porffor 0.17.0-a818353a5 → 0.17.0-b4e7f7ee0

package/compiler/pgo.js

@@ -91,7 +91,9 @@ export const run = obj => {
       },
       w: (ind, outPtr) => { // readArgv
         const pgoInd = process.argv.indexOf('--pgo');
-        const args = process.argv.slice(pgoInd).filter(x => !x.startsWith('-'));
+        let args = process.argv.slice(pgoInd);
+        args = args.slice(args.findIndex(x => !x.startsWith('-')) + 1);
+
         const str = args[ind - 1];
         if (pgoInd === -1 || !str) {
           if (Prefs.pgoLog) console.log('\nPGO warning: script was expecting arguments, please specify args to use for PGO after --pgo arg');
@@ -100,6 +102,9 @@ export const run = obj => {
 
         writeByteStr(exports.$, outPtr, str);
         return str.length;
+      },
+      q: (pathPtr, outPtr) => {
+        return -1;
       }
     }, () => {});
 
@@ -179,6 +184,9 @@ export const run = obj => {
 
   log = '';
   for (const x of funcs) {
+    // skip pgo opt for main()
+    if (x.name === 'main') continue;
+
     const wasmFunc = wasmFuncs.find(y => y.name === x.name);
 
     let targets = [];

package/compiler/types.js

@@ -20,10 +20,26 @@ export const TYPE_NAMES = {
   [TYPES.bigint]: 'BigInt'
 };
 
+// flags
+export const TYPE_FLAGS = {
+  // iterable:  0b10000000,
+  length:    0b01000000,
+};
+
+// TYPES.string |= TYPE_FLAGS.iterable;
+TYPES.string |= TYPE_FLAGS.length;
+
+export const typeHasFlag = (type, flag) => (type & flag) !== 0;
+
 export const INTERNAL_TYPE_BASE = 0x10;
 let internalTypeIndex = INTERNAL_TYPE_BASE;
-const registerInternalType = name => {
-  const n = internalTypeIndex++;
+const registerInternalType = (name, flags = []) => {
+  let n = internalTypeIndex++;
+
+  for (const x of flags) {
+    if (TYPE_FLAGS[x]) n |= TYPE_FLAGS[x];
+  }
+
   TYPES[name.toLowerCase()] = n;
   TYPE_NAMES[n] = name;
 };
@@ -31,8 +47,18 @@ const registerInternalType = name => {
 // note: when adding a new internal type, please also add a deserializer to wrap.js
 // (it is okay to add a throw todo deserializer for wips)
 
-registerInternalType('Array');
+registerInternalType('Array', ['iterable', 'length']);
 registerInternalType('RegExp');
-registerInternalType('ByteString');
+registerInternalType('ByteString', ['iterable', 'length']);
 registerInternalType('Date');
-registerInternalType('Set');
+registerInternalType('Set', ['iterable']);
+
+registerInternalType('Uint8Array', ['iterable', 'length']);
+registerInternalType('Int8Array', ['iterable', 'length']);
+registerInternalType('Uint8ClampedArray', ['iterable', 'length']);
+registerInternalType('Uint16Array', ['iterable', 'length']);
+registerInternalType('Int16Array', ['iterable', 'length']);
+registerInternalType('Uint32Array', ['iterable', 'length']);
+registerInternalType('Int32Array', ['iterable', 'length']);
+registerInternalType('Float32Array', ['iterable', 'length']);
+registerInternalType('Float64Array', ['iterable', 'length']);

package/compiler/wasmSpec.js

@@ -68,6 +68,7 @@ export const Opcodes = {
 
   i32_load: 0x28,
   i64_load: 0x29,
+  f32_load: 0x2a,
   f64_load: 0x2b,
 
   i32_load8_s: 0x2c,
@@ -82,6 +83,7 @@ export const Opcodes = {
 
   i32_store: 0x36,
   i64_store: 0x37,
+  f32_store: 0x38,
   f64_store: 0x39,
 
   i32_store8: 0x3a,

package/compiler/wrap.js

@@ -2,7 +2,7 @@ import { encodeVector, encodeLocal } from './encoding.js';
 import { importedFuncs } from './builtins.js';
 import compile from './index.js';
 import decompile from './decompile.js';
-import { TYPES } from './types.js';
+import { TYPES, TYPE_NAMES } from './types.js';
 import { log } from './log.js';
 import Prefs from './prefs.js';
 
