mindee 5.4.0 → 5.6.0

package/CHANGELOG.md

@@ -1,5 +1,26 @@
 # Mindee Node.js API Library Changelog
 
+
+## v5.6.0 - 2026-07-21
+### Changes
+* :sparkles: add support for webhook errors
+* :arrow_up: add husky
+* :arrow_up: add eslint rules
+* :arrow_up: add direct license checker
+* :recycle: replace dashes with colons in package.json script commands
+* :wrench: add secretlint
+* :wrench: :coffin: remove support for mocha
+### Fixes
+* :arrow_up: bump dependencies
+
+
+## v5.5.0 - 2026-07-01
+### Fixes
+* :memo: minor tweak to SKILL file
+* :bug: retry once on connection error
+* :recycle: rework and fix extractors (:boom: some **undocumented** functions break)
+
+
 ## v5.4.0 - 2026-06-22
 ### Fixes
 * :bug: :boom: harmonize Crop and Split extraction (now ready for public use)
@@ -78,7 +99,7 @@
 
 
 ## v5.0.0-rc1 - 2026-02-17
-### Changes
+### ¡Breaking Changes!
 * :boom: drop support for node.js 18
 * :boom: :recycle: separate polling options
 
@@ -92,7 +113,8 @@
 
 
 ## v5.0.0-alpha1 - 2026-02-13
-### :boom: :recycle: base for v5
+### ¡Breaking Changes!
+* :boom: :recycle: base for v5
 * :arrow_up: update file-type
 * :recycle: migrate to ES modules (node18)
 * :coffin: remove support for obsolete/unused products:

package/package.json

@@ -1,6 +1,6 @@
 {
     "name": "mindee",
-    "version": "5.4.0",
+    "version": "5.6.0",
     "description": "Mindee Client Library for Node.js",
     "author": {
         "name": "Mindee",
@@ -24,17 +24,23 @@
     },
     "scripts": {
         "build": "tsc --build && tsc-alias",
-        "build-for-dist": "tsc --build && tsc-alias && cp LICENSE README.md CHANGELOG.md ./dist",
+        "build:dist": "tsc --build && tsc-alias && cp LICENSE README.md CHANGELOG.md ./dist",
         "clean": "rm -rf ./dist ./docs/_build",
-        "test": "mocha --grep '#OptionalDepsRemoved' --invert 'tests/**/*.spec.ts'",
-        "test-light": "mocha --grep '#OptionalDepsRequired' --invert 'tests/**/*.spec.ts'",
-        "test-integration": "mocha --grep '#OptionalDepsRemoved' --invert 'tests/**/*.integration.ts'",
-        "test-integration-light": "mocha --grep '#OptionalDepsRequired' --invert 'tests/**/*.integration.ts'",
-        "test-v2": "tsc --noEmit && node --import tsx --test 'tests/v2/**/*.spec.ts'",
+        "test": "node tests/runner.mjs spec",
+        "test:light": "node tests/runner.mjs spec",
+        "test:integration": "node tests/runner.mjs integration",
+        "test:integration:light": "node tests/runner.mjs integration",
+        "test:v2": "tsc --noEmit && node tests/runner.mjs spec v2",
         "lint": "tsc --noEmit && eslint --report-unused-disable-directives './src/**/*.ts' './tests/**/*.ts'",
-        "lint-fix": "tsc --noEmit && eslint --fix --report-unused-disable-directives './src/**/*.ts' './tests/**/*.ts'",
+        "lint:check": "npm run lint",
+        "lint:fix": "tsc --noEmit && eslint --fix --report-unused-disable-directives './src/**/*.ts' './tests/**/*.ts'",
+        "lint:package": "npm run build:dist && publint ./dist && attw --pack ./dist --profile esm-only",
+        "lint:secrets": "secretlint '**/*'",
+        "lint:full": "npm run lint:check && npm run lint:package && npm run lint:secrets",
+        "license:check": "license-checker-rseidelsohn --onlyAllow \"MIT;Apache-2.0;BSD-2-Clause;BSD-3-Clause;ISC;LGPL-3.0-or-later;Python-2.0;BlueOak-1.0.0;CC-BY-3.0;CC0-1.0;0BSD;LGPL-3.0-only;WTFPL;Artistic-2.0;Unlicense\"",
         "docs": "typedoc --out docs/_build ./src/index.ts",
-        "docs-for-dist": "typedoc --out docs/_build ./src/index.ts && cp -r ./docs/code_samples ./docs/_build/"
+        "docs:dist": "typedoc --out docs/_build ./src/index.ts && cp -r ./docs/code_samples ./docs/_build/",
+        "prepare": "husky"
     },
     "files": [
         "src/*",
@@ -56,21 +62,27 @@
         "@cantoo/pdf-lib": "^2.5.3",
         "node-poppler": "^9.1.1",
         "pdf.js-extract": "~0.2.1",
-        "sharp": "~0.34.5"
+        "sharp": "^0.35.3"
     },
     "devDependencies": {
+        "@arethetypeswrong/cli": "^0.18.5",
         "@eslint/eslintrc": "^3.3.5",
         "@eslint/js": "^10.0.1",
-        "@types/mocha": "^10.0.10",
+        "@secretlint/secretlint-rule-preset-recommend": "^13.0.4",
         "@types/node": "^20.19.37",
         "@types/tmp": "^0.2.6",
         "@typescript-eslint/eslint-plugin": "^8.57.0",
         "@typescript-eslint/parser": "^8.58.1",
-        "eslint": "^10.0.3",
+        "eslint": "^10.6.0",
         "eslint-plugin-jsdoc": "^62.7.1",
+        "eslint-plugin-security": "^4.0.1",
+        "eslint-plugin-sonarjs": "^4.2.0",
         "globals": "^17.4.0",
-        "mocha": "^11.7.5",
-        "tsc-alias": "^1.8.16",
+        "husky": "^9.1.7",
+        "license-checker-rseidelsohn": "^4.4.2",
+        "publint": "^0.3.22",
+        "secretlint": "^13.0.4",
+        "tsc-alias": "^1.8.17",
         "tslib": "^2.8.1",
         "tsx": "^4.21.0",
         "typedoc": "^0.28.18",

