unpdf 0.5.4 → 0.6.0

package/dist/shared/{unpdf.6363738c.d.ts → unpdf.3991886a.d.cts}

@@ -1,5 +1,5 @@
-import { BinaryData, PDFDocumentProxy, DocumentInitParameters } from 'pdfjs-dist/types/src/display/api';
-import PDFJS from 'pdfjs-dist';
+import { BinaryData, PDFDocumentProxy, DocumentInitParameters } from '../types/src/display/api';
+import _PDFJS from '../types/src/pdf';
 
 declare function getPDFMeta(data: BinaryData | PDFDocumentProxy): Promise<{
     info: Record<string, any>;
@@ -15,6 +15,7 @@ declare function extractPDFText(data: BinaryData | PDFDocumentProxy, options?: {
 
 declare function getImagesFromPage(data: BinaryData | PDFDocumentProxy, pageNumber: number): Promise<ArrayBuffer[]>;
 
+type PDFJS = typeof _PDFJS;
 interface UnPDFConfiguration {
     /**
      * By default, UnPDF will use the latest version of PDF.js. If you want to
@@ -22,22 +23,23 @@ interface UnPDFConfiguration {
      * the PDF.js module.
      *
      * @example
+     * // Use the legacy build
      * () => import('pdfjs-dist/legacy/build/pdf.js')
      */
-    pdfjs?: () => Promise<typeof PDFJS>;
+    pdfjs?: () => Promise<PDFJS>;
 }
 
 /**
  * Returns a PDFDocumentProxy instance from a given binary data.
  *
  * Applies the following defaults:
- * - `useWorkerFetch: false`
  * - `isEvalSupported: false`
- * - `useSystemFonts: true`
  */
 declare function getDocumentProxy(data: BinaryData, options?: DocumentInitParameters): Promise<PDFDocumentProxy>;
 declare function defineUnPDFConfig(options: UnPDFConfiguration): Promise<void>;
-declare function getResolvedPDFJS(): Promise<typeof PDFJS>;
-declare function resolvePDFJSImports(pdfjs?: () => Promise<typeof PDFJS>): Promise<void>;
+declare function getResolvedPDFJS(): Promise<PDFJS>;
+declare function resolvePDFJSImports(pdfjsResolver?: () => Promise<PDFJS>, { force }?: {
+    force?: boolean | undefined;
+}): Promise<void>;
 
 export { getImagesFromPage as a, getDocumentProxy as b, getResolvedPDFJS as c, defineUnPDFConfig as d, extractPDFText as e, getPDFMeta as g, resolvePDFJSImports as r };

package/dist/shared/{unpdf.6363738c.d.cts → unpdf.3991886a.d.mts}

@@ -1,5 +1,5 @@
-import { BinaryData, PDFDocumentProxy, DocumentInitParameters } from 'pdfjs-dist/types/src/display/api';
-import PDFJS from 'pdfjs-dist';
+import { BinaryData, PDFDocumentProxy, DocumentInitParameters } from '../types/src/display/api';
+import _PDFJS from '../types/src/pdf';
 
 declare function getPDFMeta(data: BinaryData | PDFDocumentProxy): Promise<{
     info: Record<string, any>;
@@ -15,6 +15,7 @@ declare function extractPDFText(data: BinaryData | PDFDocumentProxy, options?: {
 
 declare function getImagesFromPage(data: BinaryData | PDFDocumentProxy, pageNumber: number): Promise<ArrayBuffer[]>;
 
+type PDFJS = typeof _PDFJS;
 interface UnPDFConfiguration {
     /**
      * By default, UnPDF will use the latest version of PDF.js. If you want to
@@ -22,22 +23,23 @@ interface UnPDFConfiguration {
      * the PDF.js module.
      *
      * @example
+     * // Use the legacy build
      * () => import('pdfjs-dist/legacy/build/pdf.js')
      */
-    pdfjs?: () => Promise<typeof PDFJS>;
+    pdfjs?: () => Promise<PDFJS>;
 }
 
