fix: add eclipse normalize

Signed-off-by: Umberto Sgueglia <usgueglia@contractor.linuxfoundation.org>

Author: ulemons

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

@@ -236,6 +236,29 @@ describe('normalizeScmUrl', () => {
     expect(normalizeScmUrl('https://gitbox.apache.org/some/other/path')).toBeNull()
   })
 
+  it('recovers git.eclipse.org cgit URLs, keeping the /c/ prefix and dropping trailing tree paths', () => {
+    expect(normalizeScmUrl('http://git.eclipse.org/c/eclipselink/javax.persistence.git')).toBe(
+      'https://git.eclipse.org/c/eclipselink/javax.persistence',
+    )
+    expect(
+      normalizeScmUrl('https://git.eclipse.org/c/eclipsescada/org.eclipse.scada.utils.git'),
+    ).toBe('https://git.eclipse.org/c/eclipsescada/org.eclipse.scada.utils')
+    expect(
+      normalizeScmUrl('http://git.eclipse.org/c/jetty/org.eclipse.jetty.project.git/tree'),
+    ).toBe('https://git.eclipse.org/c/jetty/org.eclipse.jetty.project')
+    expect(
+      normalizeScmUrl(
+        'http://git.eclipse.org/c/jetty/org.eclipse.jetty.orbit.git/tree/jetty-orbit',
+      ),
+    ).toBe('https://git.eclipse.org/c/jetty/org.eclipse.jetty.orbit')
+  })
+
+  it('returns null for git.eclipse.org URLs with no recoverable repo path', () => {
+    expect(normalizeScmUrl('https://git.eclipse.org/')).toBeNull()
+    expect(normalizeScmUrl('https://git.eclipse.org/c/')).toBeNull()
+    expect(normalizeScmUrl('https://git.eclipse.org/c/onlyowner')).toBeNull()
+  })
+
   // SCP colon form: "host:owner/repo" where the colon is a path separator, not a port
   it('recovers bare host:owner/repo SCP colon form', () => {
     expect(normalizeScmUrl('github.com:japgolly/scalacss.git')).toBe(

services/apps/packages_worker/src/maven/extract.ts

@@ -551,6 +551,28 @@ function normalizeApacheGitwebUrl(host: string, pathname: string, search: string
   return `https://${host}/repos/asf/${name}`
 }
 
+/**
+ * Eclipse's cgit instance serves repos as /c/<owner>/<repo>[.git][/tree/...], the
+ * same owner/repo shape as the generic hosts but under a fixed /c/ prefix that
+ * cgit requires for a working link (unlike GitHub-style hosts, a bare
+ * https://git.eclipse.org/<owner>/<repo> does not resolve) — so the prefix is
+ * kept in the output rather than stripped like the generic path logic would.
+ */
+const ECLIPSE_CGIT_HOSTS = new Set(['git.eclipse.org'])
+
+function normalizeEclipseCgitUrl(host: string, pathname: string): string | null {
+  if (!pathname.startsWith('/c/')) return null
+
+  const segments = pathname.slice('/c/'.length).split('/').filter(Boolean)
+  if (segments.length < 2) return null
+
+  const owner = segments[0]
+  const name = segments[1].replace(/\.git$/, '')
+  if (!owner || !name || /\$\{|%7B/i.test(owner) || /\$\{|%7B/i.test(name)) return null
+
+  return `https://${host}/c/${owner}/${name}`
+}
+
 /**
  * Converts the raw SCM URL from a POM (declared_repository_url) into a clean,
  * canonical `https://<host>/<owner>/<repo>` repository URL suitable for storage
@@ -621,6 +643,10 @@ export function normalizeScmUrl(raw: string | null): string | null {
     return normalizeApacheGitwebUrl(host, parsed.pathname, parsed.search)
   }
 
+  if (ECLIPSE_CGIT_HOSTS.has(host)) {
+    return normalizeEclipseCgitUrl(host, parsed.pathname)
+  }
+
   if (!SCM_HOSTS.has(host)) return null
 
   // Require at least owner + repo path segments