porffor 0.0.0-425ea20 → 0.0.0-48403fd

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

@@ -15,13 +15,17 @@ 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
 
 export const PrototypeFuncs = function() {
   const noUnlikelyChecks = process.argv.includes('-funsafe-no-unlikely-proto-checks');
+  let zeroChecks = process.argv.find(x => x.startsWith('-funsafe-zero-proto-checks='));
+  if (zeroChecks) zeroChecks = zeroChecks.split('=')[1].split(',').reduce((acc, x) => { acc[x.toLowerCase()] = true; return acc; }, {});
+    else zeroChecks = {};
 
   this[TYPES._array] = {
     // lX = local accessor of X ({ get, set }), iX = local index of X, wX = wasm ops of X
@@ -35,7 +39,7 @@ export const PrototypeFuncs = function() {
       [ Opcodes.i32_lt_s ],
       [ Opcodes.if, Blocktype.void ],
       [ Opcodes.local_get, iTmp ],
-      ...length.cachedI32,
+      ...length.getCachedI32(),
       [ Opcodes.i32_add ],
       [ Opcodes.local_set, iTmp ],
       [ Opcodes.end ],
@@ -46,7 +50,7 @@ export const PrototypeFuncs = function() {
       [ Opcodes.i32_lt_s ],
 
       [ Opcodes.local_get, iTmp ],
-      ...length.cachedI32,
+      ...length.getCachedI32(),
       [ Opcodes.i32_ge_s ],
       [ Opcodes.i32_or ],
 
@@ -66,7 +70,7 @@ export const PrototypeFuncs = function() {
     // todo: only for 1 argument
     push: (pointer, length, wNewMember) => [
       // get memory offset of array at last index (length)
-      ...length.cachedI32,
+      ...length.getCachedI32(),
       ...number(ValtypeSize[valtype], Valtype.i32),
       [ Opcodes.i32_mul ],
 
@@ -78,17 +82,17 @@ export const PrototypeFuncs = function() {
 
       // bump array length by 1 and return it
       ...length.setI32([
-        ...length.cachedI32,
+        ...length.getCachedI32(),
         ...number(1, Valtype.i32),
         [ Opcodes.i32_add ]
       ]),
 
-      ...length.get
+      ...length.get()
     ],
 
     pop: (pointer, length) => [
       // if length == 0, noop
-      ...length.cachedI32,
+      ...length.getCachedI32(),
       [ Opcodes.i32_eqz ],
       [ Opcodes.if, Blocktype.void ],
       ...number(UNDEFINED),
@@ -99,13 +103,13 @@ export const PrototypeFuncs = function() {
 
       // decrement length by 1
       ...length.setI32([
-        ...length.cachedI32,
+        ...length.getCachedI32(),
         ...number(1, Valtype.i32),
         [ Opcodes.i32_sub ]
       ]),
 
       // load last element
-      ...length.cachedI32,
+      ...length.getCachedI32(),
       ...number(ValtypeSize[valtype], Valtype.i32),
       [ Opcodes.i32_mul ],
 
@@ -114,7 +118,7 @@ export const PrototypeFuncs = function() {
 
     shift: (pointer, length) => [
       // if length == 0, noop
-      ...length.cachedI32,
+      ...length.getCachedI32(),
       Opcodes.i32_eqz,
       [ Opcodes.if, Blocktype.void ],
       ...number(UNDEFINED),
@@ -125,7 +129,7 @@ export const PrototypeFuncs = function() {
 
