@ekairos/dataset 1.22.51-beta.development.0 → 1.22.52-beta.development.0

package/dist/builder/materialize.js

@@ -6,22 +6,40 @@ import { buildFileDefaultInstructions, buildRawSourceInstructions, buildTransfor
 import { createOrUpdateDatasetMetadata, uploadInlineTextSource, } from "./persistence.js";
 import { getDomainDescriptor } from "./sourceRows.js";
 import { materializeQuerySource } from "./materializeQuery.js";
+import { createDatasetSandboxStep } from "../sandbox/steps.js";
 function makeIntermediateDatasetId(targetDatasetId, sourceKind, index) {
     return `${targetDatasetId}__${sourceKind}_${index}`;
 }
+async function resolveDatasetSandboxId(state, targetDatasetId) {
+    const sandboxId = String(state.sandboxId ?? "").trim();
+    if (sandboxId)
+        return sandboxId;
+    const created = await createDatasetSandboxStep({
+        runtime: state.runtime,
+        provider: "vercel",
+        sandboxRuntime: "python3.13",
+        timeoutMs: 20 * 60 * 1000,
+        resources: { vcpus: 2 },
+        purpose: "dataset.materialize",
+        params: { datasetId: targetDatasetId },
+        vercel: {
+            profile: "ephemeral",
+            deleteOnStop: true,
+        },
+    });
+    return created.sandboxId;
+}
 export async function materializeSingleFileLikeSource(state, source, targetDatasetId) {
     if (!state.reactor) {
         throw new Error("dataset_reactor_required");
     }
-    if (!state.sandboxId) {
-        throw new Error("dataset_sandbox_required");
-    }
+    const sandboxId = await resolveDatasetSandboxId(state, targetDatasetId);
     const fileId = source.kind === "file"
         ? source.fileId
         : await uploadInlineTextSource(state.runtime, targetDatasetId, source);
     await createOrUpdateDatasetMetadata(state.runtime, {
         datasetId: targetDatasetId,
-        sandboxId: state.sandboxId,
+        sandboxId,
         title: state.title ?? targetDatasetId,
         instructions: state.instructions,
         sources: [
@@ -42,7 +60,7 @@ export async function materializeSingleFileLikeSource(state, source, targetDatas
         datasetId: targetDatasetId,
         instructions: state.instructions ?? buildFileDefaultInstructions(state.outputSchema),
         reactor: state.reactor,
-        sandboxId: state.sandboxId,
+        sandboxId,
     });
     await parseContext.parse(state.runtime, { durable: state.durable });
     if (!state.outputSchema) {
@@ -84,14 +102,13 @@ export async function materializeDerivedDataset(state, targetDatasetId) {
     if (!state.reactor) {
         throw new Error("dataset_reactor_required");
     }
-    if (!state.sandboxId) {
-        throw new Error("dataset_sandbox_required");
-    }
+    const sandboxId = await resolveDatasetSandboxId(state, targetDatasetId);
+    const stateWithSandbox = { ...state, sandboxId };
     const normalizedSources = [];
-    for (let index = 0; index < state.sources.length; index++) {
-        normalizedSources.push(await normalizeSourceToDatasetId(state, state.sources[index], targetDatasetId, index));
+    for (let index = 0; index < stateWithSandbox.sources.length; index++) {
+        normalizedSources.push(await normalizeSourceToDatasetId(stateWithSandbox, stateWithSandbox.sources[index], targetDatasetId, index));
     }
-    const transformSchema = state.outputSchema ??
+    const transformSchema = stateWithSandbox.outputSchema ??
         {
             title: "DatasetRow",
             description: "One dataset row",
@@ -101,12 +118,12 @@ export async function materializeDerivedDataset(state, targetDatasetId) {
                 properties: {},
             },
         };
-    await createOrUpdateDatasetMetadata(state.runtime, {
+    await createOrUpdateDatasetMetadata(stateWithSandbox.runtime, {
         datasetId: targetDatasetId,
-        sandboxId: state.sandboxId,
-        title: state.title ?? targetDatasetId,
-        instructions: state.instructions,
-        sources: state.sources.map((source) => source.kind === "query"
+        sandboxId,
+        title: stateWithSandbox.title ?? targetDatasetId,
+        instructions: stateWithSandbox.instructions,
+        sources: stateWithSandbox.sources.map((source) => source.kind === "query"
             ? {
                 kind: "query",
                 query: source.query,
@@ -115,29 +132,29 @@ export async function materializeDerivedDataset(state, targetDatasetId) {
                 ...getDomainDescriptor(source.domain),
             }
             : source),
-        sourceKinds: state.sources.map((source) => source.kind),
+        sourceKinds: stateWithSandbox.sources.map((source) => source.kind),
         schema: transformSchema,
         status: "building",
     });
     const transformContext = createTransformDatasetContext({
         sourceDatasetIds: normalizedSources,
         outputSchema: transformSchema,
-        instructions: buildTransformInstructions(normalizedSources.length, state.instructions, state.outputSchema),
+        instructions: buildTransformInstructions(normalizedSources.length, stateWithSandbox.instructions, stateWithSandbox.outputSchema),
         datasetId: targetDatasetId,
-        reactor: state.reactor,
-        sandboxId: state.sandboxId,
+        reactor: stateWithSandbox.reactor,
+        sandboxId,
     });
-    await transformContext.transform(state.runtime, { durable: state.durable });
-    if (!state.outputSchema) {
+    await transformContext.transform(stateWithSandbox.runtime, { durable: stateWithSandbox.durable });
+    if (!stateWithSandbox.outputSchema) {
         await datasetInferAndUpdateSchemaStep({
-            runtime: state.runtime,
+            runtime: stateWithSandbox.runtime,
             datasetId: targetDatasetId,
             title: `${targetDatasetId}Row`,
             description: "One dataset row",
         });
     }
-    if (state.first) {
-        await datasetReadOneStep({ runtime: state.runtime, datasetId: targetDatasetId });
+    if (stateWithSandbox.first) {
+        await datasetReadOneStep({ runtime: stateWithSandbox.runtime, datasetId: targetDatasetId });
     }
     return targetDatasetId;
 }

