feat: add sonatype script (CM-1309)

services/apps/packages_worker/src/maven/scripts/importSonatypePopularityFromCsv.ts

@@ -0,0 +1,222 @@
+/**
+ * One-shot import of the Sonatype popularity signal into the osspckgs `packages`
+ * table.
+ *
+ * Sonatype could not deliver raw Maven download counts, so instead they provide a
+ * popularity score derived from their SCA telemetry: normalized 0-100 per
+ * ecosystem, tiered P0-P3. First delivery is Maven P0 only (top ~500 components).
+ *
+ * Safe to run multiple times — every write goes through upsertSonatypePopularity,
+ * keyed on the composite identity (ecosystem, namespace, name). Rows not yet in
+ * `packages` are inserted with ingestion_source='sonatype' (deps.dev / Maven
+ * enrichment backfills the rest later); existing rows get only their sonatype_*
+ * fields refreshed and keep their original ingestion_source.
+ *
+ * Expected CSV header (exact): format,namespace,name,component_key,popularity_score,rank,tier
+ *   - namespace = Maven groupId, name = Maven artifactId
+ *   - component_key = groupId:artifactId (used only as a sanity check)
+ *
+ * Flags:
+ *   --dry-run                 Parse + validate, print what would happen, no DB writes
+ *   --snapshot-date=YYYY-MM-DD  Snapshot timestamp; derived from the filename if omitted
+ *
+ * Usage:
+ *   tsx src/maven/scripts/importSonatypePopularityFromCsv.ts <input.csv> [--snapshot-date=YYYY-MM-DD] [--dry-run]
+ */
+import * as fs from 'fs'
+import * as path from 'path'
+
+import { upsertSonatypePopularity } from '@crowd/data-access-layer/src/osspckgs/packages'
+import { pgpQx } from '@crowd/data-access-layer/src/queryExecutor'
+
+import { getPackagesDbConnection } from '../../db'
+
+import { parseCsv } from './csv'
+
+const EXPECTED_HEADER = [
+  'format',
+  'namespace',
+  'name',
+  'component_key',
+  'popularity_score',
+  'rank',
+  'tier',
+]
+const VALID_TIERS = new Set(['P0', 'P1', 'P2', 'P3'])
+
+// packages.purl is stored version-stripped for Maven (one row per groupId:artifactId).
+// Mirror that exact format for insert-if-missing rows so a later deps.dev / Maven
+// enrichment upsert (keyed on purl) resolves to the same row instead of duplicating it.
+function buildMavenPurl(groupId: string, artifactId: string): string {
+  return `pkg:maven/${groupId}/${artifactId}`
+}
+
+function resolveSnapshotDate(args: string[], inputPath: string): Date {
+  const flag = args.find((a) => a.startsWith('--snapshot-date='))
+  const explicit = flag?.split('=')[1]
+  const raw = explicit ?? path.basename(inputPath).match(/(\d{4}-\d{2}-\d{2})/)?.[1]
+
+  if (!raw) {
+    throw new Error(
+      'Could not determine snapshot date: pass --snapshot-date=YYYY-MM-DD or include a YYYY-MM-DD in the filename',
+    )
+  }
+  const date = new Date(`${raw}T00:00:00Z`)
+  if (isNaN(date.getTime())) {
+    throw new Error(`Invalid snapshot date: "${raw}" (expected YYYY-MM-DD)`)
+  }
+  return date
+}
+
+type SonatypeRow = {
+  namespace: string
+  name: string
+  score: number
+  rank: number
+  tier: string
+}
+
+// Parse + validate one CSV record. Returns null for rows to skip (non-maven or
+// malformed), with the reason logged so nothing is dropped silently.
+function parseRow(r: Record<string, string>, lineNo: number): SonatypeRow | null {
+  if (r.format !== 'maven') return null // defensive: first delivery is maven-only
+
+  const namespace = r.namespace
+  const name = r.name
+  const score = Number(r.popularity_score)
+  const rank = Number(r.rank)
+  const tier = r.tier
+
+  if (!namespace || !name) {
+    console.warn(`  line ${lineNo}: missing namespace/name — skipped`)
+    return null
+  }
+  if (!Number.isFinite(score) || score < 0 || score > 100) {
+    console.warn(`  line ${lineNo}: invalid popularity_score "${r.popularity_score}" — skipped`)
+    return null
+  }
+  if (!Number.isInteger(rank) || rank <= 0) {
+    console.warn(`  line ${lineNo}: invalid rank "${r.rank}" — skipped`)
+    return null
+  }
+  if (!VALID_TIERS.has(tier)) {
+    console.warn(`  line ${lineNo}: invalid tier "${tier}" — skipped`)
+    return null
+  }