 /**
  * Returns a PDFDocumentProxy instance from a given binary data.
  *
  * Applies the following defaults:
- * - `useWorkerFetch: false`
  * - `isEvalSupported: false`
- * - `useSystemFonts: true`
  */
 declare function getDocumentProxy(data: BinaryData, options?: DocumentInitParameters): Promise<PDFDocumentProxy>;
 declare function defineUnPDFConfig(options: UnPDFConfiguration): Promise<void>;
-declare function getResolvedPDFJS(): Promise<typeof PDFJS>;
-declare function resolvePDFJSImports(pdfjs?: () => Promise<typeof PDFJS>): Promise<void>;
+declare function getResolvedPDFJS(): Promise<PDFJS>;
+declare function resolvePDFJSImports(pdfjsResolver?: () => Promise<PDFJS>, { force }?: {
+    force?: boolean | undefined;
+}): Promise<void>;
 
 export { getImagesFromPage as a, getDocumentProxy as b, getResolvedPDFJS as c, defineUnPDFConfig as d, extractPDFText as e, getPDFMeta as g, resolvePDFJSImports as r };

package/dist/shared/{unpdf.6363738c.d.mts → unpdf.3991886a.d.ts}

@@ -1,5 +1,5 @@
-import { BinaryData, PDFDocumentProxy, DocumentInitParameters } from 'pdfjs-dist/types/src/display/api';
-import PDFJS from 'pdfjs-dist';
+import { BinaryData, PDFDocumentProxy, DocumentInitParameters } from '../types/src/display/api';
+import _PDFJS from '../types/src/pdf';
 
 declare function getPDFMeta(data: BinaryData | PDFDocumentProxy): Promise<{
     info: Record<string, any>;
@@ -15,6 +15,7 @@ declare function extractPDFText(data: BinaryData | PDFDocumentProxy, options?: {
 
 declare function getImagesFromPage(data: BinaryData | PDFDocumentProxy, pageNumber: number): Promise<ArrayBuffer[]>;
 
+type PDFJS = typeof _PDFJS;
 interface UnPDFConfiguration {
     /**
      * By default, UnPDF will use the latest version of PDF.js. If you want to
@@ -22,22 +23,23 @@ interface UnPDFConfiguration {
      * the PDF.js module.
      *
      * @example
+     * // Use the legacy build
      * () => import('pdfjs-dist/legacy/build/pdf.js')
      */
-    pdfjs?: () => Promise<typeof PDFJS>;
+    pdfjs?: () => Promise<PDFJS>;
 }
 
 /**
  * Returns a PDFDocumentProxy instance from a given binary data.
  *
  * Applies the following defaults:
- * - `useWorkerFetch: false`
  * - `isEvalSupported: false`
- * - `useSystemFonts: true`
  */
 declare function getDocumentProxy(data: BinaryData, options?: DocumentInitParameters): Promise<PDFDocumentProxy>;
 declare function defineUnPDFConfig(options: UnPDFConfiguration): Promise<void>;
-declare function getResolvedPDFJS(): Promise<typeof PDFJS>;
-declare function resolvePDFJSImports(pdfjs?: () => Promise<typeof PDFJS>): Promise<void>;
+declare function getResolvedPDFJS(): Promise<PDFJS>;
+declare function resolvePDFJSImports(pdfjsResolver?: () => Promise<PDFJS>, { force }?: {
+    force?: boolean | undefined;
+}): Promise<void>;
 
 export { getImagesFromPage as a, getDocumentProxy as b, getResolvedPDFJS as c, defineUnPDFConfig as d, extractPDFText as e, getPDFMeta as g, resolvePDFJSImports as r };

