bea-edit/src/lib/zip-processing.ts

import { PDFDocument } from 'pdf-lib';

import { copyPagesToNewPdf } from './pdf-processing';
import { parseXJustizMetadata, type XJustizMetadata } from './services/xml-reading.service';
import { unzipArchive } from './services/zip-inflating.service';

export type ZipAttachmentKind = 'pdf' | 'image';

export type ZipAttachment = {
	name: string;
	path: string;
	kind: ZipAttachmentKind;
	data: Uint8Array;
};

export type SubDocument = {
	id: string;
	name: string;
	attachments: ZipAttachment[];
};

export type ProcessedZipArchive = {
	name: string;
	attachments: ZipAttachment[];
	subDocuments: SubDocument[];
};

export type ExportMode = 'single' | 'separate';

export type ExportUnit = {
	name: string;
	bytes: Uint8Array;
};

/**
 * Drag source/target descriptors shared between the archive editor and the
 * sub-document editor so HTML5 `dataTransfer` payloads stay consistent.
 */
export type AttachmentLocation =
	| { kind: 'loose'; index: number }
	| { kind: 'subDocument'; id: string; index: number };

const IMAGE_EXTENSIONS = new Set([
	'.avif',
	'.bmp',
	'.gif',
	'.heic',
	'.jpeg',
	'.jpg',
	'.jp2',
	'.png',
	'.tif',
	'.tiff',
	'.webp'
]);

const ZIP_META_FILE_NAME = 'xjustiz_nachricht.xml';

const getBaseName = (path: string) => path.split('/').pop() ?? path;

const getCurrentDateString = (date: Date) =>
	`${date.getFullYear()}-${String(date.getMonth() + 1).padStart(2, '0')}-${String(date.getDate()).padStart(2, '0')}`;

const createArchiveName = (date: Date, metadata: XJustizMetadata | null) => {
	const datePrefix = getCurrentDateString(date);

	if (!metadata) {
		return `${datePrefix}_unbekannt_unbekannt.pdf`;
	}

	return `${datePrefix}_${metadata.sender}_${metadata.receiver}.pdf`;
};

const getExtension = (path: string) => {
	const baseName = getBaseName(path).toLowerCase();
	const lastDotIndex = baseName.lastIndexOf('.');

	return lastDotIndex === -1 ? '' : baseName.slice(lastDotIndex);
};

const isPdf = (path: string) => getExtension(path) === '.pdf';

const isImage = (path: string) => IMAGE_EXTENSIONS.has(getExtension(path));

const isZipArchive = (path: string) => getExtension(path) === '.zip';

const normalizeKey = (value: string) => value.replaceAll('\\', '/').toLowerCase().trim();

const isMacMetadataEntry = (path: string) => {
	const normalizedPath = normalizeKey(path);
	const baseName = getBaseName(normalizedPath);

	return (
		normalizedPath.startsWith('__macosx/') ||
		normalizedPath.includes('/__macosx/') ||
		baseName === '.ds_store' ||
		baseName.startsWith('._')
	);
};
const orderAttachmentsByDocumentNames = (attachments: ZipAttachment[], documentNames: string[]) => {
	const buckets = new Map<string, ZipAttachment[]>();

	for (const attachment of attachments) {
		const keys = new Set([normalizeKey(attachment.path), normalizeKey(attachment.name)]);

		for (const key of keys) {
			const bucket = buckets.get(key) ?? [];
			bucket.push(attachment);
			buckets.set(key, bucket);
		}
	}

	const usedAttachments = new Set<ZipAttachment>();
	const orderedAttachments: ZipAttachment[] = [];

	const takeAttachment = (key: string) => {
		const bucket = buckets.get(key);

		if (!bucket) {
			return null;
		}

		const nextAttachment = bucket.find((attachment) => !usedAttachments.has(attachment)) ?? null;

		if (nextAttachment) {
			usedAttachments.add(nextAttachment);
		}

		return nextAttachment;
	};

	for (const documentName of documentNames) {
		const normalizedName = normalizeKey(documentName);
		const normalizedBaseName = normalizeKey(getBaseName(documentName));
		const matchedAttachment = takeAttachment(normalizedName) ?? takeAttachment(normalizedBaseName);

