fix: stop metadata lane overwriting 30d download mirror

Signed-off-by: anilb <epipav@gmail.com>

Author: epipav

services/apps/packages_worker/src/packagist/__tests__/dueSelection.test.ts

@@ -207,7 +207,6 @@ describe('updatePackagistPackageStats', () => {
         declaredRepositoryUrl: null,
         repositoryUrl: null,
         status: 'active',
-        downloadsLast30d: 1,
         totalDownloads: 2,
         dependentCount: 3,
       }),
@@ -229,7 +228,6 @@ describe('updatePackagistPackageStats', () => {
       declaredRepositoryUrl: 'https://github.com/a/x',
       repositoryUrl: 'https://github.com/a/x',
       status: 'active',
-      downloadsLast30d: 1,
       totalDownloads: 2,
       dependentCount: 3,
     })

services/apps/packages_worker/src/packagist/__tests__/normalize.test.ts

@@ -33,7 +33,7 @@ describe('normalizePackagistStats', () => {
     downloads: { total: 500_000_000, monthly: 10_000_000, daily: 350_000 },
   }
 
-  it('extracts description, repository, dependents and the three download counters', () => {
+  it('extracts description, repository, dependents and total downloads', () => {
     const stats = normalizePackagistStats(full)
     expect(stats).toEqual({
       name: 'monolog/monolog',
@@ -42,7 +42,8 @@ describe('normalizePackagistStats', () => {
       status: 'active',
       dependents: 5423,
       downloadsTotal: 500_000_000,
-      downloadsMonthly: 10_000_000,
+      // downloads.monthly is deliberately not extracted — packages.downloads_last_30d
+      // belongs exclusively to the dedicated downloads-30d lane
       // nameless maintainer entries are dropped
       maintainers: [{ username: 'seldaek', displayName: null, email: null, role: 'maintainer' }],
     })
@@ -76,7 +77,6 @@ describe('normalizePackagistStats', () => {
       status: 'active',
       dependents: null,
       downloadsTotal: null,
-      downloadsMonthly: null,
       maintainers: [],
     })
   })

services/apps/packages_worker/src/packagist/__tests__/persistPackageInfo.test.ts

@@ -41,7 +41,6 @@ const stats: NormalizedPackagistStats = {
   status: 'active',
   dependents: 42,
   downloadsTotal: 1000,
-  downloadsMonthly: 300,
   maintainers: [{ username: 'seldaek', displayName: null, email: null, role: 'maintainer' }],
 }
 
@@ -70,11 +69,13 @@ describe('persistPackagistPackageInfo', () => {
         purl: PURL,
         description: 'logs',
         status: 'active',
-        downloadsLast30d: 300,
         totalDownloads: 1000,
         dependentCount: 42,
       }),
     )
+    // packages.downloads_last_30d belongs exclusively to the dedicated downloads-30d
+    // lane's boundary-anchored snapshot — the metadata lane must never touch it
+    expect(mockUpdate.mock.calls[0][1]).not.toHaveProperty('downloadsLast30d')
     // canonicalized (lowercased) url + coarse host, linked with the manifest-declared convention
     expect(mockRepoGet).toHaveBeenCalledWith(qx, 'https://github.com/seldaek/monolog', 'github')
     expect(mockRepoLink).toHaveBeenCalledWith(qx, '7', '55', 'declared', 0.8)

services/apps/packages_worker/src/packagist/normalize.ts

@@ -28,9 +28,10 @@ export function normalizePackagistStats(pkg: PackagistPackageInfo): NormalizedPa
     status = 'deprecated'
   }
 
