fix: serve Studio at the contract path instead of /api/api/v1/...

PresentationWebConfig prefixes every controller mapping with ca-skeleton.presentation.api-base-path ("/api"), which is why every other controller in this repository declares its path without it — healthcheck is "/healthcheck", uploads are "/v1/uploads", files are "/v1/files". The two Studio controllers declared "/api/v1/studio/..." instead, so the prefix landed on top of a path that already had it and both operations were served at /api/api/v1/studio/... — nowhere near the address studio-v1.yaml declares (servers: "/", paths: /api/v1/studio/...). The frontend calls the contract path, so nothing connected. Verified against a running backend on PostgreSQL behind a real Keycloak: /api/v1/studio/catalog?type=TOPIC 200, 3 items (was 404) /api/api/v1/studio/catalog?type=TOPIC 404 (was 200) Why the tests were green while production was broken: the three slice tests and both nested apps in StudioContractDriftTest build contexts that never include PresentationWebConfig, so no prefix was applied and the controllers' literal "/api/v1/..." matched. They now import it and supply the same "/api" the real app uses, which makes the paths they exercise the effective ones. Re-introducing the bug fails five of them. StudioContractDriftTest needed two more repairs to stay meaningful: springdoc's own endpoint is prefixed too, so the published document is read from /api/v3/api-docs; and publishedStudioOperationsMatchTheContract skips any path not starting with /api/v1/studio/, so a missing prefix would have made it compare nothing and pass. It now asserts it compared at least one path — a gate that cannot see drift is not the same as no drift. Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
2026-08-19 18:26:34 +09:00
parent ab0447a0f9
commit e615c24152
15 changed files with 227 additions and 128 deletions
@@ -12,9 +12,8 @@ import org.springframework.boot.context.properties.ConfigurationProperties;
 *
 * <p>Binding to an enum is what makes an unknown vendor a startup failure. With a raw string the
 * two {@code @ConditionalOnProperty} vendor configurations would both stay off, and the first
- * missing SPI bean would surface as a {@code NoSuchBeanDefinitionException} naming
- * {@code OutboxClaimRepository} — a symptom several layers away from the misspelled value that
- * caused it.
+ * missing SPI bean would surface as a {@code NoSuchBeanDefinitionException} naming {@code
+ * OutboxClaimRepository} — a symptom several layers away from the misspelled value that caused it.
 */
@ConfigurationProperties(prefix = PersistenceVendorSettings.PREFIX)
 public record PersistenceVendorSettings(Vendor vendor) {
@@ -12,8 +12,8 @@ import org.jspecify.annotations.Nullable;
 * H2 atomic scope claim.
 *
 * <p>H2 has no {@code INSERT ... ON CONFLICT ... DO UPDATE ... RETURNING}, so the PostgreSQL
- * statement does not port. The standard {@code MERGE ... USING} does, and carries the same
- * meaning in one statement:
+ * statement does not port. The standard {@code MERGE ... USING} does, and carries the same meaning
+ * in one statement:
 *
 * <ul>
 *   <li>no row for the scope → {@code WHEN NOT MATCHED} inserts the claim (1 row);
@@ -15,12 +15,12 @@ import org.springframework.jdbc.core.JdbcOperations;
 *   <li><b>Session scope, not transaction scope.</b> PostgreSQL takes {@code set_config(..., true)}
 *       — a value that reverts at transaction end. H2's {@code SET} is session-wide and outlives
 *       the transaction on a pooled connection. It is not left stale in practice because the
- *       transaction port applies these before every transaction, so each one overwrites the last;
- *       a connection borrowed outside that path keeps the previous transaction's guard.
- *   <li><b>No idle-in-transaction guard.</b> H2 has no counterpart to
- *       {@code idle_in_transaction_session_timeout}, so that budget cannot be pushed into the
- *       database here. It is left to the caller-side deadline the transaction port already
- *       enforces, rather than silently reported as applied.
+ *       transaction port applies these before every transaction, so each one overwrites the last; a
+ *       connection borrowed outside that path keeps the previous transaction's guard.
+ *   <li><b>No idle-in-transaction guard.</b> H2 has no counterpart to {@code
+ *       idle_in_transaction_session_timeout}, so that budget cannot be pushed into the database
+ *       here. It is left to the caller-side deadline the transaction port already enforces, rather
+ *       than silently reported as applied.
 * </ul>
 *
 * <p>The millisecond values are inlined because H2's {@code SET} takes no bind parameter. They
@@ -15,13 +15,13 @@ import org.springframework.jdbc.core.JdbcOperations;

 /**
 * H2 vendor persistence configuration — the same four SPI beans the PostgreSQL vendor registers,
- * implemented against H2. Selected by {@code ca-skeleton.persistence.vendor=h2}, which the
- * {@code local} profile sets.
+ * implemented against H2. Selected by {@code ca-skeleton.persistence.vendor=h2}, which the {@code
+ * local} profile sets.
 *
 * <p><b>No Flyway location customizer, deliberately.</b> The PostgreSQL vendor points Flyway at
 * {@code classpath:db/migration/postgresql}; there is no H2 equivalent tree, because the local
- * profile turns Flyway off and lets Hibernate derive the schema from the entities. Two
- * consequences worth stating out loud:
+ * profile turns Flyway off and lets Hibernate derive the schema from the entities. Two consequences
+ * worth stating out loud:
 *
 * <ul>
 *   <li>Tables that exist only in migrations — the capability schema registry, the polling-delivery
@@ -30,12 +30,12 @@ import org.springframework.jdbc.core.JdbcOperations;
 *       there will fail on a missing table rather than silently misbehave.
 *   <li>A fork that enables Flyway while this vendor is selected gets no location override, so
 *       Flyway falls back to {@code classpath:db/migration} and walks the whole tree — including
- *       PostgreSQL DDL H2 cannot parse. Such a fork should register its own
- *       {@code FlywayConfigurationCustomizer} naming an H2 location.
+ *       PostgreSQL DDL H2 cannot parse. Such a fork should register its own {@code
+ *       FlywayConfigurationCustomizer} naming an H2 location.
 * </ul>
 *
- * <p>Local therefore verifies wiring and behaviour, not migrations. Migration and vendor-concurrency
- * fidelity stay with the real-PostgreSQL integration suites.
+ * <p>Local therefore verifies wiring and behaviour, not migrations. Migration and
+ * vendor-concurrency fidelity stay with the real-PostgreSQL integration suites.
 */
@Configuration(proxyBeanMethods = false)
@ConditionalOnProperty(
@@ -31,8 +31,8 @@ import org.springframework.transaction.support.TransactionTemplate;
 * USING}, and only an execution proves that the substitution kept the three outcomes intact.
 *
 * <p>In-memory and process-local, so this stays an ordinary unit test: no container, no network,
- * nothing to skip when Docker is absent. Real-PostgreSQL fidelity remains the job of the
- * {@code postgresqlIntegrationTest} source set.
+ * nothing to skip when Docker is absent. Real-PostgreSQL fidelity remains the job of the {@code
+ * postgresqlIntegrationTest} source set.
 */
 class H2ClaimSqlTest {

@@ -142,8 +142,9 @@ class H2ClaimSqlTest {

    List<OutboxEventEntity> claimed = claimEligible(now, 10);

-    assertThat(claimed).extracting(OutboxEventEntity::getEventId).containsExactly("evt-old",
-        "evt-other");
+    assertThat(claimed)
+        .extracting(OutboxEventEntity::getEventId)
+        .containsExactly("evt-old", "evt-other");
  }

  @Test