package/dist/shared/unpdf.6c0ca5b5.cjs

@@ -0,0 +1,107 @@
+'use strict';
+
+let resolvedModule;
+async function getDocumentProxy(data, options = {}) {
+  const { getDocument } = await getResolvedPDFJS();
+  const pdf = await getDocument({
+    data,
+    isEvalSupported: false,
+    ...options
+  }).promise;
+  return pdf;
+}
+async function defineUnPDFConfig(options) {
+  const { pdfjs } = { ...options };
+  if (pdfjs) {
+    await resolvePDFJSImports(pdfjs, { force: true });
+  }
+}
+async function getResolvedPDFJS() {
+  if (!resolvedModule) {
+    await resolvePDFJSImports();
+  }
+  return resolvedModule;
+}
+async function resolvePDFJSImports(pdfjsResolver, { force = false } = {}) {
+  if (resolvedModule && !force) {
+    return;
+  }
+  if (pdfjsResolver) {
+    try {
+      const _import = await pdfjsResolver();
+      resolvedModule = _import.default || _import;
+      return;
+    } catch {
+      throw new Error(
+        "Resolving failed. Please check the provided configuration."
+      );
+    }
+  }
+  try {
+    resolvedModule = await import('unpdf/pdfjs');
+  } catch {
+    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;
+}
+
+async function getPDFMeta(data) {
+  const pdf = isPDFDocumentProxy(data) ? data : await getDocumentProxy(data);
+  const meta = await pdf.getMetadata();
+  return {
+    info: meta?.info ?? {},
+    metadata: meta?.metadata?.getAll() ?? {}
+  };
+}
+
+async function extractPDFText(data, options = {}) {
+  const { mergePages = false } = { ...options };
+  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;
+}
+
+exports.defineUnPDFConfig = defineUnPDFConfig;
+exports.extractPDFText = extractPDFText;
+exports.getDocumentProxy = getDocumentProxy;
+exports.getImagesFromPage = getImagesFromPage;
+exports.getPDFMeta = getPDFMeta;
+exports.getResolvedPDFJS = getResolvedPDFJS;
+exports.resolvePDFJSImports = resolvePDFJSImports;

package/dist/shared/unpdf.ce219a40.mjs

@@ -0,0 +1,99 @@
+let resolvedModule;
+async function getDocumentProxy(data, options = {}) {
+  const { getDocument } = await getResolvedPDFJS();
+  const pdf = await getDocument({
+    data,
+    isEvalSupported: false,
+    ...options
+  }).promise;
+  return pdf;
+}
+async function defineUnPDFConfig(options) {
+  const { pdfjs } = { ...options };
+  if (pdfjs) {
+    await resolvePDFJSImports(pdfjs, { force: true });
+  }
+}
+async function getResolvedPDFJS() {
+  if (!resolvedModule) {
+    await resolvePDFJSImports();
+  }
+  return resolvedModule;
+}
+async function resolvePDFJSImports(pdfjsResolver, { force = false } = {}) {
+  if (resolvedModule && !force) {
+    return;
+  }
+  if (pdfjsResolver) {
+    try {
+      const _import = await pdfjsResolver();
+      resolvedModule = _import.default || _import;
+      return;
+    } catch {
+      throw new Error(
+        "Resolving failed. Please check the provided configuration."
+      );
+    }
+  }
+  try {
+    resolvedModule = await import('unpdf/pdfjs');
+  } catch {
+    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;
+}
+
+async function getPDFMeta(data) {
+  const pdf = isPDFDocumentProxy(data) ? data : await getDocumentProxy(data);
+  const meta = await pdf.getMetadata();
+  return {
+    info: meta?.info ?? {},
+    metadata: meta?.metadata?.getAll() ?? {}
+  };
+}
+
+async function extractPDFText(data, options = {}) {
+  const { mergePages = false } = { ...options };
+  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, getDocumentProxy as b, getResolvedPDFJS as c, defineUnPDFConfig as d, extractPDFText as e, getPDFMeta as g, resolvePDFJSImports as r };

