porffor 0.17.0-b103c8894 → 0.17.0-b598eb7bb

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

@@ -6,75 +6,56 @@ 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;
-
-let exprLastGet = false;
+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 = [];
   switch (node.type) {
     case 'Expression':
-      exprLastGet = false;
       out = [
         // set length local
         [ Opcodes.local_get, BasePointer ],
         [ 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.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 ],
+        [ Opcodes.loop, Blocktype.void ],
+          [ 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 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
+        // no match
         ...number(({
           test: 0,
           search: -1
@@ -110,12 +91,12 @@ 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 +115,81 @@ 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 +197,31 @@ 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,
@@ -206,28 +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 },
+    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 },
   }
 });

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

@@ -80,13 +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;
   });
@@ -127,12 +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 { }
 
@@ -141,7 +141,7 @@ replServer.defineCommand('asm', {
 });
 replServer.defineCommand('js', {
   help: 'Log JS being actually ran',
-  async action() {
+  action() {
     this.clearBufferedCommand();
     console.log(prev);
     this.displayPrompt();