unpdf 0.5.3 → 0.6.0

package/dist/types/web/ui_utils.d.ts

@@ -0,0 +1,288 @@
+export type GetPageSizeInchesParameters = {
+    view: number[];
+    userUnit: number;
+    rotate: number;
+};
+export type PageSize = {
+    /**
+     * - In inches.
+     */
+    width: number;
+    /**
+     * - In inches.
+     */
+    height: number;
+};
+export type GetVisibleElementsParameters = {
+    /**
+     * - A container that can possibly scroll.
+     */
+    scrollEl: HTMLElement;
+    /**
+     * - Objects with a `div` property that contains an
+     * HTMLElement, which should all be descendants of `scrollEl` satisfying the
+     * relevant layout assumptions.
+     */
+    views: any[];
+    /**
+     * - If `true`, the returned elements are
+     * sorted in descending order of the percent of their padding box that is
+     * visible. The default value is `false`.
+     */
+    sortByVisibility: boolean;
+    /**
+     * - If `true`, the elements are assumed to be
+     * laid out horizontally instead of vertically. The default value is `false`.
+     */
+    horizontal: boolean;
+    /**
+     * - If `true`, the `scrollEl` container is assumed to
+     * be in right-to-left mode. The default value is `false`.
+     */
+    rtl: boolean;
+};
+/**
+ * Promise that is resolved when DOM window becomes visible.
+ */
+export const animationStarted: Promise<any>;
+/**
+ * Converts API PageLayout values to the format used by `BaseViewer`.
+ * @param {string} mode - The API PageLayout value.
+ * @returns {Object}
+ */
+export function apiPageLayoutToViewerModes(layout: any): Object;
+/**
+ * Converts API PageMode values to the format used by `PDFSidebar`.
+ * NOTE: There's also a "FullScreen" parameter which is not possible to support,
+ *       since the Fullscreen API used in browsers requires that entering
+ *       fullscreen mode only occurs as a result of a user-initiated event.
+ * @param {string} mode - The API PageMode value.
+ * @returns {number} A value from {SidebarView}.
+ */
+export function apiPageModeToSidebarView(mode: string): number;
+/**
+ *  Approximates float number as a fraction using Farey sequence (max order
+ *  of 8).
+ *  @param {number} x - Positive float number.
+ *  @returns {Array} Estimated fraction: the first array item is a numerator,
+ *                   the second one is a denominator.
+ */
+export function approximateFraction(x: number): any[];
+export const AutoPrintRegExp: RegExp;
+/**
+ * Helper function for getVisibleElements.
+ *
+ * @param {number} index - initial guess at the first visible element
+ * @param {Array} views - array of pages, into which `index` is an index
+ * @param {number} top - the top of the scroll pane
+ * @returns {number} less than or equal to `index` that is definitely at or
+ *   before the first visible element in `views`, but not by too much. (Usually,
+ *   this will be the first element in the first partially visible row in
+ *   `views`, although sometimes it goes back one row further.)
+ */
+export function backtrackBeforeAllVisibleElements(index: number, views: any[], top: number): number;
+/**
+ * Use binary search to find the index of the first item in a given array which
+ * passes a given condition. The items are expected to be sorted in the sense
+ * that if the condition is true for one item in the array, then it is also true
+ * for all following items.
+ *
+ * @returns {number} Index of the first array element to pass the test,
+ *                   or |items.length| if no such element exists.
+ */
+export function binarySearchFirstItem(items: any, condition: any, start?: number): number;
+export namespace CursorTool {
+    let SELECT: number;
+    let HAND: number;
+    let ZOOM: number;
+}
+export const DEFAULT_SCALE: 1;
+export const DEFAULT_SCALE_DELTA: 1.1;
+export const DEFAULT_SCALE_VALUE: "auto";
+export const docStyle: CSSStyleDeclaration | null;
+/**
+ * Get the active or focused element in current DOM.
+ *
+ * Recursively search for the truly active or focused element in case there are
+ * shadow DOMs.
+ *
+ * @returns {Element} the truly active or focused element.
+ */
+export function getActiveOrFocusedElement(): Element;
+/**
+ * @typedef {Object} GetPageSizeInchesParameters
+ * @property {number[]} view
+ * @property {number} userUnit
+ * @property {number} rotate
+ */
+/**
+ * @typedef {Object} PageSize
+ * @property {number} width - In inches.
+ * @property {number} height - In inches.
+ */
+/**
+ * Gets the size of the specified page, converted from PDF units to inches.
+ * @param {GetPageSizeInchesParameters} params
+ * @returns {PageSize}
+ */
+export function getPageSizeInches({ view, userUnit, rotate }: GetPageSizeInchesParameters): PageSize;
+/**
+ * @typedef {Object} GetVisibleElementsParameters
+ * @property {HTMLElement} scrollEl - A container that can possibly scroll.
+ * @property {Array} views - Objects with a `div` property that contains an
+ *   HTMLElement, which should all be descendants of `scrollEl` satisfying the
+ *   relevant layout assumptions.
+ * @property {boolean} sortByVisibility - If `true`, the returned elements are
+ *   sorted in descending order of the percent of their padding box that is
+ *   visible. The default value is `false`.
+ * @property {boolean} horizontal - If `true`, the elements are assumed to be
+ *   laid out horizontally instead of vertically. The default value is `false`.
