feat(import): resolve references and schema composition

apps/admin-api/src/main.rs

@@ -227,6 +227,7 @@ async fn run(
     .with_artifact_store(artifact_store.clone())
     .with_public_base_url(base_url)
     .with_outbound_http_policy(outbound_http_policy)
+    .with_external_reference_import(&config.external_references)?
     .with_identity_provider(std::sync::Arc::new(identity_provider))
     .build();
     if config.demo_seed {

apps/admin-api/src/service.rs

@@ -4,6 +4,7 @@ use std::sync::Arc;
 
 use base64::{Engine as _, engine::general_purpose::URL_SAFE_NO_PAD};
 use crank_artifacts::ArtifactStore;
+use crank_config::ExternalReferenceSettings;
 use crank_core::{
     AuditActor, AuditEvent, AuditEventId, AuditSink, AuditTarget, AuditTargetKind, AuthProfile,
     CapabilityProfile, CommunityCapabilityProfile, CorrelationContext, EditionCapabilities,
@@ -19,7 +20,8 @@ use crank_registry::{
     UsageBucket,
 };
 use crank_runtime::{
-    OutboundHttpPolicy, ResolvedAuth, RuntimeError, RuntimeExecutor, SecretCrypto,
+    ExternalReferenceFetcher, OutboundHttpPolicy, ResolvedAuth, RuntimeError, RuntimeExecutor,
+    SecretCrypto,
 };
 use crank_schema::{Schema, SchemaKind};
 use crank_trace::{DbOperation, ErrorCategory, Stage, StageOutcome, observe_db_query};
@@ -68,6 +70,9 @@ pub struct AdminService {
     audit_sink: Arc<dyn AuditSink>,
     capability_profile: Arc<dyn CapabilityProfile>,
     outbound_http_policy: OutboundHttpPolicy,
+    external_reference_fetcher: ExternalReferenceFetcher,
+    external_reference_normalization: crank_import::rest::NormalizationConfig,
+    external_reference_materialization_timeout: std::time::Duration,
     public_base_url: String,
 }
 
@@ -83,6 +88,9 @@ pub struct AdminServiceBuilder {
     audit_sink: Option<Arc<dyn AuditSink>>,
     capability_profile: Option<Arc<dyn CapabilityProfile>>,
     outbound_http_policy: OutboundHttpPolicy,
+    external_reference_fetcher: ExternalReferenceFetcher,
+    external_reference_normalization: crank_import::rest::NormalizationConfig,
+    external_reference_settings: ExternalReferenceSettings,
     public_base_url: String,
 }
 
@@ -220,6 +228,22 @@ impl AdminServiceBuilder {
         secret_crypto: SecretCrypto,
         runtime: RuntimeExecutor,
     ) -> Self {
+        let outbound_http_policy = OutboundHttpPolicy::default();
+        let external_reference_settings = ExternalReferenceSettings {
+            allowed_url_prefixes: Vec::new(),
+            max_depth: 8,
+            max_documents: 32,
+            max_fetch_bytes: 256 * 1024,
+            fetch_timeout_ms: 5_000,
+            max_expanded_nodes: 10_000,
+        };
+        let external_reference_fetcher = ExternalReferenceFetcher::try_new(
+            outbound_http_policy.clone(),
+            external_reference_settings.allowed_url_prefixes.clone(),
+            external_reference_settings.max_fetch_bytes,
+            std::time::Duration::from_millis(external_reference_settings.fetch_timeout_ms),
+        )
+        .expect("static external reference defaults are valid");
         Self {
             registry,
             storage_root,
@@ -231,7 +255,10 @@ impl AdminServiceBuilder {
             policy_engine: None,
             audit_sink: None,
             capability_profile: None,
-            outbound_http_policy: OutboundHttpPolicy::default(),
+            outbound_http_policy,
+            external_reference_fetcher,
+            external_reference_normalization: Default::default(),
+            external_reference_settings,
             public_base_url: "http://localhost:3000".to_owned(),
         }
     }
@@ -242,10 +269,41 @@ impl AdminServiceBuilder {
     }
 
     pub fn with_outbound_http_policy(mut self, policy: OutboundHttpPolicy) -> Self {
+        self.external_reference_fetcher = ExternalReferenceFetcher::try_new(
+            policy.clone(),
+            self.external_reference_settings
+                .allowed_url_prefixes
+                .clone(),
+            self.external_reference_settings.max_fetch_bytes,
+            std::time::Duration::from_millis(self.external_reference_settings.fetch_timeout_ms),
+        )
+        .expect("validated external reference settings remain valid");
         self.outbound_http_policy = policy;
         self
     }
 
+    pub fn with_external_reference_import(
+        mut self,
+        settings: &ExternalReferenceSettings,
+    ) -> Result<Self, crank_runtime::ExternalReferenceFetchError> {
+        self.external_reference_fetcher = ExternalReferenceFetcher::try_new(
+            self.outbound_http_policy.clone(),
+            settings.allowed_url_prefixes.clone(),
+            settings.max_fetch_bytes,
+            std::time::Duration::from_millis(settings.fetch_timeout_ms),
+        )?;
+        self.external_reference_normalization.max_reference_depth = settings.max_depth;
+        self.external_reference_normalization
+            .max_reference_documents = settings.max_documents;
+        self.external_reference_normalization
+            .max_external_document_bytes = settings.max_fetch_bytes;
+        self.external_reference_normalization.max_expanded_nodes = settings.max_expanded_nodes;
+        self.external_reference_normalization
+            .external_references_enabled = !settings.allowed_url_prefixes.is_empty();
+        self.external_reference_settings = settings.clone();
+        Ok(self)
+    }
+
     /// Reuses the process-wide immutable artifact authority for OpenAPI
     /// ingress and reconciliation. Tests may omit this and use their private
     /// storage root instead.
@@ -299,6 +357,11 @@ impl AdminServiceBuilder {
                 .capability_profile
                 .unwrap_or_else(|| Arc::new(CommunityCapabilityProfile)),
             outbound_http_policy: self.outbound_http_policy,
+            external_reference_fetcher: self.external_reference_fetcher,
+            external_reference_normalization: self.external_reference_normalization,
+            external_reference_materialization_timeout: std::time::Duration::from_millis(
+                self.external_reference_settings.fetch_timeout_ms,
+            ),
             public_base_url: self.public_base_url,
         }
     }

