@pixdif/pdf-parser/index.js

PDF Parser

import { Parser } from '@pixdif/parser';
import PdfOutline from './base/PdfOutline.js';
import PdfPage from './base/PdfPage.js';
import { getDocument, } from './util/pdfjs.js';
export default class PdfParser extends Parser {
    document;
    async openFile() {
        if (!this.document) {
            this.document = await getDocument(this.filePath).promise;
        }
    }
    async closeFile() {
        if (!this.document) {
            throw new Error('The PDF document is not open yet.');
        }
        await this.document.cleanup(false);
        await this.document.destroy();
        delete this.document;
    }
    async getPageNum() {
        if (this.document) {
            return this.document.numPages;
        }
        return 0;
    }
    async getPage(index) {
        if (!this.document) {
            throw new Error('The PDF file is not open yet.');
        }
        if (index < 0 || index >= await this.getPageNum()) {
            throw new Error(`Invalid page index: ${index}`);
        }
        const pageIndex = index + 1;
        const page = await this.document.getPage(pageIndex);
        return new PdfPage(this.document, `Page ${pageIndex}`, page);
    }
    async getOutline() {
        const outline = await this.document?.getOutline();
        if (!outline) {
            return [];
        }
        return outline.map(this.wrapOutline);
    }
    wrapOutline = ({ title, items, }) => {
        if (!this.document) {
            throw new Error('The PDF file is not open yet.');
        }
        const children = items && items.length > 0 ? items.map(this.wrapOutline) : undefined;
        return new PdfOutline(this.document, {
            title,
            children,
            offset: 0,
            limit: 0,
        });
    };
}