@@ -115,6 +115,19 @@ const porfToJSValue = ({ memory, funcs, pages }, value, type) => {
       return Symbol(desc);
     }
 
+    case TYPES.uint8array:
+    case TYPES.int8array:
+    case TYPES.uint8clampedarray:
+    case TYPES.uint16array:
+    case TYPES.int16array:
+    case TYPES.uint32array:
+    case TYPES.int32array:
+    case TYPES.float32array:
+    case TYPES.float64array: {
+      const length = (new Int32Array(memory.buffer, value, 1))[0];
+      return new globalThis[TYPE_NAMES[type]](memory.buffer, value + 4, length);
+    }
+
     default: return value;
   }
 };
@@ -127,7 +140,7 @@ export default (source, flags = [ 'module' ], customImports = {}, print = str =>
 
   globalThis.porfDebugInfo = { funcs, globals };
 
-  if (process.argv[1].includes('/runner') && source.includes?.('export ')) flags.push('module');
+  // if (process.argv[1].includes('/runner') && source.includes?.('export ')) flags.push('module');
 
   // fs.writeFileSync('out.wasm', Buffer.from(wasm));
 
@@ -235,8 +248,10 @@ export default (source, flags = [ 'module' ], customImports = {}, print = str =>
         y: () => {},
         z: () => {},
         w: (ind, outPtr) => { // readArgv
-          const args = process.argv.slice(2).filter(x => !x.startsWith('-'));
-          const str = args[ind];
+          let args = process.argv.slice(2);
+          args = args.slice(args.findIndex(x => !x.startsWith('-')) + 1);
+
+          const str = args[ind - 1];
           if (!str) return -1;
 
           writeByteStr(memory, outPtr, str);
@@ -244,7 +259,7 @@ export default (source, flags = [ 'module' ], customImports = {}, print = str =>
         },
         q: (pathPtr, outPtr) => { // readFile
           try {
-            const path = readByteStr(memory, pathPtr);
+            const path = pathPtr === 0 ? 0 : readByteStr(memory, pathPtr);
             const contents = fs.readFileSync(path, 'utf8');
             writeByteStr(memory, outPtr, contents);
             return contents.length;

package/package.json

@@ -1,7 +1,7 @@
 {
   "name": "porffor",
   "description": "a basic experimental wip aot optimizing js -> wasm engine/compiler/runtime in js",
-  "version": "0.17.0-a818353a5",
+  "version": "0.17.0-b4e7f7ee0",
   "author": "CanadaHonk",
   "license": "MIT",
   "scripts": {},

package/rhemyn/README.md

@@ -24,13 +24,16 @@ Made for use with Porffor but could possibly be adapted, implementation/library
   - 🟢 digit, not digit (eg `\d\D`)
   - 🟢 word, not word (eg `\w\W`)
   - 🟢 whitespace, not whitespace (eg `\s\S`)
-- 🟠 quantifiers
-  - 🟠 star (eg `a*`)
-  - 🟠 plus (eg `a+`)
-  - 🟠 optional (eg `a?`)
+- 🟡 quantifiers
+  - 🟡 star (eg `a*`)
+  - 🟡 plus (eg `a+`)
+  - 🟡 optional (eg `a?`)
   - 🟠 lazy modifier (eg `a*?`)
   - 🔴 n repetitions (eg `a{4}`)
   - 🔴 n-m repetitions (eg `a{2,4}`)
+- 🟠 groups
+  - 🟠 capturing groups (`(a)`)
+  - 🔴 non-capturing groups (`(?:a)`)
 - 🔴 assertions
   - 🔴 beginning (eg `^a`)
   - 🔴 end (eg `a$`)

package/rhemyn/compile.js

@@ -12,13 +12,12 @@ const Counter = 3; // what char we are running on
 const Pointer = 4; // next char BYTE pointer
 const Length = 5;
 const Tmp = 6;
+const QuantifierTmp = 7; // the temporary variable used for quanitifers
 
