perf: incremental parent-map + esprima2 6.0.0 optional-chain support

[itamarga]

TLDR

Two changes: (1) a quadratic parent-map rebuild that made pyjsclear.deobfuscate() run for 15+ minutes on large obfuscator.io-style samples is now an O(1) incremental update; (2) full support for esprima2 6.0.0, which wraps optional chains in an ESTree ChainExpression node.

Changes

• base.py: new record_replacement(...) — an O(1) patch of the cached parent map instead of a full invalidation. ObjectSimplifier, ClassStaticResolver, StringRevealer now call it instead of invalidate_parent_map().
• generator.py: handle ChainExpression (emit the wrapped optional chain) and parenthesize ?? when adjacent to ||/&&.
• ast_helpers.py: register ChainExpression's child key so traversal descends into it and identifiers inside optional chains still resolve.
• Depend on esprima2==6.0.0 (was capped below 5.0.2) and bump to 0.1.6.
• New/updated tests for the incremental parent map, ChainExpression traversal, and ??/||/&& parenthesization; snapshot regenerated (the old one encoded the pre-fix nullish-mixing bug).

Full description

Performance: incremental parent map

Root cause. Every node replacement in ObjectSimplifier, ClassStaticResolver, and StringRevealer called invalidate_parent_map(), so the next find_parent rebuilt the entire parent map with a full-AST walk. With thousands of proxy-object references that is O(n²). Profiling 0.1.5 on a 606KB sample shows build_parent_map called 12,023 times, consuming 980.8s of the 1011.9s total (97%) — 1.76 billion dict.get calls.

Evidence (same machine, both versions under cProfile):

Sample	0.1.5	this branch (cProfile)	this branch (no profiler)
606KB obfuscator.io-style JS	1011.9s	10.4s	3.1s
753KB obfuscator.io-style JS	1134.9s	27.5s	8.3s

Fix. record_replacement(replacement, parent, key, index) patches the cached parent map in O(1). This is safe because the replacement swaps a node in place: in-place swaps don't shift list indices, and replaced subtrees are detached and never looked up again, so only the replacement's own entry changes. The incremental map is behaviourally identical to a full rebuild. The parent-map test asserts build_parent_map is called exactly once across many in-place replacements — a deterministic check rather than a timing threshold.

esprima2 6.0.0 support

esprima2 5.0.2+ changed optional-chaining parsing to the ESTree ChainExpression wrapper (the ?. flag lives on the inner MemberExpression/CallExpression), and now correctly rejects ?? combined with ||/&& without parentheses (older versions leniently accepted it). Three fixes make the latest version work:

• Generator ChainExpression handler — emits the wrapped expression, so optional chains round-trip instead of rendering as /* unknown: ChainExpression */.
• Traversal child key — ChainExpression was not in the static child-key map, and the generic fallback deliberately skips the expression key for non-ExpressionStatement nodes, so transforms never descended into a chain and left identifiers there obfuscated. Registering its child key fixes this.
• Nullish-mixing parentheses — ?? shares a precedence level with ||, so the precedence check alone never parenthesized (a ?? b) || c, producing invalid JS. The generator now adds parentheses whenever ?? is adjacent to ||/&&.

The snapshot changed on exactly two lines: the old output mis-grouped a nullish/logical expression (a ?? (true || …)) — the regenerated snapshot renders the true AST correctly ((a ?? true) || …).

Tests. Full unit + regression suite green: 1759 passed, 3 skipped (fuzz runs separately in CI).

🤖 Generated with Claude Code