Move string functions to centralised lstring.js

12 files changed, 90 insertions, 72 deletions

diff --git a/src/lapi.js b/src/lapi.js
index aa8cf04..fa6da20 100644
--- a/src/lapi.js
+++ b/src/lapi.js
@@ -10,6 +10,7 @@ const lfunc     = require('./lfunc.js');
 const llex      = require('./llex.js');
 const lobject   = require('./lobject.js');
 const lstate    = require('./lstate.js');
+const lstring   = require('./lstring.js');
 const ltm       = require('./ltm.js');
 const luaconf   = require('./luaconf.js');
 const lvm       = require('./lvm.js');
@@ -216,7 +217,7 @@ const lua_pushlstring = function(L, s, len) {
     assert(Array.isArray(s), "lua_pushlstring expects array of byte");
     assert(typeof len === "number");
 
-    let ts = len === 0 ? L.l_G.intern(defs.to_luastring("", true)) : new TValue(CT.LUA_TLNGSTR, s.slice(0, len));
+    let ts = new TValue(CT.LUA_TLNGSTR, lstring.luaS_bless(L, s.slice(0, len)));
     L.stack[L.top++] = ts;
 
     assert(L.top <= L.ci.top, "stack overflow");
@@ -230,7 +231,7 @@ const lua_pushstring = function (L, s) {
     if (s === undefined || s === null)
         L.stack[L.top] = new TValue(CT.LUA_TNIL, null);
     else {
-        L.stack[L.top] = new TValue(CT.LUA_TLNGSTR, s);
+        L.stack[L.top] = new TValue(CT.LUA_TLNGSTR, lstring.luaS_new(L, s));
     }
 
     L.top++;
@@ -255,7 +256,7 @@ const lua_pushliteral = function (L, s) {
     if (s === undefined || s === null)
         L.stack[L.top] = new TValue(CT.LUA_TNIL, null);
     else {
-        let ts = L.l_G.intern(defs.to_luastring(s));
+        let ts = new TValue(CT.LUA_TLNGSTR, lstring.luaS_newliteral(L, s));
         L.stack[L.top] = ts;
     }
 
@@ -330,7 +331,7 @@ const lua_pushglobaltable = function(L) {
 const auxsetstr = function(L, t, k) {
     assert(Array.isArray(k), "key must be an array of bytes");
 
-    let str = L.l_G.intern(k);
+    let str = new TValue(CT.LUA_TLNGSTR, lstring.luaS_new(L, k));
 
     assert(1 < L.top - L.ci.funcOff, "not enough elements in the stack");
 
@@ -429,7 +430,7 @@ const lua_rawsetp = function(L, idx, p) {
 const auxgetstr = function(L, t, k) {
     assert(Array.isArray(k), "key must be an array of bytes");
 
-    let str = L.l_G.intern(k);
+    let str = new TValue(CT.LUA_TLNGSTR, lstring.luaS_new(L, k));
 
