Add cs.money worker stack with per-worker IPRoyal residential proxy

Brings up the pull-model scraper: the .NET C2 hands skin+wear jobs to Python nodriver workers that scrape cs.money and post results back, plus the supporting Core/EFCore data model, migrations, and docker-compose orchestration.

IPRoyal proxying lets workers scale horizontally with a distinct residential exit IP each: every worker process mints its own sticky session at startup, and an in-process forwarding proxy injects the gateway auth so Chromium talks only to an auth-free localhost endpoint (zero CDP). On a Cloudflare challenge a worker rotates to a fresh session/IP and re-warms. Verified end-to-end against live IPRoyal: distinct US residential exits per worker and IP rotation on demand.

Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
This commit is contained in:
bob
2026-05-31 15:03:31 -05:00
parent eb5fb0dac7
commit dc7c3f99ae
82 changed files with 8354 additions and 571 deletions

View File

@@ -0,0 +1,13 @@
<Project Sdk="Microsoft.NET.Sdk.Web">
<ItemGroup>
<ProjectReference Include="..\BlueLaminate.Core\BlueLaminate.Core.csproj" />
</ItemGroup>
<PropertyGroup>
<TargetFramework>net10.0</TargetFramework>
<Nullable>enable</Nullable>
<ImplicitUsings>enable</ImplicitUsings>
</PropertyGroup>
</Project>

View File

@@ -0,0 +1,19 @@
using BlueLaminate.Core.CsMoney;
namespace BlueLaminate.C2;
/// <summary>A unit of scrape work handed to a worker: one skin+wear, as a search.</summary>
/// <param name="JobId">Opaque id the worker echoes back when posting results.</param>
/// <param name="SkinId">Catalogue skin this job targets.</param>
/// <param name="ConditionId">Wear band (skin_conditions row), or null for a whole skin.</param>
/// <param name="Search">Free-text market search, e.g. "M4A4 Cyber Security ft".</param>
/// <param name="MaxPages">Safety cap on page fetches (60 items each). The worker
/// paginates by walking the float axis, so a skin+wear needs ceil(listings/60) pages.</param>
public sealed record ScrapeJobDto(string JobId, int SkinId, int? ConditionId, string Search, int MaxPages);
/// <summary>A worker's results for a claimed job: the listings it scraped.</summary>
/// <param name="Items">All sell-order items gathered across pages (raw cs.money shape).</param>
/// <param name="Pages">How many pages the worker fetched.</param>
/// <param name="StoppedReason">Why it stopped. "completed" = full sweep (authoritative);
/// anything else (fetch-cap / challenged / stuck-float-tie) is partial.</param>
public sealed record ScrapeResultDto(List<CsMoneyItem> Items, int Pages, string? StoppedReason);

View File

@@ -0,0 +1,24 @@
# Build context is the REPO ROOT (so Central Package Management's Directory.*.props
# at the root are available). Build with:
# docker compose build (compose sets the context)
FROM mcr.microsoft.com/dotnet/sdk:10.0 AS build
WORKDIR /src
# Restore against the full solution sources the C2 transitively needs.
COPY Directory.Build.props Directory.Packages.props ./
COPY BlueLaminate/ BlueLaminate/
RUN dotnet restore BlueLaminate/BlueLaminate.C2/BlueLaminate.C2.csproj
RUN dotnet publish BlueLaminate/BlueLaminate.C2/BlueLaminate.C2.csproj \
-c Release -o /app --no-restore
FROM mcr.microsoft.com/dotnet/aspnet:10.0 AS runtime
# NOTE: deliberately do NOT install libgssapi-krb5-2. Without it Npgsql logs a
# harmless "cannot load libgssapi_krb5.so.2" line and falls back to password auth;
# WITH it, a failed/misconfigured connection attempt segfaults during GSS negotiation
# (observed: container exit 139 / crash-loop). Graceful failure beats the segfault.
WORKDIR /app
COPY --from=build /app ./
# Bind all interfaces inside the container (overrides appsettings' localhost binding).
ENV ASPNETCORE_URLS=http://+:5080
EXPOSE 5080
ENTRYPOINT ["dotnet", "BlueLaminate.C2.dll"]

