porffor 0.17.0-b4e7f7ee0 → 0.17.0-b598eb7bb

package/compiler/codegen.js

@@ -1590,9 +1590,9 @@ const generateCall = (scope, decl, _global, _name, unusedValue = false) => {
   if (!name && decl.callee.type === 'MemberExpression') {
     // megahack for /regex/.func()
     const funcName = decl.callee.property.name;
-    if (decl.callee.object.regex && Object.hasOwn(Rhemyn, funcName)) {
+    if (decl.callee.object.regex && ['test'].includes(funcName)) {
       const regex = decl.callee.object.regex.pattern;
-      const rhemynName = `regex_${funcName}_${regex}`;
+      const rhemynName = `regex_${funcName}_${sanitize(regex)}`;
 
       if (!funcIndex[rhemynName]) {
         const func = Rhemyn[funcName](regex, currentFuncIndex++, rhemynName);
@@ -1613,7 +1613,7 @@ const generateCall = (scope, decl, _global, _name, unusedValue = false) => {
         [ Opcodes.call, idx ],
         Opcodes.i32_from_u,
 
-        ...setLastType(scope, TYPES.boolean)
+        ...setLastType(scope, Rhemyn.types[funcName])
       ];
     }
 
@@ -1622,24 +1622,44 @@ const generateCall = (scope, decl, _global, _name, unusedValue = false) => {
     target = decl.callee.object;
   }
 
-  // if (protoName && baseType === TYPES.string && Rhemyn[protoName]) {
-  //   const func = Rhemyn[protoName](decl.arguments[0].regex.pattern, currentFuncIndex++);
+  if (protoName) {
+    if (['search'].includes(protoName)) {
+      const regex = decl.arguments[0]?.regex?.pattern;
+      if (!regex) return [
+        // no/bad regex arg, return -1/0 for now
+        ...generate(scope, target),
+        [ Opcodes.drop ],
 
-  //   funcIndex[func.name] = func.index;
-  //   funcs.push(func);
+        ...number(Rhemyn.types[protoName] === TYPES.number ? -1 : 0),
+        ...setLastType(scope, Rhemyn.types[protoName])
+      ];
 
-  //   return [
-  //     generate(scope, decl.callee.object)
+      const rhemynName = `regex_${protoName}_${sanitize(regex)}`;
 
-  //     // call regex func
-  //     [ Opcodes.call, func.index ],
-  //     Opcodes.i32_from_u
-  //   ];
-  // }
+      if (!funcIndex[rhemynName]) {
+        const func = Rhemyn[protoName](regex, currentFuncIndex++, rhemynName);
+        func.internal = true;
 
-  if (protoName) {
-    const protoBC = {};
+        funcIndex[func.name] = func.index;
+        funcs.push(func);
+      }
 
+      const idx = funcIndex[rhemynName];
+      return [
+        // make string arg
+        ...generate(scope, target),
+        Opcodes.i32_to_u,
+        ...getNodeType(scope, target),
+
+        // call regex func
+        [ Opcodes.call, idx ],
+        Opcodes.i32_from,
+
+        ...setLastType(scope, Rhemyn.types[protoName])
+      ];
+    }
+
+    const protoBC = {};
     const builtinProtoCands = Object.keys(builtinFuncs).filter(x => x.startsWith('__') && x.endsWith('_prototype_' + protoName));
 
     if (!decl._protoInternalCall && builtinProtoCands.length > 0) {
@@ -2053,6 +2073,16 @@ const DEFAULT_VALUE = {
   name: 'undefined'
 };
 
+const codeToSanitizedStr = code => {
+  let out = '';
+  while (code > 0) {
+    out += String.fromCharCode(97 + code % 26);
+    code -= 26;
+  }
+  return out;
+};
+const sanitize = str => str.replace(/[^0-9a-zA-Z_]/g, _ => codeToSanitizedStr(_.charCodeAt(0)));
+
 const unhackName = name => {
   if (name.startsWith('__')) return name.slice(2).replaceAll('_', '.');
   return name;

