mindee 4.25.0 → 4.27.0

package/CHANGELOG.md

@@ -1,5 +1,18 @@
 # CHANGELOG
 
+## v4.27.0 - 2025-04-23
+### Changes
+* :sparkles: add support for RAG polling
+* :sparkles: add extras accessor from inference
+### Fixes
+* :bug: fix improper deserialization of extras in some instances
+
+
+## v4.26.0 - 2025-04-16
+### Changes
+* :sparkles: add support for rag param in workflow executions
+
+
 ## v4.25.0 - 2025-04-08
 ### Changes
 * :sparkles: add support for Financial Document V1.12

package/package.json

@@ -1,6 +1,6 @@
 {
     "name": "mindee",
-    "version": "4.25.0",
+    "version": "4.27.0",
     "description": "Mindee Client Library for Node.js",
     "main": "src/index.js",
     "bin": "bin/mindee.js",
@@ -32,7 +32,7 @@
     },
     "author": {
         "name": "Mindee",
-        "email": "devrel@mindee.com",
+        "email": "opensource@mindee.com",
         "url": "https://mindee.com/"
     },
     "bugs": {

package/src/client.d.ts

@@ -17,6 +17,10 @@ interface BaseOptions {
      * This is done before sending the file to the server and is useful to avoid page limitations.
      */
     pageOptions?: PageOptions;
+    /**
+     * If set, will enable Retrieval-Augmented Generation (only works if a valid workflowId is set).
+     */
+    rag?: boolean;
 }
 /**
  * Options relating to predictions.
@@ -34,6 +38,10 @@ export interface PredictOptions extends BaseOptions {
      * Whether to include cropper results for each page.
      */
     cropper?: boolean;
+    /**
+     * The ID of the workflow.
+     */
+    workflowId?: string;
 }
 /**
  * Options relating to workflows.

package/src/client.js

@@ -81,6 +81,9 @@ class Client {
      */
     async enqueue(productClass, inputSource, params = {}) {
         const endpoint = params?.endpoint ?? __classPrivateFieldGet(this, _Client_instances, "m", _Client_initializeOTSEndpoint).call(this, productClass);
+        if (params.workflowId) {
+            endpoint.useWorkflowId(params.workflowId);
+        }
         if (inputSource === undefined) {
             throw new Error("The 'parse' function requires an input document.");
         }
@@ -90,6 +93,8 @@ class Client {
             fullText: this.getBooleanParam(params.fullText),
             pageOptions: params?.pageOptions,
             cropper: this.getBooleanParam(params.cropper),
+            workflowId: params.workflowId,
+            rag: this.getBooleanParam(params.rag)
         });
         return new common_1.AsyncPredictResponse(productClass, rawResponse.data);
     }

package/src/http/endpoint.d.ts

@@ -15,6 +15,11 @@ export declare class Endpoint extends BaseEndpoint {
     /** Product's version, as a string. */
     version: string;
     constructor(urlName: string, owner: string, version: string, settings: ApiSettings);
+    /**
+     * Changes the url to a workflow ID.
+     * @param workflowId
+     */
+    useWorkflowId(workflowId: string): void;
     /**
      * Sends a document to the API and parses out the result.
      * Throws an error if the server's response contains one.
@@ -58,6 +63,7 @@ export declare class Endpoint extends BaseEndpoint {
      * @param includeWords
      * @param fullText
      * @param cropper
+     * @param rag
      */
-    protected sendFileForPrediction(input: InputSource, predictUrl: string, includeWords?: boolean, fullText?: boolean, cropper?: boolean): Promise<EndpointResponse>;
+    protected sendFileForPrediction(input: InputSource, predictUrl: string, includeWords?: boolean, fullText?: boolean, cropper?: boolean, rag?: boolean): Promise<EndpointResponse>;
 }

