provider.getResolver(name) not working for unicode characters (valid ENS names)

[lucashenning]

Describe the bug
The provider.getResolver(name) function does not work with all unicode characters. According to the definition of ENS names, unicode characters are valid. The function should resolve names including unicode characters, e.g. emoticons like 👋

Reproduction steps

const r1 = await provider.getResolver("foo.eth"); // this works
console.log(r1?.address); // output: 0x4976fb03C32e5B8cfe2b6cCB31c09Ba78EBaBa41
const r2 = await provider.getResolver("🚀🚀🚀.eth"); // this doesn't work, throws error
console.log(r2?.address); 

Stack Trace

(node:39562) UnhandledPromiseRejectionWarning: Error: STRINGPREP_CONTAINS_UNASSIGNED
    at /ens-test/node_modules/@ethersproject/strings/src.ts/idna.ts:193:19
    at Array.forEach (<anonymous>)
    at nameprep (/ens-test/node_modules/@ethersproject/strings/src.ts/idna.ts:191:11)
    at namehash (/ens-test/node_modules/@ethersproject/hash/src.ts/namehash.ts:40:43)
    at InfuraProvider.<anonymous> (/ens-test/node_modules/@ethersproject/providers/src.ts/base-provider.ts:1701:43)
    at step (/ens-test/node_modules/@ethersproject/providers/lib/base-provider.js:48:23)
    at Object.next (/ens-test/node_modules/@ethersproject/providers/lib/base-provider.js:29:53)
    at fulfilled (/ens-test/node_modules/@ethersproject/providers/lib/base-provider.js:20:58)
    at processTicksAndRejections (internal/process/task_queues.js:95:5)

Affected function

ethers.js/packages/providers/src.ts/base-provider.ts

Lines 1783 to 1809 in 33a029e

	async _getResolver(name: string, operation?: string): Promise<string> {
	if (operation == null) { operation = "ENS"; }
	const network = await this.getNetwork();
	// No ENS...
	if (!network.ensAddress) {
	logger.throwError(
	"network does not support ENS",
	Logger.errors.UNSUPPORTED_OPERATION,
	{ operation, network: network.name }
	);
	}
	try {
	// keccak256("resolver(bytes32)")
	const addrData = await this.call({
	to: network.ensAddress,
	data: ("0x0178b8bf" + namehash(name).substring(2))
	});
	return this.formatter.callAddress(addrData);
	} catch (error) {
	// ENS registry cannot throw errors on resolver(bytes32)
	}
	return null;
	}

The error seems to be generated here:

ethers.js/packages/strings/src.ts/idna.ts

Lines 190 to 195 in ce8f1e4

	// Prohibit Unassigned Code Points (Table A.1)
	codes.forEach((code) => {
	if (_nameprepTableA1(code)) {
	throw new Error("STRINGPREP_CONTAINS_UNASSIGNED");
	}
	});

Environment:

• node v14.17.1
• ethers v5.5.4

Search Terms
ENS name resolution resolve unicode characters

Related issues

• avatar record doesn't work for domains with unicode characters ensdomains/ens-metadata-service#50

[ricmoo]

This is a known issue, and is pending the updated stringprep-like normalization from ENS regarding UTF-8 joiners.

Closing as a duplicate of #2376, which is being kept updated. :)

[ricmoo]

This has been address in v5.7.0. See the related issue for details.

[xwartz]

@ricmoo v5.7.0 still has problems parsing ens with ZWSP, such as vitalik‌.eth.