package/src/geometry/bbox.d.ts

@@ -1,9 +1,14 @@
 /** A simple bounding box defined by 4 coordinates: xMin, yMin, xMax, yMax */
 export declare class BBox {
+    /** Minimum X coordinate. */
     xMin: number;
+    /** Minimum Y coordinate. */
     yMin: number;
+    /** Maximum X coordinate. */
     xMax: number;
+    /** Maximum Y coordinate. */
     yMax: number;
     constructor(xMin: number, yMin: number, xMax: number, yMax: number);
+    /** Returns a bounding box that contains this box and another one. */
     mergeBbox(bbox: BBox): BBox;
 }

package/src/geometry/bbox.js

@@ -7,6 +7,7 @@ export class BBox {
         this.xMax = xMax;
         this.yMax = yMax;
     }
+    /** Returns a bounding box that contains this box and another one. */
     mergeBbox(bbox) {
         return mergeBbox(this, bbox);
     }

package/src/geometry/minMax.d.ts

@@ -1,4 +1,7 @@
+/** Numeric range boundaries. */
 export type MinMax = {
+    /** Lower bound. */
     min: number;
+    /** Upper bound. */
     max: number;
 };

package/src/geometry/polygonUtils.d.ts

@@ -40,8 +40,31 @@ export declare function isPointInPolygonX(centroid: Point, polygon: Array<Point>
  * Can be used to order (sort) words in the same line.
  */
 export declare function relativeX(polygon: Array<Point>): number;
+/**
+ * Get the minimum Y coordinate in a given list of Points.
+ * @param polygon An array of points or polygon.
+ */
 export declare function getMinYCoordinate(polygon: Array<Point>): number;
+/**
+ * Get the minimum X coordinate in a given list of Points.
+ * @param polygon An array of points or polygon.
+ */
 export declare function getMinXCoordinate(polygon: Array<Point>): number;
+/**
+ * Compare two polygons on their minimum Y coordinate.
+ * @param polygon1 An array of points or polygon.
+ * @param polygon2 An array of points or polygon.
+ */
 export declare function compareOnY(polygon1: Array<Point>, polygon2: Array<Point>): number;
+/**
+ * Compare two polygons on their minimum X coordinate.
+ * @param polygon1 An array of points or polygon.
+ * @param polygon2 An array of points or polygon.
+ */
 export declare function compareOnX(polygon1: Array<Point>, polygon2: Array<Point>): number;
+/**
+ * Adjust a polygon for a given rotation.
+ * @param polygon An array of points or polygon.
+ * @param orientation The rotation angle in degrees.
+ */
 export declare function adjustForRotation(polygon: Array<Point>, orientation: number): Array<Point>;