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-b4e7f7ee0",
+  "version": "0.17.0-b598eb7bb",
   "author": "CanadaHonk",
   "license": "MIT",
   "scripts": {},

package/rhemyn/compile.js

@@ -6,13 +6,11 @@ import { TYPES } from '../compiler/types.js';
 
 // local indexes
 const BasePointer = 0; // base string pointer
-const IterPointer = 1; // this iteration base pointer
-const EndPointer = 2; // pointer for the end
-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
+const Counter = 2; // what char we are running on
+const Pointer = 3; // next char pointer
+const Length = 4;
+const Tmp = 5;
+const QuantifierTmp = 6; // the temporary variable used for quanitifers
 
 const generate = (node, negated = false, get = true, stringSize = 2, func = 'test') => {
   let out = [];
@@ -24,17 +22,13 @@ const generate = (node, negated = false, get = true, stringSize = 2, func = 'tes
         [ Opcodes.i32_load, Math.log2(ValtypeSize.i32) - 1, 0 ],
         [ Opcodes.local_set, Length ],
 
-        // set iter pointer local as base + sizeof i32 initially
+        // pointer = base + sizeof i32
         [ Opcodes.local_get, BasePointer ],
         ...number(ValtypeSize.i32, Valtype.i32),
         [ Opcodes.i32_add ],
-        [ Opcodes.local_set, IterPointer ],
+        [ Opcodes.local_set, Pointer ],
 
         [ 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 => generate(x, negated, true, stringSize, func)),
@@ -49,12 +43,6 @@ const generate = (node, negated = false, get = true, stringSize = 2, func = 'tes
             [ 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),
@@ -67,7 +55,7 @@ const generate = (node, negated = false, get = true, stringSize = 2, func = 'tes
           [ Opcodes.br_if, 0 ],
         [ Opcodes.end ],
 
-        // no match, return 0
+        // no match
         ...number(({
           test: 0,
           search: -1
@@ -269,7 +257,7 @@ const wrapFunc = (regex, func, name, index) => {
   const parsed = parse(regex);
 
   return outputFunc([
-    [ Opcodes.local_get, IterPointer ],
+    [ Opcodes.local_get, 1 ],
     ...number(TYPES.string, Valtype.i32),
     [ Opcodes.i32_eq ],
     [ Opcodes.if, Valtype.i32 ],
@@ -279,29 +267,33 @@ const wrapFunc = (regex, func, name, index) => {
     // bytestring
     ...generate(parsed, false, true, 1, func),
     [ Opcodes.end ]
-  ], name, index);
+  ], name, index, types[func]);
 };
 
 export const test = (regex, index = 0, name = 'regex_test_' + regex) => wrapFunc(regex, 'test', name, index);
 export const search = (regex, index = 0, name = 'regex_search_' + regex) => wrapFunc(regex, 'search', name, index);
 
-const outputFunc = (wasm, name, index) => ({
+export const types = {
+  test: TYPES.boolean,
+  search: TYPES.number
+};
+
+const outputFunc = (wasm, name, index, returnType) => ({
   name,
   index,
   wasm,
+  returnType,
 
   export: true,
   params: [ Valtype.i32, Valtype.i32 ],
   returns: [ Valtype.i32 ],
-  returnType: TYPES.boolean,
   locals: {
     basePointer: { idx: 0, type: Valtype.i32 },
-    iterPointer: { idx: 1, type: Valtype.i32 },
-    endPointer: { idx: 2, type: Valtype.i32 },
-    counter: { idx: 3, type: Valtype.i32 },
-    pointer: { idx: 4, type: Valtype.i32 },
-    length: { idx: 5, type: Valtype.i32 },
-    tmp: { idx: 6, type: Valtype.i32 },
-    quantifierTmp: { idx: 7, type: Valtype.i32 },
+    inputType: { idx: 1, type: Valtype.i32 },
+    counter: { idx: 2, type: Valtype.i32 },
+    pointer: { idx: 3, type: Valtype.i32 },
+    length: { idx: 4, type: Valtype.i32 },
+    tmp: { idx: 5, type: Valtype.i32 },
+    quantifierTmp: { idx: 6, type: Valtype.i32 },
   }
 });