WhatsApp-Chatexport einlesen und Videos chronologisch ordnen
Erkennt iPhone- und Android-Exporte auf Deutsch und Englisch, ordnet Videos nach Zeitstempel statt nach Chatreihenfolge, meldet fehlende Dateien, markiert doppelt gepostete Videos und filtert auf Montag bis Freitag der Vorwoche. Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
This commit is contained in:
ferdi2goandClaude Opus 5.5 committed
1 parent
f7d249f28b
commit
899b212cba
2 files changed
+362
No files matched your search
@@ -0,0 +1,209 @@
|
|||||||
|
// Liest WhatsApp-Chatexporte (iPhone und Android, deutsch und englisch) und
|
||||||
|
// ordnet die Videos des Exports chronologisch. Läuft im Browser und in Node.
|
||||||
|
|
||||||
|
// Unsichtbare Richtungszeichen, die WhatsApp vor Zeilen und Namen setzt.
|
||||||
|
const INVISIBLE = /[\u200e\u200f\u202a-\u202e\u2066-\u2069]/g;
|
||||||
|
const ODD_SPACES = /[\u00a0\u202f]/g;
|
||||||
|
|
||||||
|
// iPhone: "[24.09.26, 16:36:19] Name: Text"
|
||||||
|
const IOS_HEADER =
|
||||||
|
/^\[(\d{1,2})([./])(\d{1,2})\2(\d{2,4}),? (\d{1,2}):(\d{2})(?::(\d{2}))?(?: ?([AP]M))?\] (.*)$/i;
|
||||||
|
// Android: "24.09.26, 16:36 - Name: Text"
|
||||||
|
const ANDROID_HEADER =
|
||||||
|
/^(\d{1,2})([./])(\d{1,2})\2(\d{2,4}),? (\d{1,2}):(\d{2})(?::(\d{2}))?(?: ?([AP]M))? - (.*)$/i;
|
||||||
|
|
||||||
|
// "<Anhang: 00001338-VIDEO-2026-09-24-16-36-19.mp4>" bzw. "<attached: …>"
|
||||||
|
const TAGGED_ATTACHMENT = /<[^<>:]{1,40}:\s*([^<>]+?\.[A-Za-z0-9]{2,5})>/g;
|
||||||
|
// "VID-20260924-WA0012.mp4 (Datei angehängt)" bzw. "(file attached)"
|
||||||
|
const ANDROID_ATTACHMENT = /([^\s<>]+\.[A-Za-z0-9]{2,5}) \((?:Datei angehängt|file attached)\)/g;
|
||||||
|
|
||||||
|
const VIDEO_EXTENSION = /\.(mp4|mov|m4v|3gp|avi|mkv|webm)$/i;
|
||||||
|
|
||||||
|
const pad = (value) => String(value).padStart(2, '0');
|
||||||
|
|
||||||
|
function cleanLine(line) {
|
||||||
|
return line.replace(INVISIBLE, '').replace(ODD_SPACES, ' ');
|
||||||
|
}
|
||||||
|
|
||||||
|
export function isVideoFile(name) {
|
||||||
|
return VIDEO_EXTENSION.test(name);
|
||||||
|
}
|
||||||
|
|
||||||
|
export function baseName(path) {
|
||||||
|
return path.split(/[\\/]/).pop();
|
||||||
|
}
|
||||||
|
|
||||||
|
// Bei Datumsangaben mit Schrägstrich ist die Reihenfolge nicht eindeutig
|
||||||
|
// (US: M/D, sonst D/M). Ein Tag über 12 irgendwo im Chat entscheidet.
