fin.cx/einvoice
2
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

fix(core): Refactor module imports to use the centralized plugins module and update relative paths across the codebase. Also remove the obsolete test file (test/test.other-formats-corpus.ts) and update file metadata in test outputs.

Browse Source
This commit is contained in:
Philipp Kunz
2025-04-03 21:07:21 +00:00
parent 5014a447a3
commit 8668ac8555
22 changed files with 70 additions and 237 deletions
Download Patch File Download Diff File Expand all files Collapse all files

22
ts/formats/pdf/extractors/associated.extractor.ts
View File

@ -1,4 +1,4 @@
import { PDFDocument, PDFDict, PDFName, PDFRawStream, PDFArray, PDFString } from 'pdf-lib';
import { PDFDocument, PDFDict, PDFName, PDFRawStream, PDFArray, PDFString } from '../../../plugins.js';
import { BaseXMLExtractor } from './base.extractor.js';
/**
@ -15,48 +15,48 @@ export class AssociatedFilesExtractor extends BaseXMLExtractor {
public async extractXml(pdfBuffer: Uint8Array | Buffer): Promise<string | null> {
try {
const pdfDoc = await PDFDocument.load(pdfBuffer);
// Try to find associated files via the AF entry in the catalog
const afArray = pdfDoc.catalog.lookup(PDFName.of('AF'));
if (!(afArray instanceof PDFArray)) {
console.warn('No AF (Associated Files) entry found in PDF catalog');
return null;
}
// Process each associated file
for (let i = 0; i < afArray.size(); i++) {
const fileSpec = afArray.lookup(i);
if (!(fileSpec instanceof PDFDict)) {
continue;
}
// Get the file name
const fileNameObj = fileSpec.lookup(PDFName.of('F')) || fileSpec.lookup(PDFName.of('UF'));
if (!(fileNameObj instanceof PDFString)) {
continue;
}
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') ||
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) {
// Get the embedded file dictionary
const efDict = fileSpec.lookup(PDFName.of('EF'));
if (!(efDict instanceof PDFDict)) {
continue;
}
// Get the file stream
const fileStream = efDict.lookup(PDFName.of('F'));
if (fileStream instanceof PDFRawStream) {
@ -67,7 +67,7 @@ export class AssociatedFilesExtractor extends BaseXMLExtractor {
}
}
}
console.warn('No valid XML found in associated files');
return null;
} catch (error) {

10
ts/formats/pdf/extractors/standard.extractor.ts
View File

@ -1,4 +1,4 @@
import { PDFDocument, PDFDict, PDFName, PDFRawStream, PDFArray, PDFString } from 'pdf-lib';
import { PDFDocument, PDFDict, PDFName, PDFRawStream, PDFArray, PDFString } from '../../../plugins.js';
import { BaseXMLExtractor } from './base.extractor.js';
/**
@ -47,19 +47,19 @@ export class StandardXMLExtractor extends BaseXMLExtractor {
// 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') ||
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)) {

2
ts/formats/pdf/pdf.embedder.ts
View File

@ -1,4 +1,4 @@
import { PDFDocument, AFRelationship } from 'pdf-lib';
import { PDFDocument, AFRelationship } from '../../plugins.js';
import type { IPdf } from '../../interfaces/common.js';
/**