+ * @property {boolean} rtl - If `true`, the `scrollEl` container is assumed to
+ *   be in right-to-left mode. The default value is `false`.
+ */
+/**
+ * Generic helper to find out what elements are visible within a scroll pane.
+ *
+ * Well, pretty generic. There are some assumptions placed on the elements
+ * referenced by `views`:
+ *   - If `horizontal`, no left of any earlier element is to the right of the
+ *     left of any later element.
+ *   - Otherwise, `views` can be split into contiguous rows where, within a row,
+ *     no top of any element is below the bottom of any other element, and
+ *     between rows, no bottom of any element in an earlier row is below the
+ *     top of any element in a later row.
+ *
+ * (Here, top, left, etc. all refer to the padding edge of the element in
+ * question. For pages, that ends up being equivalent to the bounding box of the
+ * rendering canvas. Earlier and later refer to index in `views`, not page
+ * layout.)
+ *
+ * @param {GetVisibleElementsParameters}
+ * @returns {Object} `{ first, last, views: [{ id, x, y, view, percent }] }`
+ */
+export function getVisibleElements({ scrollEl, views, sortByVisibility, horizontal, rtl, }: GetVisibleElementsParameters): Object;
+export function isPortraitOrientation(size: any): boolean;
+export function isValidRotation(angle: any): boolean;
+export function isValidScrollMode(mode: any): boolean;
+export function isValidSpreadMode(mode: any): boolean;
+export const MAX_AUTO_SCALE: 1.25;
+export const MAX_SCALE: 10;
+export const MIN_SCALE: 0.1;
+/**
+ * Event handler to suppress context menu.
+ */
+export function noContextMenuHandler(evt: any): void;
+export function normalizeWheelEventDelta(evt: any): number;
+export function normalizeWheelEventDirection(evt: any): number;
+/**
+ * Scale factors for the canvas, necessary with HiDPI displays.
+ */
+export class OutputScale {
+    /**
+     * @type {number} Horizontal scale.
+     */
+    sx: number;
+    /**
+     * @type {number} Vertical scale.
+     */
+    sy: number;
+    /**
+     * @type {boolean} Returns `true` when scaling is required, `false` otherwise.
+     */
+    get scaled(): boolean;
+}
+/**
+ * Helper function to parse query string (e.g. ?param1=value&param2=...).
+ * @param {string}
+ * @returns {Map}
+ */
+export function parseQueryString(query: any): Map<any, any>;
+export namespace PresentationModeState {
+    let UNKNOWN: number;
+    let NORMAL: number;
+    let CHANGING: number;
+    let FULLSCREEN: number;
+}
+export class ProgressBar {
+    constructor(bar: any);
+    set percent(arg: number);
+    get percent(): number;
+    setWidth(viewer: any): void;
+    setDisableAutoFetch(delay?: number): void;
+    hide(): void;
+    show(): void;
+    #private;
+}
+/**
+ * @param {string} str
+ * @param {boolean} [replaceInvisible]
+ */
+export function removeNullCharacters(str: string, replaceInvisible?: boolean | undefined): string;
+export namespace RenderingStates {
+    let INITIAL: number;
+    let RUNNING: number;
+    let PAUSED: number;
+    let FINISHED: number;
+}
+export function roundToDivide(x: any, div: any): any;
+export const SCROLLBAR_PADDING: 40;
+/**
+ * Scrolls specified element into view of its parent.
+ * @param {HTMLElement} element - The element to be visible.
+ * @param {Object} [spot] - An object with optional top and left properties,
+ *   specifying the offset from the top left edge.
+ * @param {number} [spot.left]
+ * @param {number} [spot.top]
+ * @param {boolean} [scrollMatches] - When scrolling search results into view,
+ *   ignore elements that either: Contains marked content identifiers,
+ *   or have the CSS-rule `overflow: hidden;` set. The default value is `false`.
+ */
+export function scrollIntoView(element: HTMLElement, spot?: {
+    left?: number | undefined;
+    top?: number | undefined;
+} | undefined, scrollMatches?: boolean | undefined): void;
+export namespace ScrollMode {
+    let UNKNOWN_1: number;
+    export { UNKNOWN_1 as UNKNOWN };
+    export let VERTICAL: number;
+    export let HORIZONTAL: number;
+    export let WRAPPED: number;
+    export let PAGE: number;
+}
+export namespace SidebarView {
+    let UNKNOWN_2: number;
+    export { UNKNOWN_2 as UNKNOWN };
+    export let NONE: number;
+    export let THUMBS: number;
+    export let OUTLINE: number;
+    export let ATTACHMENTS: number;
+    export let LAYERS: number;
+}
+export namespace SpreadMode {
+    let UNKNOWN_3: number;
+    export { UNKNOWN_3 as UNKNOWN };
+    let NONE_1: number;
+    export { NONE_1 as NONE };
+    export let ODD: number;
+    export let EVEN: number;
+}
+export namespace TextLayerMode {
+    let DISABLE: number;
+    let ENABLE: number;
+    let ENABLE_PERMISSIONS: number;
+}
+export function toggleCheckedBtn(button: any, toggle: any, view?: null): void;
+export function toggleExpandedBtn(button: any, toggle: any, view?: null): void;
+export const UNKNOWN_SCALE: 0;
+export const VERTICAL_PADDING: 5;
+/**
+ * Helper function to start monitoring the scroll event and converting them into
+ * PDF.js friendly one: with scroll debounce and scroll direction.
+ */
+export function watchScroll(viewAreaElement: any, callback: any): {
+    right: boolean;
+    down: boolean;
+    lastX: any;
+    lastY: any;
+    _eventHandler: (evt: any) => void;
+};