package/dist/types/src/display/annotation_layer.d.ts

@@ -0,0 +1,203 @@
+export type PDFPageProxy = import("./api").PDFPageProxy;
+export type PageViewport = import("./display_utils").PageViewport;
+export type IDownloadManager = any;
+export type IPDFLinkService = import("../../web/interfaces").IPDFLinkService;
+export type AnnotationElementParameters = {
+    data: Object;
+    layer: HTMLDivElement;
+    linkService: IPDFLinkService;
+    downloadManager: any;
+    annotationStorage?: AnnotationStorage | undefined;
+    /**
+     * - Path for image resources, mainly
+     * for annotation icons. Include trailing slash.
+     */
+    imageResourcesPath?: string | undefined;
+    renderForms: boolean;
+    svgFactory: Object;
+    enableScripting?: boolean | undefined;
+    hasJSActions?: boolean | undefined;
+    fieldObjects?: Object | undefined;
+};
+export type AnnotationLayerParameters = {
+    viewport: PageViewport;
+    div: HTMLDivElement;
+    annotations: any[];
+    page: PDFPageProxy;
+    linkService: IPDFLinkService;
+    downloadManager: any;
+    annotationStorage?: AnnotationStorage | undefined;
+    /**
+     * - Path for image resources, mainly
+     * for annotation icons. Include trailing slash.
+     */
+    imageResourcesPath?: string | undefined;
+    renderForms: boolean;
+    /**
+     * - Enable embedded script execution.
+     */
+    enableScripting?: boolean | undefined;
+    /**
+     * - Some fields have JS actions.
+     * The default value is `false`.
+     */
+    hasJSActions?: boolean | undefined;
+    fieldObjects?: {
+        [x: string]: Object[];
+    } | null | undefined;
+    annotationCanvasMap?: Map<string, HTMLCanvasElement> | undefined;
+    accessibilityManager?: any;
+};
+/**
+ * @typedef {Object} AnnotationLayerParameters
+ * @property {PageViewport} viewport
+ * @property {HTMLDivElement} div
+ * @property {Array} annotations
+ * @property {PDFPageProxy} page
+ * @property {IPDFLinkService} linkService
+ * @property {IDownloadManager} downloadManager
+ * @property {AnnotationStorage} [annotationStorage]
+ * @property {string} [imageResourcesPath] - Path for image resources, mainly
+ *   for annotation icons. Include trailing slash.
+ * @property {boolean} renderForms
+ * @property {boolean} [enableScripting] - Enable embedded script execution.
+ * @property {boolean} [hasJSActions] - Some fields have JS actions.
+ *   The default value is `false`.
+ * @property {Object<string, Array<Object>> | null} [fieldObjects]
+ * @property {Map<string, HTMLCanvasElement>} [annotationCanvasMap]
+ * @property {TextAccessibilityManager} [accessibilityManager]
+ */
+/**
+ * Manage the layer containing all the annotations.
+ */
+export class AnnotationLayer {
+    constructor({ div, accessibilityManager, annotationCanvasMap, l10n, page, viewport, }: {
+        div: any;
+        accessibilityManager: any;
+        annotationCanvasMap: any;
+        l10n: any;
+        page: any;
+        viewport: any;
+    });
+    div: any;
+    l10n: any;
+    page: any;
+    viewport: any;
+    zIndex: number;
+    popupShow: any[] | undefined;
+    /**
+     * Render a new annotation layer with all annotation elements.
+     *
+     * @param {AnnotationLayerParameters} params
+     * @memberof AnnotationLayer
+     */
+    render(params: AnnotationLayerParameters): Promise<void>;
+    /**
+     * Update the annotation elements on existing annotation layer.
+     *
+     * @param {AnnotationLayerParameters} viewport
+     * @memberof AnnotationLayer
+     */
+    update({ viewport }: AnnotationLayerParameters): void;
+    getEditableAnnotations(): any[];
+    getEditableAnnotation(id: any): any;
+    #private;
+}
+export class FreeTextAnnotationElement extends AnnotationElement {
+    constructor(parameters: any);
+    textContent: any;
+    textPosition: any;
+    annotationEditorType: number;
+    render(): HTMLElement | undefined;
+}
+export class InkAnnotationElement extends AnnotationElement {
+    constructor(parameters: any);
+    containerClassName: string;
+    svgElementName: string;
+    annotationEditorType: number;
+    render(): HTMLElement | undefined;
+    getElementsToTriggerPopup(): any[];
+    #private;
+}
+export class StampAnnotationElement extends AnnotationElement {
+    constructor(parameters: any);
+    render(): HTMLElement | undefined;
+}
+import { AnnotationStorage } from "./annotation_storage.js";
+declare class AnnotationElement {
+    constructor(parameters: any, { isRenderable, ignoreBorder, createQuadrilaterals, }?: {
+        isRenderable?: boolean | undefined;
+        ignoreBorder?: boolean | undefined;
+        createQuadrilaterals?: boolean | undefined;
+    });
+    isRenderable: boolean;
+    data: any;
+    layer: any;
+    linkService: any;
+    downloadManager: any;
+    imageResourcesPath: any;
+    renderForms: any;
+    svgFactory: any;
+    annotationStorage: any;
+    enableScripting: any;
+    hasJSActions: any;
+    _fieldObjects: any;
+    parent: any;
+    container: HTMLElement | undefined;
+    /**
+     * Create an empty container for the annotation's HTML element.
+     *
+     * @private
+     * @param {boolean} ignoreBorder
+     * @memberof AnnotationElement
+     * @returns {HTMLElement} A section element.
+     */
+    private _createContainer;
+    setRotation(angle: any, container?: HTMLElement | undefined): void;
+    get _commonActions(): any;
+    _dispatchEventFromSandbox(actions: any, jsEvent: any): void;
+    _setDefaultPropertiesFromJS(element: any): void;
+    /**
+     * Create quadrilaterals from the annotation's quadpoints.
+     *
+     * @private
+     * @memberof AnnotationElement
+     */
+    private _createQuadrilaterals;
+    /**
+     * Create a popup for the annotation's HTML element. This is used for
+     * annotations that do not have a Popup entry in the dictionary, but
+     * are of a type that works with popups (such as Highlight annotations).
+     *
+     * @private
+     * @memberof AnnotationElement
+     */
+    private _createPopup;
+    /**
+     * Render the annotation's HTML element(s).
+     *
+     * @public
+     * @memberof AnnotationElement
+     */
+    public render(): void;
+    /**
+     * @private
+     * @returns {Array}
+     */
+    private _getElementsByName;
+    show(): void;
+    hide(): void;
+    /**
+     * Get the HTML element(s) which can trigger a popup when clicked or hovered.
+     *
+     * @public
+     * @memberof AnnotationElement
+     * @returns {Array<HTMLElement>|HTMLElement} An array of elements or an
+     *          element.
+     */
+    public getElementsToTriggerPopup(): Array<HTMLElement> | HTMLElement;
+    addHighlightArea(): void;
+    _editOnDoubleClick(): void;
+    #private;
+}
+export {};