|
||||||
|
function detectDayFirst(lines) {
|
||||||
|
for (const line of lines) {
|
||||||
|
const match = line.match(IOS_HEADER) || line.match(ANDROID_HEADER);
|
||||||
|
if (match && match[2] === '/' && Number(match[1]) > 12) return true;
|
||||||
|
}
|
||||||
|
return false;
|
||||||
|
}
|
||||||
|
|
||||||
|
function toTimestamp(match, dayFirst) {
|
||||||
|
const [, first, separator, second, rawYear, rawHour, minute, second_ = '0', meridiem] = match;
|
||||||
|
let day = Number(first);
|
||||||
|
let month = Number(second);
|
||||||
|
if (separator === '/' && !dayFirst) [day, month] = [month, day];
|
||||||
|
let year = Number(rawYear);
|
||||||
|
if (year < 100) year += 2000;
|
||||||
|
let hour = Number(rawHour);
|
||||||
|
if (meridiem) {
|
||||||
|
const pm = meridiem.toUpperCase() === 'PM';
|
||||||
|
if (hour === 12) hour = pm ? 12 : 0;
|
||||||
|
else if (pm) hour += 12;
|
||||||
|
}
|
||||||
|
if (month < 1 || month > 12 || day < 1 || day > 31 || hour > 23) return null;
|
||||||
|
return `${year}-${pad(month)}-${pad(day)}T${pad(hour)}:${minute}:${pad(second_)}`;
|
||||||
|
}
|
||||||
|
|
||||||
|
function splitSender(rest) {
|
||||||
|
const index = rest.indexOf(': ');
|
||||||
|
if (index > 0) return { sender: rest.slice(0, index).trim(), text: rest.slice(index + 2) };
|
||||||
|
if (rest.endsWith(':')) return { sender: rest.slice(0, -1).trim(), text: '' };
|
||||||
|
// Systemmeldung ohne Absender, z. B. "Max hat die Gruppe verlassen"
|
||||||
|
return { sender: null, text: rest };
|
||||||
|
}
|
||||||
|
|
||||||
|
export function findAttachments(text) {
|
||||||
|
const names = [];
|
||||||
|
for (const pattern of [TAGGED_ATTACHMENT, ANDROID_ATTACHMENT]) {
|
||||||
|
for (const match of text.matchAll(pattern)) names.push(match[1].trim());
|
||||||
|
}
|
||||||
|
return names;
|
||||||
|
}
|
||||||
|
|
||||||
|
function captionOf(text) {
|
||||||
|
return text.replace(TAGGED_ATTACHMENT, '').replace(ANDROID_ATTACHMENT, '').trim();
|
||||||
|
}
|
||||||
|
|
||||||
|
export function parseChat(rawText) {
|
||||||
|
const lines = rawText.replace(/\r\n?/g, '\n').split('\n').map(cleanLine);
|
||||||
|
const dayFirst = detectDayFirst(lines);
|
||||||
|
const messages = [];
|
||||||
|
|
||||||
|
for (const line of lines) {
|
||||||
|
const match = line.match(IOS_HEADER) || line.match(ANDROID_HEADER);
|
||||||
|
const timestamp = match && toTimestamp(match, dayFirst);
|
||||||
|
if (timestamp) {
|
||||||
|
messages.push({ index: messages.length, timestamp, ...splitSender(match[9]) });
|
||||||
|
} else if (messages.length > 0) {
|
||||||
|
messages[messages.length - 1].text += `\n${line}`;
|
||||||
|
}
|
||||||
|
}
|
||||||
|
|
||||||
|
return messages.map((message) => ({
|
||||||
|
...message,
|
||||||
|
text: message.text.trim(),
|
||||||
|
attachments: findAttachments(message.text),
|
||||||
|
}));
|
||||||
|
}
|
||||||
|
|
||||||
|
// Fallback für Videos ohne Chatzeile: der Dateiname trägt den Zeitpunkt.
|
||||||
|
export function timestampFromFileName(name) {
|
||||||
|
const ios = name.match(/(\d{4})-(\d{2})-(\d{2})-(\d{2})-(\d{2})-(\d{2})\.\w+$/);
|
||||||
|
if (ios) return `${ios[1]}-${ios[2]}-${ios[3]}T${ios[4]}:${ios[5]}:${ios[6]}`;
|
||||||
|
const android = name.match(/-(\d{4})(\d{2})(\d{2})-WA\d+/);
|
||||||
|
if (android) return `${android[1]}-${android[2]}-${android[3]}T00:00:00`;
|
||||||
|
return null;
|
||||||
|
}
|
||||||
|
|
||||||
|
function compareClips(a, b) {
|
||||||
|
return (
|
||||||
|
a.timestamp.localeCompare(b.timestamp) ||
|
||||||
|
a.order - b.order ||
|
||||||
|
a.file.localeCompare(b.file)
|
||||||
|
);
|
||||||
|
}
|
||||||
|
|
||||||
|
// Doppelt gepostete Videos sind bytegleich; die Dateigröße genügt als Merkmal.