apps/admin-api/src/service/imports.rs

@@ -11,7 +11,7 @@ use crank_import::rest::{
 };
 use crank_registry::{
     ApplyImportJobRequest, ArtifactSourceId, ArtifactSourceSensitivity,
-    CreateArtifactSourceRequest, CreateImportJobRequest, DetachArtifactSourceRequest,
+    CreateArtifactSourceRequest, CreateImportJobRequest, FinishImportJobRequest,
     ImportConflictMode, ImportJobApplyResult, ImportJobId, ImportJobKind, ImportJobSourceEnvelope,
     ImportJobStatus, ImportOperationDraft, RegistryError,
 };
@@ -29,6 +29,13 @@ use crate::{
     },
 };
 
+mod external_references;
+mod job_contract;
+use external_references::ImportReplayContext;
+use job_contract::{
+    SourceDetachGuard, import_job_dependencies, import_job_normalization_config, import_job_source,
+};
+
 const IMPORT_JOB_TTL_HOURS: i64 = 24;
 const OPENAPI_PARSE_DEADLINE: std::time::Duration = std::time::Duration::from_secs(30);
 const OPENAPI_PARSE_CONCURRENCY: usize = 4;
@@ -95,9 +102,14 @@ impl AdminService {
             detach_guard.detach_now().await;
             return Err(ApiError::openapi_upload(locale, "source_integrity"));
         }
+        let mut dependencies = self
+            .materialize_external_reference_snapshots(workspace_id, &verified.bytes, now)
+            .await;
         let parsed = match parse_verified_preview(
             verified.bytes,
             artifact.artifact_ref().digest_hex().to_owned(),
+            dependencies.snapshots.clone(),
+            self.external_reference_normalization.clone(),
             locale,
             false,
         )
@@ -105,6 +117,7 @@ impl AdminService {
         {
             Ok(preview) => preview,
             Err(error) => {
+                dependencies.detach_all().await;
                 detach_guard.detach_now().await;
                 return Err(error);
             }
@@ -115,6 +128,7 @@ impl AdminService {
             .iter()
             .all(|group| group.operations.is_empty())
         {
+            dependencies.detach_all().await;
             detach_guard.detach_now().await;
             return Err(ApiError::openapi_upload(locale, "no_methods"));
         }
@@ -130,12 +144,15 @@ impl AdminService {
                 "source_id": source_envelope.source_id.as_str(),
                 "digest": source_envelope.digest.as_str(),
             },
+            "dependencies": dependencies.payload(),
+            "dependency_snapshots": dependencies.snapshot_payload(),
             "preview": preview_value,
             "preview_digest": preview_digest,
             "normalization": {
                 "normalizer_version": NORMALIZER_VERSION,
                 "projection_version": PROJECTION_VERSION,
                 "ir_fingerprint": parsed.ir_fingerprint,
+                "config": self.external_reference_normalization,
             },
         });
 
@@ -155,9 +172,11 @@ impl AdminService {
             })
             .await
         {
+            dependencies.detach_all().await;
             detach_guard.detach_now().await;
             return Err(ApiError::from(error));
         }
+        dependencies.disarm();
         detach_guard.disarm();
 
         Ok(OpenApiImportPreviewResponse {
@@ -234,6 +253,13 @@ impl AdminService {
         } else {
             ImportConflictMode::Rename
         };
+        let replay_context = ImportReplayContext {
+            workspace_id,
+            job_id,
+            application_key: &application_key,
+            conflict_mode,
+            finished_at: &finished_at,
+        };
         if job.status == ImportJobStatus::Completed {
             let applied = self
                 .registry
@@ -250,7 +276,18 @@ impl AdminService {
             return Ok(openapi_import_response(applied));
         }
 
