@fin.cx/einvoice
Version:
A TypeScript module for creating, manipulating, and embedding XML data within PDF files specifically tailored for electronic invoice (einvoice) packages.
74 lines • 7.23 kB
JavaScript
import { PDFDocument, PDFDict, PDFName, PDFRawStream, PDFArray, PDFString } from '../../../plugins.js';
import { BaseXMLExtractor } from './base.extractor.js';
/**
* Standard PDF XML extractor that extracts XML from embedded files
* Works with PDF/A-3 documents that follow the standard for embedding files
*/
export class StandardXMLExtractor extends BaseXMLExtractor {
/**
* Extract XML from a PDF buffer using standard PDF/A-3 embedded files
* @param pdfBuffer PDF buffer
* @returns XML content or null if not found
*/
async extractXml(pdfBuffer) {
try {
const pdfDoc = await PDFDocument.load(pdfBuffer);
// Get the document's metadata dictionary
const namesDictObj = pdfDoc.catalog.lookup(PDFName.of('Names'));
if (!(namesDictObj instanceof PDFDict)) {
console.warn('No Names dictionary found in PDF! This PDF does not contain embedded files.');
return null;
}
// Get the embedded files dictionary
const embeddedFilesDictObj = namesDictObj.lookup(PDFName.of('EmbeddedFiles'));
if (!(embeddedFilesDictObj instanceof PDFDict)) {
console.warn('No EmbeddedFiles dictionary found! This PDF does not contain embedded files.');
return null;
}
// Get the names array
const filesSpecObj = embeddedFilesDictObj.lookup(PDFName.of('Names'));
if (!(filesSpecObj instanceof PDFArray)) {
console.warn('No files specified in EmbeddedFiles dictionary!');
return null;
}
// Try to find an XML file in the embedded files
for (let i = 0; i < filesSpecObj.size(); i += 2) {
const fileNameObj = filesSpecObj.lookup(i);
const fileSpecObj = filesSpecObj.lookup(i + 1);
if (!(fileNameObj instanceof PDFString) || !(fileSpecObj instanceof PDFDict)) {
continue;
}
// Get the filename as string
const fileName = fileNameObj.decodeText();
// Check if it's a known invoice XML file name
const isKnownFileName = this.knownFileNames.some(knownName => fileName.toLowerCase() === knownName.toLowerCase());
// Check if it's any XML file or has invoice-related keywords
const isXmlFile = fileName.toLowerCase().endsWith('.xml') ||
fileName.toLowerCase().includes('zugferd') ||
fileName.toLowerCase().includes('factur-x') ||
fileName.toLowerCase().includes('xrechnung') ||
fileName.toLowerCase().includes('invoice');
if (isKnownFileName || isXmlFile) {
const efDictObj = fileSpecObj.lookup(PDFName.of('EF'));
if (!(efDictObj instanceof PDFDict)) {
continue;
}
const fileStream = efDictObj.lookup(PDFName.of('F'));
if (fileStream instanceof PDFRawStream) {
const xmlContent = await this.extractXmlFromStream(fileStream, fileName);
if (xmlContent) {
return xmlContent;
}
}
}
}
console.warn('No valid XML found in embedded files');
return null;
}
catch (error) {
console.error('Error in standard extraction:', error);
return null;
}
}
}
//# sourceMappingURL=data:application/json;base64,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