porffor 0.0.0-ebc0491 → 0.0.0-f74a73a

package/compiler/codeGen.js

@@ -5,6 +5,7 @@ import { BuiltinFuncs, BuiltinVars, importedFuncs, NULL, UNDEFINED } from "./bui
 import { PrototypeFuncs } from "./prototype.js";
 import { number, i32x4 } from "./embedding.js";
 import parse from "./parse.js";
+import * as Rhemyn from "../rhemyn/compile.js";
 
 let globals = {};
 let globalInd = 0;
@@ -108,8 +109,8 @@ const generate = (scope, decl, global = false, name = undefined) => {
     case 'WhileStatement':
       return generateWhile(scope, decl);
 
-    /* case 'ForOfStatement':
-      return generateForOf(scope, decl); */
+    case 'ForOfStatement':
+      return generateForOf(scope, decl);
 
     case 'BreakStatement':
       return generateBreak(scope, decl);
@@ -731,21 +732,15 @@ const performOp = (scope, op, left, right, leftType, rightType, _global = false,
   ];
 };
 
-let binaryExpDepth = 0;
 const generateBinaryExp = (scope, decl, _global, _name) => {
-  binaryExpDepth++;
-
-  const out = [
-    ...performOp(scope, decl.operator, generate(scope, decl.left), generate(scope, decl.right), getNodeType(scope, decl.left), getNodeType(scope, decl.right), _global, _name)
-  ];
+  const out = performOp(scope, decl.operator, generate(scope, decl.left), generate(scope, decl.right), getNodeType(scope, decl.left), getNodeType(scope, decl.right), _global, _name);
 
   if (valtype !== 'i32' && ['==', '===', '!=', '!==', '>', '>=', '<', '<='].includes(decl.operator)) out.push(Opcodes.i32_from_u);
 
-  binaryExpDepth--;
   return out;
 };
 
-const asmFunc = (name, { wasm, params, locals: localTypes, globals: globalTypes = [], globalInits, returns, returnType, memory, localNames = [], globalNames = [] }) => {
+const asmFunc = (name, { wasm, params, locals: localTypes, globals: globalTypes = [], globalInits, returns, returnType, localNames = [], globalNames = [] }) => {
   const existing = funcs.find(x => x.name === name);
   if (existing) return existing;
 
@@ -781,7 +776,6 @@ const asmFunc = (name, { wasm, params, locals: localTypes, globals: globalTypes
     returns,
     returnType: TYPES[returnType ?? 'number'],
     wasm,
-    memory,
     internal: true,
     index: currentFuncIndex++
   };
@@ -814,7 +808,8 @@ const TYPES = {
   bigint: 0xffffffffffff7,
 
   // these are not "typeof" types but tracked internally
-  _array: 0xffffffffffff8
+  _array: 0xfffffffffff0f,
+  _regexp: 0xfffffffffff1f
 };
 
 const TYPE_NAMES = {
@@ -849,6 +844,8 @@ const getType = (scope, _name) => {
 const getNodeType = (scope, node) => {
   if (node.type === 'Literal') {
     if (['number', 'boolean', 'string', 'undefined', 'object', 'function', 'symbol', 'bigint'].includes(node.value)) return TYPES.number;
+    if (node.regex) return TYPES._regexp;
+
     return TYPES[typeof node.value];
   }
 
@@ -882,6 +879,11 @@ const getNodeType = (scope, node) => {
 
     // literal.func()
     if (!name && node.callee.type === 'MemberExpression') {
+      if (node.callee.object.regex) {
+        const funcName = node.callee.property.name;
+        return Rhemyn[funcName] ? TYPES.boolean : TYPES.undefined;
+      }
+
       const baseType = getNodeType(scope, node.callee.object);
 
       const func = node.callee.property.name;
@@ -925,6 +927,11 @@ const getNodeType = (scope, node) => {
 const generateLiteral = (scope, decl, global, name) => {
   if (decl.value === null) return number(NULL);
 
+  if (decl.regex) {
+    scope.regex[name] = decl.regex;
+    return number(1);
+  }
+
   switch (typeof decl.value) {
     case 'number':
       return number(decl.value);
@@ -1084,6 +1091,25 @@ const generateCall = (scope, decl, _global, _name) => {
 
