Bank Statement Parser: PDF to JSON & CSV, Totals Checked
Pricing
$50.00 / 1,000 statement parseds
Bank statement PDF to JSON and CSV rows for bookkeeping and agent tools. Every statement total is checked against the rows: a failed check returns no rows and costs nothing. Chase, Bank of America, Wells Fargo, Capital One, TD Bank. Digital PDFs only. $0.05 per passed statement.
node_modulesstorageapify_storagetest.gitnode_modulesstorageapify_storage.DS_Store{ "tabWidth": 4, "printWidth": 140, "singleQuote": true}FROM apify/actor-node:22
COPY package*.json ./RUN npm --quiet set progress=false \ && npm install --omit=dev --omit=optional --no-audit --no-fund \ && rm -rf ~/.npm
COPY . ./
CMD ["npm", "start", "--silent"]
1<svg xmlns="http://www.w3.org/2000/svg" width="512" height="512" viewBox="0 0 512 512">2 <rect width="512" height="512" fill="#1b4632"/>3 <!-- a statement page with rows, and a check mark: the totals matched -->4 <rect x="112" y="60" width="288" height="392" rx="20" fill="#f3eee0"/>5 <g fill="#b9b2a0">6 <rect x="148" y="108" width="128" height="20" rx="6"/>7 <rect x="148" y="160" width="146" height="14" rx="5"/>8 <rect x="148" y="200" width="146" height="14" rx="5"/>9 <rect x="148" y="240" width="146" height="14" rx="5"/>10 </g>11 <g fill="#2f6b4f">12 <rect x="316" y="160" width="48" height="14" rx="5"/>13 <rect x="316" y="200" width="48" height="14" rx="5"/>14 <rect x="316" y="240" width="48" height="14" rx="5"/>15 <rect x="148" y="284" width="216" height="6" rx="3"/>16 </g>17 <circle cx="256" cy="366" r="56" fill="#2f6b4f"/>18 <path d="M228 366l20 21 38-42" fill="none" stroke="#f3eee0" stroke-width="15" stroke-linecap="round" stroke-linejoin="round"/>19</svg>{ "name": "bank-statement-parser", "version": "1.0.0", "description": "Bank statement parser: digital statement PDF to JSON rows, with every statement total checked against the rows.", "type": "module", "private": true, "engines": { "node": ">=20" }, "main": "src/main.js", "scripts": { "start": "node src/main.js", "test": "node --test --test-concurrency=1 test/lib.test.js && node --test --test-concurrency=1 test/http.test.js && node --test --test-concurrency=1 test/main.test.js" }, "dependencies": { "apify": "3.7.2", "pdfjs-dist": "6.4.299" }}{ "defaultRunOptions": { "build": "latest", "timeoutSecs": 600, "memoryMbytes": 512 }, "pricingInfos": [ { "pricingModel": "PAY_PER_EVENT", "pricingPerEvent": { "actorChargeEvents": { "statement-parsed": { "eventTitle": "Statement parsed", "eventDescription": "Charged once for each statement whose rows match all of its own totals. Scanned PDFs, failed checks, unsupported layouts and failures are free.", "eventPriceUsd": 0.05, "isPrimaryEvent": true } } }, "minimalMaxTotalChargeUsd": 0.5 } ]}1# Changelog2
3## 0.1, 2026-10-304
5- First release: digital statement PDFs to JSON rows, with every statement total checked. Chase checking and credit card, Bank of America checking, Wells Fargo checking, Capital One credit card, TD Bank checking. $0.05 per passed statement; everything else is free.6- Build 0.1.3: a Transactions view and a Transactions CSV output link, one line for each transaction of every statement that passed. Parsing is unchanged.{ "actorSpecification": 1, "name": "bank-statement-parser", "title": "Bank Statement Parser: PDF to JSON & CSV, Totals Checked", "description": "Bank statement PDF to JSON and CSV rows for bookkeeping and agent tools. Every statement total is checked against the rows: a failed check returns no rows and costs nothing. Chase, Bank of America, Wells Fargo, Capital One, TD Bank. Digital PDFs only. $0.05 per passed statement.", "version": "0.1", "buildTag": "latest", "defaultMemoryMbytes": 512, "minMemoryMbytes": 256, "maxMemoryMbytes": 2048, "input": "./input_schema.json", "output": "./output_schema.json", "dockerfile": "../Dockerfile", "readme": "../README.md", "storages": { "dataset": { "actorSpecification": 1, "fields": { "$schema": "http://json-schema.org/draft-07/schema#", "type": "object", "additionalProperties": true, "properties": { "url": { "type": ["string", "null"], "description": "Input URL this item is for." }, "status": { "type": ["string", "null"], "description": "passed, scanned, unsupported, check_failed, password, too_long, unreadable or download_failed. Only passed has rows and a charge." }, "charged": { "type": ["boolean", "null"], "description": "True when this statement was charged." }, "layout": { "type": ["string", "null"], "description": "The statement layout read: chase-checking, chase-checking-sections, chase-card, bofa-checking, wells-checking, capone-card or td-checking." }, "period": { "type": ["object", "null"], "description": "Statement period: start and end, as YYYY-MM-DD." }, "rowCount": { "type": ["number", "null"], "description": "Number of rows returned. 