Files
multica/server/internal/handler/plugin_package_concurrency_test.go
T
f45b591623 MUL-6469: host plugin artifacts, bind installs to immutable versions (#7321)
* feat(plugins): host plugin artifacts, bind installs to immutable versions

A plugin used to be a URL. The manifest was frozen at install, but the surface
script was fetched from the author's server every time a panel opened, with no
integrity check — so an administrator consented to a manifest while the browser
ran whatever that host served that day, inside the scopes already granted. The
author's uptime was our uptime, every panel open leaked the reader's IP and
"who read which issue" to the author, and publishing at all required a public
HTTPS domain, which is why the repo's own deploy-sentinel example never ran.

The author now uploads an artifact bundle and Multica stores it:

- plugin_package / plugin_package_version / plugin_package_file. A version is
  insert-only, and the (package_id, version) unique index is what makes
  immutability a database rule rather than a convention.
- An installation names one version. Publishing another changes nothing for a
  workspace until an administrator upgrades, which is a second consent.
- The bundle is validated once, at publish: the manifest parses, every file it
  declares is present and loadable, and a surface entry with a top-level import
  is refused with the line named. That failure used to surface months later in
  a reader's browser.
- The host inlines the surface script into the document it generates, so the
  CSP names no remote script origin at all. `connect-src 'none'` now means it:
  a surface with no net: scope cannot reach anywhere, including its author.
- Install-by-URL is gone rather than kept alongside. Two paths would make "is
  this code frozen?" depend on how the plugin was installed.
- MULTICA_PLUGIN_DIR stays the development channel, but publishes an ordinary
  immutable version instead of being a second render path; re-publishing an
  unchanged version lands as `+dev.N`.

Hook endpoints and MCP servers remain on the author's own infrastructure.

Migration 376 drops existing installations: they name a source URL that is no
longer a concept, and the code they would run was never published here.
plugins_v1 has never left its feature flag and holds no production data — the
same call migration 344 made.

Closes MUL-6469

Co-authored-by: multica-agent <[email protected]>

* fix(plugins): serialize publish/install/delete, correct the isolation claim

Review of #7321 found five problems. Four are fixed here; the fifth is real,
needs its own design pass, and is now tracked and described honestly in the code
instead of claimed away.

Migration prefix collision. main took 376 for agent_task_durable_work_dir while
this branch was open, so `migration prefix 376 is reused` failed backend-tests.
Rebased and renumbered the six migrations to 377-382.

TOCTOU between delete and install/publish. Relationships are application-owned
by repository policy, so nothing made "the version this installation names still
exists" true across statements: `delete counts zero installs` -> `install reads
the version` -> `delete commits` -> `install commits` left an installation whose
panel 404s forever, and a publish racing a delete left a version whose package
row was gone. Publish, install and delete now hold a (workspace, plugin key)
advisory lock for their whole transaction, and delete counts installations
inside it. Both interleavings have regression tests that reproduce the bug with
the lock removed.

Failed publishes moved package state. upsertPackage ran before the version
transaction, so a republish that lost the unique index had already renamed the
package, and a first publish that failed afterwards left a plugin with zero
versions. It now runs inside the same transaction.

The top-level import check was wrong in BOTH directions. It read line prefixes,
so it missed `  import x` and `/* c */ import x` — and it refused a valid file
whose template literal contained the word at a line start, which is ordinary in
a surface that renders code samples. A refusal blocks a publish with no way
around it, so that false positive was the worse half. Replaced with a scanner
that skips comments and string/template literals; where it is imprecise it is
imprecise safely (`/` always reads as division), so a mistake can only cost a
detection, never invent one.

Sandbox self-navigation is NOT fixed. `sandbox="allow-scripts"` permits `_self`
navigation by design, no shipped CSP directive covers it, and contentWindow is
unchanged across it — so a hostile artifact can still reach its author once and
hand the replacement document the bridge port. Closing it needs an embedder
frame-src policy or a handshake a navigated document cannot complete; both are
decisions of their own. What lands here is damage control, labelled as such: the
generated document beacons on pagehide from a listener the plugin cannot detach,
and the embedder drops the bridge and unmounts the frame. The "a surface cannot
reach its author" wording is removed from the SDK README and the tests.

Closes MUL-6469
Refs MUL-6485

Co-authored-by: multica-agent <[email protected]>

* fix(plugins): harden surface validation and error reporting

* fix(migrations): renumber plugin package migrations

* fix(plugins): validate classic surfaces and reset failures

---------

Co-authored-by: Lambda <[email protected]>
Co-authored-by: multica-agent <[email protected]>
2026-08-21 16:03:44 +08:00