		if (matchedAttachment) {
			orderedAttachments.push(matchedAttachment);
		}
	}

	for (const attachment of attachments) {
		if (!usedAttachments.has(attachment)) {
			orderedAttachments.push(attachment);
		}
	}

	return orderedAttachments;
};

const shouldSkipZipEntry = (path: string) => isMacMetadataEntry(path);

const isMetadataFile = (path: string) => getBaseName(path).toLowerCase() === ZIP_META_FILE_NAME;

const createAttachment = (path: string, data: Uint8Array): ZipAttachment => ({
	name: getBaseName(path),
	path,
	kind: isPdf(path) ? 'pdf' : 'image',
	data
});

const extractZipEntries = async (archiveBytes: Uint8Array): Promise<ProcessedZipArchive[]> => {
	const files = await unzipArchive(archiveBytes);
	const attachments: ZipAttachment[] = [];
	const nestedArchives: ProcessedZipArchive[] = [];
	let xjustizNachrichtXml: Uint8Array | null = null;

	for (const [path, data] of Object.entries(files)) {
		if (shouldSkipZipEntry(path)) {
			continue;
		}

		if (isMetadataFile(path)) {
			xjustizNachrichtXml = data;
			continue;
		}

		if (isZipArchive(path)) {
			const childArchives = await extractZipEntries(data);
			nestedArchives.push(...childArchives);
			continue;
		}

		if (isPdf(path) || isImage(path)) {
			attachments.push(createAttachment(path, data));
		}
	}

	const metadata = xjustizNachrichtXml ? parseXJustizMetadata(xjustizNachrichtXml) : null;
	const documentNames = metadata?.documentNames ?? [];
	const orderedAttachments = orderAttachmentsByDocumentNames(attachments, documentNames);
	const archiveName = createArchiveName(new Date(), metadata);
	const archives: ProcessedZipArchive[] = [];

	if (orderedAttachments.length > 0) {
		archives.push({
			name: archiveName,
			attachments: orderedAttachments,
			subDocuments: []
		});
	}

	return [...archives, ...nestedArchives];
};

export const extractZipArchives = async (file: File): Promise<ProcessedZipArchive[]> => {
	const archiveBytes = new Uint8Array(await file.arrayBuffer());

	return extractZipEntries(archiveBytes);
};

const loadImageElement = (src: string) =>
	new Promise<HTMLImageElement>((resolve, reject) => {
		const image = new Image();
		image.decoding = 'async';
		image.onload = () => resolve(image);
		image.onerror = () => reject(new Error(`Unable to decode image asset: ${src}`));
		image.src = src;
	});

const toArrayBuffer = (bytes: Uint8Array) =>
	(() => {
		const copy = new Uint8Array(bytes.byteLength);
		copy.set(bytes);
		return copy.buffer;
	})();

const convertImageBytesToPng = async (imageBytes: Uint8Array) => {
	if (typeof document === 'undefined') {
		throw new Error('Image conversion requires a browser environment.');
	}

	const objectUrl = URL.createObjectURL(new Blob([toArrayBuffer(imageBytes)]));

	try {
		const image = await loadImageElement(objectUrl);
		const canvas = document.createElement('canvas');
		canvas.width = image.naturalWidth || image.width;
		canvas.height = image.naturalHeight || image.height;

		const context = canvas.getContext('2d');

		if (!context) {
			throw new Error('Could not create a canvas context for image conversion.');
		}

		context.drawImage(image, 0, 0);

		const pngBlob = await new Promise<Blob>((resolve, reject) => {
			canvas.toBlob((blob) => {
				if (blob) {
					resolve(blob);
					return;
				}

				reject(new Error('Could not encode image asset as PNG.'));
			}, 'image/png');
		});

		return new Uint8Array(await pngBlob.arrayBuffer());
	} finally {
		URL.revokeObjectURL(objectUrl);
	}
};

const embedAttachmentImage = async (pdfDocument: PDFDocument, attachment: ZipAttachment) => {
	const extension = getExtension(attachment.path);
	const imageData = attachment.data;

	if (extension === '.jpg' || extension === '.jpeg') {
		return pdfDocument.embedJpg(imageData);
	}