package/dist/dataset.js

@@ -128,18 +128,12 @@ export function dataset(runtime, options = {}) {
                 return finalizeOutputResult(await finalizeBuildResult(effectiveState.runtime, targetDatasetId, effectiveState.first), effectiveState.output);
             }
             if (isSingleSource && (onlySource.kind === "file" || onlySource.kind === "text")) {
-                if (!effectiveState.sandboxId) {
-                    throw new Error("dataset_sandbox_required");
-                }
                 if (!effectiveState.reactor) {
                     throw new Error("dataset_reactor_required");
                 }
                 await getDatasetAgentMaterializers().materializeSingleFileLikeSource(effectiveState, onlySource, targetDatasetId);
                 return finalizeOutputResult(await finalizeBuildResult(effectiveState.runtime, targetDatasetId, effectiveState.first), effectiveState.output);
             }
-            if (!effectiveState.sandboxId) {
-                throw new Error("dataset_sandbox_required");
-            }
             if (!effectiveState.reactor) {
                 throw new Error("dataset_reactor_required");
             }

package/dist/sandbox/steps.d.ts

@@ -1,13 +1,7 @@
+import { type SandboxConfig } from "@ekairos/sandbox";
 export type DatasetSandboxId = string;
-export type CreateDatasetSandboxParams = {
+export type CreateDatasetSandboxParams = Pick<SandboxConfig, "provider" | "timeoutMs" | "ports" | "resources" | "purpose" | "params" | "env" | "domain" | "dataset" | "vercel"> & {
     sandboxRuntime?: string;
-    timeoutMs?: number;
-    ports?: number[];
-    resources?: {
-        vcpus?: number;
-    };
-    purpose?: string;
-    params?: Record<string, any>;
 };
 export type DatasetSandboxRunCommandResult = {
     exitCode: number;

package/dist/sandbox/steps.js

@@ -72,7 +72,10 @@ export async function createDatasetSandboxStep(params) {
     }
     const db = await getRuntimeDb(params.runtime);
     const service = new SandboxService(db);
-    const sandboxParams = { ...params, runtime: params.sandboxRuntime };
+    const sandboxParams = {
+        ...params,
+        ...(params.sandboxRuntime ? { runtime: params.sandboxRuntime } : {}),
+    };
     delete sandboxParams.sandboxRuntime;
     const created = await service.createSandbox(sandboxParams);
     if (!created.ok)

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@ekairos/dataset",
-  "version": "1.22.51-beta.development.0",
+  "version": "1.22.52-beta.development.0",
   "description": "Pulzar Dataset Tools",
   "type": "module",
   "main": "dist/index.js",
@@ -65,9 +65,9 @@
     "test:ai-sdk:instant": "vitest run -c vitest.codex.config.mts src/tests/materializeDataset.ai-sdk.instant.test.ts"
   },
   "dependencies": {
-    "@ekairos/domain": "^1.22.51-beta.development.0",
-    "@ekairos/events": "^1.22.51-beta.development.0",
-    "@ekairos/sandbox": "^1.22.51-beta.development.0",
+    "@ekairos/domain": "^1.22.52-beta.development.0",
+    "@ekairos/events": "^1.22.52-beta.development.0",
+    "@ekairos/sandbox": "^1.22.52-beta.development.0",
     "@instantdb/admin": "0.22.158",
     "@instantdb/core": "0.22.142",
     "ai": "^5.0.44",