     L.stack[L.top++] = str;
     assert(L.top <= L.ci.top, "stack overflow");
@@ -952,7 +953,7 @@ const lua_concat = function(L, n) {
     if (n >= 2)
         lvm.luaV_concat(L, n);
     else if (n === 0) {
-        L.stack[L.top++] = L.l_G.intern(defs.to_luastring("", true));
+        L.stack[L.top++] = new TValue(CT.LUA_TLNGSTR, lstring.luaS_newliteral(L, []));
         assert(L.top <= L.ci.top, "stack overflow");
     }
 };
diff --git a/src/lcode.js b/src/lcode.js
index fac7d37..814498c 100644
--- a/src/lcode.js
+++ b/src/lcode.js
@@ -7,6 +7,7 @@ const llex     = require('./llex.js');
 const lobject  = require('./lobject.js');
 const lopcode  = require('./lopcodes.js');
 const lparser  = require('./lparser.js');
+const lstring  = require('./lstring.js');
 const ltable   = require('./ltable.js');
 require('./lstate.js'); /* XXX: if this isn't here then things break on require("ltm") */
 const ltm      = require('./ltm.js');
@@ -514,7 +515,7 @@ const luaK_stringK = function(fs, s) {
 */
 const luaK_intK = function(fs, n) {
     /* FIXME: shouldn't use string as key. need to use pointer */
-    let k = new TValue(CT.LUA_TLNGSTR, defs.to_luastring(`${n}`));
+    let k = new TValue(CT.LUA_TLNGSTR, lstring.luaS_bless(fs.L, defs.to_luastring(`${n}`)));
     let o = new TValue(CT.LUA_TNUMINT, n);
     return addk(fs, k, o);
 };
diff --git a/src/ldo.js b/src/ldo.js
index 5865f96..39ba4dc 100644
--- a/src/ldo.js
+++ b/src/ldo.js
@@ -13,6 +13,7 @@ const llimit         = require('./llimit.js');
 const lobject        = require('./lobject.js');
 const lparser        = require('./lparser.js');
 const lstate         = require('./lstate.js');
+const lstring        = require('./lstring.js');
 const ltm            = require('./ltm.js');
 const lvm            = require('./lvm.js');
 const CT             = defs.constant_types;
@@ -23,11 +24,11 @@ const TValue         = lobject.TValue;
 const seterrorobj = function(L, errcode, oldtop) {
     switch (errcode) {
         case TS.LUA_ERRMEM: {
-            L.stack[oldtop] = L.l_G.intern(defs.to_luastring("not enough memory", true));
+            L.stack[oldtop] = new TValue(CT.LUA_TLNGSTR, lstring.luaS_newliteral(L, "not enough memory"));
             break;
         }
         case TS.LUA_ERRERR: {
-            L.stack[oldtop] = L.l_G.intern(defs.to_luastring("error in error handling", true));
+            L.stack[oldtop] = new TValue(CT.LUA_TLNGSTR, lstring.luaS_newliteral(L, "error in error handling"));
             break;
         }
         default: {
@@ -401,7 +402,7 @@ const recover = function(L, status) {
 */
 const resume_error = function(L, msg, narg) {
     L.top -= narg;  /* remove args from the stack */
-    L.stack[L.top++] = L.l_G.intern(defs.to_luastring(msg));  /* push error message */
+    L.stack[L.top++] = new TValue(CT.LUA_TLNGSTR, lstring.luaS_newliteral(L, msg));  /* push error message */
     assert(L.top <= L.ci.top, "stack overflow");
     return TS.LUA_ERRRUN;
 };
diff --git a/src/llex.js b/src/llex.js
index 5591054..e690ea2 100644
--- a/src/llex.js
+++ b/src/llex.js
@@ -7,6 +7,7 @@ const ldebug   = require('./ldebug.js');
 const ldo      = require('./ldo.js');
 const ljstype  = require('./ljstype');
 const lobject  = require('./lobject');
+const lstring  = require('./lstring.js');
 const llimit   = require('./llimit.js');
 const TS       = defs.thread_status;
 const char     = defs.char;
@@ -231,7 +232,7 @@ const luaX_setinput = function(L, ls, z, source, firstchar) {
     ls.linenumber = 1;
     ls.lastline = 1;
     ls.source = source;
-    ls.envn = defs.to_luastring("_ENV", true);
+    ls.envn = lstring.luaS_newliteral(L, "_ENV");
 };
 
 const check_next1 = function(ls, c) {
@@ -368,7 +369,7 @@ const read_long_string = function(ls, seminfo, sep) {
     }
 
     if (seminfo)
-        seminfo.ts = ls.buff.buffer.slice(2 + sep, 2 + sep - 2 * (2 + sep));
+        seminfo.ts = lstring.luaS_bless(ls.L, ls.buff.buffer.slice(2 + sep, 2 + sep - 2 * (2 + sep)));
 };
 
 const esccheck = function(ls, c, msg) {
@@ -493,7 +494,7 @@ const read_string = function(ls, del, seminfo) {
     }
     save_and_next(ls);  /* skip delimiter */
 
