@vertesia/common 0.50.1 → 0.52.0

package/src/store/doc-analyzer.ts

@@ -0,0 +1,152 @@
+import { WorkflowExecutionPayload, WorkflowRunStatus } from "./workflow.js";
+
+export interface PdfToRichtextOptions {
+    features: string[];
+    debug?: boolean;
+    [key: string]: any;
+}
+
+export interface PdfToRichTextWorkflowPayload extends Omit<WorkflowExecutionPayload, "vars"> {
+    vars: PdfToRichtextOptions;
+}
+
+/**
+ * Represents a image in a document that has been analyzed
+ */
+export interface DocImage {
+    id?: string;
+    page_number?: number;
+    description?: string;
+    is_meaningful?: boolean;
+    width?: number;
+    height?: number;
+}
+
+/**
+ * The export type formats for tables.
+ */
+export type ExportTableFormats = 'json' | 'csv' | 'xml';
+
+/**
+ * Represents a table in a document that has been analyzed
+ */
+export interface DocTable {
+    page_number?: number;
+    table_number?: number;
+    title?: string;
+    format: "application/csv" | "application/json";
+}
+
+/**
+ * Represents a table in a document that has been analyzed in CSV format
+ */
+export interface DocTableCsv extends DocTable {
+    format: "application/csv";
+    title?: string;
+    data: string;
+}
+
+/**
+ * Represents a table in a document that has been analyzed in JSON format
+ */
+export interface DocTableJson extends DocTable {
+    format: "application/json";
+    title?: string;
+    data: Object[];
+}
+
+/**
+ * Represents a document analysis run status
+ */
+export interface DocAnalyzeRunStatusResponse extends WorkflowRunStatus {
+    progress?: DocAnalyzerProgress;
+}
+
+export interface DocAnalyzerResultResponse {
+    document?: string;
+    tables?: DocTableCsv[] | DocTableJson[];
+    images?: DocImage[];
+    annotated?: string | null;
+}
+
+export interface DocAnalyzerProgress {
+    pages: DocAnalyzerProgressStatus;
+    images: DocAnalyzerProgressStatus;
+    tables: DocAnalyzerProgressStatus;
+    visuals: DocAnalyzerProgressStatus;
+    started_at?: number;
+    percent: number;
+}
+
+interface DocAnalyzerProgressStatus {
+    total: number;
+    processed: number;
+    success: number;
+    failed: number;
+}
+
+/**
+ * Adapt Tables Parameters, part of the request
+ */
+export interface AdaptTablesParams {
+    /**
+     * JSON Schema to to convert the table into
+     */
+    target_schema: string;
+
+    /**
+     * Natural language description of the type item the table are composed of
+     */
+    item_name: string;
+
+    /**
+     * Natural language description of the type of table or item to convert
+     */
+    instructions?: string;
+
+    /**
+     * Format to return the data in (csv, json)
+     */
+    format?: "csv" | "json";
+
+    /**
+     * Process the table as a whole or by page
+     */
+    process_as?: "page" | "table";
+
+    /**
+     * Process the table as a CSV file
+     */
+    process_as_csv?: boolean;
+}
+
+interface DocAnalyzerRequestBase {
+    synchroneous?: boolean;
+
+    notify_endpoints?: string[];
+
+    /**
+     * What environmenet to use to run the request
+     * If none specified the project embedded environment will be used
+     */
+    environment?: string;
+}
+
+export interface AdaptTablesRequest extends DocAnalyzerRequestBase, AdaptTablesParams {}
+
+/**
+ * Get Adapted Tables Request
+ * @param raw If true, the raw data will be returned
+ * @param format The format to return the data in (csv, json)
+ */
+export interface GetAdaptedTablesRequestQuery {
+    raw?: boolean;
+    format: "csv" | "json";
+}
+/**
+ * The adapted table result format
+ */
+export interface AdaptedTable {
+    comment?: string;
+    data: Record<string, any>[];
+}

package/src/store/index.ts

@@ -1,8 +1,11 @@
 export * from "./activity-catalog.js";
 export * from "./agent.js";
+export * from "./collections.js";
 export * from "./common.js";
+export * from "./doc-analyzer.js";
 export * from "./dsl-workflow.js";
 export * from "./object-types.js";
 export * from "./store.js";
 export * from "./temporalio.js";
 export * from "./workflow.js";
+export * from "./signals.js";

package/src/store/signals.ts

@@ -0,0 +1,4 @@
+export interface UserInputSignal {
+    message: string;
+    metadata?: Record<string, any>;
+}

package/src/store/store.ts

@@ -48,11 +48,13 @@ export interface GenerationRunMetadata {
     target?: string;
 }
 