-  // Extract downloads, defaulting to null if not a number
+  // Extract downloads, defaulting to null if not a number. downloads.monthly is
+  // deliberately not extracted here — packages.downloads_last_30d belongs exclusively
+  // to the dedicated downloads-30d lane's boundary-anchored snapshot.
   const downloadsTotal = typeof pkg.downloads?.total === 'number' ? pkg.downloads.total : null
-  const downloadsMonthly = typeof pkg.downloads?.monthly === 'number' ? pkg.downloads.monthly : null
 
   const dependents = typeof pkg.dependents === 'number' ? pkg.dependents : null
 
@@ -55,7 +56,6 @@ export function normalizePackagistStats(pkg: PackagistPackageInfo): NormalizedPa
     status,
     dependents,
     downloadsTotal,
-    downloadsMonthly,
     maintainers,
   }
 }

services/apps/packages_worker/src/packagist/types.ts

@@ -66,7 +66,6 @@ export interface NormalizedPackagistStats {
   status: 'active' | 'deprecated'
   dependents: number | null
   downloadsTotal: number | null
-  downloadsMonthly: number | null
   maintainers: PackageMaintainerInput[]
 }
 

services/apps/packages_worker/src/packagist/upsertPackageInfo.ts

@@ -50,7 +50,6 @@ export async function persistPackagistPackageInfo(
       declaredRepositoryUrl: stats.repositoryUrl,
       repositoryUrl: trustedRepo?.url ?? null,
       status: stats.status,
-      downloadsLast30d: stats.downloadsMonthly,
       totalDownloads: stats.downloadsTotal,
       dependentCount: stats.dependents,
     })

services/libs/data-access-layer/src/packages/packages.ts

@@ -259,11 +259,14 @@ export interface PackagistStatsUpdateInput {
   declaredRepositoryUrl: string | null
   repositoryUrl: string | null
   status: string
-  downloadsLast30d: number | null
   totalDownloads: number | null
   dependentCount: number | null
 }
 
+// downloads_last_30d is NOT written here — it belongs exclusively to the dedicated
+// downloads-30d lane's boundary-anchored, insert-if-absent snapshot. The metadata
+// lane's live rolling `downloads.monthly` value is a different, unanchored reading;
+// writing it here would silently overwrite that snapshot every weekly refresh.
 export async function updatePackagistPackageStats(
   qx: QueryExecutor,
   input: PackagistStatsUpdateInput,
@@ -272,7 +275,7 @@ export async function updatePackagistPackageStats(
     await qx.selectOneOrNone(
       `WITH old AS (
          SELECT description, declared_repository_url, repository_url, status,
-                downloads_last_30d, total_downloads, dependent_count, ingestion_source
+                total_downloads, dependent_count, ingestion_source
            FROM packages WHERE purl = $(purl) AND ecosystem = 'packagist'
        ),
        ins AS (
@@ -281,21 +284,19 @@ export async function updatePackagistPackageStats(
            declared_repository_url   = $(declaredRepositoryUrl),
            repository_url            = $(repositoryUrl),
            status                    = $(status),
-           downloads_last_30d        = COALESCE($(downloadsLast30d), downloads_last_30d),
            total_downloads           = COALESCE($(totalDownloads), total_downloads),
            dependent_count           = COALESCE($(dependentCount), dependent_count),
            last_synced_at            = NOW()
          WHERE purl = $(purl) AND ecosystem = 'packagist'
          RETURNING id, is_critical, description, declared_repository_url, repository_url, status,
-                   downloads_last_30d, total_downloads, dependent_count, ingestion_source
+                   total_downloads, dependent_count, ingestion_source
        )
        SELECT ins.id::text AS id, ins.is_critical,
               array_remove(ARRAY[
                 CASE WHEN o.description               IS DISTINCT FROM ins.description               THEN 'packages.description' END,
                 CASE WHEN o.declared_repository_url   IS DISTINCT FROM ins.declared_repository_url   THEN 'packages.declared_repository_url' END,
                 CASE WHEN o.repository_url            IS DISTINCT FROM ins.repository_url            THEN 'packages.repository_url' END,
                 CASE WHEN o.status                    IS DISTINCT FROM ins.status                    THEN 'packages.status' END,
-                CASE WHEN o.downloads_last_30d        IS DISTINCT FROM ins.downloads_last_30d        THEN 'packages.downloads_last_30d' END,
                 CASE WHEN o.total_downloads           IS DISTINCT FROM ins.total_downloads           THEN 'packages.total_downloads' END,
                 CASE WHEN o.dependent_count           IS DISTINCT FROM ins.dependent_count           THEN 'packages.dependent_count' END,
                 CASE WHEN o.ingestion_source          IS DISTINCT FROM ins.ingestion_source          THEN 'packages.ingestion_source' END