js-confuser-vm 0.0.6 → 0.0.7
This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
- package/CHANGELOG.md +55 -0
- package/README.MD +101 -37
- package/dist/build-runtime.js +15 -2
- package/dist/compiler.js +98 -51
- package/dist/runtime.js +5 -1
- package/dist/transforms/bytecode/aliasedOpcodes.js +2 -8
- package/dist/transforms/bytecode/macroOpcodes.js +21 -19
- package/dist/transforms/bytecode/microOpcodes.js +236 -0
- package/dist/transforms/bytecode/resolveContants.js +5 -11
- package/dist/transforms/bytecode/resolveLabels.js +5 -3
- package/dist/transforms/bytecode/specializedOpcodes.js +21 -16
- package/dist/transforms/runtime/internalVariables.js +202 -0
- package/dist/transforms/runtime/macroOpcodes.js +30 -18
- package/dist/transforms/runtime/microOpcodes.js +76 -0
- package/dist/transforms/runtime/specializedOpcodes.js +20 -18
- package/dist/utils/op-utils.js +15 -8
- package/index.ts +3 -2
- package/jest.config.js +2 -0
- package/package.json +1 -1
- package/src/build-runtime.ts +18 -3
- package/src/compiler.ts +152 -65
- package/src/options.ts +1 -0
- package/src/runtime.ts +5 -1
- package/src/transforms/bytecode/aliasedOpcodes.ts +2 -12
- package/src/transforms/bytecode/macroOpcodes.ts +28 -29
- package/src/transforms/bytecode/microOpcodes.ts +291 -0
- package/src/transforms/bytecode/resolveContants.ts +6 -13
- package/src/transforms/bytecode/resolveLabels.ts +5 -4
- package/src/transforms/bytecode/specializedOpcodes.ts +38 -28
- package/src/transforms/runtime/internalVariables.ts +270 -0
- package/src/transforms/runtime/macroOpcodes.ts +47 -20
- package/src/transforms/runtime/microOpcodes.ts +93 -0
- package/src/transforms/runtime/specializedOpcodes.ts +27 -32
- package/src/types.ts +1 -1
- package/src/utils/op-utils.ts +21 -8
- package/src/utilts.ts +0 -3
|
@@ -0,0 +1,236 @@
|
|
|
1
|
+
import { parse } from "@babel/parser";
|
|
2
|
+
import traverseImport from "@babel/traverse";
|
|
3
|
+
import * as t from "@babel/types";
|
|
4
|
+
import { ok } from "assert";
|
|
5
|
+
import { VM_RUNTIME, SOURCE_NODE_SYM } from "../../compiler.js";
|
|
6
|
+
import { nextFreeSlot } from "../../utils/op-utils.js";
|
|
7
|
+
import { nSizedOps } from "./specializedOpcodes.js";
|
|
8
|
+
const traverse = traverseImport.default || traverseImport;
|
|
9
|
+
|
|
10
|
+
// Extract the real statement list from a SwitchCase consequent.
|
|
11
|
+
function extractCaseBody(switchCase) {
|
|
12
|
+
let stmts;
|
|
13
|
+
if (switchCase.consequent.length === 1 && t.isBlockStatement(switchCase.consequent[0])) {
|
|
14
|
+
stmts = switchCase.consequent[0].body;
|
|
15
|
+
} else {
|
|
16
|
+
stmts = switchCase.consequent;
|
|
17
|
+
}
|
|
18
|
+
return stmts.filter(s => !t.isBreakStatement(s) && !t.isEmptyStatement(s));
|
|
19
|
+
}
|
|
20
|
+
|
|
21
|
+
// Count how many IR-level operands a single statement consumes.
|
|
22
|
+
// Returns null if the statement is ineligible (contains a loop, or has
|
|
23
|
+
// _operand()/_constant() calls inside a conditional branch).
|
|
24
|
+
function countStatementOperands(stmt) {
|
|
25
|
+
let count = 0;
|
|
26
|
+
let ineligible = false;
|
|
27
|
+
const file = t.file(t.program([t.cloneNode(stmt, true)]));
|
|
28
|
+
traverse(file, {
|
|
29
|
+
enter(path) {
|
|
30
|
+
if (ineligible) {
|
|
31
|
+
path.stop();
|
|
32
|
+
return;
|
|
33
|
+
}
|
|
34
|
+
const nodeType = path.node.type;
|
|
35
|
+
|
|
36
|
+
// Don't traverse into nested functions
|
|
37
|
+
if (nodeType === "FunctionDeclaration" || nodeType === "FunctionExpression" || nodeType === "ArrowFunctionExpression") {
|
|
38
|
+
path.skip();
|
|
39
|
+
return;
|
|
40
|
+
}
|
|
41
|
+
|
|
42
|
+
// Count _operand() and _constant() calls
|
|
43
|
+
if (nodeType === "CallExpression") {
|
|
44
|
+
const call = path.node;
|
|
45
|
+
const callee = call.callee;
|
|
46
|
+
if (t.isMemberExpression(callee) && t.isThisExpression(callee.object) && t.isIdentifier(callee.property)) {
|
|
47
|
+
const name = callee.property.name;
|
|
48
|
+
const operandsConsumed = name === "_operand" ? 1 : name === "_constant" ? 2 : null;
|
|
49
|
+
if (operandsConsumed) {
|
|
50
|
+
// You are not allowed to use _operand() in loops or branches
|
|
51
|
+
const ancestors = path.getAncestry();
|
|
52
|
+
if (ancestors.find(t => t.isLoop() || t.isIfStatement() || t.isSwitchStatement() || t.isConditionalExpression() || t.isLogicalExpression())) {
|
|
53
|
+
ineligible = true;
|
|
54
|
+
path.stop();
|
|
55
|
+
return;
|
|
56
|
+
}
|
|
57
|
+
count += operandsConsumed;
|
|
58
|
+
}
|
|
59
|
+
}
|
|
60
|
+
}
|
|
61
|
+
}
|
|
62
|
+
});
|
|
63
|
+
return ineligible ? null : count;
|
|
64
|
+
}
|
|
65
|
+
|
|
66
|
+
// Analyse the VM runtime's @SWITCH statement to build a per-opcode map of
|
|
67
|
+
// { stmtIndex → irOperandCount } for every case that can be split.
