
 194
 195 local function def_istrue(def)
 196   return def == "INT_MAX > 2147483640L" or
 197          def == "LUAI_BITSINT >= 32" or
 198          def == "SIZE_Bx < LUAI_BITSINT-1" or
 199          def == "cast" or
 200          def == "defined(LUA_CORE)" or
 201          def == "MINSTRTABSIZE" or
 202          def == "LUA_MINBUFFER" or
 203          def == "HARDSTACKTESTS" or
 204          def == "UNUSED"
 205 end
 206
 207 local head, defs = {[[
 208 #ifdef _MSC_VER
 209 typedef unsigned __int64 U64;
 210 #else
 211 typedef unsigned long long U64;
 212 #endif
 213 int _CRT_glob = 0;
 214 ]]}, {}
 215
 216 local function preprocess(src)
 217   local t = { match(src, "^(.-)#") }
 218   local lvl, on, oldon = 0, true, {}
 219   for pp, def, txt in string.gmatch(src, "#(%w+) *([^\n]*)\n([^#]*)") do
 220     if pp == "if" or pp == "ifdef" or pp == "ifndef" then
 221       lvl = lvl + 1
 222       oldon[lvl] = on
 223       on = def_istrue(def)
 224     elseif pp == "else" then
 225       if oldon[lvl] then
 226         if on == false then on = true else on = false end
 227       end
 228     elseif pp == "elif" then
 229       if oldon[lvl] then
 230         on = def_istrue(def)
 231       end
 232     elseif pp == "endif" then
 233       on = oldon[lvl]
 234       lvl = lvl - 1
 235     elseif on then
 236       if pp == "include" then
 237         if not head[def] and not REMOVE_EXTINC[def] then
 238           head[def] = true
 239           head[#head+1] = "#include "..def.."\n"
 240         end
 241       elseif pp == "define" then
 242         local k, sp, v = match(def, "([%w_]+)(%s*)(.*)")
 243         if k and not (sp == "" and sub(v, 1, 1) == "(") then
 244           defs[k] = gsub(v, "%a[%w_]*", function(tok)
 245             return defs[tok] or tok
 246           end)
 247         else
 248           t[#t+1] = "#define "..def.."\n"
 249         end
 250       elseif pp ~= "undef" then
 251         error("unexpected directive: "..pp.." "..def)
 252       end
 253     end
 254     if on then t[#t+1] = txt end
 255   end
 256   return gsub(table.concat(t), "%a[%w_]*", function(tok)
 257     return defs[tok] or tok
 258   end)
 259 end
 260
 261 local function merge_header(src, license)
 262   local hdr = string.format([[
 263 /* This is a heavily customized and minimized copy of Lua %s. */
 264 /* It's only used to build LuaJIT. It does NOT have all standard functions! */
 265 ]], LUA_VERSION)
 266   return hdr..license..table.concat(head)..src
 267 end
 268
 269 local function strip_unused1(src)
 270   return gsub(src, '(  {"?([%w_]+)"?,%s+%a[%w_]*},\n)', function(line, func)
 271     return REMOVE_LIB[func] and "" or line
 272   end)
 273 end
 274
 275 local function strip_unused2(src)
 276   return gsub(src, "Symbolic Execution.-}=", "")
 277 end
 278
 279 local function strip_unused3(src)
 280   src = gsub(src, "extern", "static")
 281   src = gsub(src, "\nstatic([^\n]-)%(([^)]*)%)%(", "\nstatic%1 %2(")
 282   src = gsub(src, "#define lua_assert[^\n]*\n", "")
 283   src = gsub(src, "lua_assert%b();?", "")
 284   src = gsub(src, "default:\n}", "default:;\n}")
 285   src = gsub(src, "lua_lock%b();", "")
 286   src = gsub(src, "lua_unlock%b();", "")
 287   src = gsub(src, "luai_threadyield%b();", "")
 288   src = gsub(src, "luai_userstateopen%b();", "{}")
 289   src = gsub(src, "luai_userstate%w+%b();", "")
 290   src = gsub(src, "%(%(c==.*luaY_parser%)", "luaY_parser")
 291   src = gsub(src, "trydecpoint%(ls,seminfo%)",
 292                   "luaX_lexerror(ls,\"malformed number\",TK_NUMBER)")
 293   src = gsub(src, "int c=luaZ_lookahead%b();", "")
 294   src = gsub(src, "luaL_register%(L,[^,]*,co_funcs%);\nreturn 2;",
 295                   "return 1;")
 296   src = gsub(src, "getfuncname%b():", "NULL:")
 297   src = gsub(src, "getobjname%b():", "NULL:")
 298   src = gsub(src, "if%([^\n]*hookmask[^\n]*%)\n[^\n]*\n", "")
 299   src = gsub(src, "if%([^\n]*hookmask[^\n]*%)%b{}\n", "")