-let exprLastGet = false;
 const generate = (node, negated = false, get = true, stringSize = 2, func = 'test') => {
   let out = [];
   switch (node.type) {
     case 'Expression':
-      exprLastGet = false;
       out = [
         // set length local
         [ Opcodes.local_get, BasePointer ],
@@ -32,46 +31,40 @@ const generate = (node, negated = false, get = true, stringSize = 2, func = 'tes
         [ Opcodes.local_set, IterPointer ],
 
         [ Opcodes.loop, Blocktype.void ],
-
-        // reset pointer as iter pointer
-        [ Opcodes.local_get, IterPointer ],
-        [ Opcodes.local_set, Pointer ],
-
-        [ Opcodes.block, Blocktype.void ],
-
-        // generate checks
-        ...node.body.flatMap((x, i) => {
-          exprLastGet = x.type !== 'Group' && i === (node.body.length - 1);
-          return generate(x, negated, true, stringSize, func);
-        }),
-
-        // reached end without branching out, successful match
-        ...({
-          test: number(1, Valtype.i32),
-          search: [
-            [ Opcodes.local_get, Counter ]
-          ]
-        })[func],
-        [ Opcodes.return ],
-
-        [ Opcodes.end ],
-
-        // increment iter pointer by string size
-        [ Opcodes.local_get, IterPointer ],
-        ...number(stringSize, Valtype.i32),
-        [ Opcodes.i32_add ],
-        [ Opcodes.local_set, IterPointer ],
-
-        // increment counter by 1, check if eq length, if not loop
-        [ Opcodes.local_get, Counter ],
-        ...number(1, Valtype.i32),
-        [ Opcodes.i32_add ],
-        [ Opcodes.local_tee, Counter ],
-
-        [ Opcodes.local_get, Length ],
-        [ Opcodes.i32_ne ],
-
-        [ Opcodes.br_if, 0 ],
+          // reset pointer as iter pointer
+          [ Opcodes.local_get, IterPointer ],
+          [ Opcodes.local_set, Pointer ],
+
+          [ Opcodes.block, Blocktype.void ],
+            // generate checks
+            ...node.body.flatMap(x => generate(x, negated, true, stringSize, func)),
+
+            // reached end without branching out, successful match
+            ...({
+              test: number(1, Valtype.i32),
+              search: [
+                [ Opcodes.local_get, Counter ]
+              ]
+            })[func],
+            [ Opcodes.return ],
+          [ Opcodes.end ],
+
+          // increment iter pointer by string size
+          [ Opcodes.local_get, IterPointer ],
+          ...number(stringSize, Valtype.i32),
+          [ Opcodes.i32_add ],
+          [ Opcodes.local_set, IterPointer ],
+
+          // increment counter by 1, check if eq length, if not loop
+          [ Opcodes.local_get, Counter ],
+          ...number(1, Valtype.i32),
+          [ Opcodes.i32_add ],
+          [ Opcodes.local_tee, Counter ],
+
+          [ Opcodes.local_get, Length ],
+          [ Opcodes.i32_ne ],
+
+          [ Opcodes.br_if, 0 ],
         [ Opcodes.end ],
 
         // no match, return 0
@@ -110,12 +103,12 @@ const generate = (node, negated = false, get = true, stringSize = 2, func = 'tes
   return out;
 };
 
-const getNextChar = (stringSize) => [
+const getNextChar = (stringSize, peek = false) => [
   // get char from pointer
   [ Opcodes.local_get, Pointer ],
   [ stringSize == 2 ? Opcodes.i32_load16_u : Opcodes.i32_load8_u, 0, 0 ],
 