package/dist/types/web/xfa_layer_builder.d.ts

@@ -0,0 +1,41 @@
+export type PDFPageProxy = import("../src/display/api").PDFPageProxy;
+export type PageViewport = import("../src/display/display_utils").PageViewport;
+export type IPDFLinkService = import("./interfaces").IPDFLinkService;
+export type XfaLayerBuilderOptions = {
+    pageDiv: HTMLDivElement;
+    pdfPage: PDFPageProxy;
+    annotationStorage?: any;
+    linkService: IPDFLinkService;
+    xfaHtml?: Object | undefined;
+};
+/**
+ * @typedef {Object} XfaLayerBuilderOptions
+ * @property {HTMLDivElement} pageDiv
+ * @property {PDFPageProxy} pdfPage
+ * @property {AnnotationStorage} [annotationStorage]
+ * @property {IPDFLinkService} linkService
+ * @property {Object} [xfaHtml]
+ */
+export class XfaLayerBuilder {
+    /**
+     * @param {XfaLayerBuilderOptions} options
+     */
+    constructor({ pageDiv, pdfPage, annotationStorage, linkService, xfaHtml, }: XfaLayerBuilderOptions);
+    pageDiv: HTMLDivElement;
+    pdfPage: import("../src/display/api").PDFPageProxy;
+    annotationStorage: any;
+    linkService: import("./interfaces").IPDFLinkService;
+    xfaHtml: Object;
+    div: HTMLDivElement | null;
+    _cancelled: boolean;
+    /**
+     * @param {PageViewport} viewport
+     * @param {string} intent (default value is 'display')
+     * @returns {Promise<Object | void>} A promise that is resolved when rendering
+     *   of the XFA layer is complete. The first rendering will return an object
+     *   with a `textDivs` property that can be used with the TextHighlighter.
+     */
+    render(viewport: PageViewport, intent?: string): Promise<Object | void>;
+    cancel(): void;
+    hide(): void;
+}