       // decrement length by 1
       ...length.setI32([
-        ...length.cachedI32,
+        ...length.getCachedI32(),
         ...number(1, Valtype.i32),
         [ Opcodes.i32_sub ]
       ]),
@@ -139,14 +143,69 @@ export const PrototypeFuncs = function() {
       ...number(pointer + ValtypeSize.i32 + ValtypeSize[valtype], Valtype.i32), // src = base array index + length size + an index
       ...number(pageSize - ValtypeSize.i32 - ValtypeSize[valtype], Valtype.i32), // size = PageSize - length size - an index
       [ ...Opcodes.memory_copy, 0x00, 0x00 ]
+    ],
+
+    fill: (pointer, length, wElement, iTmp) => [
+      ...wElement,
+      [ Opcodes.local_set, iTmp ],
+
+      // use cached length i32 as pointer
+      ...length.getCachedI32(),
+
+      // length - 1 for indexes
+      ...number(1, Valtype.i32),
+      [ Opcodes.i32_sub ],
+
+      // * sizeof value
+      ...number(ValtypeSize[valtype], Valtype.i32),
+      [ Opcodes.i32_mul ],
+
+      ...length.setCachedI32(),
+
+      ...(noUnlikelyChecks ? [] : [
+        ...length.getCachedI32(),
+        ...number(0, Valtype.i32),
+        [ Opcodes.i32_lt_s ],
+        [ Opcodes.if, Blocktype.void ],
+        ...number(pointer),
+        [ Opcodes.br, 1 ],
+        [ Opcodes.end ]
+      ]),
+
+      [ Opcodes.loop, Blocktype.void ],
+
+      // set element using pointer
+      ...length.getCachedI32(),
+      [ Opcodes.local_get, iTmp ],
+      [ Opcodes.store, Math.log2(ValtypeSize[valtype]) - 1, ...unsignedLEB128(pointer + ValtypeSize.i32) ],
+
+      // pointer - sizeof value
+      ...length.getCachedI32(),
+      ...number(ValtypeSize[valtype], Valtype.i32),
+      [ Opcodes.i32_sub ],
+
+      ...length.setCachedI32(),
+
+      // if pointer >= 0, loop
+      ...length.getCachedI32(),
+      ...number(0, Valtype.i32),
+      [ Opcodes.i32_ge_s ],
+      [ Opcodes.br_if, 0 ],
+
+      [ Opcodes.end ],
+
+      // return this array
+      ...number(pointer)
     ]
   };
 
   this[TYPES._array].at.local = Valtype.i32;
   this[TYPES._array].push.noArgRetLength = true;
+  this[TYPES._array].fill.local = valtypeBinary;
+  this[TYPES._array].fill.returnType = TYPES._array;
 
   this[TYPES.string] = {
-    at: (pointer, length, wIndex, iTmp, arrayShell) => {
+    at: (pointer, length, wIndex, iTmp, _, arrayShell) => {
       const [ newOut, newPointer ] = arrayShell(1, 'i16');
 
       return [
@@ -165,7 +224,7 @@ export const PrototypeFuncs = function() {
         [ Opcodes.i32_lt_s ],
         [ Opcodes.if, Blocktype.void ],
         [ Opcodes.local_get, iTmp ],
-        ...length.cachedI32,
+        ...length.getCachedI32(),
         [ Opcodes.i32_add ],
         [ Opcodes.local_set, iTmp ],
         [ Opcodes.end ],
@@ -176,7 +235,7 @@ export const PrototypeFuncs = function() {
         [ Opcodes.i32_lt_s ],
 
         [ Opcodes.local_get, iTmp ],
-        ...length.cachedI32,
+        ...length.getCachedI32(),
         [ Opcodes.i32_ge_s ],
         [ Opcodes.i32_or ],
 
@@ -201,7 +260,7 @@ export const PrototypeFuncs = function() {
     },
 
     // todo: out of bounds properly
-    charAt: (pointer, length, wIndex, _, arrayShell) => {
+    charAt: (pointer, length, wIndex, _1, _2, arrayShell) => {
       const [ newOut, newPointer ] = arrayShell(1, 'i16');
 
       return [
@@ -232,39 +291,120 @@ export const PrototypeFuncs = function() {
       return [
         ...wIndex,
         Opcodes.i32_to,
-        [ Opcodes.local_set, iTmp ],
 
-        // index < 0
-        ...(noUnlikelyChecks ? [] : [
+        ...(zeroChecks.charcodeat ? [] : [
+          [ Opcodes.local_set, iTmp ],
+
+          // index < 0
+          ...(noUnlikelyChecks ? [] : [
+            [ Opcodes.local_get, iTmp ],
+            ...number(0, Valtype.i32),
+            [ Opcodes.i32_lt_s ],
+          ]),
+
+          // index >= length
+          [ Opcodes.local_get, iTmp ],
+          ...length.getCachedI32(),
+          [ Opcodes.i32_ge_s ],
+
+          ...(noUnlikelyChecks ? [] : [ [ Opcodes.i32_or ] ]),
+          [ Opcodes.if, Blocktype.void ],
+          ...number(NaN),
+          [ Opcodes.br, 1 ],
+          [ Opcodes.end ],
+
           [ Opcodes.local_get, iTmp ],
-          ...number(0, Valtype.i32),
-          [ Opcodes.i32_lt_s ],
         ]),
 