+  // component_key is groupId:artifactId — cross-check against namespace/name.
+  if (r.component_key && r.component_key !== `${namespace}:${name}`) {
+    console.warn(
+      `  line ${lineNo}: component_key "${r.component_key}" != "${namespace}:${name}" — skipped`,
+    )
+    return null
+  }
+
+  return { namespace, name, score, rank, tier }
+}
+
+async function main() {
+  const args = process.argv.slice(2)
+  const dryRun = args.includes('--dry-run')
+  const inputPath = args.find((a) => !a.startsWith('--'))
+
+  if (!inputPath) {
+    console.error(
+      'Usage: tsx src/maven/scripts/importSonatypePopularityFromCsv.ts <input.csv> [--snapshot-date=YYYY-MM-DD] [--dry-run]',
+    )
+    process.exit(1)
+  }
+
+  const snapshotAt = resolveSnapshotDate(args, inputPath)
+
+  console.log(`Input:        ${inputPath}`)
+  console.log(`Snapshot:     ${snapshotAt.toISOString()}`)
+  console.log(`Mode:         ${dryRun ? 'DRY RUN (no DB writes)' : 'LIVE'}`)
+  console.log()
+
+  const content = fs.readFileSync(inputPath, 'utf-8')
+  const records = parseCsv(content)
+
+  if (records.length === 0) {
+    throw new Error('CSV is empty (no data rows)')
+  }
+
+  // Validate the header exactly — order-independent, no missing/extra columns.
+  const header = Object.keys(records[0])
+  const missing = EXPECTED_HEADER.filter((h) => !header.includes(h))
+  const extra = header.filter((h) => !EXPECTED_HEADER.includes(h))
+  if (missing.length > 0 || extra.length > 0) {
+    throw new Error(
+      `Unexpected CSV header.\n  expected: ${EXPECTED_HEADER.join(', ')}\n  got:      ${header.join(', ')}` +
+        (missing.length ? `\n  missing:  ${missing.join(', ')}` : '') +
+        (extra.length ? `\n  extra:    ${extra.join(', ')}` : ''),
+    )
+  }
+
+  let skipped = 0
+  const rows: SonatypeRow[] = []
+  records.forEach((r, i) => {
+    const parsed = parseRow(r, i + 2) // +2: 1-based, and header is line 1
+    if (parsed) rows.push(parsed)
+    else skipped++
+  })
+
+  console.log(`Parsed ${records.length} data rows — ${rows.length} valid maven, ${skipped} skipped`)
+
+  if (dryRun) {
+    rows.slice(0, 10).forEach((r) => {
+      console.log(`  ${r.namespace}:${r.name}  score=${r.score} rank=${r.rank} tier=${r.tier}`)
+    })
+    if (rows.length > 10) console.log(`  ... (${rows.length - 10} more)`)
+    console.log('\nDry run complete — no changes made.')
+    return
+  }
+
+  const conn = await getPackagesDbConnection()
+
+  const insertedRows: SonatypeRow[] = []
+  let updated = 0
+
+  // Single transaction: the dataset is small (hundreds of rows) and a one-shot
+  // import should be all-or-nothing.
+  await conn.tx(async (t) => {
+    const tqx = pgpQx(t)
+    for (const r of rows) {
+      const { inserted: wasInserted } = await upsertSonatypePopularity(tqx, {
+        purl: buildMavenPurl(r.namespace, r.name),
+        ecosystem: 'maven',
+        namespace: r.namespace,
+        name: r.name,
+        sonatypePopularityScore: r.score,
+        sonatypeRank: r.rank,
+        sonatypeTier: r.tier,
+        sonatypeSnapshotAt: snapshotAt,
+      })
+      if (wasInserted) insertedRows.push(r)
+      else updated++
+    }
+  })
+
+  await (conn as unknown as { $pool: { end: () => Promise<void> } }).$pool.end()
+
+  console.log(`\n✓ Done.`)
+  console.log(`  Inserted (new packages): ${insertedRows.length}`)
+  console.log(`  Updated (existing):      ${updated}`)
+  console.log(`  Skipped (non-maven / invalid): ${skipped}`)
+
+  // List the newly-inserted packages — these are the Sonatype rows that were not
+  // in `packages` yet (deps.dev hasn't backfilled them). Small set, worth an audit
+  // trail so the overlap can be reported without re-querying the DB.
+  if (insertedRows.length > 0) {
+    console.log(`\n  New packages inserted (ingestion_source='sonatype'):`)
+    insertedRows
+      .sort((a, b) => a.rank - b.rank)
+      .forEach((r) =>
+        console.log(`    rank ${r.rank}\t${r.tier}\tscore ${r.score}\t${r.namespace}:${r.name}`),
+      )
+  }
+}
+
+main().catch((err) => {
+  console.error(err)
+  process.exit(1)
+})