   // literal.func()
   if (!name && decl.callee.type === 'MemberExpression') {
+    // megahack for /regex/.func()
+    if (decl.callee.object.regex) {
+      const funcName = decl.callee.property.name;
+      const func = Rhemyn[funcName](decl.callee.object.regex.pattern, currentFuncIndex++);
+
+      funcIndex[func.name] = func.index;
+      funcs.push(func);
+
+      return [
+        // make string arg
+        ...generate(scope, decl.arguments[0]),
+
+        // call regex func
+        Opcodes.i32_to_u,
+        [ Opcodes.call, func.index ],
+        Opcodes.i32_from
+      ];
+    }
+
     baseType = getNodeType(scope, decl.callee.object);
 
     const func = decl.callee.property.name;
@@ -1096,6 +1122,31 @@ const generateCall = (scope, decl, _global, _name) => {
     baseName = [...arrays.keys()].pop();
   }
 
+  if (protoName && baseType === TYPES.string && Rhemyn[protoName]) {
+    const func = Rhemyn[protoName](decl.arguments[0].regex.pattern, currentFuncIndex++);
+
+    funcIndex[func.name] = func.index;
+    funcs.push(func);
+
+    const pointer = arrays.get(baseName);
+    const [ local, isGlobal ] = lookupName(scope, baseName);
+
+    return [
+      ...out,
+
+      ...(pointer == null ? [
+        [ isGlobal ? Opcodes.global_get : Opcodes.local_get, local.idx ],
+        Opcodes.i32_to_u,
+      ] : [
+        ...number(pointer, Valtype.i32)
+      ]),
+
+      // call regex func
+      [ Opcodes.call, func.index ],
+      Opcodes.i32_from
+    ];
+  }
+
   if (protoFunc) {
     let pointer = arrays.get(baseName);
 
@@ -1510,7 +1561,7 @@ const generateUpdate = (scope, decl) => {
 };
 
 const generateIf = (scope, decl) => {
-  const out = truthy(scope, generate(scope, decl.test), decl.test);
+  const out = truthy(scope, generate(scope, decl.test), getNodeType(scope, decl.test));
 
   out.push(Opcodes.i32_to, [ Opcodes.if, Blocktype.void ]);
   depth.push('if');
@@ -1603,18 +1654,106 @@ const generateWhile = (scope, decl) => {
 const generateForOf = (scope, decl) => {
   const out = [];
 
+  const rightType = getNodeType(scope, decl.right);
+  const valtypeSize = rightType === TYPES._array ? ValtypeSize[valtype] : ValtypeSize.i16; // presume array (:()
+
+  // todo: for of inside for of might fuck up?
+  const pointer = localTmp(scope, 'forof_base_pointer', Valtype.i32);
+  const length = localTmp(scope, 'forof_length', Valtype.i32);
+  const counter = localTmp(scope, 'forof_counter', Valtype.i32);
+
+  out.push(
+    // set pointer as right
+    ...generate(scope, decl.right),
+    Opcodes.i32_to_u,
+    [ Opcodes.local_set, pointer ],
+
+    // get length
+    [ Opcodes.local_get, pointer ],
+    [ Opcodes.i32_load, Math.log2(ValtypeSize.i32) - 1, 0 ],
+    [ Opcodes.local_set, length ]
+  );
+
   out.push([ Opcodes.loop, Blocktype.void ]);
-  depth.push('while');
+  depth.push('forof');
 
-  out.push(...generate(scope, decl.test));
-  out.push(Opcodes.i32_to, [ Opcodes.if, Blocktype.void ]);
-  depth.push('if');
+  // setup local for left
+  generate(scope, decl.left);
 
-  out.push(...generate(scope, decl.body));
+  const leftName = decl.left.declarations[0].id.name;
 