package/src/geometry/polygonUtils.js

@@ -73,6 +73,10 @@ export function relativeX(polygon) {
         .reduce((prev, cur) => prev + cur);
     return polygon.length * sum;
 }
+/**
+ * Get the minimum Y coordinate in a given list of Points.
+ * @param polygon An array of points or polygon.
+ */
 export function getMinYCoordinate(polygon) {
     return polygon.sort((point1, point2) => {
         if (point1[1] < point2[1]) {
@@ -84,6 +88,10 @@ export function getMinYCoordinate(polygon) {
         return 0;
     })[0][1];
 }
+/**
+ * Get the minimum X coordinate in a given list of Points.
+ * @param polygon An array of points or polygon.
+ */
 export function getMinXCoordinate(polygon) {
     return polygon.sort((point1, point2) => {
         if (point1[0] < point2[0]) {
@@ -95,6 +103,11 @@ export function getMinXCoordinate(polygon) {
         return 0;
     })[0][0];
 }
+/**
+ * Compare two polygons on their minimum Y coordinate.
+ * @param polygon1 An array of points or polygon.
+ * @param polygon2 An array of points or polygon.
+ */
 export function compareOnY(polygon1, polygon2) {
     const sort = getMinYCoordinate(polygon1) - getMinYCoordinate(polygon2);
     if (sort === 0) {
@@ -102,6 +115,11 @@ export function compareOnY(polygon1, polygon2) {
     }
     return sort < 0 ? -1 : 1;
 }
+/**
+ * Compare two polygons on their minimum X coordinate.
+ * @param polygon1 An array of points or polygon.
+ * @param polygon2 An array of points or polygon.
+ */
 export function compareOnX(polygon1, polygon2) {
     const sort = getMinXCoordinate(polygon1) - getMinXCoordinate(polygon2);
     if (sort === 0) {
@@ -109,6 +127,11 @@ export function compareOnX(polygon1, polygon2) {
     }
     return sort < 0 ? -1 : 1;
 }
+/**
+ * Adjust a polygon for a given rotation.
+ * @param polygon An array of points or polygon.
+ * @param orientation The rotation angle in degrees.
+ */
 export function adjustForRotation(polygon, orientation) {
     if (orientation === 90) {
         return polygon.map(([x, y]) => [y, 1 - x]);

package/src/http/apiCore.d.ts

@@ -8,6 +8,7 @@ export interface RequestOptions {
     timeoutSecs: number;
     headers: any;
     body?: FormData | string;
+    queryParams?: Record<string, string>;
 }
 export interface BaseHttpResponse {
     messageObj: any;

package/src/http/apiCore.js

@@ -20,14 +20,32 @@ export async function cutDocPages(inputDoc, pageOptions) {
  * @returns the processed request.
  */
 export async function sendRequestAndReadResponse(dispatcher, options, url) {
-    url ?? (url = `https://${options.hostname}${options.path}`);
-    const response = await request(url, {
+    const requestUrl = new URL(url ?? `https://${options.hostname}${options.path}`);
+    if (options.queryParams) {
+        for (const [key, value] of Object.entries(options.queryParams)) {
+            requestUrl.searchParams.set(key, value);
+        }
+    }
+    const requestParams = {
         method: options.method,
         headers: options.headers,
         headersTimeout: options.timeoutSecs * 1000,
         body: options.body,
         dispatcher: dispatcher,
-    });
+    };
+    let response;
+    try {
+        response = await request(requestUrl, requestParams);
+    }
+    catch (err) {
+        // shenanigans in networking or freezing/thawing of the process in serverless environments
+        if (err.code === "UND_ERR_SOCKET" || err.code === "UND_ERR_CONNECT_TIMEOUT" || err.code === "ECONNRESET") {
+            logger.warn(`Socket error (${err.code}), retrying with a fresh connection...`);
+            response = await request(requestUrl, requestParams);
+        }
+        else
+            throw err;
+    }
     logger.debug("Parsing the response ...");
     let responseBody = await response.body.text();
     // handle empty responses from server, for example, in the case of redirects