	if (extension === '.png') {
		return pdfDocument.embedPng(imageData);
	}

	const pngBytes = await convertImageBytesToPng(imageData);

	return pdfDocument.embedPng(pngBytes);
};

const mergeAttachmentIntoPdf = async (mergedPdf: PDFDocument, attachment: ZipAttachment) => {
	if (attachment.kind === 'pdf') {
		const sourcePdf = await PDFDocument.load(attachment.data, { ignoreEncryption: true });
		const copiedPages = await mergedPdf.copyPages(sourcePdf, sourcePdf.getPageIndices());

		for (const page of copiedPages) {
			mergedPdf.addPage(page);
		}

		return;
	}

	try {
		const embeddedImage = await embedAttachmentImage(mergedPdf, attachment);
		const page = mergedPdf.addPage([embeddedImage.width, embeddedImage.height]);

		page.drawImage(embeddedImage, {
			x: 0,
			y: 0,
			width: page.getWidth(),
			height: page.getHeight()
		});
	} catch (error) {
		throw error;
	}
};

export const mergeAttachments = async (attachments: ZipAttachment[]): Promise<Uint8Array> => {
	if (attachments.length === 0) {
		throw new Error('Cannot create a PDF from an empty attachment list.');
	}

	const mergedPdf = await PDFDocument.create();

	for (const attachment of attachments) {
		await mergeAttachmentIntoPdf(mergedPdf, attachment);
	}

	return mergedPdf.save();
};

export const mergeProcessedZipArchive = async (archive: ProcessedZipArchive) => {
	if (archive.attachments.length === 0 && archive.subDocuments.length === 0) {
		throw new Error('Cannot create a PDF from an empty archive.');
	}

	return mergeAttachments(flattenArchiveForMerge(archive));
};

/**
 * Flatten the archive into a single ordered attachment list, preserving
 * sub-document order first and appending loose attachments at the end.
 */
export const flattenArchiveForMerge = (archive: ProcessedZipArchive): ZipAttachment[] => {
	const ordered: ZipAttachment[] = [];

	for (const subDocument of archive.subDocuments) {
		ordered.push(...subDocument.attachments);
	}

	ordered.push(...archive.attachments);

	return ordered;
};

/**
 * Create a new sub-document with a stable id from a list of attachments.
 */
export const createSubDocument = (name: string, attachments: ZipAttachment[]): SubDocument => ({
	id: crypto.randomUUID(),
	name,
	attachments: attachments.map((attachment) => ({ ...attachment }))
});

/**
 * Reorder sub-documents inside an archive. Matches the semantics of a single
 * `splice(fromIndex, 1)` -> `splice(toIndex, 0, moved)` move.
 */
export const moveSubDocument = (
	archive: ProcessedZipArchive,
	fromIndex: number,
	toIndex: number
): ProcessedZipArchive => {
	if (
		fromIndex === toIndex ||
		fromIndex < 0 ||
		toIndex < 0 ||
		fromIndex >= archive.subDocuments.length ||
		toIndex >= archive.subDocuments.length
	) {
		return archive;
	}

	const nextSubDocuments = [...archive.subDocuments];
	const [movedSubDocument] = nextSubDocuments.splice(fromIndex, 1);

	nextSubDocuments.splice(toIndex, 0, movedSubDocument);

	return { ...archive, subDocuments: nextSubDocuments };
};

const removeAt = <T>(list: T[], index: number): T | null => {
	if (index < 0 || index >= list.length) {
		return null;
	}

	const [item] = list.splice(index, 1);

	return item ?? null;
};

const insertAt = <T>(list: T[], index: number, item: T) => {
	const clampedIndex = Math.max(0, Math.min(index, list.length));
	list.splice(clampedIndex, 0, item);
};