-  ...(exprLastGet ? [] : [
+  ...(peek ? [] : [
     // pointer += string size
     [ Opcodes.local_get, Pointer ],
     ...number(stringSize, Valtype.i32),
@@ -134,11 +127,81 @@ const checkFailure = () => [
   [ Opcodes.br_if, 0 ]
 ];
 
-const generateChar = (node, negated, get, stringSize) => {
+const wrapQuantifier = (node, method, get, stringSize) => {
+  const [ min, max ] = node.quantifier;
   return [
-    ...(get ? getNextChar(stringSize) : []),
+    // initalize our temp value (number of matched characters)
+    ...number(0, Valtype.i32),
+    [Opcodes.local_set, QuantifierTmp],
+
+    // start loop
+    [Opcodes.loop, Blocktype.void],
+      [ Opcodes.block, Blocktype.void ],
+        // if counter + tmp == length, break
+        [ Opcodes.local_get, Counter ],
+        [ Opcodes.local_get, QuantifierTmp ],
+        [ Opcodes.i32_add ],
+        [ Opcodes.local_get, Length ],
+        [ Opcodes.i32_eq ],
+        [ Opcodes.br_if, 0 ],
+
+        // if doesn't match, break
+        ...method,
+        [Opcodes.br_if, 0 ],
+        ...(get ? [
+          // pointer += stringSize
+          [ Opcodes.local_get, Pointer ],
+          ...number(stringSize, Valtype.i32),
+          [ Opcodes.i32_add ],
+          [ Opcodes.local_set, Pointer ]
+        ] : []),
+
+        // if maximum was reached, break
+        ...(max ? [
+          [ Opcodes.local_get, QuantifierTmp ],
+          ...number(max, Valtype.i32),
+          [ Opcodes.i32_eq ],
+          [ Opcodes.br_if, 0 ]
+        ] : []),
+
+        [ Opcodes.local_get, QuantifierTmp ],
+        ...number(1, Valtype.i32),
+        [ Opcodes.i32_add ],
+        [ Opcodes.local_set, QuantifierTmp ],
+        [ Opcodes.br, 1 ],
+      [ Opcodes.end ],
+    [ Opcodes.end ],
+
+    // if less than minimum, fail
+    [Opcodes.local_get, QuantifierTmp],
+    ...number(min, Valtype.i32),
+    [Opcodes.i32_lt_s],
+    ...(get ? checkFailure(): []),
+
+    // counter += tmp - 1
+    [ Opcodes.local_get, QuantifierTmp ],
+    ...number(1, Valtype.i32),
+    [ Opcodes.i32_sub ],
+    [ Opcodes.local_get, Counter ],
+    [ Opcodes.i32_add ],
+    [ Opcodes.local_set, Counter ]
+  ];
+}
+
+const generateChar = (node, negated, get, stringSize) => {
+  const hasQuantifier = !!node.quantifier;
+  const out = [
+    ...(get ? getNextChar(stringSize, hasQuantifier) : []),
     ...number(node.char.charCodeAt(0), Valtype.i32),
     negated ? [ Opcodes.i32_eq ] : [ Opcodes.i32_ne ],
+  ];
+
+  if (node.quantifier) {
+    return wrapQuantifier(node, out, get, stringSize);
+  }
+
+  return [
+    ...out,
     ...(get ? checkFailure(): [])
   ];
 };
@@ -146,21 +209,31 @@ const generateChar = (node, negated, get, stringSize) => {
 const generateSet = (node, negated, get, stringSize) => {
   // for a single char we do not need a tmp, it is like just
   const singleChar = node.body.length === 1 && node.body[0].type === 'Character';
+  if (singleChar) return generateChar(node.body[0], negated, get, stringSize)
 
-  let out = [
-    ...(get ? getNextChar(stringSize) : []),
-    ...(singleChar ? [] : [ [ Opcodes.local_set, Tmp ] ]),
+  const hasQuantifier = !!node.quantifier;
+
+  const out = [
+    ...(get ? getNextChar(stringSize, hasQuantifier) : []),
+    [ Opcodes.local_set, Tmp ],
   ];
 
   for (const x of node.body) {
-    out = [
-      ...out,
-      ...(singleChar ? [] : [ [ Opcodes.local_get, Tmp ] ]),
+    out.push(
+      [ Opcodes.local_get, Tmp ],
       ...generate(x, negated, false, stringSize)
-    ];
+    );
   }
 