-  out.push([ Opcodes.br, 1 ]);
-  out.push([ Opcodes.end ], [ Opcodes.end ]);
-  depth.pop(); depth.pop();
+  // set type for local
+  typeStates[leftName] = rightType === TYPES._array ? TYPES.number : TYPES.string;
+
+  const [ local, isGlobal ] = lookupName(scope, leftName);
+
+  if (rightType === TYPES._array) { // array
+    out.push(
+      [ Opcodes.local_get, pointer ],
+      [ Opcodes.load, Math.log2(valtypeSize) - 1, ...unsignedLEB128(ValtypeSize.i32) ]
+    );
+  } else { // string
+    const [ newOut, newPointer ] = makeArray(scope, {
+      rawElements: new Array(1)
+    }, isGlobal, leftName, true, 'i16');
+
+    out.push(
+      // setup new/out array
+      ...newOut,
+      [ Opcodes.drop ],
+
+      ...number(0, Valtype.i32), // base 0 for store after
+
+      // load current string ind {arg}
+      [ Opcodes.local_get, pointer ],
+      [ Opcodes.i32_load16_u, Math.log2(ValtypeSize.i16) - 1, ...unsignedLEB128(ValtypeSize.i32) ],
+
+      // store to new string ind 0
+      [ Opcodes.i32_store16, Math.log2(ValtypeSize.i16) - 1, ...unsignedLEB128(newPointer + ValtypeSize.i32) ],
+
+      // return new string (page)
+      ...number(newPointer)
+    );
+  }
+
+  // set left value
+  out.push([ isGlobal ? Opcodes.global_set : Opcodes.local_set, local.idx ]);
+
+  out.push(
+    [ Opcodes.block, Blocktype.void ],
+    [ Opcodes.block, Blocktype.void ]
+  );
+  depth.push('block');
+  depth.push('block');
+
+  out.push(
+    ...generate(scope, decl.body),
+    [ Opcodes.end ]
+  );
+  depth.pop();
+
+  out.push(
+    // increment iter pointer by valtype size
+    [ Opcodes.local_get, pointer ],
+    ...number(valtypeSize, Valtype.i32),
+    [ Opcodes.i32_add ],
+    [ Opcodes.local_set, pointer ],
+
+    // increment counter by 1
+    [ Opcodes.local_get, counter ],
+    ...number(1, Valtype.i32),
+    [ Opcodes.i32_add ],
+    [ Opcodes.local_tee, counter ],
+
+    // loop if counter != length
+    [ Opcodes.local_get, length ],
+    [ Opcodes.i32_ne ],
+    [ Opcodes.br_if, 1 ],
+
+    [ Opcodes.end ], [ Opcodes.end ]
+  );
+  depth.pop();
+  depth.pop();
 
   return out;
 };
@@ -1705,19 +1844,19 @@ const generateAssignPat = (scope, decl) => {
 };
 
 let pages = new Map();
-const allocPage = reason => {
-  if (pages.has(reason)) return pages.get(reason);
+const allocPage = (reason, type) => {
+  if (pages.has(reason)) return pages.get(reason).ind;
 
-  let ind = pages.size;
-  pages.set(reason, ind);
+  const ind = pages.size;
+  pages.set(reason, { ind, type });
 
-  if (allocLog) log('alloc', `allocated new page of memory (${ind}) | ${reason}`);
+  if (allocLog) log('alloc', `allocated new page of memory (${ind}) | ${reason} (type: ${type})`);
 
   return ind;
 };
 
 const freePage = reason => {
-  let ind = pages.get(reason);
+  const { ind } = pages.get(reason);
   pages.delete(reason);
 
   if (allocLog) log('alloc', `freed page of memory (${ind}) | ${reason}`);
@@ -1749,7 +1888,7 @@ const makeArray = (scope, decl, global = false, name = '$undeclared', initEmpty
   if (!arrays.has(name) || name === '$undeclared') {
     // todo: can we just have 1 undeclared array? probably not? but this is not really memory efficient
     const uniqueName = name === '$undeclared' ? name + Math.random().toString().slice(2) : name;
-    arrays.set(name, allocPage(`${itemType === 'i16' ? 'string' : 'array'}: ${uniqueName}`) * pageSize);
+    arrays.set(name, allocPage(`${itemType === 'i16' ? 'string' : 'array'}: ${uniqueName}`, itemType) * pageSize);
   }
 
   const pointer = arrays.get(name);
@@ -1929,7 +2068,6 @@ const generateFunc = (scope, decl) => {
     localInd: 0,
     returns: [ valtypeBinary ],
     returnType: null,
-    memory: false,
     throws: false,
     name
   };

package/compiler/decompile.js

@@ -57,7 +57,7 @@ export default (wasm, name = '', ind = 0, locals = {}, params = [], returns = []
       out += ` ;; label @${depth}`;
     }
 
