@bentopdf/pymupdf-wasm 0.1.1

package/package.json

@@ -0,0 +1,48 @@
+{
+    "name": "@bentopdf/pymupdf-wasm",
+    "version": "0.1.1",
+    "description": "PyMuPDF compiled to WebAssembly - Full PDF manipulation in the browser",
+    "type": "module",
+    "main": "dist/index.js",
+    "module": "dist/index.js",
+    "types": "types/index.d.ts",
+    "exports": {
+        ".": {
+            "import": "./dist/index.js",
+            "types": "./types/index.d.ts"
+        }
+    },
+    "files": [
+        "dist",
+        "types",
+        "assets"
+    ],
+    "scripts": {
+        "build": "esbuild src/index.ts --bundle --format=esm --outfile=dist/index.js --external:./assets/* --external:@okathira/ghostpdl-wasm",
+        "dev": "esbuild src/index.ts --bundle --format=esm --outfile=dist/index.js --watch --external:@okathira/ghostpdl-wasm",
+        "typecheck": "tsc --noEmit"
+    },
+    "repository": {
+        "type": "git",
+        "url": "git+https://github.com/alam00000/bentopdf-pymupdf-wasm.git"
+    },
+    "keywords": [
+        "pdf",
+        "pymupdf",
+        "wasm",
+        "webassembly",
+        "pyodide",
+        "document",
+        "convert",
+        "merge",
+        "split",
+        "extract"
+    ],
+    "author": "BentoPDF",
+    "license": "AGPL-3.0",
+    "bugs": "https://github.com/alam00000/bentopdf-pymupdf-wasm/issues",
+    "devDependencies": {
+        "esbuild": "^0.21.2",
+        "typescript": "^5.3.3"
+    }
+}

package/types/index.d.ts

@@ -0,0 +1,269 @@
+/**
+ * Type declarations for @bentopdf/pymupdf-wasm
+ */
+
+export interface PyMuPDFOptions {
+    assetPath?: string;
+}
+
+export interface Point { x: number; y: number; }
+export interface Rect { x0: number; y0: number; x1: number; y1: number; }
+export interface Color { r: number; g: number; b: number; }
+
+export interface DocumentMetadata {
+    title?: string;
+    author?: string;
+    subject?: string;
+    keywords?: string;
+    creator?: string;
+    producer?: string;
+    creationDate?: string;
+    modDate?: string;
+}
+
+export type TextExtractionFormat = 'text' | 'blocks' | 'words' | 'dict' | 'json' | 'html' | 'xhtml' | 'xml';
+
+export interface TextBlock {
+    type: 'text' | 'image';
+    bbox: Rect;
+    text?: string;
+}
+
+export interface ImageInfo {
+    xref: number;
+    width: number;
+    height: number;
+    bpc: number;
+    colorspace: string;
+    size: number;
+}
+
+export interface ExtractedImage extends ImageInfo {
+    data: Uint8Array;
+    ext: string;
+}
+
+export type AnnotationType = 'Text' | 'FreeText' | 'Line' | 'Square' | 'Circle' | 'Polygon' | 'PolyLine' | 'Highlight' | 'Underline' | 'StrikeOut' | 'Squiggly' | 'Stamp' | 'Caret' | 'Ink' | 'FileAttachment' | 'Link';
+
+export interface AnnotationInfo {
+    type: AnnotationType;
+    rect: Rect;
+    content?: string;
+    author?: string;
+    color?: Color;
+}
+
+export interface FormField {
+    name: string;
+    type: 'text' | 'checkbox' | 'radio' | 'choice' | 'button' | 'signature';
+    value: string | boolean | string[];
+    rect: Rect;
+    readonly?: boolean;
+}
+
+export interface SearchResult {
+    page: number;
+    rect: Rect;
+    text: string;
+}
+
+export interface RenderOptions {
+    dpi?: number;
+    alpha?: boolean;
+    rotation?: number;