-  if (node.body.length > 0) out = out.concat(new Array(node.body.length - 1).fill(negated ? [ Opcodes.i32_or ] : [ Opcodes.i32_and ]));
+  if (node.body.length > 0) {
+    for (let i = 0; i < node.body.length - 1; i++) {
+      out.push(negated ? [ Opcodes.i32_or ] : [ Opcodes.i32_and ])
+    }
+  };
+
+  if (hasQuantifier) {
+    return wrapQuantifier(node, out, get, stringSize);
+  }
 
   return [
     ...out,
@@ -196,7 +269,7 @@ const wrapFunc = (regex, func, name, index) => {
   const parsed = parse(regex);
 
   return outputFunc([
-    [ Opcodes.local_get, 1 ],
+    [ Opcodes.local_get, IterPointer ],
     ...number(TYPES.string, Valtype.i32),
     [ Opcodes.i32_eq ],
     [ Opcodes.if, Valtype.i32 ],
@@ -229,5 +302,6 @@ const outputFunc = (wasm, name, index) => ({
     pointer: { idx: 4, type: Valtype.i32 },
     length: { idx: 5, type: Valtype.i32 },
     tmp: { idx: 6, type: Valtype.i32 },
+    quantifierTmp: { idx: 7, type: Valtype.i32 },
   }
 });

package/runner/debug.js

@@ -43,7 +43,7 @@ let lastLine;
 let output = '';
 
 try {
-  const { exports } = await compile(source, process.argv.includes('--module') ? [ 'module' ] : [], {
+  const { exports } = compile(source, process.argv.includes('--module') ? [ 'module' ] : [], {
     y: n => {
       if (callStarts[callStarts.length - 1] === n - 1) {
         // end of call

package/runner/index.js

@@ -130,14 +130,14 @@ const print = str => {
 let runStart;
 try {
   if (process.argv.includes('-b')) {
-    const { wasm, exports } = await compile(source, process.argv.includes('--module') ? [ 'module' ] : [], {}, print);
+    const { wasm, exports } = compile(source, process.argv.includes('--module') ? [ 'module' ] : [], {}, print);
 
     runStart = performance.now();
     if (!process.argv.includes('--no-run')) exports.main();
 
     console.log(`\n\nwasm size: ${wasm.byteLength} bytes`);
   } else {
-    const { exports } = await compile(source, process.argv.includes('--module') ? [ 'module' ] : [], {}, print);
+    const { exports } = compile(source, process.argv.includes('--module') ? [ 'module' ] : [], {}, print);
 
     runStart = performance.now();
     if (!process.argv.includes('--no-run')) exports.main();
@@ -146,7 +146,7 @@ try {
 } catch (e) {
   // if (cache) process.stdout.write(cache);
   let out = e;
-  if (!process.argv.includes('-i') && e.__proto__.message != null) out = `${e.constructor.name}${e.message != null ? `: ${e.message}` : ''}`;
+  if (!process.argv.includes('-i') && Object.getPrototypeOf(e).message != null) out = `${e.constructor.name}${e.message != null ? `: ${e.message}` : ''}`;
   console.error(out);
 }
 

package/runner/profile.js

@@ -20,7 +20,7 @@ let spin = 0;
 let last = 0;
 
 try {
-  const { exports } = await compile(source, process.argv.includes('--module') ? [ 'module' ] : [], {
+  const { exports } = compile(source, process.argv.includes('--module') ? [ 'module' ] : [], {
     y: n => {
       tmp[n] = performance.now();
     },

package/runner/repl.js

@@ -40,10 +40,9 @@ let lastMemory, lastPages;
 const PageSize = 65536;
 const memoryToString = mem => {
   let out = '';
-  const pages = lastPages.length;
   const wasmPages = mem.buffer.byteLength / PageSize;
 
