@ekairos/dataset 1.22.54-beta.development.0 → 1.22.55-beta.development.0

package/dist/file/file-dataset.agent.d.ts

@@ -16,7 +16,7 @@ export type FileParseContext = {
     filePreview?: FilePreviewContext;
 };
 export type FileParseContextParams = {
-    fileId: string;
+    fileId?: string;
     instructions?: string;
     sandboxId?: string;
     datasetId?: string;
@@ -71,3 +71,9 @@ export declare function createFileParseContext<Env extends {
     }>;
     context: any;
 };
+export declare function registerFileParseContext<Env extends {
+    orgId: string;
+}>(opts?: {
+    model?: string;
+    reactor?: ContextReactor<any, any>;
+}): void;

package/dist/file/file-dataset.agent.js

@@ -20,6 +20,7 @@ async function awaitContextRun(run) {
     await run;
 }
 async function initializeSandbox(runtime, sandboxId, datasetId, fileId, state) {
+    "use step";
     if (state.initialized) {
         return state.filePath;
     }
@@ -76,17 +77,26 @@ async function initializeSandbox(runtime, sandboxId, datasetId, fileId, state) {
  * Internamente corre un Context (`createContext("file.parse")`) que itera hasta que se ejecuta el tool `completeDataset`.
  */
 function createFileParseContextDefinition(params) {
-    const datasetId = params.datasetId ?? id();
+    const fallbackDatasetId = params.datasetId;
     const model = params.model ?? "openai/gpt-5";
     let contextBuilder = createContext("file.parse")
         .context(async (stored, _env, runtime) => {
         const previous = stored?.content ?? {};
         const sandboxState = previous?.sandboxState ?? { initialized: false, filePath: "" };
+        const datasetId = previous?.datasetId ?? fallbackDatasetId ?? "";
+        const fileId = previous?.fileId ?? params.fileId ?? "";
+        const instructions = previous?.instructions ?? params.instructions ?? "";
         const sandboxId = previous?.sandboxId ?? params.sandboxId ?? "";
+        if (!datasetId) {
+            throw new Error("dataset_id_required");
+        }
+        if (!fileId) {
+            throw new Error("dataset_file_id_required");
+        }
         if (!sandboxId) {
             throw new Error("dataset_sandbox_required");
         }
-        const sandboxFilePath = await initializeSandbox(runtime, sandboxId, datasetId, params.fileId, sandboxState);
+        const sandboxFilePath = await initializeSandbox(runtime, sandboxId, datasetId, fileId, sandboxState);
         let filePreview = undefined;
         try {
             filePreview = await generateFilePreview(runtime, sandboxId, sandboxFilePath, datasetId);
@@ -100,8 +110,8 @@ function createFileParseContextDefinition(params) {
             schema = datasetResult.data.schema;
         const ctx = {
             datasetId,
-            fileId: params.fileId,
-            instructions: params.instructions ?? "",
+            fileId,
+            instructions,
             sandboxConfig: { filePath: sandboxFilePath },
             analysis: [],
             schema,
@@ -114,8 +124,8 @@ function createFileParseContextDefinition(params) {
         return {
             ...previous,
             datasetId,
-            fileId: params.fileId,
-            instructions: params.instructions ?? "",
+            fileId,
+            instructions,
             sandboxId,
             sandboxState,
             ctx,
@@ -138,27 +148,36 @@ function createFileParseContextDefinition(params) {
     })
         .actions(async (_stored, _env, runtime) => {
         const existingSchema = _stored?.content?.ctx?.schema?.schema;
+        const datasetId = _stored?.content?.datasetId ?? fallbackDatasetId ?? "";
+        const fileId = _stored?.content?.fileId ?? params.fileId ?? "";
+        const sandboxId = _stored?.content?.sandboxId ?? params.sandboxId ?? "";
+        if (!datasetId)
+            throw new Error("dataset_id_required");
+        if (!fileId)
+            throw new Error("dataset_file_id_required");
+        if (!sandboxId)
+            throw new Error("dataset_sandbox_required");
         const actions = {
             executeCommand: createExecuteCommandTool({
                 datasetId,
-                sandboxId: _stored?.content?.sandboxId ?? params.sandboxId ?? "",
+                sandboxId,
                 runtime,
             }),
             completeDataset: createCompleteDatasetTool({
                 datasetId,
-                sandboxId: _stored?.content?.sandboxId ?? params.sandboxId ?? "",
+                sandboxId,
                 runtime,
             }),
             clearDataset: createClearDatasetTool({
                 datasetId,
-                sandboxId: _stored?.content?.sandboxId ?? params.sandboxId ?? "",
+                sandboxId,
                 runtime,
             }),
         };
         if (!existingSchema) {
             actions.generateSchema = createGenerateSchemaTool({
                 datasetId,
-                fileId: params.fileId,
+                fileId,
                 runtime,
             });
         }
@@ -174,7 +193,7 @@ function createFileParseContextDefinition(params) {
         contextBuilder = contextBuilder.model(model);
     }
     const context = contextBuilder.build();
-    return { datasetId, context };
+    return { datasetId: fallbackDatasetId ?? "", context };
 }
 /**
  * Factory (DX-first):
@@ -187,15 +206,16 @@ function createFileParseContextDefinition(params) {
  * - `parse()` is the entrypoint; it calls `context.react(...)` internally.
  */
 export function createFileParseContext(fileId, opts) {
+    const datasetId = opts?.datasetId ?? id();
     const params = {
         fileId,
         instructions: opts?.instructions,
         sandboxId: opts?.sandboxId,
-        datasetId: opts?.datasetId,
+        datasetId,
         model: opts?.model,
         reactor: opts?.reactor,
     };
-    const { datasetId, context } = createFileParseContextDefinition(params);
+    const { context } = createFileParseContextDefinition(params);
     return {
         datasetId,
         async parse(runtime, options = {}) {
@@ -213,6 +233,13 @@ export function createFileParseContext(fileId, opts) {
                 context: { key: `dataset:${datasetId}` },
                 durable: options.durable ?? false,
                 options: { silent: true, preventClose: true, sendFinish: false, maxIterations: 20, maxModelSteps: 5 },
+                __initialContent: {
+                    datasetId,
+                    fileId,
+                    instructions: opts?.instructions ?? "",
+                    sandboxId: opts?.sandboxId ?? "",
+                    sandboxState: { initialized: false, filePath: "" },
+                },
             });
             await awaitContextRun(shell.run);
             return { datasetId };
@@ -221,3 +248,10 @@ export function createFileParseContext(fileId, opts) {
         context,
     };
 }
+export function registerFileParseContext(opts) {
+    createFileParseContextDefinition({
+        model: opts?.model,
+        reactor: opts?.reactor,
+    }).context;
+}
+registerFileParseContext();

