feat: resolve rubygems owner handles via github contributor verification [CM-1341] (#4363)

Signed-off-by: Mouad BANI <mouad-mb@outlook.com>

Author: mbani01

services/apps/packages_worker/src/security-contacts/extractors/registry/index.ts

@@ -29,6 +29,7 @@ const FETCHERS: Record<string, EcosystemFetcher> = {
 export const extractManifest: Extractor = async (target, deps) => {
   const contacts: RawContact[] = []
   const policies: Partial<RepoPolicies> = {}
+  const candidatesByHandle = new Map<string, RawContact>()
   const seenPurls = new Set<string>()
 
   for (const pkg of target.packages) {
@@ -44,6 +45,15 @@ export const extractManifest: Extractor = async (target, deps) => {
     try {
       const result = await fetcher(parsed, deps.fetchTimeoutMs, deps.userAgent)
       contacts.push(...result.contacts)
+      for (const candidate of result.handleCandidates ?? []) {
+        const key = candidate.value.toLowerCase()
+        const existing = candidatesByHandle.get(key)
+        if (existing) {
+          existing.provenance.push(...candidate.provenance)
+        } else {
+          candidatesByHandle.set(key, { ...candidate, provenance: [...candidate.provenance] })
+        }
+      }
       for (const [key, value] of Object.entries(result.policies)) {
         if (!(policies as Record<string, unknown>)[key] && value != null) {
           ;(policies as Record<string, unknown>)[key] = value
@@ -55,5 +65,5 @@ export const extractManifest: Extractor = async (target, deps) => {
     }
   }
 
-  return { contacts, policies }
+  return { contacts, policies, handleCandidates: [...candidatesByHandle.values()] }
 }

services/apps/packages_worker/src/security-contacts/extractors/registry/rubygems.ts

@@ -67,6 +67,35 @@ export function mapRubygemsOwners(
   return contacts
 }
 
+// RubyGems accounts are independent of GitHub, so a handle matching a GitHub login is only a
+// guess — emitted as candidates that must pass repo-contributor corroboration before use.
+export function mapRubygemsOwnerHandles(
+  owners: unknown,
+  sourceUrl: string,
+  fetchedAt: string,
+): RawContact[] {
+  if (!Array.isArray(owners)) return []
+
+  const candidates: RawContact[] = []
+  const seen = new Set<string>()
+  for (const owner of owners) {
+    const o = (owner ?? {}) as Record<string, unknown>
+    if (typeof o.email === 'string' && isEmail(o.email)) continue
+    if (typeof o.handle !== 'string' || o.handle.length === 0) continue
+    const key = o.handle.toLowerCase()
+    if (seen.has(key)) continue
+    seen.add(key)
+    candidates.push({
+      channel: 'github-handle',
+      value: o.handle,
+      role: 'maintainer',
+      tier: 'B',
+      provenance: [{ source: SOURCE, sourceTier: 'B', path: sourceUrl, fetchedAt }],
+    })
+  }
+  return candidates
+}
+
 export async function fetchRubygems(
   parsed: ParsedPurl,
   timeoutMs: number,
@@ -91,5 +120,9 @@ export async function fetchRubygems(
     ...mapRubygemsOwners(ownersResult.json, ownersUrl, fetchedAt),
   ]
 
-  return { contacts, policies: {} }
+  return {
+    contacts,
+    policies: {},
+    handleCandidates: mapRubygemsOwnerHandles(ownersResult.json, ownersUrl, fetchedAt),
+  }
 }

services/apps/packages_worker/src/security-contacts/processBatch.ts

@@ -26,6 +26,7 @@ import {
   RepoPolicies,
   RepoTarget,
 } from './types'
+import { verifyHandleCandidates } from './verifyHandleCandidates'
 import { writeContactsBatch } from './writeContacts'
 
 const log = getServiceChildLogger('security-contacts')
@@ -150,9 +151,11 @@ export async function processRepo(
 