|
||||||
|
function markDuplicates(clips) {
|
||||||
|
const firstBySize = new Map();
|
||||||
|
for (const clip of clips) {
|
||||||
|
if (!clip.size) continue;
|
||||||
|
const first = firstBySize.get(clip.size);
|
||||||
|
if (first) clip.duplicateOf = first.file;
|
||||||
|
else firstBySize.set(clip.size, clip);
|
||||||
|
}
|
||||||
|
}
|
||||||
|
|
||||||
|
// files: [{ name, size }] aus dem Export (Pfade erlaubt).
|
||||||
|
// Liefert die Videos chronologisch, dazu im Chat genannte, aber fehlende Dateien.
|
||||||
|
export function collectClips(messages, files) {
|
||||||
|
const videos = new Map();
|
||||||
|
for (const file of files) {
|
||||||
|
if (isVideoFile(file.name)) videos.set(baseName(file.name), file);
|
||||||
|
}
|
||||||
|
|
||||||
|
const clips = [];
|
||||||
|
const missing = [];
|
||||||
|
const used = new Set();
|
||||||
|
|
||||||
|
for (const message of messages) {
|
||||||
|
for (const attachment of message.attachments) {
|
||||||
|
if (!isVideoFile(attachment)) continue;
|
||||||
|
const name = baseName(attachment);
|
||||||
|
const file = videos.get(name);
|
||||||
|
if (!file) {
|
||||||
|
missing.push({ file: name, timestamp: message.timestamp, sender: message.sender });
|
||||||
|
continue;
|
||||||
|
}
|
||||||
|
if (used.has(name)) continue;
|
||||||
|
used.add(name);
|
||||||
|
clips.push({
|
||||||
|
file: name,
|
||||||
|
path: file.name,
|
||||||
|
size: file.size ?? 0,
|
||||||
|
timestamp: message.timestamp,
|
||||||
|
sender: message.sender,
|
||||||
|
caption: captionOf(message.text),
|
||||||
|
order: message.index,
|
||||||
|
});
|
||||||
|
}
|
||||||
|
}
|
||||||
|
|
||||||
|
for (const [name, file] of videos) {
|
||||||
|
if (used.has(name)) continue;
|
||||||
|
const timestamp = timestampFromFileName(name);
|
||||||
|
if (!timestamp) continue;
|
||||||
|
clips.push({
|
||||||
|
file: name,
|
||||||
|
path: file.name,
|
||||||
|
size: file.size ?? 0,
|
||||||
|
timestamp,
|
||||||
|
sender: null,
|
||||||
|
caption: '',
|
||||||
|
order: Number.MAX_SAFE_INTEGER,
|
||||||
|
});
|
||||||
|
}
|
||||||
|
|
||||||
|
clips.sort(compareClips);
|
||||||
|
markDuplicates(clips);
|
||||||
|
return { clips, missing };
|
||||||
|
}
|
||||||
|
|
||||||
|
// from/to als "JJJJ-MM-TT", beide Tage eingeschlossen.
|
||||||
|
export function clipsInRange(clips, from, to) {
|
||||||
|
return clips.filter((clip) => {
|
||||||
|
const day = clip.timestamp.slice(0, 10);
|
||||||
|
return (!from || day >= from) && (!to || day <= to);
|
||||||
|
});
|
||||||
|
}
|
||||||
|
|
||||||
|
export function toDateString(date) {
|
||||||
|
return `${date.getFullYear()}-${pad(date.getMonth() + 1)}-${pad(date.getDate())}`;
|
||||||
|
}
|
||||||
|
|
||||||
|
// Montag bis Freitag der Kalenderwoche vor der von `today`.