View File

@@ -0,0 +1,88 @@
using System.Collections.Concurrent;
using BlueLaminate.Core.CsMoney;
using BlueLaminate.EFCore.Data;
using Microsoft.EntityFrameworkCore;
namespace BlueLaminate.C2;
/// <summary>
/// Hands out scrape jobs to workers, one skin+wear at a time, driven directly by the
/// catalogue's per-band checkpoints (<c>SkinCondition.ListingsSweptAt</c>) rather than
/// a pre-built queue. Each claim picks the stalest band (never-swept first), leases it
/// in memory so two workers can't get the same one, and builds a free-text search. On
/// completion the ingest stamps <c>ListingsSweptAt</c>, so the band drops to the back —
/// the sweep loops the whole catalogue continuously and resumes cleanly after restarts.
/// </summary>
public sealed class JobQueue
{
// A leased condition can't be re-handed-out until released or the lease expires
// (so a crashed worker's band returns to the pool instead of stalling forever).
private static readonly TimeSpan LeaseTtl = TimeSpan.FromMinutes(15);
private const int CandidateBatch = 100;
private readonly SemaphoreSlim _gate = new(1, 1);
private readonly ConcurrentDictionary<int, DateTimeOffset> _leases = new(); // conditionId -> leasedAt
private readonly ConcurrentDictionary<string, JobMapping> _inFlight = new(); // jobId -> mapping
public async Task<ScrapeJobDto?> ClaimNextAsync(SkinTrackerDbContext db, int maxPages, CancellationToken ct)
{
await _gate.WaitAsync(ct);
try
{
// Reclaim expired leases first.
var cutoff = DateTimeOffset.UtcNow - LeaseTtl;
foreach (var (cid, at) in _leases)
{
if (at < cutoff)
{
_leases.TryRemove(cid, out _);
}
}
// Stalest bands first (never-swept null sorts before any timestamp).
var candidates = await db.SkinConditions
.OrderBy(c => c.ListingsSweptAt.HasValue)
.ThenBy(c => c.ListingsSweptAt)
.Select(c => new Candidate(
c.Id, c.SkinId, c.Skin.Weapon.Name, c.Skin.Name, c.Condition))
.Take(CandidateBatch)
.ToListAsync(ct);
var pick = candidates.FirstOrDefault(c => !_leases.ContainsKey(c.ConditionId));
if (pick is null)
{
return null; // everything in the stalest batch is already in flight
}
_leases[pick.ConditionId] = DateTimeOffset.UtcNow;
var jobId = Guid.NewGuid().ToString("N");
_inFlight[jobId] = new JobMapping(pick.SkinId, pick.ConditionId);
var code = Wear.ToCode(pick.Condition) ?? pick.Condition;
var search = $"{pick.Weapon} {pick.SkinName} {code}".Trim();
return new ScrapeJobDto(jobId, pick.SkinId, pick.ConditionId, search, maxPages);
}
finally
{
_gate.Release();
}
}
/// <summary>Resolve a posted job to its skin+condition and release its lease.</summary>
public JobMapping? Complete(string jobId)
{
if (_inFlight.TryRemove(jobId, out var mapping))
{
_leases.TryRemove(mapping.ConditionId, out _);
return mapping;
}
return null;
}
public int InFlight => _inFlight.Count;
public sealed record JobMapping(int SkinId, int ConditionId);
private sealed record Candidate(int ConditionId, int SkinId, string Weapon, string SkinName, string Condition);
}

