Neue Repos importieren, Abgleich ohne Subdomains, Selbstheilung nur für gehostete Projekte

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

server/services/giteaAdmin.js

@@ -15,6 +15,13 @@ import {
 } from './previewDeploy.js'
 import { giteaFetch } from './giteaClient.js'
 import { slugify } from './slug.js'
+import { withLocks } from './keyedLock.js'
+import { repoLockKey } from './repoMove.js'
+
+// Sicherheitsnetz wie bei listRecentPushes (siehe pushHistory.js, Controller-Ruling R8): sollte
+// Gitea "page" ignorieren und immer dieselbe volle Seite liefern, blaettert die Schleife nicht
+// endlos weiter (40 x 50 = 2000 Repos).
+const MAX_REPO_PAGES = 40
 
 export async function listAllAccessibleRepos() {
   const repos = []
@@ -27,40 +34,67 @@ export async function listAllAccessibleRepos() {
     if (data.length === 0) break
     repos.push(...data)
     if (data.length < limit) break
+    if (page >= MAX_REPO_PAGES) {
+      console.warn(
+        `[giteaAdmin] mehr als ${MAX_REPO_PAGES} Seiten Repos – weitere Repos werden nicht gelistet`
+      )
+      break
+    }
     page += 1
   }
 
   return repos
 }
 
-async function resolveSubdomainForSync(repo, existingDoc) {
-  if (existingDoc?.subdomain) return existingDoc.subdomain
+/**
+ * Findet Repos, die in Gitea liegen, aber noch kein Appwrite-Projektdokument haben, und legt fuer
+ * jedes nicht archivierte ein minimales Dokument an (status "synced", ohne Hosting/Subdomain) -
+ * ohne dass dafuer manuell "Abgleich" gedrueckt werden muesste. Wird vom Projekt-Dialog bei jedem
+ * Oeffnen aufgerufen, laeuft also oft: die aeussere `withLocks(['import'])`-Sperre serialisiert
+ * parallele Laeufe komplett (ein zweiter Aufruf sieht danach garantiert den Stand des ersten), die
+ * innere `withLocks([repoLockKey(...)])` je neuem Repo schuetzt zusaetzlich gegen andere
+ * gleichzeitige Schreiber desselben Projektdokuments (z. B. Umzug/Zuweisung).
+ */
+export async function importNewRepos() {
+  return withLocks(['import'], async () => {
+    const [repos, docs] = await Promise.all([
+      listAllAccessibleRepos(),
+      listDocuments(config.collections.websiteProjects, [Query.limit(5000)]),
+    ])
 
-  const candidate = slugify(repo.name) || slugify(repo.full_name?.split('/').pop())
-  if (!candidate) return ''
+    // Ohne Ruecksicht auf Gross-/Kleinschreibung vergleichen (Controller-Ruling).
+    const existingNames = new Set(
+      docs.map((d) => String(d.repoFullName || '').toLowerCase()).filter(Boolean)
+    )
 
-  const conflict = await listDocuments(config.collections.websiteProjects, [
-    Query.equal('subdomain', candidate),
-    Query.limit(1),
-  ])
+    let created = 0
+    for (const repo of repos) {
+      const fullName = repo.full_name
+      if (!fullName || repo.archived) continue
+      if (existingNames.has(fullName.toLowerCase())) continue
 
-  if (!conflict[0] || conflict[0].repoFullName === repo.full_name) {
-    return candidate
-  }
+      await withLocks([repoLockKey(fullName)], async () => {
+        const recheck = await listDocuments(config.collections.websiteProjects, [
+          Query.equal('repoFullName', fullName),
+          Query.limit(1),
+        ])
+        if (recheck[0]) return
 
-  const ownerSlug = slugify(repo.full_name?.replace('/', '-'))
-  if (!ownerSlug) return ''
+        await upsertWebsiteProjectByRepo(fullName, {
+          projectName: repo.description || repo.name || fullName.split('/').pop(),
+          giteaRepoUrl: repo.html_url,
+          giteaRepoName: repo.name,
+          repoFullName: fullName,
+          archived: false,
+          status: 'synced',
+          giteaSyncedAt: new Date().toISOString(),
+        })
+        created += 1
+      })
+    }
 