-        let source = import_job_source(&job.preview_payload, locale)?;
+        let source = match import_job_source(&job.preview_payload, locale) {
+            Ok(source) => source,
+            Err(error) => {
+                return self
+                    .fail_openapi_import_or_replay(
+                        &replay_context,
+                        "import_source_verification_failed",
+                        error,
+                    )
+                    .await;
+            }
+        };
         let verified = match self
             .registry
             .read_artifact_source(
@@ -264,44 +301,123 @@ impl AdminService {
             Err(
                 error @ (RegistryError::SourceNotFound { .. } | RegistryError::SourceUnavailable),
             ) => {
-                // Another request may have finished and detached the source
-                // after our unlocked Pending read. Re-read the job and ask
-                // the registry for its canonical, locked replay instead of
-                // exposing a false `source_unavailable` outcome.
-                let latest = self.registry.get_import_job(workspace_id, job_id).await?;
-                if latest.is_some_and(|latest| latest.status == ImportJobStatus::Completed) {
-                    let applied = self
-                        .registry
-                        .apply_import_job(ApplyImportJobRequest {
-                            id: job_id,
-                            workspace_id,
-                            application_key: &application_key,
-                            conflict_mode,
-                            operations: &[],
-                            pre_skipped: &[],
-                            finished_at: &finished_at,
-                        })
-                        .await?;
-                    return Ok(openapi_import_response(applied));
-                }
-                return Err(openapi_source_error(locale, error));
+                return self
+                    .fail_openapi_import_or_replay(
+                        &replay_context,
+                        "import_source_verification_failed",
+                        openapi_source_error(locale, error),
+                    )
+                    .await;
+            }
+            Err(error @ RegistryError::SourceIntegrity) => {
+                return self
+                    .fail_openapi_import_or_replay(
+                        &replay_context,
+                        "import_source_verification_failed",
+                        openapi_source_error(locale, error),
+                    )
+                    .await;
             }
             Err(error) => return Err(openapi_source_error(locale, error)),
         };
         if verified.source.blob.artifact_ref != source.digest {
-            return Err(ApiError::openapi_upload(locale, "source_integrity"));
+            return self
+                .fail_openapi_import_or_replay(
+                    &replay_context,
+                    "import_source_verification_failed",
+                    ApiError::openapi_upload(locale, "source_integrity"),
+                )
+                .await;
         }
-        let legacy_v1 = job.preview_payload.get("normalization").is_none();
-        let parsed = parse_verified_preview(
-            verified.bytes,
-            source.digest.digest_hex().to_owned(),
-            locale,
-            legacy_v1,
-        )
-        .await?;
-        verify_preview_contract(&job.preview_payload, &parsed, locale)?;
-        if preview_has_blocker(&parsed.preview) {
-            return Err(ApiError::openapi_upload(locale, "invalid_document"));
+        let replay_contract = match import_replay_contract(&job.preview_payload, locale) {
+            Ok(contract) => contract,
+            Err(error) => {
+                return self
+                    .fail_openapi_import_or_replay(
+                        &replay_context,
+                        "import_replay_verification_failed",
+                        error,
+                    )
+                    .await;
+            }
+        };
+        let parsed = match replay_contract {
+            ImportReplayContract::PersistedV2(parsed) => parsed,
+            ImportReplayContract::LegacyV1 | ImportReplayContract::Current => {
+                let legacy_v1 = matches!(replay_contract, ImportReplayContract::LegacyV1);
+                let dependency_snapshots = match self
+                    .read_import_job_dependencies(workspace_id, &job.preview_payload, locale)
+                    .await
+                {
+                    Ok(snapshots) => snapshots,
+                    Err(error) => {
+                        return self
+                            .fail_openapi_import_or_replay(
+                                &replay_context,
+                                "import_dependency_verification_failed",
+                                error,
+                            )
+                            .await;
+                    }
+                };
+                let normalization_config = match import_job_normalization_config(
+                    &job.preview_payload,
+                    &self.external_reference_normalization,
+                    legacy_v1,
+                    locale,
+                ) {
+                    Ok(config) => config,
+                    Err(error) => {
+                        return self
+                            .fail_openapi_import_or_replay(
+                                &replay_context,
+                                "import_replay_verification_failed",
+                                error,
+                            )
+                            .await;
+                    }
+                };
+                let parsed = match parse_verified_preview(
+                    verified.bytes,
+                    source.digest.digest_hex().to_owned(),
+                    dependency_snapshots,
+                    normalization_config,
+                    locale,
+                    legacy_v1,
+                )
+                .await
+                {
+                    Ok(parsed) => parsed,
+                    Err(error) => {
+                        return self
+                            .fail_openapi_import_or_replay(
+                                &replay_context,
+                                "import_replay_verification_failed",
+                                error,
+                            )
+                            .await;
+                    }
+                };
+                if let Err(error) = verify_preview_contract(&job.preview_payload, &parsed, locale) {
+                    return self
+                        .fail_openapi_import_or_replay(
+                            &replay_context,
+                            "import_replay_verification_failed",
+                            error,
+                        )
+                        .await;
+                }
+                parsed
+            }
+        };
+        if preview_has_blocker(&parsed.preview, &selected) {
+            return self
+                .fail_openapi_import_or_replay(
+                    &replay_context,
+                    "reference_resolution_blocked",
+                    ApiError::openapi_upload(locale, "invalid_document"),
+                )
+                .await;
         }
         let mut candidates = BTreeMap::new();
         for group in &parsed.preview.groups {
@@ -372,6 +488,24 @@ impl AdminService {
     }
 }
 
+fn canonical_external_document_uri(base: Option<&str>, reference: &str) -> Option<String> {
+    if reference.starts_with('#') {
+        return None;
+    }
+    let mut url = match base {
+        Some(base) => url::Url::parse(base).ok()?.join(reference).ok()?,
+        None => url::Url::parse(reference).ok()?,
+    };
+    if !matches!(url.scheme(), "http" | "https")
+        || !url.username().is_empty()
+        || url.password().is_some()
+    {
+        return None;
+    }
+    url.set_fragment(None);
+    Some(url.to_string())
+}
+
 fn openapi_import_response(applied: ImportJobApplyResult) -> OpenApiImportCreateResponse {
     let created = applied
         .created
@@ -470,9 +604,72 @@ struct ParsedOpenApiPreview {
     ir_fingerprint: String,
 }
 