|
|
68
|
+
// Returns a map: opValue → array of per-statement operand counts (null if ineligible).
|
|
69
|
+
function analyzeRuntimeCases(compiler) {
|
|
70
|
+
// Parse the runtime source
|
|
71
|
+
const ast = parse(VM_RUNTIME, {
|
|
72
|
+
sourceType: "unambiguous"
|
|
73
|
+
});
|
|
74
|
+
|
|
75
|
+
// Build reverse name→opValue map from original OPs only
|
|
76
|
+
const nameToOp = new Map();
|
|
77
|
+
for (const [name, val] of Object.entries(compiler.OP)) {
|
|
78
|
+
if (val !== undefined) nameToOp.set(name, val);
|
|
79
|
+
}
|
|
80
|
+
let switchStatement = null;
|
|
81
|
+
traverse(ast, {
|
|
82
|
+
SwitchStatement(path) {
|
|
83
|
+
if (path.node.leadingComments?.some(c => c.value.includes("@SWITCH"))) {
|
|
84
|
+
switchStatement = path.node;
|
|
85
|
+
path.stop();
|
|
86
|
+
}
|
|
87
|
+
}
|
|
88
|
+
});
|
|
89
|
+
ok(switchStatement, "Could not find @SWITCH statement for micro opcodes");
|
|
90
|
+
const result = new Map();
|
|
91
|
+
for (const sc of switchStatement.cases) {
|
|
92
|
+
const test = sc.test;
|
|
93
|
+
if (!test || !t.isMemberExpression(test) || !t.isIdentifier(test.object, {
|
|
94
|
+
name: "OP"
|
|
95
|
+
}) || !t.isIdentifier(test.property)) {
|
|
96
|
+
continue;
|
|
97
|
+
}
|
|
98
|
+
const opName = test.property.name;
|
|
99
|
+
const opVal = nameToOp.get(opName);
|
|
100
|
+
if (opVal === undefined) continue;
|
|
101
|
+
const stmts = extractCaseBody(sc);
|
|
102
|
+
if (stmts.length < 2) continue; // need at least 2 statements to split
|
|
103
|
+
|
|
104
|
+
const counts = [];
|
|
105
|
+
let allEligible = true;
|
|
106
|
+
|
|
107
|
+
// Banned patterns:
|
|
108
|
+
// Return statements (Control flow isn't remembered)
|
|
109
|
+
traverse(t.file(t.program(stmts)), {
|
|
110
|
+
ReturnStatement(path) {
|
|
111
|
+
path.stop();
|
|
112
|
+
allEligible = false;
|
|
113
|
+
}
|
|
114
|
+
});
|
|
115
|
+
for (const stmt of stmts) {
|
|
116
|
+
const c = countStatementOperands(stmt);
|
|
117
|
+
if (c === null) {
|
|
118
|
+
allEligible = false;
|
|
119
|
+
break;
|
|
120
|
+
}
|
|
121
|
+
if (t.isDebuggerStatement(stmt) || t.isThrowStatement(stmt)) {
|
|
122
|
+
allEligible = false;
|
|
123
|
+
break;
|
|
124
|
+
}
|
|
125
|
+
counts.push(c);
|
|
126
|
+
}
|
|
127
|
+
if (!allEligible) continue;
|
|
128
|
+
|
|
129
|
+
// Verify that the total operand count matches the instruction size expectation
|
|
130
|
+
// (just store for now; bytecode pass validates operands match)
|
|
131
|
+
result.set(opVal, counts);
|
|
132
|
+
}
|
|
133
|
+
return result;
|
|
134
|
+
}
|
|
135
|
+
|
|
136
|
+
// Main bytecode transform: split frequently-used opcodes into per-statement
|
|
137
|
+
// micro-opcodes so each sub-instruction is as small as possible.