package/package.json

@@ -1,72 +1,98 @@
 {
   "name": "unpdf",
   "type": "module",
-  "version": "0.5.3",
-  "packageManager": "pnpm@8.6.12",
-  "description": "Utilities to work with PDFs, like extracting text",
+  "version": "0.6.0",
+  "packageManager": "pnpm@8.7.0",
+  "description": "Utilities to work with PDFs in Node.js, browser and workers",
   "author": "Johann Schopplich <pkg@johannschopplich.com>",
   "license": "MIT",
-  "homepage": "https://github.com/johannschopplich/unpdf#readme",
+  "homepage": "https://github.com/unjs/unpdf#readme",
   "repository": {
     "type": "git",
-    "url": "git+https://github.com/johannschopplich/unpdf.git"
+    "url": "git+https://github.com/unjs/unpdf.git"
   },
   "bugs": {
-    "url": "https://github.com/johannschopplich/unpdf/issues"
+    "url": "https://github.com/unjs/unpdf/issues"
   },
   "keywords": [
-    "pdf",
-    "parse"
+    "parse",
+    "pdfjs-dist",
+    "pdf"
   ],
   "sideEffects": false,
   "exports": {
-    "browser": "./dist/index.web.mjs",
-    "bun": "./dist/index.worker.mjs",
-    "deno": "./dist/index.worker.mjs",
-    "edge-light": "./dist/index.worker.mjs",
-    "edge-routine": "./dist/index.worker.mjs",
-    "lagon": "./dist/index.worker.mjs",
-    "netlify": "./dist/index.worker.mjs",
-    "react-native": "./dist/index.worker.mjs",
-    "wintercg": "./dist/index.worker.mjs",
-    "worker": "./dist/index.worker.mjs",
-    "workerd": "./dist/index.worker.mjs",
-    "node": {
-      "types": "./dist/index.node.d.ts",
-      "import": "./dist/index.node.mjs",
-      "require": "./dist/index.node.cjs"
+    ".": {
+      "browser": "./dist/index.web.mjs",
+      "bun": "./dist/index.worker.mjs",
+      "deno": "./dist/index.worker.mjs",
+      "edge-light": "./dist/index.worker.mjs",
+      "edge-routine": "./dist/index.worker.mjs",
+      "lagon": "./dist/index.worker.mjs",
+      "netlify": "./dist/index.worker.mjs",
+      "react-native": "./dist/index.worker.mjs",
+      "wintercg": "./dist/index.worker.mjs",
+      "worker": "./dist/index.worker.mjs",
+      "workerd": "./dist/index.worker.mjs",
+      "node": {
+        "types": "./dist/index.node.d.ts",
+        "import": "./dist/index.node.mjs",
+        "require": "./dist/index.node.cjs"
+      },
+      "types": "./dist/index.web.d.ts",
+      "import": "./dist/index.web.mjs",
+      "require": "./dist/index.web.cjs"
     },
-    "types": "./dist/index.web.d.ts",
-    "import": "./dist/index.web.mjs",
-    "require": "./dist/index.web.cjs"
+    "./pdfjs": {
+      "types": "./dist/pdfjs.d.ts",
+      "import": "./dist/pdfjs.mjs"
+    }
   },
   "main": "./dist/index.node.cjs",
   "module": "./dist/index.web.mjs",
-  "browser": "./dist/index.web.mjs",
   "types": "./dist/index.web.d.ts",
+  "typesVersions": {
+    "*": {
+      "*": [
+        "*",
+        "dist/*",
+        "dist/*.d.ts"
+      ]
+    }
+  },
   "files": [
     "dist"
   ],
   "scripts": {
-    "build": "unbuild",
+    "build": "unbuild && node scripts/post-build.mjs",
+    "build:pdfjs": "rollup --config pdfjs.rollup.config.ts --configPlugin typescript && cp -R node_modules/pdfjs-dist/types dist",
     "dev": "unbuild --stub",
-    "lint": "eslint .",
-    "lint:fix": "eslint . --fix",
+    "lint": "eslint \"**/*.ts\" --ignore-path .gitignore",
+    "lint:fix": "npm run lint -- --fix",
+    "format": "prettier \"**/*.{css,html,json,md,mjs,ts,vue,yml}\" --write",
+    "format:check": "prettier \"**/*.{css,html,json,md,mjs,ts,vue,yml}\" --check",
     "release": "bumpp --commit --push --tag",
-    "test": "tsx test/index.ts",
+    "test": "vitest",
     "test:types": "tsc --noEmit"
   },
-  "peerDependencies": {
-    "pdfjs-dist": "^3.9.179"
-  },
   "devDependencies": {
-    "@antfu/eslint-config": "^0.40.2",
-    "@types/node": "^20.4.10",
-    "bumpp": "^9.1.1",
-    "eslint": "^8.47.0",
-    "pdfjs-dist": "^3.9.179",
-    "tsx": "^3.12.7",
+    "@rollup/plugin-alias": "^5.0.0",
+    "@rollup/plugin-commonjs": "^25.0.4",
+    "@rollup/plugin-inject": "^5.0.3",
+    "@rollup/plugin-node-resolve": "^15.2.1",
+    "@rollup/plugin-replace": "^5.0.2",
+    "@rollup/plugin-terser": "^0.4.3",
+    "@rollup/plugin-typescript": "^11.1.3",
+    "@types/node": "^20.5.7",
+    "bumpp": "^9.2.0",
+    "eslint": "^8.48.0",
+    "eslint-config-unjs": "^0.2.1",
+    "fast-glob": "^3.3.1",
+    "pdfjs-dist": "^3.10.111",
+    "prettier": "^3.0.2",
+    "rollup": "^3.28.1",
     "typescript": "^5.1.6",
-    "unbuild": "^2.0.0-rc.0"
+    "unbuild": "^2.0.0",
+    "unenv": "^1.7.4",
+    "vitest": "^0.34.3"
   }
 }