+enum ImportReplayContract {
+    LegacyV1,
+    PersistedV2(ParsedOpenApiPreview),
+    Current,
+}
+
+fn import_replay_contract(
+    payload: &serde_json::Value,
+    locale: OpenApiUploadLocale,
+) -> Result<ImportReplayContract, ApiError> {
+    let Some(normalization) = payload.get("normalization") else {
+        return Ok(ImportReplayContract::LegacyV1);
+    };
+    let normalizer = normalization
+        .get("normalizer_version")
+        .and_then(serde_json::Value::as_str)
+        .ok_or_else(|| ApiError::openapi_upload(locale, "source_integrity"))?;
+    let projection = normalization
+        .get("projection_version")
+        .and_then(serde_json::Value::as_str)
+        .ok_or_else(|| ApiError::openapi_upload(locale, "source_integrity"))?;
+    let ir_fingerprint = normalization
+        .get("ir_fingerprint")
+        .and_then(serde_json::Value::as_str)
+        .filter(|value| is_lower_sha256(value))
+        .ok_or_else(|| ApiError::openapi_upload(locale, "source_integrity"))?;
+    let expected_preview_digest = payload
+        .get("preview_digest")
+        .and_then(serde_json::Value::as_str)
+        .filter(|value| is_lower_sha256(value))
+        .ok_or_else(|| ApiError::openapi_upload(locale, "source_integrity"))?;
+
+    if normalizer == NORMALIZER_VERSION && projection == PROJECTION_VERSION {
+        return Ok(ImportReplayContract::Current);
+    }
+    if normalizer != "normalized-ir-v2" || projection != "preview-v2" {
+        return Err(ApiError::openapi_upload(locale, "source_integrity"));
+    }
+
+    let preview_value = payload
+        .get("preview")
+        .cloned()
+        .ok_or_else(|| ApiError::openapi_upload(locale, "source_integrity"))?;
+    if preview_digest(&preview_value)? != expected_preview_digest {
+        return Err(ApiError::openapi_upload(locale, "source_integrity"));
+    }
+    let preview = serde_json::from_value(preview_value)
+        .map_err(|_| ApiError::openapi_upload(locale, "source_integrity"))?;
+    Ok(ImportReplayContract::PersistedV2(ParsedOpenApiPreview {
+        preview,
+        ir_fingerprint: ir_fingerprint.to_owned(),
+    }))
+}
+
+fn is_lower_sha256(value: &str) -> bool {
+    value.len() == 64
+        && value
+            .bytes()
+            .all(|byte| byte.is_ascii_digit() || (b'a'..=b'f').contains(&byte))
+}
+
 async fn parse_verified_preview(
     bytes: Vec<u8>,
     digest: String,
+    snapshots: Vec<crank_import::rest::ExternalDocumentSnapshot>,
+    normalization_config: crank_import::rest::NormalizationConfig,
     locale: OpenApiUploadLocale,
     legacy_v1: bool,
 ) -> Result<ParsedOpenApiPreview, ApiError> {
@@ -498,14 +695,18 @@ async fn parse_verified_preview(
         }
         let digest = crank_import::rest::SourceDigest::parse(digest)
             .map_err(|_| ApiError::openapi_upload(locale, "source_integrity"))?;
-        let ir =
-            crank_import::rest::normalize_verified_document(document, digest, &Default::default())
-                .map_err(|error| match error {
-                    crank_import::rest::ImportParseError::NoMethods => {
-                        ApiError::openapi_upload(locale, "no_methods")
-                    }
-                    _ => ApiError::openapi_upload(locale, "invalid_document"),
-                })?;
+        let ir = crank_import::rest::normalize_verified_bundle(
+            document,
+            digest,
+            &snapshots,
+            &normalization_config,
+        )
+        .map_err(|error| match error {
+            crank_import::rest::ImportParseError::NoMethods => {
+                ApiError::openapi_upload(locale, "no_methods")
+            }
+            _ => ApiError::openapi_upload(locale, "invalid_document"),
+        })?;
         let ir_fingerprint = preview_digest(
             &serde_json::to_value(&ir)
                 .map_err(|_| ApiError::openapi_upload(locale, "invalid_document"))?,
@@ -597,113 +798,21 @@ fn verify_preview_contract(
     }
 }
 