package/src/http/endpoint.js

@@ -27,6 +27,13 @@ class Endpoint extends baseEndpoint_1.BaseEndpoint {
         this.urlName = urlName;
         this.version = version;
     }
+    /**
+     * Changes the url to a workflow ID.
+     * @param workflowId
+     */
+    useWorkflowId(workflowId) {
+        this.urlRoot = `/v1/workflows/${workflowId}`;
+    }
     /**
      * Sends a document to the API and parses out the result.
      * Throws an error if the server's response contains one.
@@ -57,7 +64,7 @@ class Endpoint extends baseEndpoint_1.BaseEndpoint {
         if (params.pageOptions !== undefined) {
             await super.cutDocPages(params.inputDoc, params.pageOptions);
         }
-        const response = await __classPrivateFieldGet(this, _Endpoint_instances, "m", _Endpoint_predictAsyncReqPost).call(this, params.inputDoc, params.includeWords, params.fullText, params.cropper);
+        const response = await __classPrivateFieldGet(this, _Endpoint_instances, "m", _Endpoint_predictAsyncReqPost).call(this, params.inputDoc, params.includeWords, params.fullText, params.cropper, params.rag);
         if (!(0, responseValidation_1.isValidAsyncResponse)(response)) {
             (0, error_1.handleError)(this.urlName, response, this.extractStatusMessage(response));
         }
@@ -129,13 +136,17 @@ class Endpoint extends baseEndpoint_1.BaseEndpoint {
      * @param includeWords
      * @param fullText
      * @param cropper
+     * @param rag
      */
-    sendFileForPrediction(input, predictUrl, includeWords = false, fullText = false, cropper = false) {
+    sendFileForPrediction(input, predictUrl, includeWords = false, fullText = false, cropper = false, rag = false) {
         return new Promise((resolve, reject) => {
             const searchParams = new url_1.URLSearchParams();
             if (cropper) {
                 searchParams.append("cropper", "true");
             }
+            if (rag) {
+                searchParams.append("rag", "true");
+            }
             if (fullText) {
                 searchParams.append("full_text_ocr", "true");
             }
@@ -173,8 +184,8 @@ class Endpoint extends baseEndpoint_1.BaseEndpoint {
 exports.Endpoint = Endpoint;
 _Endpoint_instances = new WeakSet(), _Endpoint_predictReqPost = function _Endpoint_predictReqPost(input, includeWords = false, fullText = false, cropper = false) {
     return this.sendFileForPrediction(input, "predict", includeWords, fullText, cropper);
-}, _Endpoint_predictAsyncReqPost = function _Endpoint_predictAsyncReqPost(input, includeWords = false, fullText = false, cropper = false) {
-    return this.sendFileForPrediction(input, "predict_async", includeWords, fullText, cropper);
+}, _Endpoint_predictAsyncReqPost = function _Endpoint_predictAsyncReqPost(input, includeWords = false, fullText = false, cropper = false, rag = false) {
+    return this.sendFileForPrediction(input, "predict_async", includeWords, fullText, cropper, rag);
 }, _Endpoint_documentQueueReqGet = function _Endpoint_documentQueueReqGet(queueId) {
     return new Promise((resolve, reject) => {
         const options = {

package/src/http/httpParams.d.ts

@@ -4,10 +4,12 @@ interface HTTPParams {
     inputDoc: InputSource;
     fullText: boolean;
     pageOptions?: PageOptions;
+    rag?: boolean;
 }
 export interface PredictParams extends HTTPParams {
     includeWords: boolean;
     cropper: boolean;
+    workflowId?: string;
 }
 export interface WorkflowParams extends HTTPParams {
     alias?: string;

package/src/http/workflowEndpoint.d.ts

@@ -24,6 +24,7 @@ export declare class WorkflowEndpoint extends BaseEndpoint {
      * @param priority
      * @param fullText
      * @param publicUrl
+     * @param rag
      */
-    protected sendFileForPrediction(input: InputSource, alias?: string | null, priority?: ExecutionPriority | null, fullText?: boolean, publicUrl?: string | null): Promise<EndpointResponse>;
+    protected sendFileForPrediction(input: InputSource, alias?: string | null, priority?: ExecutionPriority | null, fullText?: boolean, publicUrl?: string | null, rag?: boolean | null): Promise<EndpointResponse>;
 }