|
|
138
|
+
export function microOpcodes(bc, compiler) {
|
|
139
|
+
// ── Step 1: analyse runtime to discover splittable opcodes ──────────────────
|
|
140
|
+
const opAnalysis = analyzeRuntimeCases(compiler);
|
|
141
|
+
if (opAnalysis.size === 0) return {
|
|
142
|
+
bytecode: bc
|
|
143
|
+
};
|
|
144
|
+
|
|
145
|
+
// ── Step 2: count opcode frequency in bytecode ────────────────────────────
|
|
146
|
+
const disallowedOps = new Set(nSizedOps.map(name => compiler.OP[name]));
|
|
147
|
+
disallowedOps.add(compiler.OP.RETURN);
|
|
148
|
+
const freqMap = new Map();
|
|
149
|
+
for (const instr of bc) {
|
|
150
|
+
const op = instr[0];
|
|
151
|
+
if (op === null || !opAnalysis.has(op) || disallowedOps.has(op)) continue;
|
|
152
|
+
freqMap.set(op, (freqMap.get(op) ?? 0) + 1);
|
|
153
|
+
}
|
|
154
|
+
|
|
155
|
+
// ── Step 3: sort by frequency, keep opcodes that actually appear ─────────
|
|
156
|
+
const candidates = Array.from(freqMap.entries()).filter(([, count]) => count >= 1).sort(([, a], [, b]) => b - a).map(([op]) => op);
|
|
157
|
+
if (candidates.length === 0) return {
|
|
158
|
+
bytecode: bc
|
|
159
|
+
};
|
|
160
|
+
|
|
161
|
+
// ── Step 4: assign free opcode slots for each sub-statement ─────────────
|
|
162
|
+
// Build: originalOp → [{ microOp, irOperandCount }, ...]
|
|
163
|
+
const originalToSubOps = new Map();
|
|
164
|
+
for (const origOp of candidates) {
|
|
165
|
+
const stmtCounts = opAnalysis.get(origOp);
|
|
166
|
+
|
|
167
|
+
// Pre-allocate all needed slots; if any slot is unavailable, skip this op.
|
|
168
|
+
const slots = [];
|
|
169
|
+
for (let si = 0; si < stmtCounts.length; si++) {
|
|
170
|
+
const slot = nextFreeSlot(compiler);
|
|
171
|
+
if (slot === -1) break;
|
|
172
|
+
compiler.OP_NAME[slot] = `MICRO_${origOp}_${si}`;
|
|
173
|
+
slots.push(slot);
|
|
174
|
+
}
|
|
175
|
+
if (slots.length !== stmtCounts.length) continue;
|
|
176
|
+
const subOps = [];
|
|
177
|
+
const origName = compiler.OP_NAME[origOp] ?? `OP_${origOp}`;
|
|
178
|
+
for (let si = 0; si < stmtCounts.length; si++) {
|
|
179
|
+
const microOp = slots[si];
|
|
180
|
+
const irOperandCount = stmtCounts[si];
|
|
181
|
+
subOps.push({
|
|
182
|
+
microOp,
|
|
183
|
+
irOperandCount
|
|
184
|
+
});
|
|
185
|
+
compiler.OP_NAME[microOp] = `MICRO_${origName}_${si}`;
|
|
186
|
+
compiler.MICRO_OPS[microOp] = {
|
|
187
|
+
originalOp: origOp,
|
|
188
|
+
stmtIndex: si,
|
|
189
|
+
irOperandCount
|
|
190
|
+
};
|
|
191
|
+
}
|
|
192
|
+
originalToSubOps.set(origOp, subOps);
|
|
193
|
+
}
|
|
194
|
+
if (originalToSubOps.size === 0) return {
|
|
195
|
+
bytecode: bc
|
|
196
|
+
};
|
|
197
|
+
|
|
198
|
+
// ── Step 5: replace each matched instruction with sub-instructions ────────
|
|
199
|
+
const result = [];
|
|
200
|
+
for (const instr of bc) {
|
|
201
|
+
const op = instr[0];
|
|
202
|
+
if (op === null || !originalToSubOps.has(op)) {
|
|
203
|
+
result.push(instr);
|
|
204
|
+
continue;
|
|
205
|
+
}
|
|
206
|
+
const subOps = originalToSubOps.get(op);
|
|
207
|
+
const operands = instr.slice(1); // all operands of the original instruction
|
|
208
|
+
|
|
209
|
+
// Verify total operand count matches sum of sub-op IR operand counts
|
|
210
|
+
const expectedTotal = subOps.reduce((s, {
|
|
211
|
+
irOperandCount
|
|
212
|
+
}) => s + irOperandCount, 0);
|
|
213
|
+
if (operands.length !== expectedTotal) {
|
|
214
|
+
throw new Error(`Operand count mismatch for opcode ${compiler.OP_NAME[op]}`);
|
|
215
|
+
}
|
|
216
|
+
|
|
217
|
+
// Split operands among sub-instructions
|
|
218
|
+
let offset = 0;
|
|
219
|
+
for (const {
|
|
220
|
+
microOp,
|
|
221
|
+
irOperandCount
|
|
222
|
+
} of subOps) {
|
|
223
|
+
const subOperands = operands.slice(offset, offset + irOperandCount);
|
|
224
|
+
offset += irOperandCount;
|
|
225
|
+
const newInstr = [microOp, ...subOperands];
|
|
226
|
+
// Carry source-node info on the first sub-instruction
|
|
227
|
+
if (offset === irOperandCount) {
|
|
228
|
+
newInstr[SOURCE_NODE_SYM] = instr[SOURCE_NODE_SYM];
|
|
229
|
+
}
|
|
230
|
+
result.push(newInstr);
|
|
231
|
+
}
|
|
232
|
+
}
|
|
233
|
+
return {
|
|
234
|
+
bytecode: result
|
|
235
|
+
};
|
|
236
|
+
}
|
|
@@ -18,8 +18,7 @@ function concealString(s, key) {
|
|
|
18
18
|
return Buffer.from(bytes).toString("base64");
|
|
19
19
|
}
|
|
20
20
|
|
|
21
|
-
// Resolve all {type:"constant", value}
|
|
22
|
-
// [constPoolIndex, concealKey]
|
|
21
|
+
// Resolve all {type:"constant", value} (index) and {type:"constant", value, key: true} (key) operands
|
|
23
22
|
//
|
|
24
23
|
// constPoolIndex — index into the constants array (as before).
