fix!: conform to Delegated Routing V1 HTTP spec

[hacdias]

Conform to Routing V1 HTTP spec: https://specs.ipfs.tech/routing/http-routing-v1/

I'm giving a talk on IPFS Connect about Delegated Routing. It would be useful to have some JavaScript library with the latest. I will definitely need some pointers from TS folks.

• Client
  • GET /providers
  • GET /peers
  • GET /ipns
  • PUT /ipns
  • Tests
• Server
  • GET /providers
  • GET /peers
  • GET /ipns
  • PUT /ipns
  • Tests

[achingbrain]

Thanks for opening this, some comments inline

[achingbrain]

Can the return type be guaranteed to be BitswapRecord?

[hacdias]

@achingbrain we don't want BitswapRecords to be used, see https://specs.ipfs.tech/routing/http-routing-v1/#get-routing-v1-providers-cid - it should be a generic record. PeerRecords are recommended.

[achingbrain]

Then can the return type be guaranteed to be PeerRecord? What I'm getting at here is it makes the users life easier if they don't have to deal with a union type.

[lidel]

There could be other schemas than peer. But in practice we can limit this to PeerRecord responses: ignore unknown schemas unless they have ID and Addrs fields like the legacy bitswap schema, and interpret them aspeer schema.

For the wider context, the only /routing/v1 implementation (afaik) that returned bitswap schema was cid.contact (IPNI), and it had ID and Addrs just like peer schema does.

[achingbrain]

Ok, sounds like we can just convert any BitswapRecords to PeerRecords and drop the union type then.

[BigLep]

@achingbrain (or others on the Helia team):

1. Given work is happening here, can you please take on Clarify scope  #35 ? I'm particularly interested in how this relates to https://github.com/libp2p/js-http-v1-content-routing
2. Is this addressing "task for browser-accessible delegated routing" in [🏆 Golden path scenario] Browsers can reliably retrieve content from any modern Kubo node providing content helia#255 ? If so, can we please update the task list to reference this issue?

[hacdias]

@achingbrain I can't reproduce the failing test in ubuntu. Do you have any idea what might be causing it? Other than that, this PR is ready for review.

[lidel]

@hacdias fwiw I can reproduce error locally: the npm run test:node is failing for me with the same error as CI (nodejs v20.6.1 on linux).

It looks like

https://github.com/ipfs/helia-routing-v1-http-api/blob/1dd9e7019b0f1d647ac663090c7d9690665a33e7/packages/server/test/index.spec.ts#L280

returns some random bytes from memory instead of the valid ipns record:

> console.log(new TextDecoder('utf-8').decode(new Uint8Array(arrayBuffer)))
{"Providers":[{"Schema":"peer","Protocols":["transport-bitswap"],"ID":"12D3KooWSAPTftgtTWjvVYXVChuqXxkU8b7cx1dKbUoiHNQ9iJbc","Addrs":["/ip4/123.123.123.123/tcp/123"]},{"Schema":"peer","Protocols":�%�Z�%@port-bitswap"],"ID":"12D3KooWA6yLJzjuidJ8SCiQcESWdcfsLXdtwACsPU9iiSzZzwNh","Addrs":["/ip4/123.123.123.123/tcp/123"]}]}�֘% ��/�X-�c����3�C���VS
�%�Z�for= reamD{{{{V�֘%V{{{{{ +�r|Ȃ���`��s�Qx�C�%lQ����T��se@)-��Ҧ]��ŭ1��|#��>�Ec�]ʵ��9+�r|Ȃ���`��s�Qx�C�%lQ����T��%V���%V{{{{V{@V{{{{{{"Schema":"peer","Protocols":["transport-bitswap"],"ID":"12D3KooWRhjNhkGatKpVm7cLjC7u5oPqRsqU2sRoq2xd92daw8eP","Addrs":["/ip4/123.123.123.123/tcp/123"]}
{"Schema":"peer","Protocols":["transport-bitswap"],"ID":"12D3KooWBnEiZbtipzpfuSHtwjm52pXdSBDgJn4G6yLRpiPZsyS5","Addrs":["/ip4/123.123.123.123/tcp/123"]}
eadykpr{"Schema":"peer","Protocols":["transport-bitswap"],"ID":"12D3KooWRhjNhkGatKpVm7cLjC7u5oPqRsqU2sRoq2xd92daw8eP","Addrs":["/ip4/123.123.123.123/tcp/123"]}
{"Schema":"peer","Protocols":["transport-bitswap"],"ID":"12D3KooWBnEiZbtipzpfuSHtwjm52pXdSBDgJn4G6yLRpiPZsyS5","Addrs":["/ip4/123.123.123.123/tcp/123"]}
{"Schema":"peer","Protocols":["transport-bitswap"],"ID":"12D3KooWRhjNhkGatKpVm7cLjC7u5oPqRsqU2sRoq2xd92daw8eP","Addrs":["/ip4/123.123.123.123/tcp/123"]}
{"Schema":"peer","Protocols":["transport-bitswap"],"ID":"12D3KooWBnEiZbtipzpfuSHtwjm52pXdSBDgJn4G6yLRpiPZsyS5","Addrs":["/ip4/123.123.123.123/tcp/123"]}
�=\��n�LC��p�]�*�E$<%�h����&@@,j&.��OK�H��0lDP��o殐b��hqx;��0�}7\vZ�N��*�E$<%�h����& ��	�pOoYo
{{{{{{{{{{"Peers":[{"Schema":"peer","Protocols":["transport-bitswap"],"ID":"12D3KooWLZreCQLmSjdwhA3jGXYZGwjtooMJ38qkrGdkxoP7fyJS","Addrs":["/ip4/123.123.123.123/tcp/123"]}] ��1���Ażȸ�d=""��
%@G%��O+W(��[�S̠4��fo�1b*�^���|�{��1���Ażȸ�d=""��                                                                                                                                             ��n��k�
 �oQ�9�1c��z�au2K��0jS�Lf1�0�@F��׹������t[�2 WT�;�5�hc�oQ�9�1c��z�au2K��0jS�Lf1�0�J2023-10-19T00:59:05.157000000Z/ipfs/bafkreifjjcie6lypi6ny7amxnfftagclbuxndqonfipmb64f2km2devei4�cTTL��eValueXA/ipfs/bafkreifjjcie6lypi6ny7amxnfftagclbuxndqonfipmb64f2km2devei4hSequencehValidityX2023-10-19T00:59:05.157000000ZlValidityTypeipns-signature:ipns-signature:�cTTL��eValueXA/ipfs/bafkreifjjcie6lypi6ny7amxnfftagclbuxndqonfipmb64f2km2devei4hSequencehValidityX2023-10-19T00:59:05.157000000ZlValidityTypeF��׹������t[�2 WT�;�5�hcEOL/ipfs/bafkreifjjcie6lypi6ny7amxnfftagclbuxndqonfipmb64f2km2devei42023-10-19T00:59:05.157000000ZEOLF��׹������t[�2 WT�;�5�hc/ipfs/bafkreifjjcie6lypi6ny7amxnfftagclbuxndqonfipmb64f2km2devei42023-10-19T00:59:05.157000000Z
A/ipfs/bafkreifjjcie6lypi6ny7amxnfftagclbuxndqonfipmb64f2km2devei4@��Qm�Φ�ڷ-1x�_�-�D?��s��%Y5Q8�
%Kf������?=;�* ��s"2023-10-19T00:59:05.157000000Z(0���/B@���x��V������7:p�2�2�%��}� Y�`�J�x���3�;�^͇��5G�J��cTTL��eValueXA/ipfs/bafkreifjjcie6lypi6ny7amxnfftagclbuxndqonfipmb64f2km2devei4hSequencehValidityX2023-10-19T00:59:05.157000000ZlValidityType/ipns/ �oQ�9�1c��z�au2K��0jS�Lf1�0�JH�֘%V*�������2�������K���f�����������
                                                                                                           ���q����������C�������������������P�֘%V��֘%V�֘%V�
                                                                                                                                                         �֘%V�֘%V��֘%V�.C�%V����%VP�֘%Vh�֘%V��֘%V�4	��֘%V�֘%V�4	�֘%V��֘%V�a ر֘%V�֘%V��֘%VH�֘%VX/�����֘%`�֘%Vp�֘%V��֘%V���%V��֘%V��%V��֘%VX�֘%Vx�֘%��֘%V8�֘%VP�֘%V`�֘%V��������֘%V�4	��֘%V��֘%V��G֘%V���֘%Vس֘%V��֘%V�2��֘%V�	�֘%0C�%V����%VP�֘%V�֘%V�֘%V�-C�%V����8�֘%V��֘%V�֘%Vp�֘%V@�֘%V�֘%V�
                                                                                                                                                                                                 �֘%Vp�֘%V�֘%V�.C�%V���������4	%V�֘%V8�֘%�֘%V��֘%V��֘%V!4	��֘%V��֘%V'��ܲ�%V(,��֘%V�֘%V3)�/� �֘%V8�֘%VH�֘%VX/����֘%��֘%V��֘%V��֘%V6C%����ȷ֘%V�4	x�֘%V���,8�֘%V�4	h�֘%V�4	 �֘%V����h�֘%V?�8ܲ�%V8��֘%VA%V�.C�%V����%V%V�֘%V8�֘%V��֘%Vж֘%V�֘%V�-C�%V����%V%V��֘%V��֘%V(�֘%Vж֘%V
 H�֘%V��֘%V�֘%V�.C�%V6C����������֘%V��֘%Vp�֘%V��֘%Vt4	�֘%Vt�֘%V��֘%VH�֘%V�G֘%Vи֘%V~Y֘%V}))�֘%V@�֘%V}	P�֘%V����G��֘%Vh�֘%V�4	и֘%V�и֘%Vx�֘%VH�֘%V�ܲ�%V��֘%V�Y֘%V�)��֘%V�֘%V�	�֘%V����GX�֘%V(�֘%V�4	��֘%V���֘%V8�֘%VH�֘%V�G֘%VP�֘%V�Y�)%V`�֘%V��֘%V�	й֘%V����G�֘%V�֘%V�4	P�֘%V�P�֘%VȾ֘%VH�֘%V H֘%V`�֘%V�֘%V�4	��֘%V(�֘%V���ܲ�%V�,��֘%V��֘%V �֘%V��֘%V�	��֘%V�����֘%Vغ֘%V���֘%V�֘%V�4	��֘%V@�֘%V�4	P�֘%V��֘%V��(�֘%V@�֘%V��X�֘%V�4	��֘%V��֘%V�4	��֘%V��֘%V�,��֘%V��֘%V�)%V����֘%VȻ֘%V��֘%Vػ֘%Vh-C�%V��%�֘%V��֘%V��֘%V�֘%�֘%VH�֘%VX/���ȼ֘%x�֘%V��֘%V��֘%V�!��%V�)%V`�֘%�֘%V0�֘%V��;{����X�֘%V�	��֘%V�֘%$6��֘%V	��֘%V�	�֘%V��֘%V$6�֘%V	��֘%V�-C�%V����%V��֘%V �֘%V��֘%V�֘%Vн֘%�֘%V�.C�%V�!�֘%V,��֘%V�֘%Vh�֘%Vp4`�֘%Vp`�֘%V��֘%V�H֘%V��֘%V�X�=��G֘%V��֘%V�q[?�H֘%V��֘%V�gY	�E֘%V��֘%V�%V�G֘%V�֘%Vd�-%V H֘%VP�֘%V�,
%V�H֘%V��֘%Vgp�+%V�I֘%VQ�:�I֘%V��֘%V*Sd7%Vм֘%�֘%VHH֘%V`�֘%V��F                                    %VI֘%V`�֘%V�~I
%V%VX�֘%V�ܲ�%Vи֘%V.��2%V�ݲ�%V0�֘%VM�)%VH�֘%VHH֘%V��֘%V�֘%Vy4	P�֘%V��֘%V�4	��֘%V��֘%V~[��֘%V��֘%V�4	 �֘%V��֘%V����֘%V��֘%V�)[����֘%�֘%V�)��֘%V0�֘%V��֘%V@�֘%Vw	X�֘%V����G��֘%Vp�֘%V�4	��֘%V

[hacdias]

@lidel what is odd to me is that this is only happening on linux then, and looking at what you posted, it seems the odd bytes are also in other messages, not just the getIpns. Then why do those even work? Maybe the ndjson package ignores invalid json.

[lidel]

@hacdias I don't see windows and macos passing on CI – they get cancelled when linux fails, and linux is the fastest runtime.
Could it be fetch in NodeJS being broken again? (I don't see any explicit polyfill in tests, is it finally built-in?)

[hacdias]

@lidel oh, I didn't notice. They all pass locally for me though so I'd assume at least macOS works well (Nodejs 18.18.2, same as CI)

[SgtPooki]

Could it be fetch in NodeJS being broken again? (I don't see any explicit polyfill in tests, is it finally built-in?)

It is built in now, using https://github.com/nodejs/undici

[achingbrain]

This is the cause of the CI failure: fastify/fastify#5118

[github-actions]

🎉 This PR is included in version 1.0.0 🎉

The release is available on GitHub release

Your semantic-release bot 📦🚀

[github-actions]

🎉 This PR is included in version @helia/delegated-routing-v1-http-api-interop-v1.0.0 🎉

The release is available on GitHub release

Your semantic-release bot 📦🚀