-fn preview_has_blocker(preview: &crank_import::rest::ImportPreview) -> bool {
+fn preview_has_blocker(
+    preview: &crank_import::rest::ImportPreview,
+    selected_operation_keys: &BTreeSet<String>,
+) -> bool {
     preview
         .findings
         .iter()
-        .chain(
-            preview
-                .groups
-                .iter()
-                .flat_map(|group| group.operations.iter())
-                .flat_map(|operation| operation.findings.iter()),
-        )
         .any(|finding| finding.severity == ImportFindingSeverity::Error)
-}
-
-fn import_job_source(
-    payload: &serde_json::Value,
-    locale: OpenApiUploadLocale,
-) -> Result<ImportJobSourceEnvelope, ApiError> {
-    let source = payload
-        .get("source")
-        .ok_or_else(|| ApiError::openapi_upload(locale, "source_unavailable"))?;
-    let source_id = source
-        .get("source_id")
-        .and_then(serde_json::Value::as_str)
-        .filter(|value| value.len() <= 132)
-        .ok_or_else(|| ApiError::openapi_upload(locale, "source_unavailable"))?;
-    let digest = source
-        .get("digest")
-        .and_then(serde_json::Value::as_str)
-        .and_then(|value| value.parse().ok())
-        .ok_or_else(|| ApiError::openapi_upload(locale, "source_integrity"))?;
-    Ok(ImportJobSourceEnvelope {
-        source_id: ArtifactSourceId::new(source_id),
-        digest,
-    })
-}
-
-struct SourceDetachGuard {
-    registry: crank_registry::PostgresRegistry,
-    workspace_id: WorkspaceId,
-    source_id: ArtifactSourceId,
-    expected_updated_at: OffsetDateTime,
-    armed: bool,
-}
-
-impl SourceDetachGuard {
-    fn new(
-        registry: crank_registry::PostgresRegistry,
-        workspace_id: WorkspaceId,
-        source_id: ArtifactSourceId,
-        expected_updated_at: OffsetDateTime,
-    ) -> Self {
-        Self {
-            registry,
-            workspace_id,
-            source_id,
-            expected_updated_at,
-            armed: true,
-        }
-    }
-
-    fn disarm(&mut self) {
-        self.armed = false;
-    }
-
-    async fn detach_now(&mut self) {
-        if !self.armed {
-            return;
-        }
-        self.armed = false;
-        // This is the normal error path: await the database transition so a
-        // caller receives a completed cleanup before it can retry. `Drop`
-        // remains only the cancellation/shutdown safety net.
-        let _ = self
-            .registry
-            .detach_artifact_source(DetachArtifactSourceRequest {
-                workspace_id: &self.workspace_id,
-                source_id: &self.source_id,
-                expected_updated_at: Some(self.expected_updated_at),
-                detached_at: OffsetDateTime::now_utc(),
-            })
-            .await;
-    }
-}
-
-impl Drop for SourceDetachGuard {
-    fn drop(&mut self) {
-        if !self.armed {
-            return;
-        }
-        let registry = self.registry.clone();
-        let workspace_id = self.workspace_id.clone();
-        let source_id = self.source_id.clone();
-        let expected_updated_at = Some(self.expected_updated_at);
-        if let Ok(handle) = tokio::runtime::Handle::try_current() {
-            handle.spawn(async move {
-                let _ = registry
-                    .detach_artifact_source(DetachArtifactSourceRequest {
-                        workspace_id: &workspace_id,
-                        source_id: &source_id,
-                        expected_updated_at,
-                        detached_at: OffsetDateTime::now_utc(),
-                    })
-                    .await;
-            });
-        }
-    }
+        || preview
+            .groups
+            .iter()
+            .flat_map(|group| group.operations.iter())
+            .filter(|operation| selected_operation_keys.contains(&operation.key))
+            .flat_map(|operation| operation.findings.iter())
+            .any(|finding| finding.severity == ImportFindingSeverity::Error)
 }
 
 fn openapi_application_key(payload: &OpenApiImportCreatePayload) -> Result<String, ApiError> {