package/dist/shared/unpdf.61a2a4e1.cjs

@@ -1,56 +0,0 @@
-'use strict';
-
-let resolvedModule;
-async function getDocumentProxy(data, options = {}) {
-  const { getDocument } = await getResolvedPDFJS();
-  const pdf = await getDocument({
-    data,
-    useWorkerFetch: false,
-    isEvalSupported: false,
-    useSystemFonts: true,
-    ...options
-  }).promise;
-  return pdf;
-}
-async function defineUnPDFConfig({ pdfjs }) {
-  if (pdfjs) {
-    try {
-      const { default: mod } = await pdfjs();
-      resolvedModule = mod;
-    } catch (error) {
-      throw new Error("Resolving the PDF.js module failed. Please check your configuration.");
-    }
-  }
-}
-async function getResolvedPDFJS() {
-  if (!resolvedModule)
-    await resolvePDFJSImports();
-  return resolvedModule;
-}
-async function resolvePDFJSImports(pdfjs) {
-  if (resolvedModule)
-    return;
-  if (pdfjs) {
-    try {
-      const { default: mod } = await pdfjs();
-      resolvedModule = mod;
-    } catch (error) {
-      throw new Error("Resolving the PDF.js module failed. Please check your configuration.");
-    }
-  }
-  try {
-    const { default: mod } = await import('pdfjs-dist');
-    resolvedModule = mod;
-  } catch (error) {
-    throw new Error("PDF.js is not available. Please add the package as a dependency.");
-  }
-}
-function isPDFDocumentProxy(data) {
-  return typeof data === "object" && data !== null && "_pdfInfo" in data;
-}
-
-exports.defineUnPDFConfig = defineUnPDFConfig;
-exports.getDocumentProxy = getDocumentProxy;
-exports.getResolvedPDFJS = getResolvedPDFJS;
-exports.isPDFDocumentProxy = isPDFDocumentProxy;
-exports.resolvePDFJSImports = resolvePDFJSImports;

package/dist/shared/unpdf.9511ac30.cjs

@@ -1,52 +0,0 @@
-'use strict';
-
-const utils = require('./unpdf.61a2a4e1.cjs');
-
-async function getPDFMeta(data) {
-  const pdf = utils.isPDFDocumentProxy(data) ? data : await utils.getDocumentProxy(data);
-  const meta = await pdf.getMetadata().catch(() => null);
-  return {
-    info: meta?.info ?? {},
-    metadata: meta?.metadata?.getAll() ?? {}
-  };
-}
-
-async function extractPDFText(data, { mergePages = false } = {}) {
-  const pdf = utils.isPDFDocumentProxy(data) ? data : await utils.getDocumentProxy(data);
-  const texts = await Promise.all(
-    Array.from({ length: pdf.numPages }, (_, i) => getPageText(pdf, i + 1))
-  );
-  const filteredTexts = texts.filter(Boolean);
-  return {
-    totalPages: pdf.numPages,
-    text: mergePages ? filteredTexts.join("\n\n") : filteredTexts
-  };
-}
-async function getPageText(pdf, pageNumber) {
-  const page = await pdf.getPage(pageNumber);
-  const content = await page.getTextContent();
-  const items = content.items;
-  if (items.length === 0)
-    return;
-  return items.map(({ str }) => str).join("\n");
-}
-
-async function getImagesFromPage(data, pageNumber) {
-  const pdf = utils.isPDFDocumentProxy(data) ? data : await utils.getDocumentProxy(data);
-  const page = await pdf.getPage(pageNumber);
-  const operatorList = await page.getOperatorList();
-  const { OPS } = await utils.getResolvedPDFJS();
-  const images = [];
-  for (const op of operatorList.fnArray) {
-    if (op !== OPS.paintImageXObject)
-      continue;
-    const image = await page.objs.get(operatorList.argsArray[op][0]);
-    if (image.data)
-      images.push(image.data.buffer);
-  }
-  return images;
-}
-
-exports.extractPDFText = extractPDFText;
-exports.getImagesFromPage = getImagesFromPage;
-exports.getPDFMeta = getPDFMeta;