package/src/http/baseSettings.d.ts

@@ -4,11 +4,16 @@ export interface MindeeApiConstructorProps {
     dispatcher?: Dispatcher;
 }
 export declare abstract class BaseSettings {
+    /** API key used for authenticated requests. */
     apiKey: string;
+    /** API hostname used for requests. */
     hostname: string;
+    /** Request timeout in seconds. */
     timeoutSecs: number;
+    /** HTTP dispatcher used by undici. */
     dispatcher: Dispatcher;
     protected constructor(apiKey?: string, dispatcher?: Dispatcher);
+    /** Builds a default user-agent string for outgoing requests. */
     protected getUserAgent(): string;
     protected abstract apiKeyFromEnv(): string;
     protected abstract hostnameFromEnv(): string;

package/src/http/baseSettings.js

@@ -1,4 +1,5 @@
 import { getGlobalDispatcher } from "undici";
+// eslint-disable-next-line no-restricted-imports
 import packageJson from "../../package.json" with { type: "json" };
 import * as os from "os";
 import { TIMEOUT_SECS_DEFAULT } from "./apiCore.js";
@@ -16,6 +17,7 @@ export class BaseSettings {
             ? parseInt(process.env.MINDEE_REQUEST_TIMEOUT)
             : TIMEOUT_SECS_DEFAULT;
     }
+    /** Builds a default user-agent string for outgoing requests. */
     getUserAgent() {
         let platform = os.type().toLowerCase();
         if (platform.includes("darwin")) {

package/src/image/extractedImage.d.ts

@@ -1,14 +1,18 @@
 import { Buffer } from "node:buffer";
 import { BufferInput } from "../input/index.js";
 /**
- * Generic class for image extraction
+ * Generic class for image extraction.
  */
 export declare class ExtractedImage {
+    /** Raw file bytes for the extracted image artifact. */
     buffer: Buffer;
+    /** Filename used when exporting the extracted image. */
     filename: string;
-    pageId?: number;
-    elementId?: number;
-    constructor(buffer: Uint8Array, fileName: string, pageId?: number, elementId?: number);
+    /** 1-based source page identifier. */
+    readonly pageId: number;
+    /** Identifier of the extracted element on the page. */
+    readonly elementId: number;
+    constructor(buffer: Uint8Array, fileName: string, pageId: number, elementId: number);
     /**
      * Saves the document to a file.
      *
@@ -27,5 +31,5 @@ export declare class ExtractedImage {
      *
      * @returns A BufferInput source.
      */
-    asSource(): BufferInput;
+    asInputSource(): BufferInput;
 }

package/src/image/extractedImage.js

@@ -7,7 +7,7 @@ import { BufferInput, MIMETYPES } from "../input/index.js";
 import { logger } from "../logger.js";
 import { loadOptionalDependency } from "../dependency/index.js";
 /**
- * Generic class for image extraction
+ * Generic class for image extraction.
  */
 export class ExtractedImage {
     constructor(buffer, fileName, pageId, elementId) {
@@ -92,7 +92,7 @@ export class ExtractedImage {
      *
      * @returns A BufferInput source.
      */
-    asSource() {
+    asInputSource() {
         return new BufferInput({
             buffer: this.buffer,
             filename: this.filename,

package/src/image/extractedImages.d.ts

@@ -1,4 +1,5 @@
 import { ExtractedImage } from "../image/index.js";
+/** Collection of extracted image artifacts. */
 export declare class ExtractedImages extends Array<ExtractedImage> {
     constructor(...items: ExtractedImage[]);
 }

