Files
SocialPub/PrivaPub/Federation/Objects/InstanceDescriber.cs
T
thepraandClaude Opus 5.5 dcd024100a
Build / Build (push) Successful in 36s
Every remote object keeps its raw form and how it reached us, for the client's details view
- ObjectRecord, one per stored remote object: the raw JSON (up to 256 KB, always hashed), delivered or fetched,
  refetched from origin or not, the activity that brought it (or caused the fetch), shared or personal inbox, the
  signature's key, algorithm and signed headers, received time, published and updated, the delivering activity's
  @context, and up to ten later revisions from Update.
- Delivery details travel from InboxReceiver through the inbox job to the handlers as Arrival.Current.
- A host is described from its NodeInfo when we first hear from it, at most weekly (DescribeInstance job), never when
  someone opens the details view.
- GET /api/privapub/v1/statuses/:id/provenance and /api/privapub/v1/instances/:host, with the extensions an object
  used detected from its raw form (044f quotes, interaction policies, contexts, proofs, Misskey fields, MFM, FEP-8967
  links, emoji, polls, language maps, url variants, Markdown content).

Checked live: a GoToSocial reply shows as delivered to the shared inbox, signed hs2019 with GoToSocial's fragment-less
key id, with its interaction policy detected, and gts.test is described as gotosocial 0.22.1.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_012CzABvBkbcFqoHdmi8b9WB
2026-10-01 17:49:22 +02:00

85 lines
3.5 KiB
C#

using MongoDB.Entities;
using PrivaPub.Infrastructure.Http;
using PrivaPub.Infrastructure.Jobs;
using PrivaPub.Models.Jobs;
using System.Text.Json;
namespace PrivaPub.Federation.Objects
{
public class InstanceDescriber : IJobHandler
{
const int MaxNodeInfoChars = 64 * 1024;
static readonly string[] Schemas =
{
"http://nodeinfo.diaspora.software/ns/schema/2.1",
"http://nodeinfo.diaspora.software/ns/schema/2.0"
};
readonly IFederationHttp _http;
public InstanceDescriber(IFederationHttp http) => _http = http;
public JobKind Kind => JobKind.DescribeInstance;
public int Concurrency => 1;
public int MaxAttempts => 2;
public int PerHostLimit => 1;
public async Task<JobOutcome> Handle(Job job, CancellationToken token)
{
var host = job.Payload;
using var links = await _http.GetJson($"https://{host}/.well-known/nodeinfo", "application/json", sign: default, token);
var href = links == default ? default : Schemas.Select(schema => LinkFor(links.Root, schema)).FirstOrDefault(link => link != default);
if (href == default)
return await Described(host, default, "no NodeInfo", token);
using var nodeInfo = await _http.GetJson(href, "application/json", sign: default, token);
return nodeInfo == default ? await Described(host, default, "NodeInfo unreadable", token) : await Described(host, nodeInfo.Root, default, token);
}
static async Task<JobOutcome> Described(string host, JsonElement? nodeInfo, string error, CancellationToken token)
{
var update = DB.Default.Update<RemoteInstance>()
.Match(i => i.Host == host)
.Modify(i => i.DescribedAt, DateTime.UtcNow)
.Modify(i => i.DescriptionError, error)
.Option(o => o.IsUpsert = true);
if (nodeInfo is { } root)
{
var raw = root.GetRawText();
update = update
.Modify(i => i.Software, Text(root, "software", "name"))
.Modify(i => i.SoftwareVersion, Text(root, "software", "version"))
.Modify(i => i.NodeName, Text(root, "metadata", "nodeName"))
.Modify(i => i.Protocols, root.TryGetProperty("protocols", out var protocols) && protocols.ValueKind == JsonValueKind.Array
? protocols.EnumerateArray().Where(p => p.ValueKind == JsonValueKind.String).Select(p => p.GetString()).Take(16).ToList()
: new List<string>())
.Modify(i => i.OpenRegistrations, root.TryGetProperty("openRegistrations", out var open) && open.ValueKind is JsonValueKind.True or JsonValueKind.False
? open.GetBoolean()
: default(bool?))
.Modify(i => i.NodeInfo, raw.Length <= MaxNodeInfoChars ? raw : default);
}
await update.ExecuteAsync(token);
return JobOutcome.Done;
}
static string LinkFor(JsonElement root, string schema) =>
root.TryGetProperty("links", out var links) && links.ValueKind == JsonValueKind.Array
? links.EnumerateArray()
.Where(l => l.ValueKind == JsonValueKind.Object && l.TryGetProperty("rel", out var rel) && rel.ValueKind == JsonValueKind.String && rel.GetString() == schema)
.Select(l => l.TryGetProperty("href", out var href) && href.ValueKind == JsonValueKind.String ? href.GetString() : default)
.FirstOrDefault(h => h != default && h.StartsWith("https://", StringComparison.OrdinalIgnoreCase))
: default;
static string Text(JsonElement root, string parent, string name)
{
if (!root.TryGetProperty(parent, out var section) || section.ValueKind != JsonValueKind.Object
|| !section.TryGetProperty(name, out var value) || value.ValueKind != JsonValueKind.String)
return default;
var text = value.GetString();
return text.Length <= 200 ? text : text[..200];
}
}
}