apps/admin-api/src/service/imports/external_references.rs

@@ -0,0 +1,405 @@
+use super::*;
+use std::collections::VecDeque;
+use tracing::warn;
+
+pub(super) struct MaterializedDependencies {
+    pub(super) snapshots: Vec<crank_import::rest::ExternalDocumentSnapshot>,
+    envelopes: Vec<ImportJobSourceEnvelope>,
+    canonical_uris: Vec<String>,
+    guards: Vec<SourceDetachGuard>,
+}
+
+pub(super) struct ImportReplayContext<'a> {
+    pub(super) workspace_id: &'a WorkspaceId,
+    pub(super) job_id: &'a ImportJobId,
+    pub(super) application_key: &'a str,
+    pub(super) conflict_mode: ImportConflictMode,
+    pub(super) finished_at: &'a OffsetDateTime,
+}
+
+impl MaterializedDependencies {
+    fn empty() -> Self {
+        Self {
+            snapshots: Vec::new(),
+            envelopes: Vec::new(),
+            canonical_uris: Vec::new(),
+            guards: Vec::new(),
+        }
+    }
+
+    pub(super) fn payload(&self) -> Vec<serde_json::Value> {
+        self.envelopes
+            .iter()
+            .zip(&self.canonical_uris)
+            .map(|(dependency, canonical_uri)| {
+                json!({
+                    "source_id": dependency.source_id.as_str(),
+                    "digest": dependency.digest.as_str(),
+                    "canonical_uri": canonical_uri,
+                })
+            })
+            .collect()
+    }
+
+    pub(super) fn snapshot_payload(&self) -> Vec<serde_json::Value> {
+        self.snapshots
+            .iter()
+            .filter_map(|snapshot| {
+                self.envelopes
+                    .iter()
+                    .find(|dependency| dependency.digest.digest_hex() == snapshot.digest.as_str())
+                    .map(|dependency| {
+                        json!({
+                            "source_id": dependency.source_id.as_str(),
+                            "digest": dependency.digest.as_str(),
+                            "canonical_uri": snapshot.canonical_uri,
+                        })
+                    })
+            })
+            .collect()
+    }
+
+    pub(super) fn disarm(&mut self) {
+        for guard in &mut self.guards {
+            guard.disarm();
+        }
+    }
+
+    pub(super) async fn detach_all(&mut self) {
+        for guard in &mut self.guards {
+            guard.detach_now().await;
+        }
+    }
+}
+
+impl AdminService {
+    pub(super) async fn fail_openapi_import_or_replay(
+        &self,
+        context: &ImportReplayContext<'_>,
+        error_text: &'static str,
+        error: ApiError,
+    ) -> Result<OpenApiImportCreateResponse, ApiError> {
+        let empty = json!([]);
+        self.registry
+            .finish_import_job(FinishImportJobRequest {
+                id: context.job_id,
+                status: ImportJobStatus::Failed,
+                created_operation_ids: &empty,
+                error_text: Some(error_text),
+                finished_at: context.finished_at,
+            })
+            .await?;
+
+        // `finish_import_job` preserves an already Completed row under its
+        // lock. This read distinguishes that race from the terminal Failed
+        // transition and returns the immutable canonical application result.
+        let latest = self
+            .registry
+            .get_import_job(context.workspace_id, context.job_id)
+            .await?;
+        if latest.is_some_and(|job| job.status == ImportJobStatus::Completed) {
+            let applied = self
+                .registry
+                .apply_import_job(ApplyImportJobRequest {
+                    id: context.job_id,
+                    workspace_id: context.workspace_id,
+                    application_key: context.application_key,
+                    conflict_mode: context.conflict_mode,
+                    operations: &[],
+                    pre_skipped: &[],
+                    finished_at: context.finished_at,
+                })
+                .await?;
+            return Ok(openapi_import_response(applied));
+        }
+
+        Err(error)
+    }
+
+    pub(super) async fn materialize_external_reference_snapshots(
+        &self,
+        workspace_id: &WorkspaceId,
+        primary_bytes: &[u8],
+        created_at: OffsetDateTime,
+    ) -> MaterializedDependencies {
+        let materialized_count = std::sync::atomic::AtomicUsize::new(0);
+        match tokio::time::timeout(
+            self.external_reference_materialization_timeout,
+            self.materialize_external_reference_snapshots_inner(
+                workspace_id,
+                primary_bytes,
+                created_at,
+                &materialized_count,
+            ),
+        )
+        .await
+        {
+            Ok(dependencies) => dependencies,
+            Err(_) => {
+                materialization_failure(
+                    "chain",
+                    "timeout",
+                    materialized_count.load(std::sync::atomic::Ordering::Relaxed),
+                );
+                MaterializedDependencies::empty()
+            }
+        }
+    }
+
+    async fn materialize_external_reference_snapshots_inner(
+        &self,
+        workspace_id: &WorkspaceId,
+        primary_bytes: &[u8],
+        created_at: OffsetDateTime,
+        materialized_count: &std::sync::atomic::AtomicUsize,
+    ) -> MaterializedDependencies {
+        if !self
+            .external_reference_normalization
+            .external_references_enabled
+        {
+            return MaterializedDependencies::empty();
+        }
+        let Ok(primary) = std::str::from_utf8(primary_bytes) else {
+            materialization_failure("primary_decode", "invalid_utf8", 0);
+            return MaterializedDependencies::empty();
+        };
+        let Ok(primary_references) =
+            crank_import::rest::reference_uris(primary, &self.external_reference_normalization)
+        else {
+            materialization_failure("primary_scan", "invalid_document", 0);
+            return MaterializedDependencies::empty();
+        };
+        let mut queue = VecDeque::new();
+        for reference in primary_references {
+            if let Some(uri) = canonical_external_document_uri(None, &reference) {
+                queue.push_back((uri, 1usize));
+            }
+        }
+        let mut seen = BTreeSet::new();
+        let mut result = MaterializedDependencies::empty();
+        let mut dependency_by_digest = BTreeMap::<String, usize>::new();
+        while let Some((canonical_uri, depth)) = queue.pop_front() {
+            if !seen.insert(canonical_uri.clone())
+                || depth > self.external_reference_normalization.max_reference_depth
+                || seen.len()
+                    > self
+                        .external_reference_normalization
+                        .max_reference_documents
+            {
+                continue;
+            }
+            let bytes = match self.external_reference_fetcher.get(&canonical_uri).await {
+                Ok(bytes) => bytes,
+                Err(error) => {
+                    materialization_failure(
+                        "fetch",
+                        external_fetch_error_code(&error),
+                        result.snapshots.len(),
+                    );
+                    continue;
+                }
+            };
+            let Ok(document) = std::str::from_utf8(&bytes).map(str::to_owned) else {
+                materialization_failure(
+                    "dependency_decode",
+                    "invalid_utf8",
+                    result.snapshots.len(),
+                );
+                continue;
+            };
+            let Ok(references) = crank_import::rest::external_reference_uris(
+                &document,
+                &self.external_reference_normalization,
+            ) else {
+                materialization_failure(
+                    "dependency_scan",
+                    "invalid_document",
+                    result.snapshots.len(),
+                );
+                continue;
+            };
+            let store = self.artifact_store.clone();
+            let artifact_bytes = bytes.clone();
