test(clickhouse): seed nested metadata (#13635)

maxdeichmann · web-flow · commit da8a4f54c92b · 2026-05-14T10:20:06.000Z
diff --git a/packages/shared/scripts/seeder/utils/clickhouse-builder.ts b/packages/shared/scripts/seeder/utils/clickhouse-builder.ts
@@ -24,6 +24,21 @@ export class ClickHouseQueryBuilder {
     return str.replace(/'/g, "''");
   }
 
+  private buildNestedMetadataMapSql(
+    baseEntries: string[],
+    rowExpression: string = "number",
+  ): string {
+    return `map(
+          ${baseEntries.join(",\n          ")},
+          'customer.id', concat('customer_', toString(${rowExpression} % 100)),
+          'customer.plan', arrayElement(['free', 'pro', 'enterprise'], 1 + (${rowExpression} % 3)),
+          'customer.region.code', arrayElement(['eu-central-1', 'us-east-1', 'ap-south-1'], 1 + (${rowExpression} % 3)),
+          'routing.queue', arrayElement(['support-chat', 'sales-chat', 'ops-chat'], 1 + (${rowExpression} % 3)),
+          'routing.priority', arrayElement(['low', 'normal', 'high'], 1 + (${rowExpression} % 3)),
+          'flags.beta', if(${rowExpression} % 2 = 0, 'true', 'false')
+        )`;
+  }
+
   /**
    * Creates INSERT query for trace data using VALUES syntax.
    * Use for: Small datasets, detailed trace objects with all fields populated.
@@ -93,7 +108,7 @@ export class ClickHouseQueryBuilder {
         toDateTime(now() - randUniform(0, ${opts.numberOfDays} * 24 * 60 * 60)) AS timestamp,
         concat('trace-', toString(number % 10)) AS name,
         if(randUniform(0, 1) < 0.3, concat('user_', toString(rand() % 1000)), NULL) AS user_id,
-        map('generated', 'bulk') AS metadata,
+        ${this.buildNestedMetadataMapSql(["'generated'", "'bulk'"])} AS metadata,
         NULL AS release,
         NULL AS version,
         '${projectId}' AS project_id,
@@ -162,7 +177,7 @@ export class ClickHouseQueryBuilder {
           when type = 'SPAN' then concat('span-', toString(number % 10))
           else concat('event-', toString(number % 10))
         end AS name,
-        map('key', 'value') AS metadata,
+        ${this.buildNestedMetadataMapSql(["'key'", "'value'"])} AS metadata,
         if(randUniform(0, 1) < 0.85, 'DEFAULT', if(randUniform(0, 1) < 0.7, 'DEBUG', if(randUniform(0, 1) < 0.3, 'ERROR', 'WARNING'))) AS level,
         NULL AS status_message,
         NULL AS version,
diff --git a/packages/shared/scripts/seeder/utils/data-generators.ts b/packages/shared/scripts/seeder/utils/data-generators.ts
@@ -72,6 +72,28 @@ export class DataGenerator {
     return Math.floor(Math.random() * (max - min + 1)) + min;
   }
 
+  private buildNestedSeedMetadata(
+    source: string,
+    index: number,
+    overrides: Record<string, string> = {},
+  ): Record<string, string> {
+    const plans = ["free", "pro", "enterprise"];
+    const regions = ["eu-central-1", "us-east-1", "ap-south-1"];
+    const queues = ["support-chat", "sales-chat", "ops-chat"];
+    const priorities = ["low", "normal", "high"];
+
+    return {
+      source,
+      "customer.id": `customer_${index % 100}`,
+      "customer.plan": plans[index % plans.length],
+      "customer.region.code": regions[index % regions.length],
+      "routing.queue": queues[index % queues.length],
+      "routing.priority": priorities[index % priorities.length],
+      "flags.beta": index % 2 === 0 ? "true" : "false",
+      ...overrides,
+    };
+  }
+
   /**
    * Creates dataset run items for dataset runs.
    * Use for: Dataset experiment scenarios.
@@ -299,7 +321,9 @@ export class DataGenerator {
           ? `session_${this.randomInt(1, 100)}`
           : undefined,
         environment: "default",
-        metadata: { generated: "synthetic" },
+        metadata: this.buildNestedSeedMetadata("synthetic", i, {
+          generated: "synthetic",
+        }),
         tags: this.randomBoolean(0.3) ? ["production", "ai-agent"] : [],
         public: this.randomBoolean(0.8),
         bookmarked: this.randomBoolean(0.1),
@@ -549,6 +573,14 @@ export class DataGenerator {
                 ? "WARNING"
                 : "ERROR",
           environment: trace.environment,
+          metadata: this.buildNestedSeedMetadata(
+            "synthetic-observation",
+            traceIndex * observationsPerTrace + i,
+            {
+              "observation.type": obsType,
+              "workflow.step": String(i + 1),
+            },
+          ),
         });
 
         observations.push(observation);
@@ -1039,7 +1071,10 @@ export class DataGenerator {
       timestamp: now + index * 1000,
       name: "SupportChatSession",
       user_id: null,
-      metadata: { scenario: "support-chat" },
+      metadata: this.buildNestedSeedMetadata("support-chat", index, {
+        scenario: "support-chat",
+        "routing.queue": "membership-support",
+      }),
       release: null,
       version: null,
       project_id: projectId,