package/src/image/extractedImages.js

@@ -1,3 +1,4 @@
+/** Collection of extracted image artifacts. */
 export class ExtractedImages extends Array {
     constructor(...items) {
         super(...items);

package/src/image/imageCompressor.d.ts

@@ -6,4 +6,4 @@
  * @param maxWidth Maximum bound for width.
  * @param maxHeight Maximum bound for height.
  */
-export declare function compressImage(imageBuffer: Buffer, quality?: number, maxWidth?: number | null, maxHeight?: number | null): Promise<Buffer<ArrayBufferLike>>;
+export declare function compressImage(imageBuffer: Buffer, quality?: number, maxWidth?: number | null, maxHeight?: number | null): Promise<Buffer<ArrayBuffer>>;

package/src/image/imageExtractor.d.ts

@@ -1,5 +1,5 @@
 import { Polygon } from "../geometry/index.js";
-import { ExtractedImage } from "../image/extractedImage.js";
+import { ExtractedImages } from "../image/extractedImages.js";
 import { LocalInputSource } from "../input/index.js";
 import type * as pdfLibTypes from "@cantoo/pdf-lib";
 /**
@@ -8,13 +8,12 @@ import type * as pdfLibTypes from "@cantoo/pdf-lib";
  * @param polygonsPerPage List of polygons to extract from per page.
  * @param quality JPEG quality of extracted images.
  */
-export declare function extractImagesFromPolygon(inputSource: LocalInputSource, polygonsPerPage: Map<number, Polygon[]>, quality?: number): Promise<ExtractedImage[]>;
+export declare function extractImagesFromPolygon(inputSource: LocalInputSource, polygonsPerPage: Map<number, Polygon[]>, quality?: number): Promise<ExtractedImages>;
 /**
- * Extracts elements from a page based off of a list of bounding boxes.
- *
- * @param pdfPage PDF Page to extract from.
- * @param polygons List of coordinates to pull the elements from.
- * @param asImage Whether to return the extracted elements as images.
- * @param quality JPEG quality of extracted images, given as number between 0 and 1.
+ * Extract images from a PDF page.
+ * @param pdfPage The PDF page to extract images from.
+ * @param polygons The polygons to extract images from.
+ * @param asImage Whether to return the extracted images as images or as raw data.
+ * @param quality The quality of the extracted images.
  */
 export declare function extractFromPage(pdfPage: pdfLibTypes.PDFPage, polygons: Polygon[], asImage?: boolean, quality?: number): Promise<Uint8Array<ArrayBufferLike>[]>;