+    clip?: Rect;
+}
+
+export interface EncryptionOptions {
+    userPassword?: string;
+    ownerPassword: string;
+    permissions?: {
+        print?: boolean;
+        copy?: boolean;
+        annotate?: boolean;
+        modify?: boolean;
+    };
+}
+
+export interface LinkInfo {
+    rect: Rect;
+    uri?: string;
+    page?: number;
+    dest?: Point;
+}
+
+export interface TocEntry {
+    level: number;
+    title: string;
+    page: number;
+    dest?: Point;
+}
+
+export declare class PyMuPDFPage {
+    readonly pageNumber: number;
+    readonly rect: Rect;
+    readonly width: number;
+    readonly height: number;
+    readonly rotation: number;
+
+    setRotation(angle: number): void;
+
+    // Text
+    getText(format?: TextExtractionFormat): string | TextBlock[];
+    searchFor(text: string, quads?: boolean): Rect[];
+    insertText(point: Point, text: string, options?: { fontsize?: number; fontname?: string; color?: Color; rotate?: number }): void;
+
+    // Images
+    getImages(): ImageInfo[];
+    extractImage(xref: number): ExtractedImage | null;
+    insertImage(rect: Rect, imageData: Uint8Array, options?: { overlay?: boolean; keepProportion?: boolean }): void;
+
+    // Annotations
+    getAnnotations(): AnnotationInfo[];
+    addHighlight(rect: Rect, color?: Color): void;
+    addTextAnnotation(point: Point, text: string, icon?: string): void;
+    addRectAnnotation(rect: Rect, color?: Color, fill?: Color): void;
+    deleteAnnotations(): void;
+
+    // Links
+    getLinks(): LinkInfo[];
+    insertLink(rect: Rect, uri: string): void;
+
+    // Rendering
+    toImage(options?: RenderOptions): Promise<Uint8Array>;
+    toSvg(): string;
+
+    // Redaction
+    addRedaction(rect: Rect, text?: string, fill?: Color): void;
+    applyRedactions(): void;
+
+    // Drawing
+    drawLine(from: Point, to: Point, color?: Color, width?: number): void;
+    drawRect(rect: Rect, color?: Color, fill?: Color, width?: number): void;
+    drawCircle(center: Point, radius: number, color?: Color, fill?: Color): void;
+}
+
+export declare class PyMuPDFDocument {
+    readonly pageCount: number;
+    readonly isPdf: boolean;
+    readonly isEncrypted: boolean;
+    readonly needsPass: boolean;
+    readonly metadata: DocumentMetadata;
+    readonly isFormPdf: boolean;
+
+    setMetadata(metadata: Partial<DocumentMetadata>): void;
+
+    // Pages
+    getPage(index: number): PyMuPDFPage;
+    pages(): Generator<PyMuPDFPage>;
+    deletePage(index: number): void;
+    deletePages(indices: number[]): void;
+    insertBlankPage(index: number, width?: number, height?: number): PyMuPDFPage;
+    movePage(from: number, to: number): void;
+    copyPage(from: number, to: number): void;
+    selectPages(indices: number[]): void;
+
+    // Merging
+    insertPdf(sourceDoc: PyMuPDFDocument, options?: { fromPage?: number; toPage?: number; startAt?: number; rotate?: number }): void;
+
+    // Conversion
+    convertToPdf(): Uint8Array;
+
+    // Search
+    searchText(query: string): SearchResult[];
+
+    // TOC
+    getToc(): TocEntry[];
+    setToc(toc: TocEntry[]): void;
+
+    // Forms
+    getFormFields(): FormField[];
+    setFormField(name: string, value: string | boolean): void;
+
+    // Security
+    authenticate(password: string): boolean;
+
+    // Save
