mindee 4.31.0 → 4.32.0

package/CHANGELOG.md

@@ -1,5 +1,15 @@
 # CHANGELOG
 
+## v4.32.0 - 2025-10-13
+### Changes
+* :sparkles: add getting page count of a local source
+
+
+## v4.31.1 - 2025-10-08
+### Fixes
+* :bug: make sure input sources are only initialized once
+
+
 ## v4.31.0 - 2025-10-07
 ### Changes
 * :sparkles: allow comparing v2 field confidence

package/README.md

@@ -1,4 +1,4 @@
-[![License: MIT](https://img.shields.io/github/license/mindee/mindee-api-nodejs)](https://opensource.org/licenses/MIT) [![GitHub Workflow Status](https://img.shields.io/github/actions/workflow/status/mindee/mindee-api-nodejs/test.yml)](https://github.com/mindee/mindee-api-nodejs) [![NPM Version](https://img.shields.io/npm/v/mindee)](https://www.npmjs.com/package/mindee) [![Downloads](https://img.shields.io/npm/dm/mindee)](https://www.npmjs.com/package/mindee)
+[![License: MIT](https://img.shields.io/github/license/mindee/mindee-api-nodejs)](https://opensource.org/licenses/MIT) [![GitHub Workflow Status](https://img.shields.io/github/actions/workflow/status/mindee/mindee-api-nodejs/push-main-branch.yml)](https://github.com/mindee/mindee-api-nodejs) [![NPM Version](https://img.shields.io/npm/v/mindee)](https://www.npmjs.com/package/mindee) [![Downloads](https://img.shields.io/npm/dm/mindee)](https://www.npmjs.com/package/mindee)
 
 # Mindee API Helper Library for Node.js
 Quickly and easily connect to Mindee's API services using Node.js.

package/package.json

@@ -1,6 +1,6 @@
 {
     "name": "mindee",
-    "version": "4.31.0",
+    "version": "4.32.0",
     "description": "Mindee Client Library for Node.js",
     "main": "src/index.js",
     "bin": "bin/mindee.js",
@@ -44,16 +44,16 @@
         "@types/mocha": "^10.0.10",
         "@types/node": "^18.15.13",
         "@types/tmp": "^0.2.6",
-        "@typescript-eslint/eslint-plugin": "^8.40.0",
-        "@typescript-eslint/parser": "^8.40.0",
-        "chai": "^4.3.10",
-        "eslint": "^9.15.0",
-        "eslint-plugin-jsdoc": "^50.5.0",
-        "mocha": "^11.7.1",
+        "@typescript-eslint/eslint-plugin": "^8.42.1",
+        "@typescript-eslint/parser": "^8.42.1",
+        "chai": "^4.4.1",
+        "eslint": "^9.20.1",
+        "eslint-plugin-jsdoc": "^50.6.17",
+        "mocha": "^11.7.4",
         "nock": "^13.5.6",
         "ts-node": "^10.9.2",
-        "typedoc": "~0.28.7",
-        "typescript": "^5.6.3"
+        "typedoc": "~0.28.14",
+        "typescript": "^5.7.3"
     },
     "dependencies": {
         "@cantoo/pdf-lib": "^2.3.2",

package/src/clientV2.js

@@ -46,9 +46,7 @@ class ClientV2 {
         if (inputSource === undefined) {
             throw new Error("The 'enqueue' function requires an input document.");
         }
-        if (!inputSource.isInitialized()) {
-            await inputSource.init();
-        }
+        await inputSource.init();
         return await this.mindeeApi.reqPostInferenceEnqueue(inputSource, params);
     }
     /**

package/src/input/sources/base64Input.js

@@ -3,6 +3,7 @@ Object.defineProperty(exports, "__esModule", { value: true });
 exports.Base64Input = void 0;
 const localInputSource_1 = require("./localInputSource");
 const inputSource_1 = require("./inputSource");
+const logger_1 = require("../../logger");
 class Base64Input extends localInputSource_1.LocalInputSource {
     constructor({ inputString, filename }) {
         super({
@@ -13,6 +14,10 @@ class Base64Input extends localInputSource_1.LocalInputSource {
         this.inputString = inputString;
     }
     async init() {
+        if (this.initialized) {
+            return;
+        }
+        logger_1.logger.debug("Loading from base64");
         this.fileObject = Buffer.from(this.inputString, "base64");
         this.mimeType = await this.checkMimetype();
         // clear out the string