package/dist/types/src/display/annotation_storage.d.ts

@@ -0,0 +1,89 @@
+/**
+ * Key/value storage for annotation data in forms.
+ */
+export class AnnotationStorage {
+    onSetModified: any;
+    onResetModified: any;
+    onAnnotationEditor: any;
+    /**
+     * Get the value for a given key if it exists, or return the default value.
+     * @param {string} key
+     * @param {Object} defaultValue
+     * @returns {Object}
+     */
+    getValue(key: string, defaultValue: Object): Object;
+    /**
+     * Get the value for a given key.
+     * @param {string} key
+     * @returns {Object}
+     */
+    getRawValue(key: string): Object;
+    /**
+     * Remove a value from the storage.
+     * @param {string} key
+     */
+    remove(key: string): void;
+    /**
+     * Set the value for a given key
+     * @param {string} key
+     * @param {Object} value
+     */
+    setValue(key: string, value: Object): void;
+    /**
+     * Check if the storage contains the given key.
+     * @param {string} key
+     * @returns {boolean}
+     */
+    has(key: string): boolean;
+    /**
+     * @returns {Object | null}
+     */
+    getAll(): Object | null;
+    /**
+     * @param {Object} obj
+     */
+    setAll(obj: Object): void;
+    get size(): number;
+    resetModified(): void;
+    /**
+     * @returns {PrintAnnotationStorage}
+     */
+    get print(): PrintAnnotationStorage;
+    /**
+     * PLEASE NOTE: Only intended for usage within the API itself.
+     * @ignore
+     */
+    get serializable(): Readonly<{
+        map: null;
+        hash: "";
+        transfers: undefined;
+    }> | {
+        map: Map<any, any>;
+        hash: string;
+        transfers: any[];
+    };
+    #private;
+}
+/**
+ * A special `AnnotationStorage` for use during printing, where the serializable
+ * data is *frozen* upon initialization, to prevent scripting from modifying its
+ * contents. (Necessary since printing is triggered synchronously in browsers.)
+ */
+export class PrintAnnotationStorage extends AnnotationStorage {
+    constructor(parent: any);
+    /**
+     * PLEASE NOTE: Only intended for usage within the API itself.
+     * @ignore
+     */
+    get serializable(): {
+        map: any;
+        hash: any;
+        transfers: any;
+    };
+    #private;
+}
+export const SerializableEmpty: Readonly<{
+    map: null;
+    hash: "";
+    transfers: undefined;
+}>;