+            let Ok(Ok(artifact)) =
+                tokio::task::spawn_blocking(move || store.put_registered(&artifact_bytes)).await
+            else {
+                materialization_failure(
+                    "artifact_store",
+                    "storage_unavailable",
+                    result.snapshots.len(),
+                );
+                continue;
+            };
+            let digest_key = artifact.artifact_ref().as_str().to_owned();
+            let envelope_index = if let Some(index) = dependency_by_digest.get(&digest_key) {
+                *index
+            } else {
+                let source_id = ArtifactSourceId::new(new_prefixed_id("src_openapi_dep"));
+                let Ok(source) = self
+                    .registry
+                    .create_artifact_source(CreateArtifactSourceRequest {
+                        workspace_id,
+                        source_id: &source_id,
+                        artifact: &artifact,
+                        mime_type: "application/octet-stream",
+                        sensitivity: ArtifactSourceSensitivity::Internal,
+                        created_at,
+                    })
+                    .await
+                else {
+                    materialization_failure(
+                        "source_create",
+                        "registry_unavailable",
+                        result.snapshots.len(),
+                    );
+                    continue;
+                };
+                let mut guard = SourceDetachGuard::new(
+                    self.registry.clone(),
+                    workspace_id.clone(),
+                    source_id.clone(),
+                    source.updated_at,
+                );
+                let Ok(verified) = self
+                    .registry
+                    .read_artifact_source(
+                        std::sync::Arc::clone(&self.artifact_store),
+                        workspace_id,
+                        &source_id,
+                    )
+                    .await
+                else {
+                    materialization_failure(
+                        "source_verify",
+                        "source_unavailable",
+                        result.snapshots.len(),
+                    );
+                    guard.detach_now().await;
+                    continue;
+                };
+                if verified.source.blob.artifact_ref != *artifact.artifact_ref() {
+                    materialization_failure(
+                        "source_verify",
+                        "source_integrity",
+                        result.snapshots.len(),
+                    );
+                    guard.detach_now().await;
+                    continue;
+                }
+                let index = result.envelopes.len();
+                result.envelopes.push(ImportJobSourceEnvelope {
+                    source_id: source_id.clone(),
+                    digest: artifact.artifact_ref().clone(),
+                });
+                result.guards.push(guard);
+                dependency_by_digest.insert(digest_key, index);
+                index
+            };
+            let dependency = &result.envelopes[envelope_index];
+            let snapshot_digest = match crank_import::rest::SourceDigest::parse(
+                dependency.digest.digest_hex().to_owned(),
+            ) {
+                Ok(digest) => digest,
+                Err(_) => {
+                    materialization_failure("snapshot", "source_integrity", result.snapshots.len());
+                    continue;
+                }
+            };
+            result
+                .snapshots
+                .push(crank_import::rest::ExternalDocumentSnapshot {
+                    canonical_uri: canonical_uri.clone(),
+                    digest: snapshot_digest,
+                    document,
+                });
+            materialized_count.store(result.snapshots.len(), std::sync::atomic::Ordering::Relaxed);
+            result.canonical_uris.push(canonical_uri.clone());
+            for reference in references {
+                if let Some(uri) = canonical_external_document_uri(Some(&canonical_uri), &reference)
+                {
+                    queue.push_back((uri, depth.saturating_add(1)));
+                }
+            }
+        }
+        // Registry ownership is one envelope per immutable digest, while the
+        // resolver may map several canonical URIs to that same snapshot. Keep
+        // payload rows aligned with envelopes and add aliases separately.
+        if result.envelopes.len() != result.canonical_uris.len() {
+            let mut canonical_by_digest = BTreeMap::new();
+            for snapshot in &result.snapshots {
+                canonical_by_digest
+                    .entry(snapshot.digest.as_str().to_owned())
+                    .or_insert_with(|| snapshot.canonical_uri.clone());
+            }
+            result.canonical_uris = result
+                .envelopes
+                .iter()
+                .map(|envelope| {
+                    canonical_by_digest
+                        .get(envelope.digest.digest_hex())
+                        .cloned()
+                        .unwrap_or_default()
+                })
+                .collect();
+        }
+        result
+    }
+
+    pub(super) async fn read_import_job_dependencies(
+        &self,
+        workspace_id: &WorkspaceId,
+        payload: &serde_json::Value,
+        locale: OpenApiUploadLocale,
+    ) -> Result<Vec<crank_import::rest::ExternalDocumentSnapshot>, ApiError> {
+        let dependencies = import_job_dependencies(payload, locale)?;
+        let mut snapshots = Vec::with_capacity(dependencies.len());
+        for (canonical_uri, dependency) in dependencies {
+            let verified = self
+                .registry
+                .read_artifact_source(
+                    std::sync::Arc::clone(&self.artifact_store),
+                    workspace_id,
+                    &dependency.source_id,
+                )
+                .await
+                .map_err(|error| openapi_source_error(locale, error))?;
+            if verified.source.blob.artifact_ref != dependency.digest {
+                return Err(ApiError::openapi_upload(locale, "source_integrity"));
+            }
+            let document = String::from_utf8(verified.bytes)
+                .map_err(|_| ApiError::openapi_upload(locale, "source_integrity"))?;
+            let digest =
+                crank_import::rest::SourceDigest::parse(dependency.digest.digest_hex().to_owned())
+                    .map_err(|_| ApiError::openapi_upload(locale, "source_integrity"))?;
+            snapshots.push(crank_import::rest::ExternalDocumentSnapshot {
+                canonical_uri,
+                digest,
+                document,
+            });
+        }
+        Ok(snapshots)
+    }
+}
+fn external_fetch_error_code(error: &crank_runtime::ExternalReferenceFetchError) -> &'static str {
+    match error {
+        crank_runtime::ExternalReferenceFetchError::Disabled => "disabled",
+        crank_runtime::ExternalReferenceFetchError::InvalidUrl => "invalid_url",
+        crank_runtime::ExternalReferenceFetchError::TargetNotAllowed => "target_not_allowed",
+        crank_runtime::ExternalReferenceFetchError::RedirectNotAllowed => "redirect_not_allowed",
+        crank_runtime::ExternalReferenceFetchError::ResponseTooLarge { .. } => "response_too_large",
+        crank_runtime::ExternalReferenceFetchError::UnexpectedStatus { .. } => "unexpected_status",
+        crank_runtime::ExternalReferenceFetchError::Transport { timeout: true, .. } => "timeout",
+        crank_runtime::ExternalReferenceFetchError::Transport { .. } => "transport",
+        crank_runtime::ExternalReferenceFetchError::InvalidConfiguration => "invalid_configuration",
+    }
+}
+
+fn materialization_failure(stage: &'static str, error_code: &'static str, count: usize) {
+    warn!(
+        name: "admin.openapi_import.materialization_failed",
+        stage,
+        error_code,
+        count,
+        "external OpenAPI materialization failed"
+    );
+}