+
+
 export interface ContentMetadata {
     // Common fields for all media types
     type?: ContentNature;
     size?: number;      // in bytes
-    language?: string;
+    languages?: string[];
     location?: Location;
     generation_runs: GenerationRunMetadata[];
     etag?: string;
@@ -81,6 +83,15 @@ export interface VideoMetadata extends TemporalMediaMetadata {
 export interface DocumentMetadata extends ContentMetadata {
     type: 'document';
     page_count?: number;
+    content_processor?: {
+        type?: string;
+        features_requested?: string[];
+        zones_requested?: string[];
+        table_count?: number;
+        image_count: number;
+        zone_count: number;
+        needs_ocr_count?: number;
+    }
 }
 
 export interface Transcript {
@@ -94,8 +105,8 @@ export interface TranscriptSegment {
     text: string
     speaker?: number
     end?: number
-    language?: string
     confidence?: number
+    language?: string
 }
 
 export interface ContentSource {
@@ -129,7 +140,6 @@ export interface ContentObjectItem<T = any> extends BaseObject {
         encoding: string; // the encoding used to calculate the tokens
         etag: string; //the etag of the text used for the token count
     };
-    run?: string; // the ID of the interaction run that created the object
 }
 
 /**
@@ -176,20 +186,19 @@ export interface ColumnLayout {
     default?: any;
 }
 export interface ContentObjectType extends ContentObjectTypeItem {
-    object_schema?: Record<string, any>; // an optional JSON schema for the object properties.
-    table_layout?: ColumnLayout[]; // an optional table layout for the object properties.
 }
 export interface ContentObjectTypeItem extends BaseObject {
     is_chunkable?: boolean;
-}
-/**
- * Used to list types with their table layout if any
- */
-export interface ContentObjectTypeLayout {
-    id: string;
-    name: string;
-    description?: string;
+    /**
+     * This is only included in ContentObjectTypeItem if explicitely requested
+     * It is always included in ContentObjectType
+     */
     table_layout?: ColumnLayout[];
+    /**
+     * this is only included in ContentObjectTypeItem if explicitely requested
+     * It is always included in ContentObjectType
+     */
+    object_schema?: Record<string, any>; // an optional JSON schema for the object properties.
 }
 
 export interface CreateContentObjectTypePayload extends Omit<ContentObjectType, 'id' | 'created_at' | 'updated_at' | 'created_by' | 'updated_by'> {

package/src/store/workflow.ts

@@ -1,4 +1,3 @@
-
 export enum ContentEventName {
     create = "create",
     change_type = "change_type",
@@ -6,19 +5,14 @@ export enum ContentEventName {
     delete = "delete",
     workflow_finished = "workflow_finished",
     workflow_execution_request = "workflow_execution_request",
+    api_request = "api_request",
 }
 
-
-export interface WorkflowExecutionPayload {
-    /**
-     * The event which started the workflow who created the activity.
-     */
-    event: ContentEventName;
-
+export interface WorkflowExecutionBaseParams<T = Record<string, any>> {
     /**
      * The ref of the user who initiated the workflow.
      */
-    initiated_by: string;
+    initiated_by?: string;
 
     /**
      * The account ID of the user who created the activity.
@@ -31,17 +25,6 @@ export interface WorkflowExecutionPayload {
      */
     project_id: string;
 
-    /**
-     * The Unix timestamp when the workflow was started.
-     */
-    timestamp: number;
-
-    /*
-     * The Workflow Rule ID if any. If the workflow was started by a rule this field will contain the rule ID
-     * otherwise if the workflow was started on demand the property will be undefined.
-     */
-    wf_rule_name?: string;
-
     /**
      * The vars field is mainly used to pass the user input to the workflow.
      * The user input ar custom user options that can be used to configure the workflow.
@@ -51,17 +34,12 @@ export interface WorkflowExecutionPayload {
      *
      * In case of dsl workflows the worflow execution payload vars will be applied over the default vars values stored in the DSL vars field.
      */
-    vars: Record<string, any>;
+    vars: T;
 