|
||||||
|
export function previousWorkWeek(today = new Date()) {
|
||||||
|
const date = new Date(today.getFullYear(), today.getMonth(), today.getDate());
|
||||||
|
const daysSinceMonday = (date.getDay() + 6) % 7;
|
||||||
|
const monday = new Date(date);
|
||||||
|
monday.setDate(date.getDate() - daysSinceMonday - 7);
|
||||||
|
const friday = new Date(monday);
|
||||||
|
friday.setDate(monday.getDate() + 4);
|
||||||
|
return { from: toDateString(monday), to: toDateString(friday) };
|
||||||
|
}
|
||||||
@@ -0,0 +1,153 @@
|
|||||||
|
import assert from 'node:assert/strict';
|
||||||
|
import { test } from 'node:test';
|
||||||
|
import {
|
||||||
|
clipsInRange,
|
||||||
|
collectClips,
|
||||||
|
parseChat,
|
||||||
|
previousWorkWeek,
|
||||||
|
timestampFromFileName,
|
||||||
|
} from '../src/lib/whatsapp.js';
|
||||||
|
|
||||||
|
// Ausgedachte Zeilen im Format eines deutschen iPhone-Exports mit Medien,
|
||||||
|
// samt der unsichtbaren Richtungszeichen (\u200e), die WhatsApp einfügt.
|
||||||
|
const IOS_CHAT = [
|
||||||
|
'[21.09.26, 09:00:00] Anna Beispiel: Guten Morgen',
|
||||||
|
'\u200e[21.09.26, 10:15:30] Anna Beispiel: Ordnung am Platz \u200e<Anhang: 00000101-VIDEO-2026-09-21-10-15-30.mp4>',
|
||||||
|
'[21.09.26, 10:16:00] Ben Muster: Stark!',
|
||||||
|
'Zweite Zeile der Nachricht',
|
||||||
|
'\u200e[22.09.26, 08:00:00] ~\u202fCarla: \u200e<Anhang: 00000102-VIDEO-2026-09-22-08-00-00.mp4>',
|
||||||
|
// Außer der Reihe: im Chat später, zeitlich aber früher
|
||||||
|
'\u200e[21.09.26, 07:30:00] Ben Muster: Frühaufsteher \u200e<Anhang: 00000103-VIDEO-2026-09-21-07-30-00.mp4>',
|
||||||
|
'\u200e[22.09.26, 08:00:05] ~\u202fCarla: \u200e<Anhang: 00000104-PHOTO-2026-09-22-08-00-05.jpg>',
|
||||||
|
'[22.09.26, 09:00:00] Ben Muster: \u200eDiese Nachricht wurde gelöscht.',
|
||||||
|
'\u200e[23.09.26, 11:00:00] Anna Beispiel: Fehlt im Export \u200e<Anhang: 00000105-VIDEO-2026-09-23-11-00-00.mp4>',
|
||||||
|
'\u200e[28.09.26, 12:00:00] Anna Beispiel: Nächste Woche \u200e<Anhang: 00000106-VIDEO-2026-09-28-12-00-00.mp4>',
|
||||||
|
].join('\r\n');
|
||||||
|
|
||||||
|
const IOS_FILES = [
|
||||||
|
{ name: '_chat.txt', size: 1000 },
|
||||||
|
{ name: 'WhatsApp Chat/00000101-VIDEO-2026-09-21-10-15-30.mp4', size: 5000 },
|
||||||
|
{ name: 'WhatsApp Chat/00000102-VIDEO-2026-09-22-08-00-00.mp4', size: 7000 },
|
||||||
|
{ name: 'WhatsApp Chat/00000103-VIDEO-2026-09-21-07-30-00.mp4', size: 6000 },
|
||||||
|
{ name: 'WhatsApp Chat/00000104-PHOTO-2026-09-22-08-00-05.jpg', size: 100 },
|
||||||
|
{ name: 'WhatsApp Chat/00000106-VIDEO-2026-09-28-12-00-00.mp4', size: 8000 },
|
||||||
|
// Im Chat nicht erwähnt, Zeitpunkt nur aus dem Dateinamen
|
||||||
|
{ name: 'WhatsApp Chat/00000107-VIDEO-2026-09-22-15-00-00.mp4', size: 9000 },
|
||||||
|
];
|
||||||
|
|
||||||
|
test('iPhone-Export: Nachrichten, Absender, Zeitstempel und Anhänge', () => {
|
||||||
|
const messages = parseChat(IOS_CHAT);
|
||||||
|
assert.equal(messages.length, 9);
|