|
|
25
24
|
// concealKey — XOR key used to conceal this constant.
|
|
@@ -36,7 +35,6 @@ export function resolveConstants(bc, compiler) {
|
|
|
36
35
|
const keyMap = new Map(); // pool index → conceal key
|
|
37
36
|
|
|
38
37
|
function intern(operand) {
|
|
39
|
-
const operandAsObject = typeof operand === "object" && operand ? operand : {};
|
|
40
38
|
const value = operand.value;
|
|
41
39
|
let idx = constantsMap.get(value);
|
|
42
40
|
let key = 0;
|
|
@@ -62,12 +60,10 @@ export function resolveConstants(bc, compiler) {
|
|
|
62
60
|
key = keyMap.get(idx);
|
|
63
61
|
}
|
|
64
62
|
const idxOperand = {
|
|
65
|
-
...operandAsObject,
|
|
66
63
|
type: "number",
|
|
67
64
|
resolvedValue: idx
|
|
68
65
|
};
|
|
69
66
|
const keyOperand = {
|
|
70
|
-
...operandAsObject,
|
|
71
67
|
type: "number",
|
|
72
68
|
resolvedValue: key
|
|
73
69
|
};
|
|
@@ -81,17 +77,15 @@ export function resolveConstants(bc, compiler) {
|
|
|
81
77
|
const hasConstant = operands.some(o => o !== undefined && o !== null && typeof o === "object" && o.type === "constant");
|
|
82
78
|
if (hasConstant) {
|
|
83
79
|
// 1-to-2 expansion: each {type:"constant"} becomes [constIdx, concealKey].
|
|
84
|
-
const newOperands =
|
|
85
|
-
for (const operand of operands) {
|
|
80
|
+
const newOperands = operands.map(operand => {
|
|
86
81
|
if (operand?.type === "constant") {
|
|
87
82
|
const [idxOperand, key] = intern(operand);
|
|
88
83
|
const newOperand = operand?.key ? key : idxOperand;
|
|
89
|
-
|
|
90
|
-
// newOperands.push(key); // plain number — serialized as a regular u16 slot
|
|
84
|
+
return Object.assign(operand, newOperand);
|
|
91
85
|
} else {
|
|
92
|
-
|
|
86
|
+
return operand;
|
|
93
87
|
}
|
|
94
|
-
}
|
|
88
|
+
});
|
|
95
89
|
const newInstr = [op, ...newOperands];
|
|
96
90
|
newInstr[SOURCE_NODE_SYM] = instr[SOURCE_NODE_SYM];
|
|
97
91
|
resolved.push(newInstr);
|
|
@@ -54,13 +54,15 @@ export function resolveLabels(bc, compiler) {
|
|
|
54
54
|
if (operand !== undefined && operand !== null && typeof operand === "object" && operand.type === "label") {
|
|
55
55
|
const pc = labelToPc.get(operand.label);
|
|
56
56
|
if (pc === undefined) throw new Error(`Undefined label: ${operand.label}`);
|
|
57
|
-
var operandAsObject = typeof operand === "object" && operand ? operand : {};
|
|
58
57
|
const newOperand = {
|
|
59
|
-
...operandAsObject,
|
|
60
|
-
// Preverse original operand properties
|
|
61
58
|
type: "number",
|
|
62
59
|
resolvedValue: pc + (operand.offset ?? 0)
|
|
63
60
|
};
|
|
61
|
+
|
|
62
|
+
// Mutate original object so that references are also updated
|
|
63
|
+
if (typeof operand === "object" && operand !== null) {
|
|
64
|
+
return Object.assign(operand, newOperand);
|
|
65
|
+
}
|
|
64
66
|
return newOperand;
|
|
65
67
|
}
|
|
66
68
|
return operand;
|
|
@@ -1,5 +1,6 @@
|
|
|
1
1
|
import { SOURCE_NODE_SYM } from "../../compiler.js";
|
|
2
|
-
import { getInstructionSize, nextFreeSlot
|
|
2
|
+
import { getInstructionSize, nextFreeSlot } from "../../utils/op-utils.js";
|
|
3
|
+
export const nSizedOps = ["MAKE_CLOSURE", "BUILD_ARRAY", "BUILD_OBJECT", "CALL", "CALL_METHOD", "NEW"];
|
|
3
4
|
|
|
4
5
|
// Creates specialized opcodes for the most frequent (OPCODE + single_integer_operand) pairs.
|
|
5
6
|
// Example: [OP.LOAD_CONST, 1] becomes [SPECIALIZED_LOAD_CONST_1].
|
|
@@ -7,13 +8,7 @@ import { getInstructionSize, nextFreeSlot, U16_MAX } from "../../utils/op-utils.