-    seminfo.ts = ls.buff.buffer.slice(1, ls.buff.n-1);
+    seminfo.ts = lstring.luaS_bless(ls.L, ls.buff.buffer.slice(1, ls.buff.n-1));
 };
 
 const isreserved = function(w) {
@@ -605,7 +606,7 @@ const llex = function(ls, seminfo) {
                         save_and_next(ls);
                     } while (ljstype.lislalnum(ls.current));
 
-                    let ts = ls.buff.buffer;
+                    let ts = lstring.luaS_new(ls.L, ls.buff.buffer);
                     seminfo.ts = ts;
                     let kidx = luaX_tokens.slice(0, 22).indexOf(defs.to_jsstring(ts));
                     if (kidx >= 0)  /* reserved word? */
diff --git a/src/lobject.js b/src/lobject.js
index cea9975..905eed1 100644
--- a/src/lobject.js
+++ b/src/lobject.js
@@ -6,6 +6,7 @@ const assert = require('assert');
 const defs    = require('./defs.js');
 const ljstype = require('./ljstype.js');
 const ldebug  = require('./ldebug.js');
+const lstring = require('./lstring.js');
 const luaconf = require('./luaconf.js');
 const lvm     = require('./lvm.js');
 const llimit  = require('./llimit.js');
@@ -418,7 +419,7 @@ const luaO_utf8esc = function(x) {
 };
 
 const pushstr = function(L, str) {
-    L.stack[L.top++] = L.l_G.intern(str);
+    L.stack[L.top++] = new TValue(CT.LUA_TLNGSTR, lstring.luaS_new(L, str));
 };
 
 const luaO_pushvfstring = function(L, fmt, argp) {
diff --git a/src/lparser.js b/src/lparser.js
index 566576a..c90ff01 100644
--- a/src/lparser.js
+++ b/src/lparser.js
@@ -9,6 +9,7 @@ const llex     = require('./llex.js');
 const llimit   = require('./llimit.js');
 const lobject  = require('./lobject.js');
 const lopcode  = require('./lopcodes.js');
+const lstring  = require('./lstring.js');
 const ltable   = require('./ltable.js');
 const BinOpr   = lcode.BinOpr;
 const OpCodesI = lopcode.OpCodesI;
@@ -24,6 +25,11 @@ const hasmultret = function(k) {
     return k === expkind.VCALL || k === expkind.VVARARG;
 };
 
+const eqstr = function(a, b) {
+    /* TODO: use plain equality as strings are cached */
+    return lstring.luaS_eqlngstr(a, b);
+};
+
 class BlockCnt {
     constructor() {
         this.previous = null;  /* chain */
@@ -274,7 +280,7 @@ const removevars = function(fs, tolevel) {
 const searchupvalue = function(fs, name) {
     let up = fs.f.upvalues;
     for (let i = 0; i < fs.nups; i++) {
-        if (up[i].name.join() === name.join())
+        if (eqstr(up[i].name, name))
             return i;
     }
     return -1;  /* not found */
@@ -293,7 +299,7 @@ const newupvalue = function(fs, name, v) {
 
 const searchvar = function(fs, n) {
     for (let i = fs.nactvar - 1; i >= 0; i--) {
-        if (n.join() === getlocvar(fs, i).varname.join())
+        if (eqstr(n, getlocvar(fs, i).varname))
             return i;
     }
 