     /**
      * Auth Token to access Zeno and Composable from the workers
      */
-    auth_token: string;
-
-    /**
-     * The ID of the target objects processed by the workflow.
-     */
-    objectIds: string[];
+    auth_token?: string;
 
     /**
      * The configuration for the workflow execution.
@@ -69,15 +47,46 @@ export interface WorkflowExecutionPayload {
     config?: {
         studio_url: string;
         store_url: string;
+        slack_app_url?: string;
         enabled_integrations?: string[]; //list of enabled integrations
-    }
+    };
+
+    /**
+     * The list of endpoints to notify when the workflow finishes.
+     * It is handled by a subworkflow execution, so the main workflow will not wait for the notification to be sent.
+     */
+    notify_endpoints?: string[];
 }
 
+export interface WorkflowExecutionPayload<T = Record<string, any>> extends WorkflowExecutionBaseParams<T> {
+    /**
+     * The event which started the workflow who created the activity.
+     */
+    event: ContentEventName;
+
+    /*
+     * The Workflow Rule ID if any. If the workflow was started by a rule this field will contain the rule ID
+     * otherwise if the workflow was started on demand the property will be undefined.
+     */
+    wf_rule_name?: string;
+
+    /**
+     * The ID of the target objects processed by the workflow.
+     */
+    objectIds: string[];
+
+    /**
+     * Auth Token to access Zeno and Composable from the workers
+     */
+    auth_token: string;
+}
 
 export interface ExecuteWorkflowPayload {
     task_queue?: string;
     objectIds?: string[];
     vars?: Record<string, any>;
+    unique?: boolean;
+    timeout?: number; //timeout in seconds
 }
 
 export interface ListWorkflowRunsPayload {
@@ -88,6 +97,7 @@ export interface ListWorkflowRunsPayload {
     end?: string;
     status?: string;
     search_term?: string;
+    initiated_by?: string;
 }
 
 interface WorkflowRunEvent {
@@ -101,7 +111,7 @@ interface WorkflowRunEvent {
         name: string;
         id: string;
         input?: any;
-    }
+    };
 
     error?: {
         message: string;
@@ -110,19 +120,20 @@ interface WorkflowRunEvent {
         type?: string;
     };
 
-    result?: any
-
+    result?: any;
 }
 
 export interface WorkflowRun {
-    status?: WorkflowExecutionStatus | string,
-    type?: string,
-    started_at?: number,
-    closed_at?: number,
-    execution_duration?: number,
-    run_id?: string,
-    workflow_id?: string,
-    raw?: any
+    status?: WorkflowExecutionStatus | string;
+    type?: string;
+    started_at?: number;
+    closed_at?: number;
+    execution_duration?: number;
+    run_id?: string;
+    workflow_id?: string;
+    initiated_by?: string;
+    result?: any;
+    raw?: any;
 }
 
 export interface WorkflowRunWithDetails extends WorkflowRun {
@@ -132,12 +143,12 @@ export interface ListWorkflowRunsResponse {
     runs: WorkflowRun[];
 }
 
-export interface MultiDocumentsInteractionParams extends Omit<WorkflowExecutionPayload, 'config'> {
+export interface MultiDocumentsInteractionParams extends Omit<WorkflowExecutionPayload, "config"> {
     config: {
         interactionName: string;
         action: DocumentActionConfig;
         data: Record<string, any>;
-    }
+    };
 }
 