package/src/input/sources/bufferInput.js

@@ -3,6 +3,7 @@ Object.defineProperty(exports, "__esModule", { value: true });
 exports.BufferInput = void 0;
 const localInputSource_1 = require("./localInputSource");
 const inputSource_1 = require("./inputSource");
+const logger_1 = require("../../logger");
 class BufferInput extends localInputSource_1.LocalInputSource {
     constructor({ buffer, filename }) {
         super({
@@ -12,6 +13,10 @@ class BufferInput extends localInputSource_1.LocalInputSource {
         this.filename = filename;
     }
     async init() {
+        if (this.initialized) {
+            return;
+        }
+        logger_1.logger.debug("Loading from buffer");
         this.mimeType = await this.checkMimetype();
         this.initialized = true;
     }

package/src/input/sources/bytesInput.js

@@ -3,6 +3,7 @@ Object.defineProperty(exports, "__esModule", { value: true });
 exports.BytesInput = void 0;
 const inputSource_1 = require("./inputSource");
 const localInputSource_1 = require("./localInputSource");
+const logger_1 = require("../../logger");
 class BytesInput extends localInputSource_1.LocalInputSource {
     constructor({ inputBytes, filename }) {
         super({
@@ -13,6 +14,10 @@ class BytesInput extends localInputSource_1.LocalInputSource {
         this.inputBytes = inputBytes;
     }
     async init() {
+        if (this.initialized) {
+            return;
+        }
+        logger_1.logger.debug("Loading from bytes");
         this.fileObject = Buffer.from(this.inputBytes);
         this.mimeType = await this.checkMimetype();
         this.inputBytes = new Uint8Array(0);

package/src/input/sources/localInputSource.d.ts

@@ -11,6 +11,16 @@ export declare abstract class LocalInputSource extends InputSource {
      */
     protected constructor({ inputType }: InputConstructor);
     protected checkMimetype(): Promise<string>;
+    /**
+     * Returns the file object as a Buffer.
+     * @returns Buffer representation of the file object
+     * @protected
+     */
+    protected getBuffer(): Buffer;
+    /**
+     * Determines whether the current file is a PDF.
+     * @returns {boolean} Returns true if the file is a PDF; otherwise, returns false.
+     */
     isPdf(): boolean;
     /**
      * Cut PDF pages.
@@ -41,4 +51,10 @@ export declare abstract class LocalInputSource extends InputSource {
      * @return boolean
      */
     hasSourceText(): Promise<boolean>;
+    /**
+     * Returns the number of pages in the input source.
+     * For PDFs, returns the actual page count. For images, returns 1.
+     * @return Promise<number> The number of pages
+     */
+    getPageCount(): Promise<number>;
 }

package/src/input/sources/localInputSource.js

@@ -104,6 +104,21 @@ class LocalInputSource extends inputSource_1.InputSource {
         logger_1.logger.debug(`File is of type: ${mimeType}`);
         return mimeType;
     }
+    /**
+     * Returns the file object as a Buffer.
+     * @returns Buffer representation of the file object
+     * @protected
+     */
+    getBuffer() {
+        if (typeof this.fileObject === "string") {
+            return Buffer.from(this.fileObject);
+        }
+        return this.fileObject;
+    }
+    /**
+     * Determines whether the current file is a PDF.
+     * @returns {boolean} Returns true if the file is a PDF; otherwise, returns false.
+     */
     isPdf() {
         if (!this.initialized) {
             throw new Error("The `init()` method must be called before calling `isPdf()`.");
@@ -115,13 +130,9 @@ class LocalInputSource extends inputSource_1.InputSource {
      * @param pageOptions
      */
     async applyPageOptions(pageOptions) {
-        if (!this.initialized) {
-            await this.init();
-        }
-        if (!(this.fileObject instanceof Buffer)) {
-            throw new Error(`Cannot modify an input source of type ${this.inputType}.`);
-        }
-        const processedPdf = await (0, pdf_2.extractPages)(this.fileObject, pageOptions);
+        await this.init();
+        const buffer = this.getBuffer();
+        const processedPdf = await (0, pdf_2.extractPages)(buffer, pageOptions);
         this.fileObject = processedPdf.file;
     }
     /**
@@ -145,16 +156,8 @@ class LocalInputSource extends inputSource_1.InputSource {
      * @returns A Promise that resolves when the compression is complete.
      */
     async compress(quality = 85, maxWidth = null, maxHeight = null, forceSourceText = false, disableSourceText = true) {
-        if (!this.initialized) {
-            await this.init();
-        }
-        let buffer;
-        if (typeof this.fileObject === "string") {
-            buffer = Buffer.from(this.fileObject);
-        }
-        else {
-            buffer = this.fileObject;
-        }
+        await this.init();
+        const buffer = this.getBuffer();
         if (this.isPdf()) {
             this.fileObject = await (0, pdf_1.compressPdf)(buffer, quality, forceSourceText, disableSourceText);
         }
@@ -167,14 +170,25 @@ class LocalInputSource extends inputSource_1.InputSource {
      * @return boolean
      */
     async hasSourceText() {
-        if (!this.initialized) {
-            await this.init();
-        }
+        await this.init();
         if (!this.isPdf()) {
             return false;
         }
-        const buffer = typeof this.fileObject === "string" ? Buffer.from(this.fileObject) : this.fileObject;
+        const buffer = this.getBuffer();
         return (0, pdf_2.hasSourceText)(buffer);
     }
+    /**
+     * Returns the number of pages in the input source.
+     * For PDFs, returns the actual page count. For images, returns 1.
+     * @return Promise<number> The number of pages
+     */
+    async getPageCount() {
+        await this.init();
+        if (!this.isPdf()) {
+            return 1;
+        }
+        const buffer = this.getBuffer();
+        return (0, pdf_1.countPages)(buffer);
+    }
 }
 exports.LocalInputSource = LocalInputSource;