package/dist/transform/transform-dataset.agent.d.ts

@@ -20,8 +20,8 @@ export type TransformDatasetContext = {
     instructions?: string;
 };
 export type TransformDatasetAgentParams = {
-    sourceDatasetIds: string[];
-    outputSchema: any;
+    sourceDatasetIds?: string[];
+    outputSchema?: any;
     instructions?: string;
     datasetId?: string;
     model?: string;
@@ -62,3 +62,9 @@ export declare function createTransformDatasetContext<Env extends {
     }>;
     context: any;
 };
+export declare function registerTransformDatasetContext<Env extends {
+    orgId: string;
+}>(opts?: {
+    model?: string;
+    reactor?: ContextReactor<any, any>;
+}): void;

package/dist/transform/transform-dataset.agent.js

@@ -18,6 +18,7 @@ async function awaitContextRun(run) {
     await run;
 }
 async function ensureSourcesInSandbox(runtime, sandboxId, datasetId, sourceDatasetIds, state) {
+    "use step";
     if (state.initialized) {
         return { sourcePaths: state.sourcePaths, outputPath: getDatasetOutputPath(datasetId) };
     }
@@ -39,17 +40,34 @@ async function ensureSourcesInSandbox(runtime, sandboxId, datasetId, sourceDatas
     return { sourcePaths, outputPath: getDatasetOutputPath(datasetId) };
 }
 function createTransformDatasetContextDefinition(params) {
-    const datasetId = params.datasetId ?? id();
+    const fallbackDatasetId = params.datasetId;
     const model = params.model ?? "openai/gpt-5";
     let contextBuilder = createContext("dataset.transform")
         .context(async (stored, _env, runtime) => {
         const previous = stored?.content ?? {};
         const sandboxState = previous?.sandboxState ?? { initialized: false, sourcePaths: [] };
+        const datasetId = previous?.datasetId ?? fallbackDatasetId ?? "";
+        const sourceDatasetIds = Array.isArray(previous?.sourceDatasetIds)
+            ? previous.sourceDatasetIds
+            : Array.isArray(params.sourceDatasetIds)
+                ? params.sourceDatasetIds
+                : [];
+        const outputSchema = previous?.outputSchema ?? params.outputSchema;
+        const instructions = previous?.instructions ?? params.instructions;
         const sandboxId = previous?.sandboxId ?? params.sandboxId ?? "";
+        if (!datasetId) {
+            throw new Error("dataset_id_required");
+        }
+        if (sourceDatasetIds.length === 0) {
+            throw new Error("dataset_transform_sources_required");
+        }
+        if (!outputSchema) {
+            throw new Error("dataset_transform_schema_required");
+        }
         if (!sandboxId) {
             throw new Error("dataset_sandbox_required");
         }
-        const { sourcePaths, outputPath } = await ensureSourcesInSandbox(runtime, sandboxId, datasetId, params.sourceDatasetIds, sandboxState);
+        const { sourcePaths, outputPath } = await ensureSourcesInSandbox(runtime, sandboxId, datasetId, sourceDatasetIds, sandboxState);
         const sourcePreviews = [];
         for (const sp of sourcePaths) {
             try {
@@ -64,19 +82,19 @@ function createTransformDatasetContextDefinition(params) {
         await datasetUpdateSchemaStep({
             runtime,
             datasetId,
-            schema: params.outputSchema,
+            schema: outputSchema,
             status: "schema_complete",
         });
         const promptContext = {
             datasetId,
-            sourceDatasetIds: params.sourceDatasetIds,
-            outputSchema: params.outputSchema,
+            sourceDatasetIds,
+            outputSchema,
             sandboxConfig: { sourcePaths, outputPath },
             sourcePreviews: sourcePreviews.length > 0 ? sourcePreviews : undefined,
             errors: [],
         };
         const basePrompt = buildTransformDatasetPrompt(promptContext);