-  const ownerConflict = await listDocuments(config.collections.websiteProjects, [
-    Query.equal('subdomain', ownerSlug),
-    Query.limit(1),
-  ])
-
-  if (!ownerConflict[0] || ownerConflict[0].repoFullName === repo.full_name) {
-    return ownerSlug
-  }
-
-  return ''
+    return { created }
+  })
 }
 
 export async function syncReposToAppwrite() {
@@ -77,10 +111,6 @@ export async function syncReposToAppwrite() {
       Query.limit(1),
     ])
 
-    const branch = repo.default_branch || 'main'
-    const subdomain = await resolveSubdomainForSync(repo, existing[0])
-    const previewUrl = subdomain ? buildPreviewUrl(subdomain) : ''
-
     const payload = {
       projectName: repo.description || repo.name || repoFullName.split('/').pop(),
       giteaRepoUrl: repo.html_url || `${config.gitea.baseUrl}/${repoFullName}`,
@@ -90,9 +120,6 @@ export async function syncReposToAppwrite() {
       archived: Boolean(repo.archived),
     }
 
-    if (subdomain) payload.subdomain = subdomain
-    if (previewUrl) payload.previewUrl = previewUrl
-
     if (!existing[0]) {
       payload.status = 'synced'
       created += 1
@@ -186,6 +213,20 @@ async function updatePreviewConfigFile(repoFullName, previewConfig, branch = 'ma
   })
 }
 
+/**
+ * true, wenn zu diesem Projekt wirklich ein Hosting gehoert (Preview/Website/Vorlage, oder ein
+ * bereits deployter bzw. anstehender Stand) - im Unterschied zu einem nur importierten/
+ * abgeglichenen Repo ohne eigenes Hosting (z. B. status "synced" ohne Website). Bestimmt u. a.,
+ * ob die Selbstheilung (ensureRepoPreviewConfig) fuer dieses Projekt ueberhaupt aktiv werden darf -
+ * sonst wuerde mit dem organisationsweiten Webhook jeder Push in ein bloss abgeglichenes Repo eine
+ * Preview anlegen.
+ */
+export function isHostedProject(doc) {
+  if (['preview', 'website', 'template'].includes(doc?.projectType)) return true
+  if (['deployed', 'pending'].includes(doc?.status)) return true
+  return false
+}
+
 /**
  * Selbstheilung fuer eine geloeschte/fehlende .webklar-preview.json:
  * Rekonstruiert die Hosting-Config aus dem in Appwrite gespeicherten
@@ -201,11 +242,11 @@ export async function ensureRepoPreviewConfig(repoFullName, branch = 'main') {
   ])
   const doc = docs[0]
   if (!doc || doc.archived) return null
+  if (!isHostedProject(doc)) return null
 
   const type = deriveProjectType(doc)
-  const hasWebsite = type === 'preview' || type === 'website' || type === 'template'
   const subdomain = doc.subdomain || ''
-  if (!hasWebsite || !subdomain) return null
+  if (!subdomain) return null
 
