String are represented by Array of 8-bit numbers

3 files changed, 69 insertions, 56 deletions

diff --git a/src/lstate.js b/src/lstate.js
index 2e6ebbe..519abde 100644
--- a/src/lstate.js
+++ b/src/lstate.js
@@ -69,7 +69,7 @@ class global_State {
 
     constructor(L) {
         this.mainthread = L;
-        this.strt = null; // TODO: string hash table
+        this.strt = new Map();
         this.l_registry = nil;
         this.panic = null;
         this.version = null;
@@ -77,6 +77,15 @@ class global_State {
         this.mt = new Array(LUA_NUMTAGS);
     }
 
+    intern(stringArray) {
+        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);
+    }
+
 }
 
 
diff --git a/src/lua.js b/src/lua.js
index 95d0ef8..971593d 100644
--- a/src/lua.js
+++ b/src/lua.js
@@ -132,6 +132,61 @@ class lua_Debug {
 
 }
 
+const to_luastring = function(str, maxBytesToWrite) {
+    let outU8Array = new Array(maxBytesToWrite);
+
+    if (!(maxBytesToWrite > 0)) // Parameter maxBytesToWrite is not optional. Negative values, 0, null, undefined and false each don't write out any bytes.
+      return 0;
+  
+    let outIdx = 0;
+    let startIdx = 0;
+    let endIdx = maxBytesToWrite - 1; // -1 for string null terminator.
+    for (let i = 0; i < str.length; ++i) {
+        // Gotcha: charCodeAt returns a 16-bit word that is a UTF-16 encoded code unit, not a Unicode code point of the character! So decode UTF16->UTF32->UTF8.
+        // See http://unicode.org/faq/utf_bom.html#utf16-3
+        // For UTF8 byte structure, see http://en.wikipedia.org/wiki/UTF-8#Description and https://www.ietf.org/rfc/rfc2279.txt and https://tools.ietf.org/html/rfc3629
+        let u = str.charCodeAt(i); // possibly a lead surrogate
+        if (u >= 0xD800 && u <= 0xDFFF) u = 0x10000 + ((u & 0x3FF) << 10) | (str.charCodeAt(++i) & 0x3FF);
+        if (u <= 0x7F) {
+            if (outIdx >= endIdx) break;
+            outU8Array[outIdx++] = u;
+        } else if (u <= 0x7FF) {
+            if (outIdx + 1 >= endIdx) break;
+            outU8Array[outIdx++] = 0xC0 | (u >> 6);
+            outU8Array[outIdx++] = 0x80 | (u & 63);
+        } else if (u <= 0xFFFF) {
+            if (outIdx + 2 >= endIdx) break;
+            outU8Array[outIdx++] = 0xE0 | (u >> 12);
+            outU8Array[outIdx++] = 0x80 | ((u >> 6) & 63);
+            outU8Array[outIdx++] = 0x80 | (u & 63);
+        } else if (u <= 0x1FFFFF) {
+            if (outIdx + 3 >= endIdx) break;
+            outU8Array[outIdx++] = 0xF0 | (u >> 18);
+            outU8Array[outIdx++] = 0x80 | ((u >> 12) & 63);
+            outU8Array[outIdx++] = 0x80 | ((u >> 6) & 63);
+            outU8Array[outIdx++] = 0x80 | (u & 63);
+        } else if (u <= 0x3FFFFFF) {
+            if (outIdx + 4 >= endIdx) break;
+            outU8Array[outIdx++] = 0xF8 | (u >> 24);
+            outU8Array[outIdx++] = 0x80 | ((u >> 18) & 63);
+            outU8Array[outIdx++] = 0x80 | ((u >> 12) & 63);
+            outU8Array[outIdx++] = 0x80 | ((u >> 6) & 63);
+            outU8Array[outIdx++] = 0x80 | (u & 63);
+        } else {
+            if (outIdx + 5 >= endIdx) break;
+            outU8Array[outIdx++] = 0xFC | (u >> 30);
+            outU8Array[outIdx++] = 0x80 | ((u >> 24) & 63);
+            outU8Array[outIdx++] = 0x80 | ((u >> 18) & 63);
+            outU8Array[outIdx++] = 0x80 | ((u >> 12) & 63);
+            outU8Array[outIdx++] = 0x80 | ((u >> 6) & 63);
+            outU8Array[outIdx++] = 0x80 | (u & 63);
+        }
+    }
+    // Null-terminate the pointer to the buffer.
+    outU8Array[outIdx] = 0;
+    return outU8Array;
+};
+
 module.exports.CT                      = CT;
 module.exports.FENGARI_AUTHORS         = FENGARI_AUTHORS;
 module.exports.FENGARI_COPYRIGHT       = FENGARI_COPYRIGHT;
@@ -181,4 +236,5 @@ module.exports.constant_types          = constant_types;
 module.exports.lua_Debug               = lua_Debug;
 module.exports.lua_upvalueindex        = lua_upvalueindex;
 module.exports.print_version           = print_version;
-module.exports.thread_status           = thread_status;
\ No newline at end of file
+module.exports.thread_status           = thread_status;
+module.exports.to_luastring            = to_luastring;
\ No newline at end of file
diff --git a/src/lundump.js b/src/lundump.js
index 92063a2..aa66584 100644
--- a/src/lundump.js
+++ b/src/lundump.js
@@ -13,18 +13,8 @@ const OpCodes        = require('./lopcodes.js');
 
 const LUAI_MAXSHORTLEN = 40;
 