package/src/http/workflowEndpoint.js

@@ -11,9 +11,9 @@ var _WorkflowEndpoint_instances, _WorkflowEndpoint_workflowReqPost;
 Object.defineProperty(exports, "__esModule", { value: true });
 exports.WorkflowEndpoint = void 0;
 const baseEndpoint_1 = require("./baseEndpoint");
+const input_1 = require("../input");
 const url_1 = require("url");
 const form_data_1 = __importDefault(require("form-data"));
-const input_1 = require("../input");
 const responseValidation_1 = require("./responseValidation");
 const error_1 = require("./error");
 /**
@@ -36,7 +36,7 @@ class WorkflowEndpoint extends baseEndpoint_1.BaseEndpoint {
         if (params.pageOptions !== undefined) {
             await super.cutDocPages(params.inputDoc, params.pageOptions);
         }
-        const response = await __classPrivateFieldGet(this, _WorkflowEndpoint_instances, "m", _WorkflowEndpoint_workflowReqPost).call(this, params.inputDoc, params.alias, params.priority, params.fullText, params.publicUrl);
+        const response = await __classPrivateFieldGet(this, _WorkflowEndpoint_instances, "m", _WorkflowEndpoint_workflowReqPost).call(this, params);
         if (!(0, responseValidation_1.isValidSyncResponse)(response)) {
             (0, error_1.handleError)(this.urlRoot, response, response.messageObj?.statusMessage);
         }
@@ -49,13 +49,17 @@ class WorkflowEndpoint extends baseEndpoint_1.BaseEndpoint {
      * @param priority
      * @param fullText
      * @param publicUrl
+     * @param rag
      */
-    sendFileForPrediction(input, alias = null, priority = null, fullText = false, publicUrl = null) {
+    sendFileForPrediction(input, alias = null, priority = null, fullText = false, publicUrl = null, rag = null) {
         return new Promise((resolve, reject) => {
             const searchParams = new url_1.URLSearchParams();
             if (fullText) {
                 searchParams.append("full_text_ocr", "true");
             }
+            if (rag) {
+                searchParams.append("rag", "true");
+            }
             const form = new form_data_1.default();
             if (input instanceof input_1.LocalInputSource && input.fileObject instanceof Buffer) {
                 form.append("document", input.fileObject, {
@@ -94,6 +98,6 @@ class WorkflowEndpoint extends baseEndpoint_1.BaseEndpoint {
     }
 }
 exports.WorkflowEndpoint = WorkflowEndpoint;
-_WorkflowEndpoint_instances = new WeakSet(), _WorkflowEndpoint_workflowReqPost = function _WorkflowEndpoint_workflowReqPost(input, alias = null, priority = null, fullText = false, publicUrl = null) {
-    return this.sendFileForPrediction(input, alias, priority, fullText, publicUrl);
+_WorkflowEndpoint_instances = new WeakSet(), _WorkflowEndpoint_workflowReqPost = function _WorkflowEndpoint_workflowReqPost(params) {
+    return this.sendFileForPrediction(params.inputDoc, params.alias, params.priority, params.fullText, params.publicUrl, params.rag);
 };

package/src/parsing/common/document.d.ts

@@ -11,7 +11,7 @@ export declare class Document<T extends Inference> {
     filename: string;
     /** Result of the base inference. */
     inference: T;
-    /** Id of the document as sent back by the server. */
+    /** ID of the document as sent back by the server. */
     id: string;
     /** Potential `Extras` fields sent back along the prediction. */
     extras?: Extras;

package/src/parsing/common/document.js

@@ -4,6 +4,7 @@ exports.Document = void 0;
 const extras_1 = require("./extras");
 const extras_2 = require("./extras/extras");
 const ocr_1 = require("./ocr");
+const ragExtra_1 = require("./extras/ragExtra");
 /**
  * Document prediction wrapper class. Holds the results of a parsed document.
  * @typeParam T an extension of an `Inference`. Mandatory in order to properly create an inference.
@@ -30,6 +31,9 @@ class Document {
                     case "full_text_ocr":
                         extras["fullTextOcr"] = new extras_1.FullTextOcrExtra(extraValue);
                         break;
+                    case "rag":
+                        extras["rag"] = new ragExtra_1.RAGExtra(extraValue);
+                        break;
                 }
             });
             this.extras = new extras_2.Extras(extras);
@@ -59,8 +63,7 @@ ${this.inference?.toString()}`;
             return;
         }
         const fullTextOcr = rawPrediction["inference"]["pages"].filter((e) => "extras" in e).map((e) => e["extras"]["full_text_ocr"]["content"]).join("\n");
-        // eslint-disable-next-line @typescript-eslint/naming-convention
-        const artificialTextObj = { "full_text_ocr": { "content": fullTextOcr.length > 0 ? fullTextOcr : "" } };
+        const artificialTextObj = { "content": fullTextOcr.length > 0 ? fullTextOcr : "" };
         if (!this.extras) {
             this.extras = new extras_2.Extras({ "fullTextOcr": new extras_1.FullTextOcrExtra(artificialTextObj) });
         }