||||||
|
assert.deepEqual(messages[1], {
|
||||||
|
index: 1,
|
||||||
|
timestamp: '2026-09-21T10:15:30',
|
||||||
|
sender: 'Anna Beispiel',
|
||||||
|
text: 'Ordnung am Platz <Anhang: 00000101-VIDEO-2026-09-21-10-15-30.mp4>',
|
||||||
|
attachments: ['00000101-VIDEO-2026-09-21-10-15-30.mp4'],
|
||||||
|
});
|
||||||
|
assert.equal(messages[2].text, 'Stark!\nZweite Zeile der Nachricht');
|
||||||
|
assert.equal(messages[3].sender, '~ Carla');
|
||||||
|
assert.deepEqual(messages[6].attachments, []);
|
||||||
|
});
|
||||||
|
|
||||||
|
test('Videos werden nach Zeitstempel sortiert, nicht nach Chatreihenfolge', () => {
|
||||||
|
const { clips } = collectClips(parseChat(IOS_CHAT), IOS_FILES);
|
||||||
|
assert.deepEqual(
|
||||||
|
clips.map((clip) => clip.file),
|
||||||
|
[
|
||||||
|
'00000103-VIDEO-2026-09-21-07-30-00.mp4',
|
||||||
|
'00000101-VIDEO-2026-09-21-10-15-30.mp4',
|
||||||
|
'00000102-VIDEO-2026-09-22-08-00-00.mp4',
|
||||||
|
'00000107-VIDEO-2026-09-22-15-00-00.mp4',
|
||||||
|
'00000106-VIDEO-2026-09-28-12-00-00.mp4',
|
||||||
|
],
|
||||||
|
);
|
||||||
|
});
|
||||||
|
|
||||||
|
test('Clips tragen Absender, Beschriftung und Pfad', () => {
|
||||||
|
const { clips } = collectClips(parseChat(IOS_CHAT), IOS_FILES);
|
||||||
|
const clip = clips.find((c) => c.file.startsWith('00000101'));
|
||||||
|
assert.equal(clip.sender, 'Anna Beispiel');
|
||||||
|
assert.equal(clip.caption, 'Ordnung am Platz');
|
||||||
|
assert.equal(clip.path, 'WhatsApp Chat/00000101-VIDEO-2026-09-21-10-15-30.mp4');
|
||||||
|
const orphan = clips.find((c) => c.file.startsWith('00000107'));
|
||||||
|
assert.equal(orphan.sender, null);
|
||||||
|
});
|
||||||
|
|
||||||
|
test('Im Chat genannte, aber fehlende Videos werden gemeldet', () => {
|
||||||
|
const { missing } = collectClips(parseChat(IOS_CHAT), IOS_FILES);
|
||||||
|
assert.deepEqual(missing.map((m) => m.file), ['00000105-VIDEO-2026-09-23-11-00-00.mp4']);
|
||||||
|
});
|
||||||
|
|
||||||
|
test('Bytegleiche Videos werden als Duplikat markiert', () => {
|
||||||
|
const chat = [
|
||||||
|
'\u200e[23.09.26, 10:03:49] Tom: Autostart \u200e<Anhang: 00000201-VIDEO-2026-09-23-10-03-49.mp4>',
|
||||||
|
'\u200e[23.09.26, 10:03:53] Tom: Autostart \u200e<Anhang: 00000202-VIDEO-2026-09-23-10-03-53.mp4>',
|
||||||
|
].join('\n');
|
||||||
|
const files = [
|
||||||
|
{ name: '00000201-VIDEO-2026-09-23-10-03-49.mp4', size: 4242 },
|
||||||
|
{ name: '00000202-VIDEO-2026-09-23-10-03-53.mp4', size: 4242 },
|
||||||
|
];
|
||||||
|
const { clips } = collectClips(parseChat(chat), files);
|
||||||
|
assert.equal(clips[0].duplicateOf, undefined);
|
||||||
|
assert.equal(clips[1].duplicateOf, '00000201-VIDEO-2026-09-23-10-03-49.mp4');
|
||||||
|
});
|
||||||
|
|
||||||
|
test('Android-Export (deutsch) mit Systemmeldung', () => {
|
||||||
|
const chat = [
|
||||||
|
'21.09.26, 10:15 - Anna: VID-20260921-WA0003.mp4 (Datei angehängt)',
|
||||||
|
'Ordnung am Platz',
|
||||||
|
'21.09.26, 10:20 - Ben hat die Gruppe verlassen',