-        const userInstructions = String(params.instructions ?? "").trim();
+        const userInstructions = String(instructions ?? "").trim();
         const system = userInstructions
             ? [
                 "## USER INSTRUCTIONS",
@@ -90,6 +108,9 @@ function createTransformDatasetContextDefinition(params) {
         return {
             ...previous,
             datasetId,
+            sourceDatasetIds,
+            outputSchema,
+            instructions,
             sandboxId,
             sandboxState,
             system,
@@ -100,7 +121,12 @@ function createTransformDatasetContextDefinition(params) {
         return String(stored?.content?.system ?? "");
     })
         .actions(async (stored, _env, runtime) => {
+        const datasetId = stored?.content?.datasetId ?? fallbackDatasetId ?? "";
         const sandboxId = stored?.content?.sandboxId ?? params.sandboxId ?? "";
+        if (!datasetId)
+            throw new Error("dataset_id_required");
+        if (!sandboxId)
+            throw new Error("dataset_sandbox_required");
         return {
             executeCommand: createExecuteCommandTool({
                 datasetId,
@@ -129,14 +155,15 @@ function createTransformDatasetContextDefinition(params) {
         contextBuilder = contextBuilder.model(model);
     }
     const context = contextBuilder.build();
-    return { datasetId, context };
+    return { datasetId: fallbackDatasetId ?? "", context };
 }
 export function createTransformDatasetContext(params) {
-    const { datasetId, context } = createTransformDatasetContextDefinition({
+    const datasetId = params.datasetId ?? id();
+    const { context } = createTransformDatasetContextDefinition({
         sourceDatasetIds: params.sourceDatasetIds,
         outputSchema: params.outputSchema,
         instructions: params.instructions,
-        datasetId: params.datasetId,
+        datasetId,
         model: params.model,
         sandboxId: params.sandboxId,
         reactor: params.reactor,
@@ -167,6 +194,14 @@ export function createTransformDatasetContext(params) {
                 context: { key: `dataset:${datasetId}` },
                 durable: options.durable ?? false,
                 options: { silent: true, preventClose: true, sendFinish: false, maxIterations: 20, maxModelSteps: 5 },
+                __initialContent: {
+                    datasetId,
+                    sourceDatasetIds: params.sourceDatasetIds,
+                    outputSchema: params.outputSchema,
+                    instructions: params.instructions,
+                    sandboxId: params.sandboxId ?? "",
+                    sandboxState: { initialized: false, sourcePaths: [] },
+                },
             });
             await awaitContextRun(shell.run);
             return { datasetId };
@@ -174,3 +209,10 @@ export function createTransformDatasetContext(params) {
         context,
     };
 }
+export function registerTransformDatasetContext(opts) {
+    createTransformDatasetContextDefinition({
+        model: opts?.model,
+        reactor: opts?.reactor,
+    }).context;
+}
+registerTransformDatasetContext();

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@ekairos/dataset",
-  "version": "1.22.54-beta.development.0",
+  "version": "1.22.55-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": "^1.22.54-beta.development.0",
-    "@ekairos/events": "^1.22.54-beta.development.0",
-    "@ekairos/sandbox": "^1.22.54-beta.development.0",
+    "@ekairos/domain": "^1.22.55-beta.development.0",
+    "@ekairos/events": "^1.22.55-beta.development.0",
+    "@ekairos/sandbox": "^1.22.55-beta.development.0",
     "@instantdb/admin": "0.22.158",
     "@instantdb/core": "0.22.142",
     "ai": "^5.0.44",