@@ -385,7 +391,7 @@ const closegoto = function(ls, g, label) {
     let fs = ls.fs;
     let gl = ls.dyd.gt;
     let gt = gl.arr[g];
-    assert(gt.name.join() === label.name.join());
+    assert(eqstr(gt.name, label.name));
     if (gt.nactvar < label.nactvar) {
         let vname = getlocvar(fs, gt.nactvar).varname;
         let msg = lobject.luaO_pushfstring(ls.L, defs.to_luastring("<goto %s> at line %d jumps into the scope of local '%s'"),
@@ -409,7 +415,7 @@ const findlabel = function(ls, g) {
     /* check labels in current block for a match */
     for (let i = bl.firstlabel; i < dyd.label.n; i++) {
         let lb = dyd.label.arr[i];
-        if (lb.name.join() === gt.name.join()) {  /* correct label? */
+        if (eqstr(lb.name, gt.name)) {  /* correct label? */
             if (gt.nactvar > lb.nactvar && (bl.upval || dyd.label.n > bl.firstlabel))
                 lcode.luaK_patchclose(ls.fs, gt.pc, lb.nactvar);
             closegoto(ls, g, lb);  /* close it */
@@ -438,7 +444,7 @@ const findgotos = function(ls, lb) {
     let gl = ls.dyd.gt;
     let i = ls.fs.bl.firstgoto;
     while (i < gl.n) {
-        if (gl.arr[i].name.join() === lb.name.join())
+        if (eqstr(gl.arr[i].name, lb.name))
             closegoto(ls, i, lb);
         else
             i++;
@@ -483,7 +489,7 @@ const enterblock = function(fs, bl, isloop) {
 ** create a label named 'break' to resolve break statements
 */
 const breaklabel = function(ls) {
-    let n = defs.to_luastring("break", true);
+    let n = lstring.luaS_newliteral(ls.L, "break");
     let l = newlabelentry(ls, ls.dyd.label, n, 0, ls.fs.pc);
     findgotos(ls, ls.dyd.label.arr[l]);
 };
@@ -1146,7 +1152,7 @@ const gotostat = function(ls, pc) {
         label = str_checkname(ls);
     else {
         llex.luaX_next(ls);  /* skip break */
-        label = defs.to_luastring("break", true);
+        label = lstring.luaS_newliteral(ls.L, "break");
     }
     let g = newlabelentry(ls, ls.dyd.gt, label, line, pc);
     findlabel(ls, g);  /* close it if label already defined */
@@ -1155,7 +1161,7 @@ const gotostat = function(ls, pc) {
 /* check for repeated labels on the same block */
 const checkrepeated = function(fs, ll, label) {
     for (let i = fs.bl.firstlabel; i < ll.n; i++) {
-        if (label.join() === ll.arr[i].name.join()) {
+        if (eqstr(label, ll.arr[i].name)) {
             semerror(fs.ls, defs.to_luastring(`label '${label.jsstring()}' already defined on line ${ll.arr[i].line}`));
         }
     }
@@ -1554,7 +1560,7 @@ const luaY_parser = function(L, z, buff, dyd, name, firstchar) {
     lexstate.h = new TValue(defs.CT.LUA_TTABLE, ltable.luaH_new(L));  /* create table for scanner */
     L.stack[L.top++] = lexstate.h;
     funcstate.f = cl.p = new Proto(L);
-    funcstate.f.source = name;
+    funcstate.f.source = lstring.luaS_new(L, name);
     lexstate.buff = buff;
     lexstate.dyd = dyd;
     dyd.actvar.n = dyd.gt.n = dyd.label.n = 0;
diff --git a/src/lstate.js b/src/lstate.js
index fee308c..f2e17cd 100644
--- a/src/lstate.js
+++ b/src/lstate.js
@@ -71,18 +71,6 @@ class global_State {
         this.mt = new Array(LUA_NUMTAGS);
     }
 
-    intern(stringArray) {
-        // TODO: when weak value maps are available
-        // let key = stringArray.map(e => `${e}|`).join('');
-
-        // if (!this.strt.has(key))
-        //     this.strt.set(key, new lobject.TValue(CT.LUA_TLNGSTR, stringArray));
-
-        // return this.strt.get(key);
-
-        return new lobject.TValue(CT.LUA_TLNGSTR, stringArray);
-    }
-
 }
 