   let contacts: RawContact[] = []
   const policies: Partial<RepoPolicies> = {}
+  const handleCandidates: RawContact[] = []
   for (const r of results) {
     if (r.status !== 'fulfilled') continue
     contacts.push(...r.value.contacts)
+    handleCandidates.push(...(r.value.handleCandidates ?? []))
     for (const [key, value] of Object.entries(r.value.policies)) {
       if (!(policies as Record<string, unknown>)[key] && value != null) {
         ;(policies as Record<string, unknown>)[key] = value
@@ -165,6 +168,8 @@ export async function processRepo(
     contacts = contacts.filter((c) => c.channel !== 'github-pvr')
   }
 
+  contacts.push(...(await verifyHandleCandidates(target, deps, handleCandidates)))
+
   contacts.push(...deriveGithubHandlesFromNoreplyEmails(contacts))
 
   const handleContacts = contacts.filter((c) => c.channel === 'github-handle')

services/apps/packages_worker/src/security-contacts/types.ts

@@ -46,6 +46,9 @@ export interface RepoPolicies {
 export interface ExtractorResult {
   contacts: RawContact[]
   policies: Partial<RepoPolicies>
+  /** Registry usernames that only *might* be GitHub logins — never written without
+   *  corroboration (see verifyHandleCandidates.ts). */
+  handleCandidates?: RawContact[]
 }
 
 export interface RepoPackage {

services/apps/packages_worker/src/security-contacts/verifyHandleCandidates.ts

@@ -0,0 +1,76 @@
+import { getServiceChildLogger } from '@crowd/logging'
+
+import { parseGithubUrl } from '../enricher/fetchLightRepo'
+
+import { ExtractorDeps, RawContact, RepoTarget } from './types'
+
+const log = getServiceChildLogger('security-contacts')
+
+interface ContributorEntry {
+  login?: unknown
+}
+
+/**
+ * Corroborates registry usernames that are only *guessed* to be GitHub logins (e.g. RubyGems
+ * owners). A candidate is confirmed only when the same login owns the repo or appears in its
+ * top-100 contributors — existence of a github.com/<handle> account alone is not enough, since
+ * an unrelated person or bot can hold that name.
+ */
+export async function verifyHandleCandidates(
+  target: RepoTarget,
+  deps: Pick<ExtractorDeps, 'githubGet'>,
+  candidates: RawContact[],
+): Promise<RawContact[]> {
+  if (candidates.length === 0) return []
+
+  let owner: string
+  let name: string
+  try {
+    ;({ owner, name } = parseGithubUrl(target.url))
+  } catch {
+    return []
+  }
+
+  const path = `/repos/${owner}/${name}/contributors?per_page=100`
+  const contributors = new Set<string>()
+  try {
+    const { text } = await deps.githubGet(path)
+    if (text) {
+      const entries = JSON.parse(text) as ContributorEntry[]
+      if (Array.isArray(entries)) {
+        for (const e of entries) {
+          if (typeof e.login === 'string') contributors.add(e.login.toLowerCase())
+        }
+      }
+    }
+  } catch (err) {
+    log.warn(
+      { repoId: target.repoId, errMsg: (err as Error).message },
+      'Contributor lookup failed — dropping handle candidates',
+    )
+    return []
+  }
+
+  const fetchedAt = new Date().toISOString()
+  const confirmed: RawContact[] = []
+  for (const c of candidates) {
+    const handle = c.value.toLowerCase()
+    const viaContributors = contributors.has(handle)
+    if (!viaContributors && handle !== owner.toLowerCase()) continue
+    confirmed.push({
+      ...c,
+      provenance: [
+        ...c.provenance,
+        viaContributors
+          ? {
+              source: 'github-contributors',
+              sourceTier: c.tier,
+              path: `https://api.github.com${path}`,
+              fetchedAt,
+            }
+          : { source: 'github-repo-owner', sourceTier: c.tier, path: target.url, fetchedAt },
+      ],
+    })
+  }
+  return confirmed
+}