package/src/image/imageExtractor.js

@@ -3,10 +3,14 @@ import { MindeeImageError } from "../errors/index.js";
 import { getMinMaxX, getMinMaxY } from "../geometry/index.js";
 import { adjustForRotation } from "../geometry/polygonUtils.js";
 import { ExtractedImage } from "../image/extractedImage.js";
+import { ExtractedImages } from "../image/extractedImages.js";
 import { logger } from "../logger.js";
 import { createPdfFromInputSource } from "../pdf/pdfOperation.js";
 import { rasterizePage } from "../pdf/pdfUtils.js";
 let pdfLib = null;
+/**
+ * Load the PDF library if not already loaded.
+ */
 async function getPdfLib() {
     if (!pdfLib) {
         const pdfLibImport = await loadOptionalDependency("@cantoo/pdf-lib", "Image Extraction");
@@ -21,35 +25,74 @@ async function getPdfLib() {
  * @param quality JPEG quality of extracted images.
  */
 export async function extractImagesFromPolygon(inputSource, polygonsPerPage, quality) {
-    const allExtractedImages = [];
+    const allExtractedImages = new ExtractedImages();
     const pdfDoc = await createPdfFromInputSource(inputSource);
     for (const [pageId, polygons] of polygonsPerPage) {
         logger.debug(`Extracting images from page ${pageId}`);
         const pdfPage = pdfDoc.getPage(pageId);
-        const extractions = (await extractFromPage(pdfPage, polygons, true, quality));
-        const extractedImages = extractions.map((v, i) => new ExtractedImage(v, inputSource.filename + `_page${pageId}-${i}.jpg`, pageId, i));
+        const extractions = await extractFromPage(pdfPage, polygons, true, quality);
+        const extractedImages = extractions.map((buffer, elementId) => new ExtractedImage(buffer, inputSource.filename + `_page${pageId}-${elementId}.jpg`, pageId, elementId));
         allExtractedImages.push(...extractedImages);
     }
     return allExtractedImages;
 }
 /**
- * Extracts elements from a page based off of a list of bounding boxes.
- *
- * @param pdfPage PDF Page to extract from.
- * @param polygons List of coordinates to pull the elements from.
- * @param asImage Whether to return the extracted elements as images.
- * @param quality JPEG quality of extracted images, given as number between 0 and 1.
+ * Helper function to handle the drawing math and orientation.
+ */
+function drawOrientedPage(pdfLib, samplePage, cropped, orientation, finalWidth, finalHeight, scaledWidth, scaledHeight) {
+    if (orientation === 0) {
+        samplePage.drawPage(cropped, {
+            width: scaledWidth,
+            height: scaledHeight,
+        });
+    }
+    else if (orientation === 90) {
+        samplePage.drawPage(cropped, {
+            x: 0,
+            y: finalHeight,
+            width: scaledWidth,
+            height: scaledHeight,
+            rotate: pdfLib.degrees(270),
+        });
+    }
+    else if (orientation === 180) {
+        samplePage.drawPage(cropped, {
+            x: finalWidth,
+            y: finalHeight,
+            width: scaledWidth,
+            height: scaledHeight,
+            rotate: pdfLib.degrees(180),
+        });
+    }
+    else if (orientation === 270) {
+        samplePage.drawPage(cropped, {
+            x: finalWidth,
+            y: 0,
+            width: scaledWidth,
+            height: scaledHeight,
+            rotate: pdfLib.degrees(90),
+        });
+    }
+}
+/**
+ * Extract images from a PDF page.
+ * @param pdfPage The PDF page to extract images from.
+ * @param polygons The polygons to extract images from.
+ * @param asImage Whether to return the extracted images as images or as raw data.
+ * @param quality The quality of the extracted images.
  */
 export async function extractFromPage(pdfPage, polygons, asImage = false, quality) {
     const pdfLib = await getPdfLib();
     const { width, height } = pdfPage.getSize();
     const extractedElements = [];
-    if (quality && (quality < 0)) {
-        throw new MindeeImageError("Quality must be a number between 0 and 1");
-    }
-    if (quality && quality > 1) {
-        logger.warn("Quality is greater than 1, this operation will apply a manual upscale on the output." +
-            " Use only if you know what you are doing.");
+    if (quality !== undefined) {
+        if (quality < 0) {
+            throw new MindeeImageError("Quality must be a number between 0 and 1");
+        }
+        if (quality > 1) {
+            logger.warn("Quality is greater than 1, this operation will apply a manual upscale on the output." +
+                " Use only if you know what you are doing.");
+        }
     }
     const qualityScale = quality ?? 1;
     const orientation = pdfPage.getRotation().angle;
@@ -60,72 +103,28 @@ export async function extractFromPage(pdfPage, polygons, asImage = false, qualit
         const tempPdf = await pdfLib.PDFDocument.create();
         const [copiedPage] = await tempPdf.copyPages(sourceDoc, [pageIndex]);
