ppu-pdf
Version:
Easily extract text from digital PDF files with coordinate and font size included, and optionally group text by lines or render scanned pdf to canvas/png.
104 lines (103 loc) • 5.06 kB
TypeScript
import "pdfjs-dist/build/pdf.worker.min.mjs";
import "./pdfjs-workaround.js";
import { type Canvas } from "@napi-rs/canvas";
import * as pdfjs from "pdfjs-dist/legacy/build/pdf.mjs";
import { PdfReaderCommon } from "./pdf-reader-common.js";
import { type CanvasMap, type CompactPageLines, type PageLines, type PageTexts, type PageToonLines, type PdfCompactLineAlgorithm, type PdfReaderOptions, type PdfScannedThreshold } from "./pdf.interface.js";
import { type PaddleOcrService } from "ppu-paddle-ocr";
/**
* PdfReaderLegacy class based on pdfjs-dist for reading and processing PDF documents.
*/
export declare class PdfReaderLegacy extends PdfReaderCommon {
private options;
readonly startIndex = 1;
constructor(options?: Partial<PdfReaderOptions>);
/**
* Opens a PDF document from a file path or an ArrayBuffer.
* @param filename - The file path or ArrayBuffer of the PDF document.
* @returns The opened PDFDocument instance.
*/
open(filename: string | ArrayBuffer): Promise<pdfjs.PDFDocumentProxy>;
/**
* Renders all pages of a PDF document into canvases.
* @param doc - The PDFDocumentProxy to render.
* @returns A map of page numbers to Canvas instances.
*/
renderAll(doc: pdfjs.PDFDocumentProxy): Promise<CanvasMap>;
/**
* Extracts text from scanned PDF pages using ppu-paddle-ocr package.
* @param paddleOcrService - The OCR service instance specifically from ppu-paddle-ocr to use for text recognition.
* @param canvasMap - A map of page numbers to Canvas instances representing rendered PDF pages.
* @returns A map of page numbers to extracted text data with OCR results.
*/
getTextsScanned(paddleOcrService: PaddleOcrService, canvasMap: CanvasMap): Promise<PageTexts>;
private getCanvas;
/**
* Extracts text from all pages of a PDF document.
* @param doc - The PDFDocumentProxy to extract text from.
* @returns A map of page numbers to extracted text data.
*/
getTexts(pdf: pdfjs.PDFDocumentProxy): Promise<PageTexts>;
private extractTexts;
private extractOcrTexts;
private convertOcrToPdfWords;
private mapTokenToPdfWord;
private mergeTextContent;
private filterTextContent;
/**
* Converts extracted text into structured lines.
* @param pageTexts - The extracted text data from a PDF.
* @returns A map of page numbers to structured lines.
*/
getLinesFromTexts(pageTexts: PageTexts): PageLines;
/**
* Converts extracted text into TOON format string for LLM-friendly input.
* @param pageTexts - The extracted text data from a PDF.
* @returns A string of TOON format
*/
getLinesFromTextsInToon(pageTexts: PageTexts): PageToonLines;
/**
* Converts extracted text into compact structured lines using a specified algorithm.
* @param pageTexts - The extracted text data from a PDF.
* @param algorithm - The algorithm for compacting lines (default: "middleY").
* @returns A map of page numbers to compact structured lines.
*/
getCompactLinesFromTexts(pageTexts: PageTexts, algorithm?: PdfCompactLineAlgorithm): CompactPageLines;
/**
* Determines if the PDF document is scanned based on text thresholds.
* @param pageTexts - The extracted text data from a PDF.
* @param options - The threshold options for scanned detection.
* @returns True if the document is likely scanned, false otherwise.
*/
isScanned(pageTexts: PageTexts, options?: PdfScannedThreshold): boolean;
/**
* Determines if the individual PDF page is a scanned/digital based on text thresholds.
* @param pageText - The extracted page text.
* @param options - The threshold options for scanned detection.
* @returns True if the page is likely scanned, false otherwise.
*/
isPageScanned(pageText: string, options?: PdfScannedThreshold): boolean;
/**
* Saves rendered canvases as image files.
* @param canvasMap - The map of canvases to save.
* @param filename - The base filename for the output images.
* @param foldername - The folder to save the images in (default: "out").
*/
dumpCanvasMap(canvasMap: Map<number, Canvas>, filename: string, foldername?: string): Promise<void>;
/**
* Rebuilds a scanned PDF by placing invisible text over the orginial images,
* making the PDF searchable without altering its visual appearance.
* @param doc - The PDFDocumentProxy instance to rebuild.
* @param pageTexts - The extracted text data to overlay.
* @param options - Rebuild options (optional, default font is Helvetica).
* @returns A Uint8Array containing the rebuilt PDF binary data.
*/
rebuild(doc: pdfjs.PDFDocumentProxy, pageTexts: PageTexts, options?: {
fontName?: string;
}): Promise<Uint8Array>;
/**
* Destroys the PDF document instance to free memory.
* @param doc - The PDFDocumentProxy instance to destroy.
*/
destroy(pdf: pdfjs.PDFDocumentProxy): Promise<void>;
}