-/**
- * Parse Lua 5.3 bytecode
- * @see {@link http://www.lua.org/source/5.3/lundump.c.html|lundump.c}
- */
 class BytecodeParser {
 
-    /**
-     * Initilialize bytecode parser
-     * @constructor
-     * @param {lua_State} Lua state object
-     * @param {DataView} dataView Contains the binary data
-     */
     constructor(dataView) {
         this.intSize = 4;
         this.size_tSize = 8;
@@ -79,7 +69,6 @@ class BytecodeParser {
         return number;
     }
 
-    // TODO: 8-bit clean strings
     readString(n) {
         let size = typeof n !== 'undefined' ? n : this.readByte() - 1;
 
@@ -90,10 +79,10 @@ class BytecodeParser {
             return null;
         }
 
-        let string = "";
+        let string = new Uint8Array();
 
         for (let i = 0; i < size; i++)
-            string += String.fromCharCode(this.readByte());
+            string.push(this.readByte());
 
         return string;
     }
@@ -133,8 +122,6 @@ class BytecodeParser {
                 Ax:     (ins >> o.POS_Ax) & p.MASK1(o.SIZE_Ax, 0),
                 sBx:    ((ins >> o.POS_Bx) & p.MASK1(o.SIZE_Bx, 0)) - o.MAXARG_sBx
             };
-
-            // console.log(`   [${i}]    Op: ${o.OpCodes[f.code[i].opcode]}  A: ${f.code[i].A}  B: ${f.code[i].B}  C: ${f.code[i].C}  Ax: ${f.code[i].Ax}  Bx: ${f.code[i].Bx}  sBx: ${f.code[i].sBx}`);
         }
     }
 
@@ -147,12 +134,6 @@ class BytecodeParser {
                 instack: this.readByte(),
                 idx:     this.readByte()
             };
-
-            // console.log(`
-            //     f.upvalues[${i}].name    = ${f.upvalues[i].name}
-            //     f.upvalues[${i}].instack = ${f.upvalues[i].instack}
-            //     f.upvalues[${i}].idx     = ${f.upvalues[i].idx}
-            // `);
         }
     }
 
@@ -165,24 +146,19 @@ class BytecodeParser {
             switch (t) {
             case constant_types.LUA_TNIL:
                 f.k.push(new TValue(constant_types.LUA_TNIL, null));
-                // console.log(`   LUA_TNIL     = ${f.k[f.k.length - 1].value}`);
                 break;
             case constant_types.LUA_TBOOLEAN:
                 f.k.push(new TValue(constant_types.LUA_TBOOLEAN, this.readByte()));
-                // console.log(`   LUA_TBOOLEAN = ${f.k[f.k.length - 1].value}`);
                 break;
             case constant_types.LUA_TNUMFLT:
                 f.k.push(new TValue(constant_types.LUA_TNUMFLT, this.readNumber()));
-                // console.log(`   LUA_TNUMFLT  = ${f.k[f.k.length - 1].value}`);
                 break;
             case constant_types.LUA_TNUMINT:
                 f.k.push(new TValue(constant_types.LUA_TNUMINT, this.readInteger()));
-                // console.log(`   LUA_TNUMINT  = ${f.k[f.k.length - 1].value}`);
                 break;
             case constant_types.LUA_TSHRSTR:
             case constant_types.LUA_TLNGSTR:
                 f.k.push(new TValue(constant_types.LUA_TLNGSTR, this.readString()));
-                // console.log(`   LUA_TLNGSTR  = ${f.k[f.k.length - 1].value}`);
                 break;
             default:
                 throw new Error(`unrecognized constant '${t}'`);
@@ -211,21 +187,11 @@ class BytecodeParser {
                 startpc: this.readInt(),
                 endpc:   this.readInt()
             };
-
-            // console.log(`
-            //     f.locvars[${i}].varname = ${f.locvars[i].varname}
-            //     f.locvars[${i}].startpc = ${f.locvars[i].startpc}
-            //     f.locvars[${i}].endpc   = ${f.locvars[i].endpc}
-            // `);
         }
 
         n = this.readInt();
         for (let i = 0; i < n; i++) {
             f.upvalues[i].name = this.readString();
-
-            // console.log(`
-            //     f.upvalues[${i}].name = ${f.upvalues[i].name}
-            // `);
         }
     }
 
@@ -238,16 +204,6 @@ class BytecodeParser {
         f.numparams = this.readByte();
         f.is_vararg = this.readByte();
         f.maxstacksize = this.readByte();
-
-        // console.log(`
-        //     f.source          = ${f.source}
-        //     f.linedefined     = ${f.linedefined}
-        //     f.lastlinedefined = ${f.lastlinedefined}
-        //     f.numparams       = ${f.numparams}
-        //     f.is_vararg       = ${f.is_vararg}
-        //     f.maxstacksize    = ${f.maxstacksize}
-        // `);
-
         this.readCode(f);
         this.readConstants(f);
         this.readUpvalues(f);
@@ -274,14 +230,6 @@ class BytecodeParser {
         this.integerSize     = this.readByte();
         this.numberSize      = this.readByte();
 
-        // console.log(`
-        //     intSize         = ${this.intSize}
-        //     size_tSize      = ${this.size_tSize}
-        //     instructionSize = ${this.instructionSize}
-        //     integerSize     = ${this.integerSize}
-        //     numberSize      = ${this.numberSize}
-        // `)
-
         if (this.readInteger() !== 0x5678)
             throw new Error("endianness mismatch");