|
||||||
|
'21.09.26, 11:00 - Ben: Hallo',
|
||||||
|
].join('\n');
|
||||||
|
const messages = parseChat(chat);
|
||||||
|
assert.equal(messages.length, 3);
|
||||||
|
assert.equal(messages[0].timestamp, '2026-09-21T10:15:00');
|
||||||
|
assert.deepEqual(messages[0].attachments, ['VID-20260921-WA0003.mp4']);
|
||||||
|
assert.equal(messages[1].sender, null);
|
||||||
|
|
||||||
|
const { clips } = collectClips(messages, [{ name: 'VID-20260921-WA0003.mp4', size: 1 }]);
|
||||||
|
assert.equal(clips[0].caption, 'Ordnung am Platz');
|
||||||
|
});
|
||||||
|
|
||||||
|
test('Englischer Export mit Monat zuerst und AM/PM', () => {
|
||||||
|
const messages = parseChat(
|
||||||
|
[
|
||||||
|
'9/21/26, 4:36\u202fPM - Anna: VID-20260921-WA0001.mp4 (file attached)',
|
||||||
|
'[9/22/26, 12:05:00 AM] Ben: <attached: 00000001-VIDEO-2026-09-22-00-05-00.mp4>',
|
||||||
|
].join('\n'),
|
||||||
|
);
|
||||||
|
assert.equal(messages[0].timestamp, '2026-09-21T16:36:00');
|
||||||
|
assert.equal(messages[1].timestamp, '2026-09-22T00:05:00');
|
||||||
|
assert.deepEqual(messages[1].attachments, ['00000001-VIDEO-2026-09-22-00-05-00.mp4']);
|
||||||
|
});
|
||||||
|
|
||||||
|
test('Schrägstrich-Datum mit Tag über 12 gilt als Tag zuerst', () => {
|
||||||
|
const messages = parseChat('24/09/2026, 16:36 - Anna: Hallo\n05/09/2026, 10:00 - Ben: Hi');
|
||||||
|
assert.equal(messages[0].timestamp, '2026-09-24T16:36:00');
|
||||||
|
assert.equal(messages[1].timestamp, '2026-09-05T10:00:00');
|
||||||
|
});
|
||||||
|
|
||||||
|
test('Zeitstempel aus Dateinamen', () => {
|
||||||
|
assert.equal(timestampFromFileName('00001338-VIDEO-2026-09-24-16-36-18.mp4'), '2026-09-24T16:36:18');
|
||||||
|
assert.equal(timestampFromFileName('VID-20260924-WA0012.mp4'), '2026-09-24T00:00:00');
|
||||||
|
assert.equal(timestampFromFileName('film.mp4'), null);
|
||||||
|
});
|
||||||
|
|
||||||
|
test('Zeitraumfilter schließt beide Tage ein', () => {
|
||||||
|
const { clips } = collectClips(parseChat(IOS_CHAT), IOS_FILES);
|
||||||
|
const week = clipsInRange(clips, '2026-09-21', '2026-09-25');
|
||||||
|
assert.equal(week.length, 4);
|
||||||
|
assert.ok(week.every((clip) => clip.timestamp < '2026-09-26'));
|
||||||
|
});
|
||||||
|
|
||||||
|
test('Vergangene Woche ist Montag bis Freitag der Vorwoche', () => {
|
||||||
|
// Dienstag, 29.09.2026
|
||||||
|
assert.deepEqual(previousWorkWeek(new Date(2026, 8, 29)), { from: '2026-09-21', to: '2026-09-25' });
|
||||||
|
// Montag, 28.09.2026
|
||||||
|
assert.deepEqual(previousWorkWeek(new Date(2026, 8, 28)), { from: '2026-09-21', to: '2026-09-25' });
|
||||||
|
// Sonntag, 27.09.2026 gehört noch zur Woche ab 21.09.
|
||||||
|
assert.deepEqual(previousWorkWeek(new Date(2026, 8, 27)), { from: '2026-09-14', to: '2026-09-18' });
|
||||||
|
// Über den Jahreswechsel
|
||||||
|
assert.deepEqual(previousWorkWeek(new Date(2027, 0, 5)), { from: '2026-12-28', to: '2027-01-01' });
|
||||||
|
});
|
||||||
Reference in new issue
Block a user