-    if (inst[0] === Opcodes.br) {
+    if (inst[0] === Opcodes.br || inst[0] === Opcodes.br_if) {
       out += ` ;; goto @${depth - inst[1]}`;
     }
 

package/compiler/index.js

@@ -4,6 +4,8 @@ import opt from './opt.js';
 import produceSections from './sections.js';
 import decompile from './decompile.js';
 import { BuiltinPreludes } from './builtins.js';
+import toc from './2c.js';
+
 
 globalThis.decompile = decompile;
 
@@ -15,7 +17,8 @@ const areaColors = {
   codegen: [ 20, 80, 250 ],
   opt: [ 250, 20, 80 ],
   sections: [ 20, 250, 80 ],
-  alloc: [ 250, 250, 20 ]
+  alloc: [ 250, 250, 20 ],
+  '2c': [ 20, 250, 250 ]
 };
 
 globalThis.log = (area, ...args) => console.log(`\u001b[90m[\u001b[0m${rgb(...areaColors[area], area)}\u001b[90m]\u001b[0m`, ...args);
@@ -36,10 +39,16 @@ const logFuncs = (funcs, globals, exceptions) => {
   console.log();
 };
 
+const getArg = name => process.argv.find(x => x.startsWith(`-${name}=`))?.slice(name.length + 2);
+
+const writeFileSync = (typeof process !== 'undefined' ? (await import('node:fs')).writeFileSync : undefined);
+const execSync = (typeof process !== 'undefined' ? (await import('node:child_process')).execSync : undefined);
+
 export default (code, flags) => {
   globalThis.optLog = process.argv.includes('-opt-log');
   globalThis.codeLog = process.argv.includes('-code-log');
   globalThis.allocLog = process.argv.includes('-alloc-log');
+  globalThis.regexLog = process.argv.includes('-regex-log');
 
   for (const x in BuiltinPreludes) {
     if (code.indexOf(x + '(') !== -1) code = BuiltinPreludes[x] + code;
@@ -72,5 +81,38 @@ export default (code, flags) => {
     // console.log([...pages.keys()].map(x => `\x1B[36m - ${x}\x1B[0m`).join('\n'));
   }
 
-  return { wasm: sections, funcs, globals, tags, exceptions, pages };
+  const out = { wasm: sections, funcs, globals, tags, exceptions, pages };
+
+  const target = getArg('target') ?? getArg('t') ?? 'wasm';
+  const outFile = getArg('o');
+
+  if (target === 'c') {
+    const c = toc(out);
+
+    if (outFile) {
+      writeFileSync(outFile, c);
+    } else {
+      console.log(c);
+    }
+
+    process.exit();
+  }
+
+  if (target === 'native') {
+    const compiler = getArg('compiler') ?? 'clang';
+    const cO = getArg('cO') ?? 'Ofast';
+
+    const tmpfile = 'tmp.c';
+    const args = [ compiler, tmpfile, '-o', outFile ?? (process.platform === 'win32' ? 'out.exe' : 'out'), '-' + cO, '-march=native' ];
+
+    const c = toc(out);
+    writeFileSync(tmpfile, c);
+
+    // obvious command escape is obvious
+    execSync(args.join(' '), { stdio: 'inherit' });
+
+    process.exit();
+  }
+
+  return out;
 };

package/compiler/opt.js

@@ -142,7 +142,7 @@ export default (funcs, globals) => {
             depth--;
             if (depth <= 0) break;
           }
-          if (op === Opcodes.br) {
+          if (op === Opcodes.br || op === Opcodes.br_if) {
             hasBranch = true;
             break;
           }

package/compiler/parse.js

@@ -1,3 +1,4 @@
+// import { parse } from 'acorn';
 const { parse } = (await import(globalThis.document ? 'https://esm.sh/acorn' : 'acorn'));
 
 export default (input, flags) => {

package/compiler/prototype.js

@@ -15,7 +15,8 @@ const TYPES = {
   bigint: 0xffffffffffff7,
 
   // these are not "typeof" types but tracked internally
-  _array: 0xffffffffffff8
+  _array: 0xfffffffffff0f,
+  _regexp: 0xfffffffffff1f
 };
 