-        // index >= length
+        ...number(ValtypeSize.i16, Valtype.i32),
+        [ Opcodes.i32_mul ],
+
+        // load current string ind {arg}
+        [ Opcodes.i32_load16_u, Math.log2(ValtypeSize.i16) - 1, ...unsignedLEB128(pointer + ValtypeSize.i32) ],
+        Opcodes.i32_from_u
+      ];
+    },
+
+    isWellFormed: (pointer, length, wIndex, iTmp, iTmp2, arrayShell, { wellFormed } = {}) => {
+      // aot approx metadata
+      if (wellFormed != null) return number(wellFormed ? 1 : 0);
+
+      return [
+        // note: we cannot presume it begins as 0 in case it was used previously
+        ...number(0, Valtype.i32),
+        [ Opcodes.local_set, iTmp ],
+
+        [ Opcodes.loop, Blocktype.void ],
+
+        [ Opcodes.block, Blocktype.void ],
+
         [ Opcodes.local_get, iTmp ],
-        ...length.cachedI32,
+        [ Opcodes.i32_load16_u, Math.log2(ValtypeSize.i16) - 1, ...unsignedLEB128(pointer + ValtypeSize.i32) ],
+        [ Opcodes.local_set, iTmp2 ],
+
+        // if not surrogate, continue
+        [ Opcodes.local_get, iTmp2 ],
+        ...number(0xF800, Valtype.i32),
+        [ Opcodes.i32_and ],
+        ...number(0xD800, Valtype.i32),
+        [ Opcodes.i32_ne ],
+        [ Opcodes.br_if, 0 ],
+
+        // if not leading surrogate, return false
+        [ Opcodes.local_get, iTmp2 ],
+        ...number(0xDC00, Valtype.i32),
         [ Opcodes.i32_ge_s ],
+        [ Opcodes.if, Blocktype.void ],
+        ...number(0),
+        [ Opcodes.br, 3 ],
+        [ Opcodes.end ],
 
-        ...(noUnlikelyChecks ? [] : [ [ Opcodes.i32_or ] ]),
+        // if not followed by trailing surrogate, return false
+        [ Opcodes.local_get, iTmp ],
+        [ Opcodes.i32_load16_u, Math.log2(ValtypeSize.i16) - 1, ...unsignedLEB128(pointer + ValtypeSize.i32 + ValtypeSize.i16) ],
+        ...number(0xFC00, Valtype.i32),
+        [ Opcodes.i32_and ],
+        ...number(0xDC00, Valtype.i32),
+        [ Opcodes.i32_ne ],
         [ Opcodes.if, Blocktype.void ],
-        ...number(NaN),
-        [ Opcodes.br, 1 ],
+        ...number(0),
+        [ Opcodes.br, 3 ],
         [ Opcodes.end ],
 
+        // bump index again since gone through two valid chars
         [ Opcodes.local_get, iTmp ],
         ...number(ValtypeSize.i16, Valtype.i32),
+        [ Opcodes.i32_add ],
+        [ Opcodes.local_set, iTmp ],
+
+        [ Opcodes.end ],
+
+        // bump pointer and loop if not at the end
+        [ Opcodes.local_get, iTmp ],
+        ...number(ValtypeSize.i16, Valtype.i32),
+        [ Opcodes.i32_add ],
+        [ Opcodes.local_tee, iTmp ],
+
+        ...length.getCachedI32(),
+        ...number(ValtypeSize.i16, Valtype.i32),
         [ Opcodes.i32_mul ],
+        [ Opcodes.i32_ne ],
+        [ Opcodes.br_if, 0 ],
 
-        // load current string ind {arg}
-        [ Opcodes.i32_load16_u, Math.log2(ValtypeSize.i16) - 1, ...unsignedLEB128(pointer + ValtypeSize.i32) ],
-        Opcodes.i32_from_u
-      ];
-    },
+        [ Opcodes.end ],
+
+        // return true
+        ...number(1)
+      ]
+    }
   };
 
   this[TYPES.string].at.local = Valtype.i32;
   this[TYPES.string].at.returnType = TYPES.string;
   this[TYPES.string].charAt.returnType = TYPES.string;
   this[TYPES.string].charCodeAt.local = Valtype.i32;
+
+  this[TYPES.string].isWellFormed.local = Valtype.i32;
+  this[TYPES.string].isWellFormed.local2 = Valtype.i32;
+  this[TYPES.string].isWellFormed.returnType = TYPES.boolean;
 };

package/compiler/sections.js

@@ -1,5 +1,5 @@
 import { Valtype, FuncType, Empty, ExportDesc, Section, Magic, ModuleVersion, Opcodes, PageSize } from './wasmSpec.js';
-import { encodeVector, encodeString, encodeLocal } from './encoding.js';
+import { encodeVector, encodeString, encodeLocal, unsignedLEB128, signedLEB128 } from './encoding.js';
 import { number } from './embedding.js';
 import { importedFuncs } from './builtins.js';
 