         const polygon = adjustForRotation(origPolygon, orientation);
-        const newWidth = width * (getMinMaxX(polygon).max - getMinMaxX(polygon).min);
-        const newHeight = height * (getMinMaxY(polygon).max - getMinMaxY(polygon).min);
+        const minX = getMinMaxX(polygon).min;
+        const maxX = getMinMaxX(polygon).max;
+        const minY = getMinMaxY(polygon).min;
+        const maxY = getMinMaxY(polygon).max;
+        const newWidth = width * (maxX - minX);
+        const newHeight = height * (maxY - minY);
         const cropped = await tempPdf.embedPage(copiedPage, {
-            left: getMinMaxX(polygon).min * width,
-            right: getMinMaxX(polygon).max * width,
-            top: height - (getMinMaxY(polygon).min * height),
-            bottom: height - (getMinMaxY(polygon).max * height),
+            left: minX * width,
+            right: maxX * width,
+            top: height - (minY * height),
+            bottom: height - (maxY * height),
         });
-        let finalWidth;
-        let finalHeight;
-        if (orientation === 90 || orientation === 270) {
-            finalWidth = newHeight * qualityScale;
-            finalHeight = newWidth * qualityScale;
-        }
-        else {
-            finalWidth = newWidth * qualityScale;
-            finalHeight = newHeight * qualityScale;
-        }
+        const isVertical = orientation === 90 || orientation === 270;
+        const finalWidth = (isVertical ? newHeight : newWidth) * qualityScale;
+        const finalHeight = (isVertical ? newWidth : newHeight) * qualityScale;
         const samplePage = tempPdf.addPage([finalWidth, finalHeight]);
         samplePage.drawRectangle({
-            x: 0,
-            y: 0,
-            width: finalWidth,
-            height: finalHeight,
-            color: pdfLib.rgb(1, 1, 1),
+            x: 0, y: 0, width: finalWidth, height: finalHeight, color: pdfLib.rgb(1, 1, 1),
         });
-        if (orientation === 0) {
-            samplePage.drawPage(cropped, {
-                width: newWidth * qualityScale,
-                height: newHeight * qualityScale,
-            });
-        }
-        else if (orientation === 90) {
-            samplePage.drawPage(cropped, {
-                x: 0,
-                y: finalHeight,
-                width: newWidth * qualityScale,
-                height: newHeight * qualityScale,
-                rotate: pdfLib.degrees(270),
-            });
-        }
-        else if (orientation === 180) {
-            samplePage.drawPage(cropped, {
-                x: finalWidth,
-                y: finalHeight,
-                width: newWidth * qualityScale,
-                height: newHeight * qualityScale,
-                rotate: pdfLib.degrees(180),
-            });
-        }
-        else if (orientation === 270) {
-            samplePage.drawPage(cropped, {
-                x: finalWidth,
-                y: 0,
-                width: newWidth * qualityScale,
-                height: newHeight * qualityScale,
-                rotate: pdfLib.degrees(90),
-            });
-        }
+        drawOrientedPage(pdfLib, samplePage, cropped, orientation, finalWidth, finalHeight, newWidth * qualityScale, newHeight * qualityScale);
         const pdfBuffer = Buffer.from(await tempPdf.save());
-        if (asImage) {
-            extractedElements.push(await rasterizePage(pdfBuffer, 0, 100));
-        }
-        else {
-            extractedElements.push(pdfBuffer);
-        }
+        extractedElements.push(asImage ? await rasterizePage(pdfBuffer, 0, 100) : pdfBuffer);
     }
     return extractedElements;
 }

package/src/input/base64Input.d.ts

@@ -3,10 +3,13 @@ interface Base64InputProps {
     inputString: string;
     filename: string;
 }
+/** Local input source backed by a base64-encoded payload. */
 export declare class Base64Input extends LocalInputSource {
     private inputString;
+    /** Decoded binary payload built from the base64 input. */
     fileObject: Buffer;
     constructor({ inputString, filename }: Base64InputProps);
+    /** Decodes and validates the provided base64 content. */
     init(): Promise<void>;
 }
 export {};

package/src/input/base64Input.js

@@ -1,15 +1,18 @@
 import { LocalInputSource } from "./localInputSource.js";
 import { INPUT_TYPE_BASE64 } from "./inputSource.js";
 import { logger } from "../logger.js";
+/** Local input source backed by a base64-encoded payload. */
 export class Base64Input extends LocalInputSource {
     constructor({ inputString, filename }) {
         super({
             inputType: INPUT_TYPE_BASE64,
         });
+        /** Decoded binary payload built from the base64 input. */
         this.fileObject = Buffer.alloc(0);
         this.filename = filename;
         this.inputString = inputString;
     }
+    /** Decodes and validates the provided base64 content. */
     async init() {
         if (this.initialized) {
             return;