package/dist/shared/unpdf.c7f36e56.mjs

@@ -1,50 +0,0 @@
-let resolvedModule;
-async function getDocumentProxy(data, options = {}) {
-  const { getDocument } = await getResolvedPDFJS();
-  const pdf = await getDocument({
-    data,
-    useWorkerFetch: false,
-    isEvalSupported: false,
-    useSystemFonts: true,
-    ...options
-  }).promise;
-  return pdf;
-}
-async function defineUnPDFConfig({ pdfjs }) {
-  if (pdfjs) {
-    try {
-      const { default: mod } = await pdfjs();
-      resolvedModule = mod;
-    } catch (error) {
-      throw new Error("Resolving the PDF.js module failed. Please check your configuration.");
-    }
-  }
-}
-async function getResolvedPDFJS() {
-  if (!resolvedModule)
-    await resolvePDFJSImports();
-  return resolvedModule;
-}
-async function resolvePDFJSImports(pdfjs) {
-  if (resolvedModule)
-    return;
-  if (pdfjs) {
-    try {
-      const { default: mod } = await pdfjs();
-      resolvedModule = mod;
-    } catch (error) {
-      throw new Error("Resolving the PDF.js module failed. Please check your configuration.");
-    }
-  }
-  try {
-    const { default: mod } = await import('pdfjs-dist');
-    resolvedModule = mod;
-  } catch (error) {
-    throw new Error("PDF.js is not available. Please add the package as a dependency.");
-  }
-}
-function isPDFDocumentProxy(data) {
-  return typeof data === "object" && data !== null && "_pdfInfo" in data;
-}
-
-export { getResolvedPDFJS as a, defineUnPDFConfig as d, getDocumentProxy as g, isPDFDocumentProxy as i, resolvePDFJSImports as r };

package/dist/shared/unpdf.cd67d475.mjs

@@ -1,48 +0,0 @@
-import { i as isPDFDocumentProxy, g as getDocumentProxy, a as getResolvedPDFJS } from './unpdf.c7f36e56.mjs';
-
-async function getPDFMeta(data) {
-  const pdf = isPDFDocumentProxy(data) ? data : await getDocumentProxy(data);
-  const meta = await pdf.getMetadata().catch(() => null);
-  return {
-    info: meta?.info ?? {},
-    metadata: meta?.metadata?.getAll() ?? {}
-  };
-}
-
-async function extractPDFText(data, { mergePages = false } = {}) {
-  const pdf = isPDFDocumentProxy(data) ? data : await getDocumentProxy(data);
-  const texts = await Promise.all(
-    Array.from({ length: pdf.numPages }, (_, i) => getPageText(pdf, i + 1))
-  );
-  const filteredTexts = texts.filter(Boolean);
-  return {
-    totalPages: pdf.numPages,
-    text: mergePages ? filteredTexts.join("\n\n") : filteredTexts
-  };
-}
-async function getPageText(pdf, pageNumber) {
-  const page = await pdf.getPage(pageNumber);
-  const content = await page.getTextContent();
-  const items = content.items;
-  if (items.length === 0)
-    return;
-  return items.map(({ str }) => str).join("\n");
-}
-
-async function getImagesFromPage(data, pageNumber) {
-  const pdf = isPDFDocumentProxy(data) ? data : await getDocumentProxy(data);
-  const page = await pdf.getPage(pageNumber);
-  const operatorList = await page.getOperatorList();
-  const { OPS } = await getResolvedPDFJS();
-  const images = [];
-  for (const op of operatorList.fnArray) {
-    if (op !== OPS.paintImageXObject)
-      continue;
-    const image = await page.objs.get(operatorList.argsArray[op][0]);
-    if (image.data)
-      images.push(image.data.buffer);
-  }
-  return images;
-}
-
-export { getImagesFromPage as a, extractPDFText as e, getPDFMeta as g };