|
|
|
7
8
|
// MAKE_CLOSURE and other N-sized instructions cannot be specialized
|
|
8
9
|
// Runs after selfModifying but before resolveLabels (operands stay plain numbers).
|
|
9
10
|
export function specializedOpcodes(bc, compiler) {
|
|
10
|
-
const disallowedOps = new Set(
|
|
11
|
-
|
|
12
|
-
// ── Collect used opcodes exactly as specified ─────────────────────────────
|
|
13
|
-
const usedOpcodes = new Set(Object.keys(compiler.OP_NAME).map(k => parseInt(k, 10)).filter(v => !isNaN(v)));
|
|
14
|
-
if (usedOpcodes.size > U16_MAX) return {
|
|
15
|
-
bytecode: bc
|
|
16
|
-
};
|
|
11
|
+
const disallowedOps = new Set(nSizedOps.map(name => compiler.OP[name]));
|
|
17
12
|
|
|
18
13
|
// ── Step 1: count frequency of eligible (op, operand) pairs ───────────────
|
|
19
14
|
const freqMap = new Map();
|
|
@@ -24,7 +19,21 @@ export function specializedOpcodes(bc, compiler) {
|
|
|
24
19
|
// Only supports between 1-6 operands
|
|
25
20
|
const operandCount = getInstructionSize(instr) - 1;
|
|
26
21
|
if (operandCount < 1 || operandCount > 6) continue;
|
|
27
|
-
|
|
22
|
+
|
|
23
|
+
// Convert numbers into operand objects so they can be modified elsewhere and preserved
|
|
24
|
+
const oldOperands = instr.slice(1);
|
|
25
|
+
const operands = oldOperands.map(operand => {
|
|
26
|
+
if (typeof operand === "number") {
|
|
27
|
+
return {
|
|
28
|
+
type: "number",
|
|
29
|
+
value: operand,
|
|
30
|
+
resolvedValue: operand
|
|
31
|
+
};
|
|
32
|
+
}
|
|
33
|
+
return operand;
|
|
34
|
+
});
|
|
35
|
+
instr.length = 1;
|
|
36
|
+
instr.push(...operands);
|
|
28
37
|
const operandsKey = JSON.stringify(operands);
|
|
29
38
|
const key = `${op},${operandsKey}`;
|
|
30
39
|
const entry = freqMap.get(key);
|
|
@@ -50,7 +59,7 @@ export function specializedOpcodes(bc, compiler) {
|
|
|
50
59
|
const sigToSpecial = new Map();
|
|
51
60
|
const specializedOps = {};
|
|
52
61
|
for (let i = 0; i < candidates.length; i++) {
|
|
53
|
-
const specialOp = nextFreeSlot(
|
|
62
|
+
const specialOp = nextFreeSlot(compiler);
|
|
54
63
|
if (specialOp === -1) break;
|
|
55
64
|
const {
|
|
56
65
|
op: originalOp,
|
|
@@ -92,14 +101,10 @@ export function specializedOpcodes(bc, compiler) {
|
|
|
92
101
|
const newOperands = operands.map(operand => {
|
|
93
102
|
const operandAsObject = typeof operand === "object" && operand ? operand : {
|
|
94
103
|
type: "number",
|
|
95
|
-
value: operand,
|
|
96
104
|
resolvedValue: operand
|
|
97
105
|
};
|
|
98
|
-
|
|
99
|
-
|
|
100
|
-
placeholder: true
|
|
101
|
-
};
|
|
102
|
-
return newOperand;
|
|
106
|
+
operandAsObject.placeholder = true;
|
|
107
|
+
return operandAsObject;
|
|
103
108
|
});
|
|
104
109
|
const newInstr = [specialOpCode, ...newOperands];
|
|
105
110
|
|
|
@@ -0,0 +1,202 @@
|
|
|
1
|
+
// Goes through the switch case for defined identifiers on the statement level
|
|
2
|
+
// Example:
|
|
3
|
+
// case OP.LOAD_CONST: {
|
|
4
|
+
// var dst = this._operand();
|
|
5
|
+
// frame.regs[dst] = this._constant();
|
|
6
|
+
// break;
|
|
7
|
+
// }
|
|
8
|
+
// You find "dst" is defined in this scope.