package/src/input/sources/pathInput.js

@@ -19,7 +19,10 @@ class PathInput extends localInputSource_1.LocalInputSource {
         this.filename = path_1.default.basename(this.inputPath);
     }
     async init() {
-        logger_1.logger.debug(`Loading from: ${this.inputPath}`);
+        if (this.initialized) {
+            return;
+        }
+        logger_1.logger.debug(`Loading from path: ${this.inputPath}`);
         this.fileObject = Buffer.from(await fs_1.promises.readFile(this.inputPath));
         this.mimeType = await this.checkMimetype();
         this.initialized = true;

package/src/input/sources/streamInput.js

@@ -3,6 +3,7 @@ Object.defineProperty(exports, "__esModule", { value: true });
 exports.StreamInput = void 0;
 const localInputSource_1 = require("./localInputSource");
 const inputSource_1 = require("./inputSource");
+const logger_1 = require("../../logger");
 class StreamInput extends localInputSource_1.LocalInputSource {
     constructor({ inputStream, filename }) {
         super({
@@ -13,6 +14,10 @@ class StreamInput extends localInputSource_1.LocalInputSource {
         this.inputStream = inputStream;
     }
     async init() {
+        if (this.initialized) {
+            return;
+        }
+        logger_1.logger.debug("Loading from stream");
         this.fileObject = await this.stream2buffer(this.inputStream);
         this.mimeType = await this.checkMimetype();
         this.initialized = true;

package/src/input/sources/urlInput.js

@@ -8,12 +8,17 @@ const crypto_1 = require("crypto");
 const promises_1 = require("fs/promises");
 const https_1 = require("https");
 const bytesInput_1 = require("./bytesInput");
+const logger_1 = require("../../logger");
 class UrlInput extends inputSource_1.InputSource {
     constructor({ url }) {
         super();
         this.url = url;
     }
     async init() {
+        if (this.initialized) {
+            return;
+        }
+        logger_1.logger.debug(`source URL: ${this.url}`);
         if (!this.url.toLowerCase().startsWith("https")) {
             throw new Error("URL must be HTTPS");
         }

package/src/pdf/pdfOperation.d.ts

@@ -10,4 +10,9 @@ export interface SplitPdf {
  * @returns the new cut pdf file.
  */
 export declare function extractPages(file: Buffer, pageOptions: PageOptions): Promise<SplitPdf>;
+/**
+ * Count the number of pages in a pdf file.
+ * @param file
+ * @returns the number of pages in the file.
+ */
 export declare function countPages(file: Buffer): Promise<number>;

package/src/pdf/pdfOperation.js

@@ -66,6 +66,11 @@ async function extractPages(file, pageOptions) {
     const fileBuffer = Buffer.from(await newPdf.save());
     return { file: fileBuffer, totalPagesRemoved: sumRemovedPages };
 }
+/**
+ * Count the number of pages in a pdf file.
+ * @param file
+ * @returns the number of pages in the file.
+ */
 async function countPages(file) {
     const currentPdf = await pdf_lib_1.PDFDocument.load(file, {
         ignoreEncryption: true,