 // todo: turn these into built-ins once arrays and these become less hacky

package/compiler/sections.js

@@ -76,6 +76,7 @@ export default (funcs, globals, tags, pages, flags) => {
       }
     }
   }
+  globalThis.importFuncs = importFuncs;
 
   if (optLog) log('sections', `treeshake: using ${importFuncs.length}/${importedFuncs.length} imports`);
 

package/compiler/wrap.js

@@ -1,10 +1,11 @@
 import compile from './index.js';
 import decompile from './decompile.js';
-import fs from 'node:fs';
+// import fs from 'node:fs';
 
 const bold = x => `\u001b[1m${x}\u001b[0m`;
 
 const typeBase = 0xffffffffffff0;
+const internalTypeBase = 0xfffffffffff0f;
 const TYPES = {
   [typeBase]: 'number',
   [typeBase + 1]: 'boolean',
@@ -16,7 +17,8 @@ const TYPES = {
   [typeBase + 7]: 'bigint',
 
   // internal
-  [typeBase + 8]: '_array'
+  [internalTypeBase]: '_array',
+  [internalTypeBase + 1]: '_regexp'
 };
 
 export default async (source, flags = [ 'module' ], customImports = {}, print = str => process.stdout.write(str)) => {
@@ -27,7 +29,7 @@ export default async (source, flags = [ 'module' ], customImports = {}, print =
 
   if (source.includes('export function')) flags.push('module');
 
-  fs.writeFileSync('out.wasm', Buffer.from(wasm));
+  // fs.writeFileSync('out.wasm', Buffer.from(wasm));
 
   times.push(performance.now() - t1);
   if (flags.includes('info')) console.log(bold(`compiled in ${times[0].toFixed(2)}ms`));

package/hi.c

@@ -0,0 +1,37 @@
+#include <stdio.h>
+
+double inline f64_f(double x, double y) {
+  return x - (int)(x / y) * y;
+}
+
+double isPrime(double number) {
+  double i;
+
+  if (number < 2e+0) {
+    return 0e+0;
+  }
+  i = 2e+0;
+  while (i < number) {
+    if (f64_f(number, i) == 0e+0) {
+      return 0e+0;
+    }
+    i = i + 1e+0;
+  }
+  return 1e+0;
+}
+
+int main() {
+  double sum;
+  double counter;
+
+  sum = 0e+0;
+  counter = 0e+0;
+  while (counter <= 1e+5) {
+    if (isPrime(counter) == 1e+0) {
+      sum = sum + counter;
+    }
+    counter = counter + 1e+0;
+  }
+  printf("%f\n", sum);
+}
+

package/package.json

@@ -1,7 +1,7 @@
 {
   "name": "porffor",
   "description": "a basic experimental wip aot optimizing js -> wasm engine/compiler/runtime in js",
-  "version": "0.0.0-ebc0491",
+  "version": "0.0.0-f74a73a",
   "author": "CanadaHonk",
   "license": "MIT",
   "dependencies": {

package/r.js

@@ -1 +1 @@
-/a(b)/.test('hi');
+print(performance.now());

package/rhemyn/README.md

@@ -0,0 +1,37 @@
+# rhemyn
+a basic experimental wip regex engine/aot wasm compiler in js. regex engine for porffor. uses own regex parser, no dependencies (excluding porffor internals). <br>
+age: ~1 day
+
+made for use with porffor but could possibly be adapted, implementation/library notes:
+- exposes functions for each regex "operation" (eg test, match)
+- given a regex pattern string (eg `a+`), it returns a "function" object
+- wasm function returned expects an i32 pointer to a utf-16 string (can add utf-8 option later if someone else actually wants to use this)
+
+## syntax
+🟢 supported 🟡 partial 🟠 parsed only 🔴 unsupported
+
+- 🟢 literal characters (eg `a`)
+- 🟢 escaping (eg `\.\n\cJ\x0a\u000a`)
+  - 🟢 character itself (eg `\.`)
+  - 🟢 escape sequences (eg `\n`)
+  - 🟢 control character (eg `\cJ`)
+  - 🟢 unicode code points (eg `\x00`, `\u0000`)
+- 🟢 sets (eg `[ab]`)
+  - 🟢 ranges (eg `[a-z]`)
+  - 🟢 negated sets (eg `[^ab]`)
+- 🟢 metacharacters
+  - 🟢 dot (eg `a.b`)
+  - 🟢 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?`)
+  - 🟠 lazy modifier (eg `a*?`)
+  - 🔴 n repetitions (eg `a{4}`)
+  - 🔴 n-m repetitions (eg `a{2,4}`)
+- 🔴 assertions
+  - 🔴 beginning (eg `^a`)
+  - 🔴 end (eg `a$`)
+  - 🔴 word boundary assertion (eg `\b\B`)