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

@@ -1,6 +1,6 @@
 import { QueryExecutor } from '../queryExecutor'
 
-import { IDbPackageUniverse, IDbPackageUpsert } from './types'
+import { IDbPackageUniverse, IDbPackageUpsert, IDbSonatypePopularityUpsert } from './types'
 
 export async function findPackageIdsByPurl(
   qx: QueryExecutor,
@@ -241,3 +241,63 @@ export async function upsertPackage(
   )
   return { id: row.id as number, changedFields: row.changed_fields as string[] }
 }
+
+// ─── sonatype popularity upsert ────────────────────────────────────────────────
+
+/**
+ * Upserts the Sonatype popularity signal for a Maven component, keyed by the
+ * composite identity (ecosystem, namespace, name) — not purl — so it is immune
+ * to purl format and matches the granularity of the unique index.
+ *
+ * Insert-if-missing: Sonatype's list is "what industry actually uses", so some
+ * rows may not exist in `packages` yet. Those are inserted with
+ * ingestion_source='sonatype' and only the identity + sonatype_* columns; the
+ * rest is backfilled later by deps.dev / Maven enrichment.
+ *
+ * On conflict only the sonatype_* fields are updated — ingestion_source is
+ * deliberately preserved so we never clobber how an existing row was ingested.
+ *
+ * Returns whether the row was inserted (true) or updated (false).
+ */
+export async function upsertSonatypePopularity(
+  qx: QueryExecutor,
+  item: IDbSonatypePopularityUpsert,
+): Promise<{ id: number; inserted: boolean }> {
+  const row = await qx.selectOne(
+    `
+    WITH existing AS (
+      -- Evaluated against the pre-INSERT snapshot, so it reflects whether the row
+      -- already existed. More robust than the xmax=0 trick for insert-vs-update.
+      SELECT id FROM packages
+       WHERE ecosystem = $(ecosystem)
+         AND COALESCE(namespace, '') = COALESCE($(namespace), '')
+         AND name = $(name)
+    ),
+    ins AS (
+      INSERT INTO packages (
+        purl, ecosystem, namespace, name,
+        sonatype_popularity_score, sonatype_rank, sonatype_tier,
+        sonatype_snapshot_at, sonatype_updated_at,
+        ingestion_source
+      ) VALUES (
+        $(purl), $(ecosystem), $(namespace), $(name),
+        $(sonatypePopularityScore), $(sonatypeRank), $(sonatypeTier),
+        $(sonatypeSnapshotAt), NOW(),
+        'sonatype'
+      )
+      ON CONFLICT (ecosystem, COALESCE(namespace, ''), name) DO UPDATE SET
+        sonatype_popularity_score = EXCLUDED.sonatype_popularity_score,
+        sonatype_rank             = EXCLUDED.sonatype_rank,
+        sonatype_tier             = EXCLUDED.sonatype_tier,
+        sonatype_snapshot_at      = EXCLUDED.sonatype_snapshot_at,
+        sonatype_updated_at       = NOW()
+        -- ingestion_source intentionally left untouched on update
+      RETURNING id
+    )
+    SELECT ins.id, NOT EXISTS (SELECT 1 FROM existing) AS inserted
+      FROM ins
+    `,
+    item,
+  )
+  return { id: row.id as number, inserted: row.inserted as boolean }
+}

services/libs/data-access-layer/src/osspckgs/types.ts

@@ -35,6 +35,24 @@ export type IDbPackageUpsert = {
   repositoryUrl?: string | null
 }
 
+// ─── sonatype popularity ──────────────────────────────────────────────────────
+
+/**
+ * Sonatype popularity signal for a single Maven component (one row per
+ * groupId:artifactId). Only the sonatype_* fields plus the identity columns are
+ * carried — deps.dev / Maven enrichment backfills everything else later.
+ */
+export type IDbSonatypePopularityUpsert = {
+  purl: string
+  ecosystem: string
+  namespace: string // Maven groupId
+  name: string // Maven artifactId
+  sonatypePopularityScore: number
+  sonatypeRank: number
+  sonatypeTier: string
+  sonatypeSnapshotAt: Date
+}
+
 // ─── maintainers ──────────────────────────────────────────────────────────────
 
 export type IDbMaintainerUpsert = {