package/src/parsing/common/extras/fullTextOcrExtra.js

@@ -5,9 +5,9 @@ const extras_1 = require("./extras");
 class FullTextOcrExtra extends extras_1.ExtraField {
     constructor(rawPrediction) {
         super();
-        if (rawPrediction["full_text_ocr"] !== undefined && rawPrediction["full_text_ocr"]["content"]) {
-            this.content = "content" in rawPrediction["full_text_ocr"] ? rawPrediction["full_text_ocr"]["content"] : "";
-            this.languages = "languages" in rawPrediction["full_text_ocr"] ? rawPrediction["full_text_ocr"]["languages"] : "";
+        if (rawPrediction["content"]) {
+            this.content = "content" in rawPrediction ? rawPrediction["content"] : "";
+            this.languages = "languages" in rawPrediction ? rawPrediction["languages"] : "";
         }
     }
     /**

package/src/parsing/common/extras/ragExtra.d.ts

@@ -0,0 +1,9 @@
+import { ExtraField } from "./extras";
+import { StringDict } from "../stringDict";
+export declare class RAGExtra extends ExtraField {
+    /**
+     * ID reference of the document matched by the Retrieval-Augmented Generation.
+     */
+    matchingDocumentId?: string;
+    constructor(rawPrediction: StringDict);
+}

package/src/parsing/common/extras/ragExtra.js

@@ -0,0 +1,13 @@
+"use strict";
+Object.defineProperty(exports, "__esModule", { value: true });
+exports.RAGExtra = void 0;
+const extras_1 = require("./extras");
+class RAGExtra extends extras_1.ExtraField {
+    constructor(rawPrediction) {
+        super();
+        if (rawPrediction["matching_document_id"]) {
+            this.matchingDocumentId = rawPrediction["matching_document_id"];
+        }
+    }
+}
+exports.RAGExtra = RAGExtra;

package/src/parsing/common/inference.d.ts

@@ -1,5 +1,5 @@
 import { StringDict } from "../common";
-import { ExtraField } from "./extras/extras";
+import { Extras } from "./extras/extras";
 import { Page } from "./page";
 import type { Prediction } from "./prediction";
 import { Product } from "./product";
@@ -20,7 +20,7 @@ export declare abstract class Inference<DocT extends Prediction = Prediction, Pa
     /** A document's top-level `Prediction`. */
     prediction: DocT;
     /** Extraneous fields relating to specific tools for some APIs. */
-    extras?: ExtraField[];
+    extras?: Extras;
     /** Name of a document's endpoint. Has a default value for OTS APIs. */
     endpointName?: string;
     /** A document's version. Has a default value for OTS APIs. */

package/src/parsing/common/inference.js

@@ -1,6 +1,8 @@
 "use strict";
 Object.defineProperty(exports, "__esModule", { value: true });
 exports.InferenceFactory = exports.Inference = void 0;
+const extras_1 = require("./extras");
+const ragExtra_1 = require("./extras/ragExtra");
 /**
  *
  * @typeParam DocT an extension of a `Prediction`. Is generic by default to
@@ -12,10 +14,26 @@ class Inference {
     constructor(rawPrediction) {
         /** Wrapper for a document's pages prediction. */
         this.pages = [];
-        /** Extraneous fields relating to specific tools for some APIs. */
-        this.extras = [];
         this.isRotationApplied = rawPrediction?.is_rotation_applied ?? undefined;
         this.product = rawPrediction?.product;
+        if (rawPrediction["extras"] &&
+            Object.keys(rawPrediction["extras"].length > 0)) {
+            const extras = {};
+            Object.entries(rawPrediction["extras"]).forEach(([extraKey, extraValue]) => {
+                switch (extraKey) {
+                    case "cropper":
+                        extras.cropper = new extras_1.CropperExtra(extraValue);
+                        break;
+                    case "full_text_ocr":
+                        extras.fullTextOcr = new extras_1.FullTextOcrExtra(extraValue);
+                        break;
+                    case "rag":
+                        extras.rag = new ragExtra_1.RAGExtra(extraValue);
+                        break;
+                }
+            });
+            this.extras = extras;
+        }
     }
     /**
      * Default string representation.

package/src/pdf/pdfCompressor.js

@@ -228,7 +228,7 @@ async function rasterizePage(pdfData, index, quality = 85) {
         return jpegBuffer;
     }
     catch (error) {
-        console.error("Error rasterizing PDF:", error);
+        logger_1.logger.error("Error rasterizing PDF:", error);
         throw error;
     }
     finally {