@@ -8,11 +8,26 @@ const createSection = (type, data) => [
   ...encodeVector(data)
 ];
 
-export default (funcs, globals, tags, pages, flags) => {
+const customSection = (name, data) => [
+  Section.custom,
+  ...encodeVector([...encodeString(name), ...data])
+];
+
+const chHint = (topTier, baselineTier, strategy) => {
+  // 1 byte of 4 2 bit components: spare, top tier, baseline tier, compilation strategy
+  // tiers: 0x00 = default, 0x01 = baseline (liftoff), 0x02 = optimized (turbofan)
+  // strategy: 0x00 = default, 0x01 = lazy, 0x02 = eager, 0x03 = lazy baseline, eager top tier
+  return (strategy | (baselineTier << 2) | (topTier << 4));
+};
+
+export default (funcs, globals, tags, pages, data, flags) => {
   const types = [], typeCache = {};
 
   const optLevel = parseInt(process.argv.find(x => x.startsWith('-O'))?.[2] ?? 1);
 
+  const compileHints = process.argv.includes('-compile-hints');
+  if (compileHints) log('sections', 'warning: compile hints is V8 only w/ experimental arg! (you used -compile-hints)');
+
   const getType = (params, returns) => {
     const hash = `${params.join(',')}_${returns.join(',')}`;
     if (optLog) log('sections', `getType(${JSON.stringify(params)}, ${JSON.stringify(returns)}) -> ${hash} | cache: ${typeCache[hash]}`);
@@ -61,6 +76,7 @@ export default (funcs, globals, tags, pages, flags) => {
       }
     }
   }
+  globalThis.importFuncs = importFuncs;
 
   if (optLog) log('sections', `treeshake: using ${importFuncs.length}/${importedFuncs.length} imports`);
 
@@ -74,6 +90,14 @@ export default (funcs, globals, tags, pages, flags) => {
     encodeVector(funcs.map(x => getType(x.params, x.returns))) // type indexes
   );
 
+  // compilation hints section - unspecd, v8 only
+  // https://github.com/WebAssembly/design/issues/1473#issuecomment-1431274746
+  const chSection = !compileHints ? [] : customSection(
+    'compilationHints',
+    // for now just do everything as optimise eager
+    encodeVector(funcs.map(_ => chHint(0x02, 0x02, 0x02)))
+  );
+
   const globalSection = Object.keys(globals).length === 0 ? [] : createSection(
     Section.global,
     encodeVector(Object.keys(globals).map(x => [ globals[x].type, 0x01, ...number(globals[x].init ?? 0, globals[x].type).flat(), Opcodes.end ]))
@@ -81,6 +105,9 @@ export default (funcs, globals, tags, pages, flags) => {
 
   const exports = funcs.filter(x => x.export).map((x, i) => [ ...encodeString(x.name === 'main' ? 'm' : x.name), ExportDesc.func, x.index ]);
 
+  if (process.argv.includes('-always-memory') && pages.size === 0) pages.set('-always-memory', 0);
+  if (optLevel === 0) pages.set('O0 precaution', 0);
+
   const usesMemory = pages.size > 0;
   const memorySection = !usesMemory ? [] : createSection(
     Section.memory,
@@ -122,7 +149,7 @@ export default (funcs, globals, tags, pages, flags) => {
 
       if (typeCount !== 0) localDecl.push(encodeLocal(typeCount, lastType));
 
-      return encodeVector([ ...encodeVector(localDecl), ...x.wasm.flat().filter(x => x !== null), Opcodes.end ]);
+      return encodeVector([ ...encodeVector(localDecl), ...x.wasm.flat().filter(x => x < 0xff), Opcodes.end ]);
     }))
   );
 
@@ -131,13 +158,24 @@ export default (funcs, globals, tags, pages, flags) => {
     encodeVector(types)
   );
 