/**
 * Reorder an attachment within its current location (loose list or a single
 * sub-document's attachment list). Used by the per-list drag handlers.
 */
export const moveAttachmentInPlace = (
	archive: ProcessedZipArchive,
	location: AttachmentLocation,
	toIndex: number
): ProcessedZipArchive => {
	if (location.kind === 'loose') {
		const nextAttachments = [...archive.attachments];
		const moved = removeAt(nextAttachments, location.index);

		if (!moved) {
			return archive;
		}

		insertAt(nextAttachments, toIndex, moved);

		return { ...archive, attachments: nextAttachments };
	}

	let changed = false;

	const nextSubDocuments = archive.subDocuments.map((subDocument) => {
		if (subDocument.id !== location.id) {
			return subDocument;
		}

		const nextAttachments = [...subDocument.attachments];
		const moved = removeAt(nextAttachments, location.index);

		if (!moved) {
			return subDocument;
		}

		insertAt(nextAttachments, toIndex, moved);
		changed = true;

		return { ...subDocument, attachments: nextAttachments };
	});

	return changed ? { ...archive, subDocuments: nextSubDocuments } : archive;
};

/**
 * Move an attachment from a source location to a target location. Either side
 * may be the loose list or any sub-document. The attachment moves (not copies),
 * so the total attachment count is invariant. When source and target point to
 * the same list, this degenerates to an in-place reorder (explicitly delegated
 * to `moveAttachmentInPlace` so the indices are normalized correctly).
 */
export const moveAttachmentIntoSubDocument = (
	archive: ProcessedZipArchive,
	source: AttachmentLocation,
	target: AttachmentLocation
): ProcessedZipArchive => {
	const sameList =
		source.kind === 'loose' && target.kind === 'loose'
			? true
			: source.kind === 'subDocument' && target.kind === 'subDocument'
				? source.id === target.id
				: false;

	if (sameList) {
		return moveAttachmentInPlace(archive, source, target.index);
	}

	const nextSubDocuments = archive.subDocuments.map((subDocument) => ({
		...subDocument,
		attachments: [...subDocument.attachments]
	}));
	const looseAttachments = [...archive.attachments];

	const removeFromList = (location: AttachmentLocation): ZipAttachment | null => {
		if (location.kind === 'loose') {
			return removeAt(looseAttachments, location.index);
		}

		const target_ = nextSubDocuments.find((item) => item.id === location.id);

		return target_ ? removeAt(target_.attachments, location.index) : null;
	};

	const insertIntoList = (attachment: ZipAttachment, location: AttachmentLocation) => {
		if (location.kind === 'loose') {
			insertAt(looseAttachments, location.index, attachment);
			return;
		}

		const target_ = nextSubDocuments.find((item) => item.id === location.id);

		if (!target_) {
			// fall back to loose list end if the referenced sub-document vanished
			looseAttachments.push(attachment);
			return;
		}

		insertAt(target_.attachments, location.index, attachment);
	};

	const movedAttachment = removeFromList(source);

	if (!movedAttachment) {
		return archive;
	}

	insertIntoList(movedAttachment, target);

	return {
		...archive,
		attachments: looseAttachments,
		subDocuments: nextSubDocuments
	};
};

/**
 * Remove a single attachment from a sub-document and return it to the end of
 * the loose list. Useful as a targeted "move back to loose" action.
 */
export const removeAttachmentFromSubDocument = (
	archive: ProcessedZipArchive,
	subDocumentId: string,
	attachmentIndex: number
): ProcessedZipArchive => {
	const nextSubDocuments = archive.subDocuments.map((subDocument) => ({
		...subDocument,
		attachments: [...subDocument.attachments]
	}));
	const nextLooseAttachments = [...archive.attachments];

	const target_ = nextSubDocuments.find((item) => item.id === subDocumentId);

	if (!target_) {
		return archive;
	}

	const moved = removeAt(target_.attachments, attachmentIndex);

	if (!moved) {
		return archive;
	}

	nextLooseAttachments.push(moved);

	return {
		...archive,
		attachments: nextLooseAttachments,
		subDocuments: nextSubDocuments
	};
};

/**
 * Remove an entire sub-document and return its attachments to the end of the
 * loose list. The sub-document ids of the remaining sub-documents are
 * preserved.
 */
export const dissolveSubDocument = (
	archive: ProcessedZipArchive,
	subDocumentId: string
): ProcessedZipArchive => {
	const dissolved = archive.subDocuments.find((item) => item.id === subDocumentId);

	if (!dissolved) {
		return archive;
	}

	return {
		...archive,
		attachments: [...archive.attachments, ...dissolved.attachments],
		subDocuments: archive.subDocuments.filter((item) => item.id !== subDocumentId)
	};
};

const stripPdfExtension = (name: string) =>
	name.toLowerCase().endsWith('.pdf') ? name.slice(0, -'.pdf'.length) : name;