+    save(options?: { garbage?: number; deflate?: boolean; clean?: boolean; encryption?: EncryptionOptions }): Uint8Array;
+    saveAsBlob(options?: { garbage?: number; deflate?: boolean; clean?: boolean; encryption?: EncryptionOptions }): Blob;
+
+    close(): void;
+}
+
+export declare class PyMuPDF {
+    constructor(options?: PyMuPDFOptions | string);
+
+    load(): Promise<void>;
+
+    // Document operations
+    open(input: Blob | File): Promise<PyMuPDFDocument>;
+    openUrl(url: string): Promise<PyMuPDFDocument>;
+    create(): Promise<PyMuPDFDocument>;
+
+    // PDF2DOCX conversion
+    pdfToDocx(pdf: Blob | File, pages?: number[]): Promise<Blob>;
+
+    // PDF to EPUB conversion (uses Pandoc WASM)
+    pdfToEpub(pdf: Blob | File, options?: {
+        title?: string;
+        author?: string;
+        toc?: boolean;
+        pandocAssetPath?: string;
+    }): Promise<Blob>;
+
+    // Utilities
+    merge(pdfs: (Blob | File)[]): Promise<Blob>;
+    split(pdf: Blob | File, ranges: Array<{ start: number; end: number }>): Promise<Blob[]>;
+    extractText(pdf: Blob | File): Promise<string>;
+    renderPage(pdf: Blob | File, pageIndex: number, dpi?: number): Promise<Uint8Array>;
+
+    // File to PDF conversion
+    // Supports: XPS, EPUB, MOBI, FB2, CBZ, SVG, images (JPEG, PNG, BMP, GIF, TIFF, WEBP)
+    convertToPdf(file: Blob | File, options?: { filetype?: string }): Promise<Blob>;
+    xpsToPdf(xps: Blob | File): Promise<Blob>;
+    epubToPdf(epub: Blob | File): Promise<Blob>;
+    imageToPdf(image: Blob | File, options?: { imageType?: string }): Promise<Blob>;
+    svgToPdf(svg: Blob | File): Promise<Blob>;
+    imagesToPdf(images: (Blob | File)[]): Promise<Blob>;
+
+    // PDF to other formats
+    // Images: PNG, JPEG, PNM, PGM, PBM, PPM, PAM, PSD, PS
+    // Vector: SVG | Text: plain, HTML, XML, JSON
+    pdfToImages(pdf: Blob | File, options?: {
+        format?: 'png' | 'jpeg' | 'pnm' | 'pgm' | 'pbm' | 'ppm' | 'pam' | 'psd' | 'ps';
+        dpi?: number;
+        pages?: number[];
+    }): Promise<Uint8Array[]>;
+    pdfToSvg(pdf: Blob | File, pages?: number[]): Promise<string[]>;
+    pdfToText(pdf: Blob | File): Promise<string>;
+    pdfToHtml(pdf: Blob | File): Promise<string>;
+    pdfToJson(pdf: Blob | File): Promise<object[]>;
+    pdfToXml(pdf: Blob | File): Promise<string>;
+
+    // LLM / Markdown conversion (pymupdf4llm)
+    pdfToMarkdown(pdf: Blob | File, options?: {
+        pageBreaks?: boolean;
+        includeImages?: boolean;
+        pages?: number[];
+    }): Promise<string>;
+    pdfToLlmChunks(pdf: Blob | File): Promise<Array<{
+        text: string;
+        metadata: { page?: number; heading?: string };
+    }>>;
+
+    // HTML to PDF conversion - experimental 
+    // TODO@ALAM - revisit this
+    // htmlToPdf(html: string, options?: {
+    //     css?: string;
+    //     pageSize?: 'a4' | 'letter' | 'legal' | 'a3' | 'a5';
+    //     margins?: number | { top: number; right: number; bottom: number; left: number };
+    // }): Promise<Blob>;
+
+    // Text to PDF conversion
+    textToPdf(text: string, options?: {
+        fontName?: 'helv' | 'tiro' | 'cour' | 'times';
+        fontSize?: number;
+        pageSize?: 'a4' | 'letter' | 'legal' | 'a3' | 'a5';
+        margins?: number;
+    }): Promise<Blob>;
+}