|
|
9
|
+
// You first check the compiler to see if it's already assigned an index in compiler._internals mapping varName=>index
|
|
10
|
+
// If not found, use compiler._internals.globally.size as the new index (when options.randomizeOpcodes is off, when on, choose random between 0 and 65535), and add varName=>index to compiler._internals
|
|
11
|
+
// Then replace the VariableDeclaration to an AssignmentExpression setting left this._internals[index] = init;
|
|
12
|
+
// Then replace all identifiers of "dst" to this._internals[index] as well (Updates references)
|
|
13
|
+
// Final output:
|
|
14
|
+
// case OP.LOAD_CONST: {
|
|
15
|
+
// this._internals[index] = this._operand();
|
|
16
|
+
// frame.regs[this._internals[index]] = this._constant();
|
|
17
|
+
// break;
|
|
18
|
+
// }
|
|
19
|
+
|
|
20
|
+
import * as t from "@babel/types";
|
|
21
|
+
import traverseImport from "@babel/traverse";
|
|
22
|
+
import { ok } from "assert";
|
|
23
|
+
import { getRandomInt } from "../../utils/random-utils.js";
|
|
24
|
+
import { U16_MAX } from "../../utils/op-utils.js";
|
|
25
|
+
const traverse = traverseImport.default || traverseImport;
|
|
26
|
+
export function makeInternalsAccess(index) {
|
|
27
|
+
return t.memberExpression(t.memberExpression(t.thisExpression(), t.identifier("_internals")), t.numericLiteral(index), true // computed
|
|
28
|
+
);
|
|
29
|
+
}
|
|
30
|
+
function collectUsedIndices(compiler) {
|
|
31
|
+
const used = new Set();
|
|
32
|
+
for (const v of compiler._internals.globally.values()) used.add(v);
|
|
33
|
+
for (const opMap of compiler._internals.opcodes.values()) {
|
|
34
|
+
for (const v of opMap.values()) used.add(v);
|
|
35
|
+
}
|
|
36
|
+
return used;
|
|
37
|
+
}
|
|
38
|
+
|
|
39
|
+
// Assign or look up the _internals slot index for a variable name within a
|
|
40
|
+
// specific opcode handler.
|
|
41
|
+
//
|
|
42
|
+
// _internals.opcodes[currentOpcode] is the source of truth for this opcode.
|
|
43
|
+
// _internals.globally holds the shared pool written on first sight.
|
|
44
|
+
//
|
|
45
|
+
// randomizeOpcodes OFF → always reuse / create in globally, mirror to opcodes.
|
|
46
|
+
// randomizeOpcodes ON → first time a name is seen: create global slot.
|
|
47
|
+
// subsequent opcodes: 50% reuse global, 50% create an
|
|
48
|
+
// opcode-specific random slot (NOT written to globally).
|
|
49
|
+
function assignInternalsIndex(name, compiler, currentOpcode) {
|
|
50
|
+
// Ensure per-opcode map exists
|
|
51
|
+
let opcodeMap = compiler._internals.opcodes.get(currentOpcode);
|
|
52
|
+
if (!opcodeMap) {
|
|
53
|
+
opcodeMap = new Map();
|
|
54
|
+
compiler._internals.opcodes.set(currentOpcode, opcodeMap);
|
|
55
|
+
}
|
|
56
|
+
|
|
57
|
+
// Already registered for this opcode — return immediately
|
|
58
|
+
const existing = opcodeMap.get(name);
|
|
59
|
+
if (existing !== undefined) return existing;
|
|
60
|
+
const globalIndex = compiler._internals.globally.get(name);
|
|
61
|
+
let index;
|
|
62
|
+
if (!compiler.options.randomizeOpcodes) {
|
|
63
|
+
// Non-random: always share the global sequential slot
|
|
64
|
+
if (globalIndex === undefined) {
|
|
65
|
+
index = compiler._internals.globally.size;
|
|
66
|
+
compiler._internals.globally.set(name, index);
|
|
67
|
+
} else {
|
|
68
|
+
index = globalIndex;
|
|
69
|
+
}
|
|
70
|
+
} else if (globalIndex === undefined) {
|
|
71
|
+
// First opcode to declare this variable — establish the global slot
|
|
72
|
+
const used = collectUsedIndices(compiler);
|
|
73
|
+
let candidate;
|
|
74
|
+
do {
|
|
75
|
+
candidate = getRandomInt(0, U16_MAX);
|
|
76
|
+
} while (used.has(candidate));
|
|
77
|
+
index = candidate;
|
|
78
|
+
compiler._internals.globally.set(name, index);
|
|
79
|
+
} else {
|
|
80
|
+
// Already in global: 50% chance to reuse, 50% opcode-specific new slot
|
|
81
|
+
if (Math.random() < 0.5) {
|
|
82
|
+
index = globalIndex;
|
|
83
|
+
} else {
|
|
84
|
+
const used = collectUsedIndices(compiler);
|
|
85
|
+
let candidate;
|
|
86
|
+
do {
|
|
87
|
+
candidate = getRandomInt(0, U16_MAX);
|
|
88
|
+
} while (used.has(candidate));
|
|
89
|
+
index = candidate;
|
|
90
|
+
// Intentionally NOT written to globally — this slot is opcode-specific
|
|
91
|
+
}
|
|
92
|
+
}
|
|
93
|
+
opcodeMap.set(name, index);
|
|
94
|
+
return index;
|
|
95
|
+
}
|
|
96
|
+
export function applyInternalVariablesToSwitchCase(node, compiler, currentOpcode) {
|
|
97
|
+
// Work with the actual body array (block body or flat consequent)
|
|
98
|
+
let bodyArr;
|
|
99
|
+
if (node.consequent.length === 1 && t.isBlockStatement(node.consequent[0])) {
|
|
100
|
+
bodyArr = node.consequent[0].body;
|
|
101
|
+
} else {
|
|
102
|
+
bodyArr = node.consequent;
|
|
103
|
+
}
|
|
104
|
+
|
|
105
|
+
// Single traversal: declarations and references handled in one pass.
|
|
106
|
+
//
|
|
107
|
+
// Declaration (Identifier is VariableDeclarator.id):
|
|
108
|
+
// → register/look-up slot, replace entire VariableDeclaration with
|
|
109
|
+
// AssignmentExpression (bare for ForStatement.init, else ExpressionStatement).