 
diff --git a/src/lstring.js b/src/lstring.js
new file mode 100644
index 0000000..34f988e
--- /dev/null
+++ b/src/lstring.js
@@ -0,0 +1,33 @@
+"use strict";
+
+const defs = require('./defs.js');
+
+const luaS_eqlngstr = function(a, b) {
+    return a == b || (a.length == b.length && a.join() == b.join());
+};
+
+/* converts strings (arrays) to a consistent map key */
+const luaS_hash = function(str) {
+    return str.map(e => `${e}|`).join('');
+};
+
+/* variant that takes ownership of array */
+const luaS_bless = function(L, str) {
+    return str;
+};
+
+/* makes a copy */
+const luaS_new = function(L, str) {
+    return luaS_bless(L, str.slice(0));
+};
+
+/* takes a js string */
+const luaS_newliteral = function(L, str) {
+    return luaS_bless(L, defs.to_luastring(str));
+};
+
+module.exports.luaS_eqlngstr   = luaS_eqlngstr;
+module.exports.luaS_hash       = luaS_hash;
+module.exports.luaS_bless      = luaS_bless;
+module.exports.luaS_new        = luaS_new;
+module.exports.luaS_newliteral = luaS_newliteral;
diff --git a/src/ltable.js b/src/ltable.js
index 428ee5e..de4c686 100644
--- a/src/ltable.js
+++ b/src/ltable.js
@@ -6,13 +6,9 @@ const assert = require('assert');
 const defs    = require('./defs.js');
 const ldebug  = require('./ldebug.js');
 const lobject = require('./lobject.js');
+const lstring = require('./lstring.js');
 const CT      = defs.constant_types;
 
-/* converts strings (arrays) to a consistent map key */
-const hashstr = function(str) {
-    return str.map(e => `${e}|`).join('');
-};
-
 const table_hash = function(key) {
     switch(key.type) {
     case CT.LUA_TBOOLEAN:
@@ -28,7 +24,7 @@ const table_hash = function(key) {
         return key.value;
     case CT.LUA_TSHRSTR:
     case CT.LUA_TLNGSTR:
-        return hashstr(key.value);
+        return lstring.luaS_hash(key.value);
     default:
         throw new Error("unknown key type: " + key.type);
     }
@@ -58,7 +54,7 @@ const luaH_getint = function(t, key) {
 
 const luaH_getstr = function(t, key) {
     assert(Array.isArray(key));
-    return getgeneric(t, hashstr(key));
+    return getgeneric(t, lstring.luaS_hash(key));
 };
 
 const luaH_get = function(t, key) {
diff --git a/src/ltm.js b/src/ltm.js
index 4bf37fe..3647308 100644
--- a/src/ltm.js
+++ b/src/ltm.js
@@ -7,6 +7,7 @@ const defs    = require('./defs.js');
 const lobject = require('./lobject.js');
 const ldo     = require('./ldo.js');
 const lstate  = require('./lstate.js');
+const lstring = require('./lstring.js');
 const ltable  = require('./ltable.js');
 const ldebug  = require('./ldebug.js');
 const lvm     = require('./lvm.js');
@@ -61,7 +62,7 @@ const ttypename = function(t) {
 const luaT_init = function(L) {
     L.l_G.tmname = [];
     for (let event in TMS)
-        L.l_G.tmname.push(L.l_G.intern(TMS[event]));
+        L.l_G.tmname.push(new lobject.TValue(CT.LUA_TLNGSTR, lstring.luaS_new(L, TMS[event])));
 };
 
 /*
diff --git a/src/lundump.js b/src/lundump.js
index 1570588..3e56a84 100644
--- a/src/lundump.js
+++ b/src/lundump.js
@@ -7,6 +7,7 @@ const defs     = require('./defs.js');
 const lfunc    = require('./lfunc.js');
 const lobject  = require('./lobject.js');
 const lopcodes = require('./lopcodes.js');
+const lstring  = require('./lstring.js');
 const llex     = require('./llex.js');
 
 let LUAC_DATA = [0x19, 0x93, defs.char["\r"], defs.char["\n"], 0x1a, defs.char["\n"]];
@@ -36,6 +37,7 @@ class BytecodeParser {
 
     read(size) {
         let buffer = this.buffer.read(size);
+        assert(Array.isArray(buffer));
         if (buffer.length < size) this.error("truncated");
         return buffer;
     }
@@ -74,8 +76,8 @@ class BytecodeParser {
         return this.dv.getFloat64(0, true);
     }
 