   const previewConfig = {
     enabled: true,

test/giteaAdmin.test.js

@@ -0,0 +1,110 @@
+import { test, before, after } from 'node:test'
+import assert from 'node:assert/strict'
+import { startFakeGitea } from './helpers/fakeGitea.js'
+import { startFakeAppwrite } from './helpers/fakeAppwrite.js'
+
+let gitea
+let aw
+let config
+let listAllAccessibleRepos
+let importNewRepos
+let syncReposToAppwrite
+let isHostedProject
+let ensureRepoPreviewConfig
+
+before(async () => {
+  gitea = await startFakeGitea({
+    repos: [
+      { full_name: 'knso/alt' },
+      { full_name: 'JUSN/neu', description: 'Döner Nation' },
+      { full_name: 'JUSN/archiv', archived: true },
+      { full_name: 'knso/privat', private: true },
+      { full_name: 'knso/synced-sub' },
+      { full_name: 'knso/lead' },
+    ],
+  })
+
+  aw = await startFakeAppwrite({
+    seed: {
+      websiteProjects: [
+        { repoFullName: 'knso/alt', projectName: 'alt', subdomain: 'alt' },
+        { repoFullName: 'knso/synced-sub', projectName: 'synced-sub', status: 'synced', subdomain: 'synced-sub' },
+        { repoFullName: 'knso/lead', projectName: 'lead', status: 'deployed', subdomain: 'lead' },
+      ],
+    },
+  })
+
+  Object.assign(process.env, {
+    GITEA_BASE_URL: gitea.baseUrl,
+    GITEA_API_TOKEN: 'test-token',
+    APPWRITE_ENDPOINT: aw.endpoint,
+    APPWRITE_API_KEY: 'test-key',
+    PREVIEW_DEPLOY_ROOT: '',
+    TRAEFIK_DYNAMIC_DIR: '',
+  })
+  // config.js liest die Umgebung beim Import
+  ;({ config } = await import('../server/config.js'))
+  ;({ listAllAccessibleRepos, importNewRepos, syncReposToAppwrite, isHostedProject, ensureRepoPreviewConfig } =
+    await import('../server/services/giteaAdmin.js'))
+})
+
+after(async () => {
+  await gitea.close()
+  await aw.close()
+})
+
+test('importNewRepos legt nur fehlende an', async () => {
+  assert.deepEqual(await importNewRepos(), { created: 2 })
+  const neu = aw.all('websiteProjects').find((d) => d.repoFullName === 'JUSN/neu')
+  assert.deepEqual([neu.projectName, neu.status, neu.subdomain || '', neu.previewUrl || ''], ['Döner Nation', 'synced', '', ''])
+  assert.ok(aw.all('websiteProjects').some((d) => d.repoFullName === 'knso/privat'))
+  assert.ok(!aw.all('websiteProjects').some((d) => d.repoFullName === 'JUSN/archiv'))
+  assert.equal(aw.calls.filter((c) => c.method === 'PATCH').length, 0)
+})
+test('importNewRepos läuft nie doppelt', async () => {
+  gitea.addRepo({ full_name: 'ANDJ/zweimal' })
+  await Promise.all([importNewRepos(), importNewRepos()])
+  assert.equal(aw.all('websiteProjects').filter((d) => d.repoFullName === 'ANDJ/zweimal').length, 1)
+})
+test('Abgleich vergibt keine Subdomain mehr', async () => {
+  gitea.addRepo({ full_name: 'ANDJ/sync-neu' })
+  await syncReposToAppwrite()
+  const d = aw.all('websiteProjects').find((x) => x.repoFullName === 'ANDJ/sync-neu')
+  assert.equal(d.subdomain || '', ''); assert.equal(d.previewUrl || '', '')
+  assert.equal(aw.all('websiteProjects').find((x) => x.repoFullName === 'knso/alt').subdomain, 'alt')
+})
+test('isHostedProject', () => {
+  for (const [doc, want] of [[{ projectType: 'preview' }, true], [{ projectType: 'website' }, true], [{ projectType: 'template' }, true],
+    [{ status: 'deployed' }, true], [{ status: 'pending' }, true], [{ status: 'synced', subdomain: 'x' }, false], [{}, false]]) {
+    assert.equal(isHostedProject(doc), want, JSON.stringify(doc))
+  }
+})
+test('Selbstheilung nur für gehostete Projekte', async () => {
+  assert.equal(await ensureRepoPreviewConfig('knso/synced-sub', 'main'), null)
+  assert.equal(gitea.calls.filter((c) => c.path.includes('synced-sub/contents')).length, 0)
+  const cfg = await ensureRepoPreviewConfig('knso/lead', 'main')
+  assert.equal(cfg.subdomain, 'lead')
+  assert.ok(gitea.calls.some((c) => c.method === 'POST' && c.path.includes('knso/lead/contents/.webklar-preview.json')))
+})
+test('Repo-Liste hat eine Obergrenze', async () => {
+  // R8: eigener Fake-Server + kurzzeitig umgebogene config.gitea.baseUrl (statt in `gitea` oben
+  // Tausende Repos zu seeden) - sonst wuerden importNewRepos/syncReposToAppwrite in den Tests
+  // darueber auf einmal Tausende zusaetzliche, unbenannte Repos mitverarbeiten.
+  const capRepos = []
+  for (let i = 0; i < 2005; i++) capRepos.push({ full_name: `cap/repo${i}` })
+  const capGitea = await startFakeGitea({ repos: capRepos })
+  const originalBaseUrl = config.gitea.baseUrl
+  const originalWarn = console.warn
+  console.warn = () => {}
+  config.gitea.baseUrl = capGitea.baseUrl
+  let repos
+  try {
+    repos = await listAllAccessibleRepos()
+  } finally {
+    config.gitea.baseUrl = originalBaseUrl
+    console.warn = originalWarn
+    await capGitea.close()
+  }
+  assert.equal(repos.length, 2000)
+  assert.equal(capGitea.calls.filter((c) => c.path === '/api/v1/repos/search').length, 40)
+})