/**
 * Build a single PDF export unit for one sub-document. The resulting filename
 * follows the same convention as the separate archive export:
 * `{archiveName} - {subDocumentName}` (without duplicate `.pdf` suffixes).
 */
export const buildSubDocumentExport = async (
	archiveName: string,
	subDocument: SubDocument
): Promise<ExportUnit> => {
	if (subDocument.attachments.length === 0) {
		throw new Error('Cannot create a PDF from an empty sub-document.');
	}

	return {
		name: `${stripPdfExtension(archiveName)} - ${stripPdfExtension(subDocument.name)}`,
		bytes: await mergeAttachments(subDocument.attachments)
	};
};

/**
 * Build one or more PDF export units from the archive.
 *
 * - `mode: 'single'` mirrors today's behaviour: one merged PDF covering
 *   sub-documents first, then loose attachments, named after the archive.
 * - `mode: 'separate'` produces one PDF per sub-document (named
 *   `{archiveName} - {subDocumentName}.pdf`) plus a single merged PDF for any
 *   loose attachments (named after the archive). Empty groups are skipped.
 */
export const buildArchiveExport = async (
	archive: ProcessedZipArchive,
	mode: ExportMode
): Promise<ExportUnit[]> => {
	if (mode === 'single') {
		const attachments = flattenArchiveForMerge(archive);

		if (attachments.length === 0) {
			throw new Error('Cannot create a PDF from an empty archive.');
		}

		return [
			{
				name: stripPdfExtension(archive.name),
				bytes: await mergeAttachments(attachments)
			}
		];
	}

	const separateUnits: ExportUnit[] = [];

	for (const subDocument of archive.subDocuments) {
		if (subDocument.attachments.length === 0) {
			continue;
		}

		separateUnits.push(await buildSubDocumentExport(archive.name, subDocument));
	}

	if (archive.attachments.length > 0) {
		separateUnits.push({
			name: stripPdfExtension(archive.name),
			bytes: await mergeAttachments(archive.attachments)
		});
	}

	if (separateUnits.length === 0) {
		throw new Error('Cannot create a PDF from an empty archive.');
	}

	return separateUnits;
};

/**
 * One sub-document produced by splitting a PDF attachment into page groups.
 */
export type PdfSplitGroup = {
	name: string;
	attachment: ZipAttachment;
};

/**
 * Result of a page-level PDF split. `source` is the remaining source
 * attachment after cutting out the assigned pages; `null` removes the
 * source from the archive entirely (all pages assigned or copy mode off).
 */
export type PdfSplitResult = {
	groups: PdfSplitGroup[];
	source: ZipAttachment | null;
};

/**
 * Turn a PDF file dropped into the beA workspace into a single-attachment
 * archive so the editor's splitting/grouping workflow can operate on it.
 */
export const pdfFileToArchive = async (file: File): Promise<ProcessedZipArchive> => {
	const bytes = new Uint8Array(await file.arrayBuffer());
	const sourcePdf = await PDFDocument.load(bytes);

	if (sourcePdf.getPageCount() === 0) {
		throw new Error('The PDF does not contain any pages.');
	}

	const name = file.name.toLowerCase().endsWith('.pdf') ? file.name : `${file.name}.pdf`;

	return {
		name,
		attachments: [
			{
				name,
				path: file.name,
				kind: 'pdf',
				data: bytes
			}
		],
		subDocuments: []
	};
};