 export interface DocumentActionConfig {
@@ -159,4 +170,38 @@ export enum WorkflowExecutionStatus {
     TERMINATED = 5,
     CONTINUED_AS_NEW = 6,
     TIMED_OUT = 7,
-}
+}
+
+/**
+ * Basic response for anything run with an async workflow
+ */
+export interface WorkflowRunStatus {
+    workflow_id: string | null;
+    workflow_run_id: string | null;
+    status: WorkflowExecutionStatus;
+}
+
+/**
+ * Workflow Update Message
+ */
+export interface AgentMessage {
+    timestamp: number;
+    workflow_run_id: string;
+    type: AgentMessageType;
+    message?: string;
+    details?: any;
+}
+
+export enum AgentMessageType {
+    SYSTEM = "system",
+    THOUGHT = "thought",
+    PLAN = "plan",
+    UPDATE = "update",
+    COMPLETE = "complete",
+    WARNING = "warning",
+    ERROR = "error",
+    ANSWER = "answer",
+    QUESTION = "question",
+    REQUEST_INPUT = "request_input",
+    IDLE = "idle",
+}

package/src/user.ts

@@ -31,6 +31,16 @@ export enum Datacenters {
     azure = 'azure'
 }
 
+export enum BillingMethod {
+    stripe = 'stripe',
+    invoice ='invoice'
+}
+
+
+interface AccountBilling {
+    method: BillingMethod;
+    stripe_customer_id?: string;
+}
 export interface Account {
     id: string;
     name: string;
@@ -49,6 +59,8 @@ export interface Account {
 
     datacenter: string;
 
+    billing: AccountBilling;
+
     created_by: string;
     updated_by: string;
     created_at: string;
@@ -58,6 +70,7 @@ export interface Account {
 export interface UpdateAccountPayload {
     name?: string;
     email_domains?: string[];
+    billing?: AccountBilling;
 }
 
 export interface AccountRef {
@@ -94,11 +107,9 @@ export interface UserSessionInfo extends SessionInfo<User> { }
 export interface ApiKeySessionInfo extends SessionInfo<ApiKey> { }
 
 export interface OnboardingProgress {
-    projects: boolean,
     interactions: boolean,
     prompts: boolean,
     environments: boolean,
-    runs: boolean;
     default_environment_defined: boolean
 }
 

package/src/utils/advanced-content-processing.ts

@@ -0,0 +1,32 @@
+import mime from 'mime';
+
+
+const ADVANCED_PROCESSING_PREFIX = 'magic-pdf';
+
+export interface AdvancedProcessingObjectFileParams {
+    objectId: string;
+    fileId: string;
+    prefix?: string;
+    contentType?: string;
+    path?: any;
+}
+
+/**
+ * 
+ * Advanced content processing only
+ * generate a file path based on object ID and file role
+ * 
+ * @returns 
+ */
+export function getPathForAdvancedProcessingFile({ objectId, prefix, fileId, contentType }: AdvancedProcessingObjectFileParams) {
+
+    const extension = contentType ? `.${mime.getExtension(contentType)}` : '';
+    const p = prefix ? `${prefix}/` : '';
+
+    let path = `${ADVANCED_PROCESSING_PREFIX}/${objectId}/${p}${fileId}`;
+    if (!path.endsWith(extension)) {
+        path += extension;
+    }
+    
+    return path;
+}

package/src/utils/schemas.ts

@@ -1,8 +1,25 @@
-import { PromptRole } from "@llumiverse/core";
+import { JSONSchema, PromptRole } from "@llumiverse/core";
 import { JSONSchema4 } from "json-schema";
 import { InteractionRefWithSchema, PopulatedInteraction } from "../interaction.js";
 import { PopulatedPromptSegmentDef, PromptSegmentDef, PromptSegmentDefType, PromptTemplateRefWithSchema } from "../prompt.js";
 
+
+export function mergeJSONSchemas(schemas: JSONSchema[]) {
+    const props: Record<string, JSONSchema4> = {};
+    let required: string[] = [];
+    for (const schema of schemas) {
+        if (schema.properties) {
+            if (schema.required) {
+                for (const prop of schema.required as string[]) {
+                    if (!required.includes(prop)) required.push(prop);
+                }
+            }
+            Object.assign(props, schema.properties);
+        }
+    }
+    return Object.keys(props).length > 0 ? { properties: props, required } as JSONSchema : null;
+}
+
 export function _mergePromptsSchema(prompts: PromptSegmentDef<PromptTemplateRefWithSchema>[] | PopulatedPromptSegmentDef[]) {
     const props: Record<string, JSONSchema4> = {};
     let required: string[] = [];