apps/admin-api/src/service/imports/job_contract.rs

@@ -0,0 +1,169 @@
+use std::collections::BTreeSet;
+
+use crank_core::WorkspaceId;
+use crank_registry::{ArtifactSourceId, DetachArtifactSourceRequest, ImportJobSourceEnvelope};
+use time::OffsetDateTime;
+
+use crate::{error::ApiError, service::OpenApiUploadLocale};
+
+use super::canonical_external_document_uri;
+
+const MAX_IMPORT_JOB_DOCUMENTS: usize = 32;
+
+pub(super) fn import_job_source(
+    payload: &serde_json::Value,
+    locale: OpenApiUploadLocale,
+) -> Result<ImportJobSourceEnvelope, ApiError> {
+    let source = payload
+        .get("source")
+        .ok_or_else(|| ApiError::openapi_upload(locale, "source_unavailable"))?;
+    let source_id = source
+        .get("source_id")
+        .and_then(serde_json::Value::as_str)
+        .filter(|value| value.len() <= 132)
+        .ok_or_else(|| ApiError::openapi_upload(locale, "source_unavailable"))?;
+    let digest = source
+        .get("digest")
+        .and_then(serde_json::Value::as_str)
+        .and_then(|value| value.parse().ok())
+        .ok_or_else(|| ApiError::openapi_upload(locale, "source_integrity"))?;
+    Ok(ImportJobSourceEnvelope {
+        source_id: ArtifactSourceId::new(source_id),
+        digest,
+    })
+}
+
+pub(super) fn import_job_dependencies(
+    payload: &serde_json::Value,
+    locale: OpenApiUploadLocale,
+) -> Result<Vec<(String, ImportJobSourceEnvelope)>, ApiError> {
+    let empty = Vec::new();
+    let items = payload
+        .get("dependency_snapshots")
+        .and_then(serde_json::Value::as_array)
+        .unwrap_or(&empty);
+    if items.len() > MAX_IMPORT_JOB_DOCUMENTS {
+        return Err(ApiError::openapi_upload(locale, "source_integrity"));
+    }
+    let mut canonical_uris = BTreeSet::new();
+    items
+        .iter()
+        .map(|item| {
+            let canonical_uri = item
+                .get("canonical_uri")
+                .and_then(serde_json::Value::as_str)
+                .filter(|value| {
+                    canonical_external_document_uri(None, value).as_deref() == Some(*value)
+                })
+                .ok_or_else(|| ApiError::openapi_upload(locale, "source_integrity"))?;
+            if !canonical_uris.insert(canonical_uri.to_owned()) {
+                return Err(ApiError::openapi_upload(locale, "source_integrity"));
+            }
+            let source_id = item
+                .get("source_id")
+                .and_then(serde_json::Value::as_str)
+                .filter(|value| value.len() <= 132)
+                .ok_or_else(|| ApiError::openapi_upload(locale, "source_integrity"))?;
+            let digest = item
+                .get("digest")
+                .and_then(serde_json::Value::as_str)
+                .and_then(|value| value.parse().ok())
+                .ok_or_else(|| ApiError::openapi_upload(locale, "source_integrity"))?;
+            Ok((
+                canonical_uri.to_owned(),
+                ImportJobSourceEnvelope {
+                    source_id: ArtifactSourceId::new(source_id),
+                    digest,
+                },
+            ))
+        })
+        .collect()
+}
+
+pub(super) fn import_job_normalization_config(
+    payload: &serde_json::Value,
+    current: &crank_import::rest::NormalizationConfig,
+    legacy_v1: bool,
+    locale: OpenApiUploadLocale,
+) -> Result<crank_import::rest::NormalizationConfig, ApiError> {
+    if legacy_v1 {
+        return Ok(current.clone());
+    }
+    payload
+        .pointer("/normalization/config")
+        .cloned()
+        .ok_or_else(|| ApiError::openapi_upload(locale, "source_integrity"))
+        .and_then(|value| {
+            serde_json::from_value(value)
+                .map_err(|_| ApiError::openapi_upload(locale, "source_integrity"))
+        })
+}
+
+pub(super) struct SourceDetachGuard {
+    registry: crank_registry::PostgresRegistry,
+    workspace_id: WorkspaceId,
+    source_id: ArtifactSourceId,
+    expected_updated_at: OffsetDateTime,
+    armed: bool,
+}
+
+impl SourceDetachGuard {
+    pub(super) fn new(
+        registry: crank_registry::PostgresRegistry,
+        workspace_id: WorkspaceId,
+        source_id: ArtifactSourceId,
+        expected_updated_at: OffsetDateTime,
+    ) -> Self {
+        Self {
+            registry,
+            workspace_id,
+            source_id,
+            expected_updated_at,
+            armed: true,
+        }
+    }
+
+    pub(super) fn disarm(&mut self) {
+        self.armed = false;
+    }
+
+    pub(super) async fn detach_now(&mut self) {
+        if !self.armed {
+            return;
+        }
+        self.armed = false;
+        let _ = self
+            .registry
+            .detach_artifact_source(DetachArtifactSourceRequest {
+                workspace_id: &self.workspace_id,
+                source_id: &self.source_id,
+                expected_updated_at: Some(self.expected_updated_at),
+                detached_at: OffsetDateTime::now_utc(),
+            })
+            .await;
+    }
+}
+
+impl Drop for SourceDetachGuard {
+    fn drop(&mut self) {
+        if !self.armed {
+            return;
+        }
+        let registry = self.registry.clone();
+        let workspace_id = self.workspace_id.clone();
+        let source_id = self.source_id.clone();
+        let expected_updated_at = Some(self.expected_updated_at);
+        if let Ok(handle) = tokio::runtime::Handle::try_current() {
+            handle.spawn(async move {
+                let _ = registry
+                    .detach_artifact_source(DetachArtifactSourceRequest {
+                        workspace_id: &workspace_id,
+                        source_id: &source_id,
+                        expected_updated_at,
+                        detached_at: OffsetDateTime::now_utc(),
+                    })
+                    .await;
+            });
+        }
+    }
+}