/**
 * Copy a set of pages (1-based) of a PDF attachment into a new synthetic
 * attachment. The generated `path` carries a unique suffix because `path` is
 * the identity used for keyed lists, drag payloads, and selection state.
 */
export const createAttachmentFromPages = async (
	source: ZipAttachment,
	pages: number[],
	name?: string
): Promise<ZipAttachment> => {
	if (pages.length === 0) {
		throw new Error('Cannot create an attachment from an empty page selection.');
	}

	const minPage = Math.min(...pages);
	const maxPage = Math.max(...pages);
	const baseName = stripPdfExtension(source.name);
	const attachmentName =
		name ??
		(pages.length === 1
			? `${baseName}_Seite_${pages[0]}.pdf`
			: `${baseName}_Seiten_${minPage}-${maxPage}.pdf`);
	const bytes = await copyPagesToNewPdf(
		source.data,
		pages.map((page) => page - 1)
	);

	return {
		name: attachmentName,
		path: `${source.path}#seiten-${crypto.randomUUID()}`,
		kind: 'pdf',
		data: bytes
	};
};

/**
 * Remove a set of pages (1-based) from a PDF attachment. Returns the remaining
 * attachment, `null` when no pages are left, or the unchanged source when the
 * page set removes nothing.
 */
export const removePagesFromAttachment = async (
	source: ZipAttachment,
	pages: number[]
): Promise<ZipAttachment | null> => {
	const sourcePdf = await PDFDocument.load(source.data);
	const pageCount = sourcePdf.getPageCount();
	const pagesToRemove = new Set(pages);
	const keepIndexes: number[] = [];

	for (let pageIndex = 0; pageIndex < pageCount; pageIndex += 1) {
		if (!pagesToRemove.has(pageIndex + 1)) {
			keepIndexes.push(pageIndex);
		}
	}

	if (keepIndexes.length === 0) {
		return null;
	}

	if (keepIndexes.length === pageCount) {
		return { ...source };
	}

	return {
		...source,
		data: await copyPagesToNewPdf(source.data, keepIndexes)
	};
};

/**
 * Apply a completed PDF split to the archive: the source attachment (located
 * by its unique `path`) is replaced by the remaining pages (or removed), and
 * the new sub-documents are inserted directly after the source's parent
 * sub-document so the reading order is preserved. A loose source appends the
 * new sub-documents at the end.
 */
export const applyPdfSplit = (
	archive: ProcessedZipArchive,
	sourcePath: string,
	result: PdfSplitResult
): ProcessedZipArchive => {
	let sourceSubDocumentId: string | null = null;
	let insertIndex = -1;

	const nextSubDocuments: SubDocument[] = [];

	for (const subDocument of archive.subDocuments) {
		const sourceIndex = subDocument.attachments.findIndex(
			(attachment) => attachment.path === sourcePath
		);

		if (sourceIndex === -1) {
			nextSubDocuments.push(subDocument);
			continue;
		}

		const remainingAttachments = subDocument.attachments.filter(
			(attachment) => attachment.path !== sourcePath
		);

		if (result.source) {
			remainingAttachments.splice(sourceIndex, 0, { ...result.source });
		}

		sourceSubDocumentId = subDocument.id;
		insertIndex = nextSubDocuments.length;
		nextSubDocuments.push({ ...subDocument, attachments: remainingAttachments });
	}

	const looseSourceIndex = archive.attachments.findIndex(
		(attachment) => attachment.path === sourcePath
	);

	if (looseSourceIndex === -1 && sourceSubDocumentId === null) {
		return archive;
	}

	const newSubDocuments = result.groups.map((group) =>
		createSubDocument(group.name, [group.attachment])
	);
	const nextLooseAttachments = archive.attachments.filter(
		(attachment) => attachment.path !== sourcePath
	);

	if (sourceSubDocumentId === null) {
		if (result.source) {
			nextLooseAttachments.splice(looseSourceIndex, 0, { ...result.source });
		}

		return {
			...archive,
			attachments: nextLooseAttachments,
			subDocuments: [...nextSubDocuments, ...newSubDocuments]
		};
	}

	nextSubDocuments.splice(insertIndex + 1, 0, ...newSubDocuments);

	return {
		...archive,
		attachments: nextLooseAttachments,
		subDocuments: nextSubDocuments
	};
};