View File

@@ -0,0 +1,87 @@
using BlueLaminate.C2;
using BlueLaminate.Core.CsMoney;
using BlueLaminate.Core.DependencyInjection;
using BlueLaminate.EFCore.Data;
using Microsoft.EntityFrameworkCore;
// The C2: hands cs.money scrape jobs to Python workers and ingests their results.
// Reuses the whole BlueLaminate stack (DB, ingest service) via the one composition root.
// Content root = the binary directory so appsettings.json is found regardless of the
// working directory the process is launched from (matches the CLI's approach).
var builder = WebApplication.CreateBuilder(new WebApplicationOptions
{
Args = args,
ContentRootPath = AppContext.BaseDirectory,
});
builder.Services.AddBlueLaminateCore(builder.Configuration);
builder.Services.AddSingleton<JobQueue>();
var app = builder.Build();
// Apply pending EF migrations at startup (incl. the market_listings view) so a fresh
// container is ready with one command. Disable with AutoMigrate=false if you'd rather
// run `dotnet ef database update` yourself.
if (app.Configuration.GetValue("AutoMigrate", true))
{
using var scope = app.Services.CreateScope();
var db = scope.ServiceProvider.GetRequiredService<SkinTrackerDbContext>();
db.Database.Migrate();
}
// Shared-secret gate. Workers send it as X-Worker-Token; if no token is configured
// the gate is open (local dev). Set WorkerToken (config) / WORKER_TOKEN (env) in prod.
var workerToken = builder.Configuration["WorkerToken"];
var maxPagesPerJob = builder.Configuration.GetValue("MaxPagesPerJob", 60);
app.MapGet("/health", () => Results.Ok(new { status = "ok" }));
// Operator read endpoints: "where is this listed?" across markets. Open (read-only).
app.MapGet("/market/skin/{skinId:int}", async (
int skinId, MarketPresenceService presence, CancellationToken ct) =>
Results.Ok(await presence.ForSkinAsync(skinId, ct)));
app.MapGet("/market/instance/{instanceId:int}", async (
int instanceId, MarketPresenceService presence, CancellationToken ct) =>
Results.Ok(await presence.ForInstanceAsync(instanceId, ct)));
var jobs = app.MapGroup("/jobs");
jobs.AddEndpointFilter(async (ctx, next) =>
{
if (!string.IsNullOrEmpty(workerToken)
&& ctx.HttpContext.Request.Headers["X-Worker-Token"].ToString() != workerToken)
{
return Results.Unauthorized();
}
return await next(ctx);
});
// Claim the next stalest skin+wear to scrape. 204 when nothing is currently available
// (everything in the stalest batch is already leased to other workers).
jobs.MapGet("/next", async (JobQueue queue, SkinTrackerDbContext db, CancellationToken ct) =>
{
var job = await queue.ClaimNextAsync(db, maxPagesPerJob, ct);
return job is null ? Results.NoContent() : Results.Ok(job);
});
// Post a claimed job's scraped listings. The C2 owns parsing/persistence so the
// worker stays dumb: it just forwards the raw cs.money items it gathered.
jobs.MapPost("/{jobId}/result", async (
string jobId, ScrapeResultDto result, JobQueue queue, CsMoneyIngestService ingest, CancellationToken ct) =>
{
var mapping = queue.Complete(jobId);
if (mapping is null)
{
return Results.NotFound(new { error = "unknown or expired jobId" });
}
// Only a fully-walked sweep ("completed") is authoritative. On a partial result
// (fetch-cap / challenged / float tie) we still upsert what we saw, but we must NOT
// mark unseen listings Removed or stamp the swept-checkpoint — the unseen ones may
// simply be unfetched, and the band must be re-queued and retried.
var complete = string.Equals(result.StoppedReason, "completed", StringComparison.OrdinalIgnoreCase);
var r = await ingest.IngestAsync(mapping.SkinId, mapping.ConditionId, result.Items ?? [], complete, ct);
return Results.Ok(r);
});
app.Run();

View File

@@ -0,0 +1,23 @@
{
"$schema": "https://json.schemastore.org/launchsettings.json",
"profiles": {
"http": {
"commandName": "Project",
"dotnetRunMessages": true,
"launchBrowser": true,
"applicationUrl": "http://localhost:5103",
"environmentVariables": {
"ASPNETCORE_ENVIRONMENT": "Development"
}
},
"https": {
"commandName": "Project",
"dotnetRunMessages": true,
"launchBrowser": true,
"applicationUrl": "https://localhost:7111;http://localhost:5103",
"environmentVariables": {
"ASPNETCORE_ENVIRONMENT": "Development"
}
}
}
}

View File

@@ -0,0 +1,8 @@
{
"Logging": {
"LogLevel": {
"Default": "Information",
"Microsoft.AspNetCore": "Warning"
}
}
}

View File

@@ -0,0 +1,16 @@
{
"Logging": {
"LogLevel": {
"Default": "Information",
"Microsoft.AspNetCore": "Warning",
"Microsoft.EntityFrameworkCore": "Warning"
}
},
"AllowedHosts": "*",
"Urls": "http://0.0.0.0:5080",
"ConnectionStrings": {
"SkinTracker": "Host=localhost;Port=5432;Database=skintracker;Username=postgres"
},
"WorkerToken": "dev-worker-token",
"MaxPagesPerJob": 60
}