add logic for aliasing packages discovered by the telescope that are already known to us by a different name

Author: tayloraswift

Sources/SwiftinitServer/Plugins/GitHubPlugin.RepoTelescope.swift

@@ -40,7 +40,7 @@ extension GitHubPlugin.RepoTelescope:GitHubCrawler
             with: session)
         else
         {
-            Log[.debug] = "Skipping telescope crawl: no windows left to crawl."
+            // Log[.debug] = "Skipping telescope crawl: no windows left to crawl."
             return
         }
 

Sources/UnidocDB/Packages/UnidocDatabase.Packages.swift

@@ -40,6 +40,19 @@ extension UnidocDatabase.Packages
         $0[Unidoc.PackageMetadata[.repo]] = .init { $0[.exists] = true }
     }
 
+    public static
+    let indexRepoGitHub:Mongo.CollectionIndex = .init("RepoGitHub", unique: true)
+    {
+        $0[ Unidoc.PackageMetadata[.repo] /
+            Unidoc.PackageRepo[.github] /
+            Unidoc.PackageRepo.GitHubOrigin[.id]] = (+)
+    }
+        where:
+    {
+        $0[ Unidoc.PackageMetadata[.repo] /
+            Unidoc.PackageRepo[.github]] = .init { $0[.exists] = true }
+    }
+
     public static
     let indexRealm:Mongo.CollectionIndex = .init("Realm", unique: false)
     {
@@ -64,6 +77,7 @@ extension UnidocDatabase.Packages:Mongo.CollectionModel
         [
             Self.indexLastCrawled,
             Self.indexRepoCreated,
+            Self.indexRepoGitHub,
             Self.indexRealm,
         ]
     }
@@ -80,6 +94,23 @@ extension UnidocDatabase.Packages:Mongo.RecodableModel
 }
 extension UnidocDatabase.Packages
 {
+    func findGitHub(repo id:Int32,
+        with session:Mongo.Session) async throws -> Unidoc.PackageMetadata?
+    {
+        try await session.run(
+            command: Mongo.Find<Mongo.Single<Unidoc.PackageMetadata>>.init(Self.name, limit: 1)
+            {
+                $0[.filter] = .init
+                {
+                    $0[ Unidoc.PackageMetadata[.repo] /
+                        Unidoc.PackageRepo[.github] /
+                        Unidoc.PackageRepo.GitHubOrigin[.id]] = id
+                }
+                $0[.hint] = Self.indexRepoGitHub.id
+            },
+            against: self.database)
+    }
+
     public
     func update(metadata:Unidoc.PackageMetadata,
         with session:Mongo.Session) async throws -> Bool?

Sources/UnidocDB/UnidocDatabase.swift

@@ -141,10 +141,20 @@ extension UnidocDatabase
 
     public
     func index(package:Symbol.Package,
-        repo:consuming Unidoc.PackageRepo? = nil,
+        repo:Unidoc.PackageRepo? = nil,
         mode:Unidoc.PackageIndexMode = .manual,
         with session:Mongo.Session) async throws -> (package:Unidoc.PackageMetadata, new:Bool)
     {
+        if  case .github(let origin)? = repo?.origin,
+            let existing:Unidoc.PackageMetadata = try await self.packages.findGitHub(
+                repo: origin.id,
+                with: session)
+        {
+            //  According to GitHub, this package is already known to us by a different name.
+            try await self.alias(existing: existing.symbol, package: package, with: session)
+            return (existing, false)
+        }
+
         //  Placement involves autoincrement, which is why this cannot be done in an update.
         let placement:Unidoc.Autoincrement<Unidoc.PackageMetadata> = try await session.query(
             database: self.id,