+  const dataSection = data.length === 0 ? [] : createSection(
+    Section.data,
+    encodeVector(data.map(x => [ 0x00, Opcodes.i32_const, ...signedLEB128(x.offset), Opcodes.end, ...encodeVector(x.bytes) ]))
+  );
+
+  const dataCountSection = data.length === 0 ? [] : createSection(
+    Section.data_count,
+    unsignedLEB128(data.length)
+  );
+
   if (process.argv.includes('-sections')) console.log({
     typeSection: typeSection.map(x => x.toString(16)),
     importSection: importSection.map(x => x.toString(16)),
     funcSection: funcSection.map(x => x.toString(16)),
     globalSection: globalSection.map(x => x.toString(16)),
     exportSection: exportSection.map(x => x.toString(16)),
-    codeSection: codeSection.map(x => x.toString(16))
+    codeSection: codeSection.map(x => x.toString(16)),
+    dataSection: dataSection.map(x => x.toString(16)),
   });
 
   return Uint8Array.from([
@@ -146,10 +184,13 @@ export default (funcs, globals, tags, pages, flags) => {
     ...typeSection,
     ...importSection,
     ...funcSection,
+    ...chSection,
     ...memorySection,
     ...tagSection,
     ...globalSection,
     ...exportSection,
-    ...codeSection
+    ...dataCountSection,
+    ...codeSection,
+    ...dataSection
   ]);
 };

package/compiler/wasmSpec.js

@@ -118,6 +118,7 @@ export const Opcodes = {
   i64_shl: 0x86,
   i64_shr_s: 0x87,
   i64_shr_u: 0x88,
+  i64_rotl: 0x89,
 
   f64_eq: 0x61,
   f64_ne: 0x62,
@@ -156,6 +157,8 @@ export const Opcodes = {
   f64_convert_i64_s: 0xb9,
   f64_convert_i64_u: 0xba,
 
+  f64_reinterpret_i64: 0xbf,
+
   i32_trunc_sat_f64_s: [ 0xfc, 0x02 ],
   i32_trunc_sat_f64_u: [ 0xfc, 0x03 ],
 

package/compiler/wrap.js

@@ -4,7 +4,8 @@ import fs from 'node:fs';
 
 const bold = x => `\u001b[1m${x}\u001b[0m`;
 
-const typeBase = 0xffffffffffff0;
+const typeBase = 0x00;
+const internalTypeBase = 0x10;
 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)) => {
@@ -64,11 +66,15 @@ export default async (source, flags = [ 'module' ], customImports = {}, print =
 
     exports[func.name] = function() {
       try {
-        const ret = exp.apply(this, arguments);
+        const _ret = exp.apply(this, arguments);
 
-        if (ret >= typeBase && ret <= typeBase + 8) return ret > (typeBase + 7) ? 'object' : TYPES[ret];
+        if (_ret == null) return undefined;
 
-        switch (TYPES[func.returnType]) {
+        const [ ret, type ] = _ret;
+
+        // if (ret >= typeBase && ret <= typeBase + 8) return ret > (typeBase + 7) ? 'object' : TYPES[ret];
+
+        switch (TYPES[type]) {
           case 'boolean': return Boolean(ret);
           case 'undefined': return undefined;
           case 'object': return ret === 0 ? null : {};

package/fib.js

@@ -0,0 +1,10 @@
+function fibonacci(num) {
+  if(num < 2) {
+      return num;
+  }
+  else {
+      return fibonacci(num-1) + fibonacci(num - 2);
+  }
+}
+
+console.log(fibonacci(42));

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": "0.0.0-425ea20",
+  "version": "0.0.0-48403fd",
   "author": "CanadaHonk",
   "license": "MIT",
   "dependencies": {

package/r.js

@@ -0,0 +1,39 @@
+compareArray.isSameValue = function(a, b) {
+  if (a === 0 && b === 0) return 1 / a === 1 / b;
+  if (a !== a && b !== b) return true;
+
+  return a === b;
+};
+
+function compareArray(a, b) {
+  // if either are nullish
+  if (a == null || b == null) return false;
+
+  // megahack: all arrays from now on will be >0 pointer
+  const _hack = '';
+
+  // hack: enforce type inference of being arrays
+  a ??= [];
+  b ??= [];
+
+  if (b.length !== a.length) {
+    return false;
+  }
+
+  for (var i = 0; i < a.length; i++) {
+    if (!compareArray.isSameValue(b[i], a[i])) {
+      return false;
+    }
+  }
+
+  return true;
+}
+
+console.log(compareArray(null, []));
+console.log(compareArray(undefined, []));
+
+console.log(compareArray([], []));
+console.log(compareArray([ 1 ], []));
+console.log(compareArray([ 1 ], [ 1 ]));
+console.log(compareArray([ 1, 2 ], [ 1 ]));
+console.log(compareArray([ 1, 2 ], [ 1, 2 ]));

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`)