|
|
110
|
+
//
|
|
111
|
+
// Reference (any other Identifier):
|
|
112
|
+
// → look up opcodes[currentOpcode] (source of truth) and replace if found.
|
|
113
|
+
// This handles cross-statement refs produced by micro-opcode splitting.
|
|
114
|
+
const syntheticFile = t.file(t.program(bodyArr));
|
|
115
|
+
const illegalNames = new Set(); // Nested closure names are skipped
|
|
116
|
+
|
|
117
|
+
traverse(syntheticFile, {
|
|
118
|
+
Identifier(path) {
|
|
119
|
+
const name = path.node.name;
|
|
120
|
+
if (illegalNames.has(name)) return;
|
|
121
|
+
|
|
122
|
+
// Skip non-computed property names: obj.name
|
|
123
|
+
if (t.isMemberExpression(path.parent) && !path.parent.computed && path.parent.property === path.node) {
|
|
124
|
+
return;
|
|
125
|
+
}
|
|
126
|
+
|
|
127
|
+
// Skip non-computed object-property keys: { name: value }
|
|
128
|
+
if (t.isObjectProperty(path.parent) && !path.parent.computed && path.parent.key === path.node) {
|
|
129
|
+
return;
|
|
130
|
+
}
|
|
131
|
+
|
|
132
|
+
// Don't descend into nested function scopes
|
|
133
|
+
if (path.find(p => p.isFunctionDeclaration() || p.isFunctionExpression() || p.isArrowFunctionExpression())) {
|
|
134
|
+
return;
|
|
135
|
+
}
|
|
136
|
+
|
|
137
|
+
// ── Declaration binding ──────────────────────────────────────────────
|
|
138
|
+
if (t.isVariableDeclarator(path.parent) && path.parent.id === path.node) {
|
|
139
|
+
// Verify it's not referenced in nested closure (illegal)
|
|
140
|
+
const binding = path.scope.getBinding(name);
|
|
141
|
+
if (binding?.referencePaths.some(rp => rp.findParent(p => p.isFunctionDeclaration() || p.isFunctionExpression() || p.isArrowFunctionExpression()))) {
|
|
142
|
+
illegalNames.add(name);
|
|
143
|
+
return;
|
|
144
|
+
}
|
|
145
|
+
const index = assignInternalsIndex(name, compiler, currentOpcode);
|
|
146
|
+
const init = path.parent.init;
|
|
147
|
+
const assignment = t.assignmentExpression("=", makeInternalsAccess(index), init ?? t.identifier("undefined"));
|
|
148
|
+
|
|
149
|
+
// Two levels up: VariableDeclarator → VariableDeclaration
|
|
150
|
+
const varDeclPath = path.parentPath.parentPath;
|
|
151
|
+
if (t.isForStatement(varDeclPath.parent) && varDeclPath.parent.init === varDeclPath.node) {
|
|
152
|
+
// ForStatement.init accepts an Expression directly
|
|
153
|
+
varDeclPath.replaceWith(assignment);
|
|
154
|
+
} else {
|
|
155
|
+
varDeclPath.replaceWith(t.expressionStatement(assignment));
|
|
156
|
+
}
|
|
157
|
+
return;
|
|
158
|
+
}
|
|
159
|
+
|
|
160
|
+
// ── Reference ───────────────────────────────────────────────────────
|
|
161
|
+
// Source of truth for this opcode is its own per-opcode map
|
|
162
|
+
const opcodeMap = compiler._internals.opcodes.get(currentOpcode);
|
|
163
|
+
const index = opcodeMap?.get(name);
|
|
164
|
+
if (index !== undefined) {
|
|
165
|
+
path.replaceWith(makeInternalsAccess(index));
|
|
166
|
+
path.skip();
|
|
167
|
+
}
|
|
168
|
+
}
|
|
169
|
+
});
|
|
170
|
+
}
|
|
171
|
+
|
|
172
|
+
// This takes the AST and finds the runtime switch statement via the leading
|
|
173
|
+
// comment "@SWITCH" then applies the above transformation to each switch case.
|
|
174
|
+
export function applyInteralVariablesToRuntime(ast, compiler) {
|
|
175
|
+
let switchStatement = null;
|
|
176
|
+
traverse(ast, {
|
|
177
|
+
SwitchStatement(path) {
|
|
178
|
+
if (path.node.leadingComments?.some(c => c.value.includes("@SWITCH"))) {
|
|
179
|
+
switchStatement = path.node;
|
|
180
|
+
path.stop();
|
|
181
|
+
}
|
|
182
|
+
}
|
|
183
|
+
});
|
|
184
|
+
ok(switchStatement, "Could not find @SWITCH statement for internal variables");
|
|
185
|
+
for (const sc of switchStatement.cases) {
|
|
186
|
+
const test = sc.test;
|
|
187
|
+
let currentOpcode = null;
|
|
188
|
+
if (test && t.isMemberExpression(test) && t.isIdentifier(test.object, {
|
|
189
|
+
name: "OP"
|
|
190
|
+
}) && t.isIdentifier(test.property)) {
|
|
191
|
+
// case OP.LOAD_CONST: → resolve via compiler.OP
|
|
192
|
+
const opName = test.property.name;
|
|
193
|
+
const val = compiler.OP[opName];
|
|
194
|
+
if (val !== undefined) currentOpcode = val;
|
|
195
|
+
} else if (test && t.isNumericLiteral(test)) {
|
|
196
|
+
// Already a numeric literal (e.g. generated micro-opcode cases)
|
|
197
|
+
currentOpcode = test.value;
|
|
198
|
+
}
|
|
199
|
+
if (currentOpcode === null) continue;
|
|
200
|
+
applyInternalVariablesToSwitchCase(sc, compiler, currentOpcode);
|
|
201
|
+
}
|
|
202
|
+
}
|
|
@@ -17,6 +17,29 @@ function extractCaseBody(switchCase) {
|
|
|
17
17
|
}
|
|
18
18
|
return stmts.filter(s => !t.isBreakStatement(s) && !t.isEmptyStatement(s));
|
|
19
19
|
}
|
|
20
|
+
export function getOpcodeToCaseMap(switchStatement, compiler) {
|
|
21
|
+
// Build a map opName → SwitchCase from the existing OP.xxx case tests.