-    read8bitString(n) {
-        let size = typeof n !== 'undefined' ? n : Math.max(this.readByte() - 1, 0);
+    readString() {
+        let size = Math.max(this.readByte() - 1, 0);
 
         if (size + 1 === 0xFF)
             size = this.readSize_t() - 1;
@@ -84,25 +86,7 @@ class BytecodeParser {
             return null;
         }
 
-        let string = [];
-
-        for (let i = 0; i < size; i++)
-            string.push(this.readByte());
-
-        return string;
-    }
-
-    readString(n) {
-        let size = typeof n !== 'undefined' ? n : Math.max(this.readByte() - 1, 0);
-
-        if (size + 1 === 0xFF)
-            size = this.readSize_t() - 1;
-
-        if (size === 0) {
-            return null;
-        }
-
-        return this.read(size);
+        return lstring.luaS_new(this.L, this.read(size));
     }
 
     /* creates a mask with 'n' 1 bits at position 'p' */
@@ -176,7 +160,7 @@ class BytecodeParser {
                 break;
             case defs.CT.LUA_TSHRSTR:
             case defs.CT.LUA_TLNGSTR:
-                f.k.push(this.L.l_G.intern(this.read8bitString()));
+                f.k.push(new lobject.TValue(defs.CT.LUA_TLNGSTR, this.readString()));
                 break;
             default:
                 this.error(`unrecognized constant '${t}'`);
@@ -229,9 +213,14 @@ class BytecodeParser {
         this.readDebug(f);
     }
 
+    checkliteral(s, msg) {
+        let buff = this.read(s.length);
+        if (buff.join() !== s.join())
+            this.error(msg);
+    }
+
     checkHeader() {
-        if (this.readString(3).join() !== defs.to_luastring(defs.LUA_SIGNATURE.substring(1)).join())  /* 1st char already checked */
-            this.error("bad LUA_SIGNATURE, expected '<esc>Lua'");
+        this.checkliteral(defs.to_luastring(defs.LUA_SIGNATURE.substring(1)), "bad LUA_SIGNATURE, expected '<esc>Lua'"); /* 1st char already checked */
 
         if (this.readByte() !== 0x53)
             this.error("bad Lua version, expected 5.3");
@@ -239,8 +228,7 @@ class BytecodeParser {
         if (this.readByte() !== 0)
             this.error("supports only official PUC-Rio implementation");
 
-        if (this.readString(6).join() !== LUAC_DATA.join())
-            this.error("bytecode corrupted");
+        this.checkliteral(LUAC_DATA, "bytecode corrupted");
 
         this.intSize         = this.readByte();
         this.size_tSize      = this.readByte();
diff --git a/src/lvm.js b/src/lvm.js
index ef5d370..8bfef40 100644
--- a/src/lvm.js
+++ b/src/lvm.js
@@ -9,6 +9,7 @@ const luaconf     = require('./luaconf.js');
 const lobject     = require('./lobject.js');
 const lfunc       = require('./lfunc.js');
 const lstate      = require('./lstate.js');
+const lstring     = require('./lstring.js');
 const llimit      = require('./llimit.js');
 const ldo         = require('./ldo.js');
 const ltm         = require('./ltm.js');
@@ -991,7 +992,7 @@ const tostring = function(L, i) {
     if (o.ttisstring()) return true;
 
     if (o.ttisnumber() && !isNaN(o.value)) {
-        L.stack[i] = L.l_G.intern(defs.to_luastring(`${o.value}`));
+        L.stack[i] = new lobject.TValue(CT.LUA_TLNGSTR, lstring.luaS_bless(L, defs.to_luastring(`${o.value}`)));
         return true;
     }
 
@@ -1029,7 +1030,7 @@ const luaV_concat = function(L, total) {
                 ts = ts.concat(L.stack[top - i].value);
             }
 
-            L.stack[top - n] = L.l_G.intern(ts);
+            L.stack[top - n] = new lobject.TValue(CT.LUA_TLNGSTR, lstring.luaS_bless(L, ts));
         }
         total -= n - 1; /* got 'n' strings to create 1 new */
         L.top -= n - 1; /* popped 'n' strings and pushed one */