159 lines
5.9 KiB
Go

package handler
import (
"context"
"encoding/json"
"net/http"
"net/http/httptest"
"strings"
"sync"
"testing"
)
// Publish, install and delete race each other.
//
// Relationships are application-owned by repository policy, so there is no
// foreign key making "the version this installation names still exists" true.
// The interleaving that matters is `delete counts zero installs` → `install
// reads the version` → `delete commits` → `install commits`: the installation
// survives pointing at a version that is gone, its panel 404s forever, and
// nothing in the product can explain why. A (workspace, plugin key) advisory
// lock held for each transaction is what serializes them.
//
// These run the real handlers against the real database, because the bug only
// exists at the transaction boundary — a test with a fake would assert the
// structure of the fix rather than its effect.
// TestDeleteAndInstallRaceLeavesNoDanglingInstallation is the direct regression.
// Whichever side wins, the end state has to be self-consistent: either the
// package is gone and nothing installed it, or the installation exists and its
// version is still there.
func TestDeleteAndInstallRaceLeavesNoDanglingInstallation(t *testing.T) {
withPluginsV1Flag(t, testHandler, true)
withHostCapabilities(t)
// Repeated because a lost race is a scheduling accident: a single pass can
// miss the window even when the lock is absent.
for attempt := range 12 {
cleanupPluginInstallations(t)
published := publishUploadedBundle(t, packageManifest("1.0.0"), "console.log('v1');\n")
versionID := published.Versions[0].ID
var wait sync.WaitGroup
wait.Add(2)
go func() {
defer wait.Done()
body, _ := json.Marshal(map[string]any{"version_id": versionID, "granted_scopes": []string{"issues:read"}})
recorder := httptest.NewRecorder()
testHandler.InstallPlugin(recorder,
pluginHandlerRequest(http.MethodPost, "/plugins", body, map[string]string{"id": testWorkspaceID}))
}()
go func() {
defer wait.Done()
recorder := httptest.NewRecorder()
testHandler.DeletePluginPackage(recorder, pluginHandlerRequest(http.MethodDelete, "/plugins/packages", nil,
map[string]string{"id": testWorkspaceID, "packageId": published.ID}))
}()
wait.Wait()
var dangling int
if err := testPool.QueryRow(context.Background(),
`SELECT COUNT(*) FROM plugin_installation i
WHERE i.workspace_id = $1
AND NOT EXISTS (SELECT 1 FROM plugin_package_version v WHERE v.id = i.package_version_id)`,
testWorkspaceID,
).Scan(&dangling); err != nil {
t.Fatalf("count dangling installations: %v", err)
}
if dangling != 0 {
t.Fatalf("attempt %d: %d installation(s) name a version that no longer exists", attempt, dangling)
}
}
}
// The mirror case: a publish racing a delete of the same plugin must not leave
// a version whose package row is gone. Such a version is unreachable — the
// settings page lists versions per package — so it would be storage nothing can
// find or clean up.
func TestPublishAndDeleteRaceLeavesNoOrphanVersion(t *testing.T) {
withPluginsV1Flag(t, testHandler, true)
withHostCapabilities(t)
for attempt := range 12 {
cleanupPluginInstallations(t)
published := publishUploadedBundle(t, packageManifest("1.0.0"), "console.log('v1');\n")
var wait sync.WaitGroup
wait.Add(2)
go func() {
defer wait.Done()
uploadPluginBundle(t, pluginBundleZip(t, packageManifest("2.0.0"), "console.log('v2');\n"))
}()
go func() {
defer wait.Done()
recorder := httptest.NewRecorder()
testHandler.DeletePluginPackage(recorder, pluginHandlerRequest(http.MethodDelete, "/plugins/packages", nil,
map[string]string{"id": testWorkspaceID, "packageId": published.ID}))
}()
wait.Wait()
var orphans int
if err := testPool.QueryRow(context.Background(),
`SELECT COUNT(*) FROM plugin_package_version v
WHERE v.workspace_id = $1
AND NOT EXISTS (SELECT 1 FROM plugin_package p WHERE p.id = v.package_id)`,
testWorkspaceID,
).Scan(&orphans); err != nil {
t.Fatalf("count orphan versions: %v", err)
}
if orphans != 0 {
t.Fatalf("attempt %d: %d published version(s) outlived their package", attempt, orphans)
}
}
}
// A failed publish must leave the package row exactly as it was. The display
// name follows the newest PUBLISHED version, so a conflicting publish that
// renamed the package would have the list describing an artifact that was never
// stored — and a first publish that failed after creating the row would leave a
// plugin with no versions at all.
func TestFailedPublishDoesNotMovePackageState(t *testing.T) {
withPluginsV1Flag(t, testHandler, true)
withHostCapabilities(t)
cleanupPluginInstallations(t)
publishUploadedBundle(t, packageManifest("1.0.0"), "console.log('v1');\n")
// Same version, different display name: the version insert loses the unique
// index, so nothing about this publish may survive.
renamed := packageManifest("1.0.0")
renamed = replaceOnce(t, renamed, `"name": "Published Panel"`, `"name": "Renamed Panel"`)
recorder := uploadPluginBundle(t, pluginBundleZip(t, renamed, "console.log('tampered');\n"))
if recorder.Code != http.StatusConflict {
t.Fatalf("republish status=%d body=%s", recorder.Code, recorder.Body.String())
}
var name string
var versions int
if err := testPool.QueryRow(context.Background(),
`SELECT p.name, (SELECT COUNT(*) FROM plugin_package_version v WHERE v.package_id = p.id)
FROM plugin_package p WHERE p.workspace_id = $1`, testWorkspaceID,
).Scan(&name, &versions); err != nil {
t.Fatalf("read package: %v", err)
}
if name != "Published Panel" {
t.Fatalf("a failed publish renamed the package to %q", name)
}
if versions != 1 {
t.Fatalf("versions = %d, want the one that actually published", versions)
}
}
func replaceOnce(t *testing.T, source, old, replacement string) string {
t.Helper()
if !strings.Contains(source, old) {
t.Fatalf("fixture does not contain %q", old)
}
return strings.Replace(source, old, replacement, 1)
}