/**
 * Next free "Teildokument N" number, derived from the maximum existing
 * trailing number rather than the list length, so dissolve/re-create cycles
 * cannot produce duplicate names.
 */
export const nextSubDocumentNumber = (subDocuments: SubDocument[]): number => {
	const namePattern = /^Teildokument (\d+)$/;
	let maxNumber = 0;

	for (const subDocument of subDocuments) {
		const match = namePattern.exec(subDocument.name.trim());

		if (match) {
			maxNumber = Math.max(maxNumber, Number.parseInt(match[1], 10));
		}
	}

	return maxNumber + 1;
};

const detachAttachmentsByPaths = (
	archive: ProcessedZipArchive,
	paths: Set<string>
): { extracted: ZipAttachment[]; archive: ProcessedZipArchive } => {
	const extracted: ZipAttachment[] = [];

	const subDocuments = archive.subDocuments.map((subDocument) => ({
		...subDocument,
		attachments: subDocument.attachments.filter((attachment) => {
			if (paths.has(attachment.path)) {
				extracted.push(attachment);
				return false;
			}

			return true;
		})
	}));

	const attachments = archive.attachments.filter((attachment) => {
		if (paths.has(attachment.path)) {
			extracted.push(attachment);
			return false;
		}

		return true;
	});

	return { extracted, archive: { ...archive, attachments, subDocuments } };
};

/**
 * Group the attachments identified by `paths` (loose or inside any
 * sub-document) into a new sub-document appended at the end. Returns the
 * input archive unchanged when no selected attachment exists.
 */
export const createSubDocumentFromPaths = (
	archive: ProcessedZipArchive,
	paths: Set<string>
): ProcessedZipArchive => {
	const { extracted, archive: strippedArchive } = detachAttachmentsByPaths(archive, paths);

	if (extracted.length === 0) {
		return archive;
	}

	const name = `Teildokument ${nextSubDocumentNumber(archive.subDocuments)}`;

	return {
		...strippedArchive,
		subDocuments: [...strippedArchive.subDocuments, createSubDocument(name, extracted)]
	};
};

/**
 * Move the attachments identified by `paths` (loose or inside any
 * sub-document) to the end of an existing sub-document's attachment list.
 */
export const moveAttachmentsByPathsToSubDocument = (
	archive: ProcessedZipArchive,
	paths: Set<string>,
	targetSubDocumentId: string
): ProcessedZipArchive => {
	const { extracted, archive: strippedArchive } = detachAttachmentsByPaths(archive, paths);

	if (extracted.length === 0) {
		return archive;
	}

	let targetFound = false;

	const subDocuments = strippedArchive.subDocuments.map((subDocument) => {
		if (subDocument.id !== targetSubDocumentId) {
			return subDocument;
		}

		targetFound = true;

		return {
			...subDocument,
			attachments: [
				...subDocument.attachments,
				...extracted.map((attachment) => ({ ...attachment }))
			]
		};
	});

	if (!targetFound) {
		return archive;
	}

	return { ...strippedArchive, subDocuments };
};

/**
 * Remove a single attachment from the archive, wherever it currently lives.
 * Located by its unique `path`; returns the input archive unchanged when no
 * attachment matches.
 */
export const removeAttachmentFromArchive = (
	archive: ProcessedZipArchive,
	path: string
): ProcessedZipArchive => {
	const attachments = archive.attachments.filter((attachment) => attachment.path !== path);
	let removed = attachments.length !== archive.attachments.length;
	const subDocuments = archive.subDocuments.map((subDocument) => {
		const nextAttachments = subDocument.attachments.filter(
			(attachment) => attachment.path !== path
		);

		if (nextAttachments.length !== subDocument.attachments.length) {
			removed = true;
		}

		return nextAttachments.length === subDocument.attachments.length
			? subDocument
			: { ...subDocument, attachments: nextAttachments };
	});

	return removed ? { ...archive, attachments, subDocuments } : archive;
};

/**
 * Move every loose attachment into its own sub-document named after the
 * attachment. The sub-document order matches the loose attachment order.
 */
export const groupEachAttachmentAsSubDocument = (
	archive: ProcessedZipArchive
): ProcessedZipArchive => {
	if (archive.attachments.length === 0) {
		return archive;
	}

	const newSubDocuments = archive.attachments.map((attachment) =>
		createSubDocument(stripPdfExtension(attachment.name), [attachment])
	);

	return {
		...archive,
		attachments: [],
		subDocuments: [...archive.subDocuments, ...newSubDocuments]
	};
};