 300   src = gsub(src, "if%([^\n]*hookmask[^\n]*&&\n[^\n]*%b{}\n", "")
 301   src = gsub(src, "(twoto%b()%()", "%1(size_t)")
 302   src = gsub(src, "i<sizenode", "i<(int)sizenode")
 303   return gsub(src, "\n\n+", "\n")
 304 end
 305
 306 local function strip_comments(src)
 307   return gsub(src, "/%*.-%*/", " ")
 308 end
 309
 310 local function strip_whitespace(src)
 311   src = gsub(src, "^%s+", "")
 312   src = gsub(src, "%s*\n%s*", "\n")
 313   src = gsub(src, "[ \t]+", " ")
 314   src = gsub(src, "(%W) ", "%1")
 315   return gsub(src, " (%W)", "%1")
 316 end
 317
 318 local function rename_tokens1(src)
 319   src = gsub(src, "getline", "getline_")
 320   src = gsub(src, "struct ([%w_]+)", "ZX%1")
 321   return gsub(src, "union ([%w_]+)", "ZY%1")
 322 end
 323
 324 local function rename_tokens2(src)
 325   src = gsub(src, "ZX([%w_]+)", "struct %1")
 326   return gsub(src, "ZY([%w_]+)", "union %1")
 327 end
 328
 329 local function func_gather(src)
 330   local nodes, list = {}, {}
 331   local pos, len = 1, #src
 332   while pos < len do
 333     local d, w = match(src, "^(#define ([%w_]+)[^\n]*\n)", pos)
 334     if d then
 335       local n = #list+1
 336       list[n] = d
 337       nodes[w] = n
 338     else
 339       local s
 340       d, w, s = match(src, "^(([%w_]+)[^\n]*([{;])\n)", pos)
 341       if not d then
 342         d, w, s = match(src, "^(([%w_]+)[^(]*%b()([{;])\n)", pos)
 343         if not d then d = match(src, "^[^\n]*\n", pos) end
 344       end
 345       if s == "{" then
 346         d = d..sub(match(src, "^%b{}[^;\n]*;?\n", pos+#d-2), 3)
 347         if sub(d, -2) == "{\n" then
 348           d = d..sub(match(src, "^%b{}[^;\n]*;?\n", pos+#d-2), 3)
 349         end
 350       end
 351       local k, v = nil, d
 352       if w == "typedef" then
 353         if match(d, "^typedef enum") then
 354           head[#head+1] = d
 355         else
 356           k = match(d, "([%w_]+);\n$")
 357           if not k then k = match(d, "^.-%(.-([%w_]+)%)%(") end
 358         end
 359       elseif w == "enum" then
 360         head[#head+1] = v
 361       elseif w ~= nil then
 362         k = match(d, "^[^\n]-([%w_]+)[(%[=]")
 363         if k then
 364           if w ~= "static" and k ~= "main" then v = "static "..d end
 365         else
 366           k = w
 367         end
 368       end
 369       if w and k then
 370         local o = nodes[k]
 371         if o then nodes["*"..k] = o end
 372         local n = #list+1
 373         list[n] = v
 374         nodes[k] = n
 375       end
 376     end
 377     pos = pos + #d
 378   end
 379   return nodes, list
 380 end
 381
 382 local function func_visit(nodes, list, used, n)
 383   local i = nodes[n]
 384   for m in string.gmatch(list[i], "[%w_]+") do
 385     if nodes[m] then
 386       local j = used[m]
 387       if not j then
 388         used[m] = i
 389         func_visit(nodes, list, used, m)
 390       elseif i < j then
 391         used[m] = i
 392       end
 393     end
 394   end
 395 end
 396
 397 local function func_collect(src)
 398   local nodes, list = func_gather(src)
 399   local used = {}
 400   func_visit(nodes, list, used, "main")
 401   for n,i in pairs(nodes) do
 402     local j = used[n]
 403     if j and j < i then used["*"..n] = j end
 404   end
 405   for n,i in pairs(nodes) do
 406     if not used[n] then list[i] = "" end
 407   end
 408   return table.concat(list)
 409 end
 410
 411 find_sources()
 412 local src = read_sources()
 413 src = merge_includes(src)
 414 local license = get_license(src)
 415 src = fold_lines(src)
 416 src = strip_unused1(src)
 417 src = save_strings(src)
 418 src = strip_unused2(src)
 419 src = strip_comments(src)
 420 src = preprocess(src)
 421 src = strip_whitespace(src)
 422 src = strip_unused3(src)
 423 src = rename_tokens1(src)
 424 src = func_collect(src)
 425 src = rename_tokens2(src)
 426 src = restore_strings(src)
 427 src = merge_header(src, license)
 428 io.write(src)