|
|
22
|
+
const opcodeToCaseMap = new Map();
|
|
23
|
+
for (const sc of switchStatement.cases) {
|
|
24
|
+
const test = sc.test;
|
|
25
|
+
if (!test) continue;
|
|
26
|
+
let opcode;
|
|
27
|
+
let opName;
|
|
28
|
+
if (t.isMemberExpression(test) && t.isIdentifier(test.object, {
|
|
29
|
+
name: "OP"
|
|
30
|
+
}) && t.isIdentifier(test.property)) {
|
|
31
|
+
opName = test.property.name;
|
|
32
|
+
opcode = +Object.keys(compiler.OP_NAME).find(key => compiler.OP_NAME[key] == opName);
|
|
33
|
+
} else if (t.isNumericLiteral(test)) {
|
|
34
|
+
opcode = test.value;
|
|
35
|
+
}
|
|
36
|
+
ok(typeof opcode === "number" && !Number.isNaN(opcode), `Failed to parse ${opcode} from ${opName}`);
|
|
37
|
+
if (opcode !== undefined) {
|
|
38
|
+
opcodeToCaseMap.set(opcode, sc);
|
|
39
|
+
}
|
|
40
|
+
}
|
|
41
|
+
return opcodeToCaseMap;
|
|
42
|
+
}
|
|
20
43
|
|
|
21
44
|
// Append a generated switch case for every entry in compiler.MACRO_OPS.
|
|
22
45
|
// Each case inlines the constituent case bodies directly — no operand stack,
|
|
@@ -35,17 +58,7 @@ export function applyMacroOpcodes(ast, compiler) {
|
|
|
35
58
|
}
|
|
36
59
|
});
|
|
37
60
|
ok(switchStatement, "Could not find @SWITCH statement for macro opcodes");
|
|
38
|
-
|
|
39
|
-
// Build a map opName → SwitchCase from the existing OP.xxx case tests.
|
|
40
|
-
const nameToCaseMap = new Map();
|
|
41
|
-
for (const sc of switchStatement.cases) {
|
|
42
|
-
const test = sc.test;
|
|
43
|
-
if (test && t.isMemberExpression(test) && t.isIdentifier(test.object, {
|
|
44
|
-
name: "OP"
|
|
45
|
-
}) && t.isIdentifier(test.property)) {
|
|
46
|
-
nameToCaseMap.set(test.property.name, sc);
|
|
47
|
-
}
|
|
48
|
-
}
|
|
61
|
+
const opcodeToCaseMap = getOpcodeToCaseMap(switchStatement, compiler);
|
|
49
62
|
for (const [macroOpStr, constituentOps] of Object.entries(compiler.MACRO_OPS)) {
|
|
50
63
|
const macroOpCode = Number(macroOpStr);
|
|
51
64
|
const N = constituentOps.length;
|
|
@@ -54,20 +67,19 @@ export function applyMacroOpcodes(ast, compiler) {
|
|
|
54
67
|
const constituentCases = [];
|
|
55
68
|
let allFound = true;
|
|
56
69
|
for (const opVal of constituentOps) {
|
|
57
|
-
const
|
|
58
|
-
if (!opName) {
|
|
59
|
-
allFound = false;
|
|
60
|
-
break;
|
|
61
|
-
}
|
|
62
|
-
const found = nameToCaseMap.get(opName);
|
|
70
|
+
const found = opcodeToCaseMap.get(opVal);
|
|
63
71
|
if (!found) {
|
|
64
72
|
allFound = false;
|
|
65
73
|
break;
|
|
66
74
|
}
|
|
67
75
|
constituentCases.push(found);
|
|
68
76
|
}
|
|
69
|
-
if (!allFound)
|
|
77
|
+
if (!allFound) {
|
|
78
|
+
throw new Error(`Could not find all constituent ops for macro op ${macroOpCode}`);
|
|
79
|
+
}
|
|
70
80
|
const opNames = constituentOps.map(v => compiler.OP_NAME[v] ?? `OP_${v}`);
|
|
81
|
+
let newName = opNames.join(",");
|
|
82
|
+
compiler.OP_NAME[macroOpCode] = newName;
|
|
71
83
|
|
|
72
84
|
// ── Build the macro case body ──────────────────────────────────────────
|
|
73
85
|
// Clone and inline each sub-instruction's case body directly.
|