-  out += `\x1B[1mallocated ${mem.buffer.byteLength / 1024}KB\x1B[0m for ${pages} thing${pages === 1 ? '' : 's'} using ${wasmPages} Wasm page${wasmPages === 1 ? '' : 's'}\n\n`;
+  out += `\x1B[1mallocated ${mem.buffer.byteLength / 1024}KiB\x1B[0m (using ${wasmPages} Wasm page${wasmPages === 1 ? '' : 's'})\n\n`;
 
   const buf = new Uint8Array(mem.buffer);
 
@@ -55,10 +54,16 @@ const memoryToString = mem => {
   }
 
   out += `\x1B[0m\x1B[1m  name${' '.repeat(longestName - 4)} \x1B[0m\x1B[90m│\x1B[0m\x1B[1m type${' '.repeat(longestType - 4)} \x1B[0m\x1B[90m│\x1B[0m\x1B[1m memory\x1B[0m\n`; // ─
-  for (let i = 0; i < pages; i++) {
-    const [ type, name ] = lastPages[i].split(': ');
-    // out += `\x1B[36m${lastPages[i].replace(':', '\x1B[90m:\x1B[34m')}\x1B[90m${' '.repeat(longestName - lastPages[i].length)} | \x1B[0m`;
-    out += `  \x1B[34m${name}${' '.repeat(longestName - name.length)} \x1B[90m│\x1B[0m \x1B[36m${type}${' '.repeat(longestType - type.length)} \x1B[90m│\x1B[0m `;
+  for (let i = 0; i < wasmPages; i++) {
+    if (lastPages[i]) {
+      const [ type, name ] = lastPages[i].split(': ');
+      // out += `\x1B[36m${lastPages[i].replace(':', '\x1B[90m:\x1B[34m')}\x1B[90m${' '.repeat(longestName - lastPages[i].length)} | \x1B[0m`;
+      out += `  \x1B[34m${name}${' '.repeat(longestName - name.length)} \x1B[90m│\x1B[0m \x1B[36m${type}${' '.repeat(longestType - type.length)} \x1B[90m│\x1B[0m `;
+    } else {
+      const type = '???';
+      const name = '???';
+      out += `  \x1B[34m${name}${' '.repeat(longestName - name.length)} \x1B[90m│\x1B[0m \x1B[36m${type}${' '.repeat(longestType - type.length)} \x1B[90m│\x1B[0m `;
+    }
 
     for (let j = 0; j < 40; j++) {
       const val = buf[i * pageSize + j];
@@ -75,13 +80,13 @@ const memoryToString = mem => {
 };
 
 let prev = '';
-const run = async (source, _context, _filename, callback, run = true) => {
+const run = (source, _context, _filename, callback, run = true) => {
   // hack: print "secret" before latest code ran to only enable printing for new code
 
   let toRun = (prev ? (prev + `;\nprint(-0x1337);\n`) : '') + source.trim();
 
   let shouldPrint = !prev;
-  const { exports, pages } = await compile(toRun, [], {}, str => {
+  const { exports, pages } = compile(toRun, [], {}, str => {
     if (shouldPrint) process.stdout.write(str);
     if (str === '-4919') shouldPrint = true;
   });
@@ -122,12 +127,12 @@ replServer.defineCommand('memory', {
 });
 replServer.defineCommand('asm', {
   help: 'Log Wasm decompiled bytecode',
-  async action() {
+  action() {
     this.clearBufferedCommand();
 
     try {
       process.argv.push('--opt-funcs');
-      await run('', null, null, () => {}, false);
+      run('', null, null, () => {}, false);
       process.argv.pop();
     } catch { }
 
@@ -136,7 +141,7 @@ replServer.defineCommand('asm', {
 });
 replServer.defineCommand('js', {
   help: 'Log JS being actually ran',
-  async action() {
+  action() {
     this.clearBufferedCommand();
     console.log(prev);
     this.displayPrompt();