0 unless status is passed." }, "rows": { "type": ["array", "null"], "description": "Transactions: date (YYYY-MM-DD), description, amount (money in positive, money out negative), balance (or null), section. Empty unless status is passed." }, "totals": { "type": ["array", "null"], "description": "The check: label, statement (the total printed on the statement), parsed (the sum of the rows), match." }, "problem": { "type": ["string", "null"], "description": "Why the statement did not pass. Null when it passed." }, "note": { "type": ["string", "null"], "description": "Run-level note, for example a stop at the run's maximum cost." } } }, "views": { "overview": { "title": "Statements", "transformation": { "fields": ["url", "status", "charged", "layout", "period", "rowCount", "rows", "totals", "problem", "note"] }, "display": { "component": "table", "properties": { "url": { "label": "Statement", "format": "link" } } } }, "transactions": { "title": "Transactions (CSV)", "description": "One line for each transaction, for a CSV or Excel export. Only statements that passed have rows.", "transformation": { "fields": ["rows", "url"], "unwind": ["rows"] }, "display": { "component": "table", "properties": { "date": { "label": "Date" }, "description": { "label": "Description" }, "amount": { "label": "Amount", "format": "number" }, "balance": { "label": "Balance", "format": "number" }, "section": { "label": "Section" }, "url": { "label": "Statement", "format": "link" } } } } } } }}{ "title": "Bank Statement Parser", "type": "object", "schemaVersion": 1, "properties": { "statementUrls": { "title": "Statement URLs", "type": "array", "editor": "stringList", "description": "Direct links to digital statement PDFs, up to 20 MB and 100 pages each. For a record in a private Apify key-value store, use a signed URL (getRecordPublicUrl in the Apify client). The prefilled link is a made-up sample statement.", "prefill": ["https://brasspage.com/bank-statement-to-excel/samples/chase-checking-sample.pdf"] }, "statementFile": { "title": "Upload a statement", "type": "string", "editor": "fileupload", "description": "Upload one digital statement PDF from your computer." } }}{ "actorOutputSchemaVersion": 1, "title": "Bank Statement Parser output", "properties": { "statements": { "type": "string", "title": "Statements", "description": "One item per statement: status, layout, period, the rows (only when every total matched), the totals check, and the reason for any statement that did not pass.", "template": "{{links.apiDefaultDatasetUrl}}/items?view=overview" }, "transactionsCsv": { "type": "string", "title": "Transactions CSV", "description": "One line for each transaction of every statement that passed: date, description, amount, balance, section and the statement URL. Change format=csv to format=xlsx for Excel.", "template": "{{links.apiDefaultDatasetUrl}}/items?view=transactions&format=csv" } }}1// Safe download of a user-supplied URL (trimmed copy of actors/sitemap-extractor/src/http.js, with the Apify record2// route of actors/document-ocr/src/fetch.py). Safe: http(s) only, no credentials sent, and every redirect hop may3// connect only to a public IP. The IP check runs inside the socket's DNS lookup, so the checked address is the one we4// connect to (no DNS rebinding). Sends no Accept-Encoding, so the body is the raw file and the size cap is real.5import dns from 'node:dns';6import http from 'node:http';7import https from 'node:https';8import net from 'node:net';9import { MAX_FILE_MB } from './lib.js';10
11const UA = 'Mozilla/5.0 (compatible; BankStatementParser/1.0; +https://apify.com/brasspage/bank-statement-parser)';12const MAX_BYTES = MAX_FILE_MB * 1024 * 1024;13const MAX_REDIRECTS = 5;14const TIMEOUT_MS = 60000;15
16// Every IPv4 range that is not public unicast: this-network, private, CGNAT, loopback, link-local,17// IETF, documentation, 6to4 relay, benchmarking, multicast, reserved and broadcast.18const V4_BLOCKED = new net.BlockList();19for (const [ip, bits] of [20 ['0.0.0.0', 8],21 ['10.0.0.0', 8],22 ['100.64.0.0', 10],23 ['127.0.0.0', 8],24 ['169.254.0.0', 16],25 ['172.16.0.0', 12],26 ['192.0.0.0', 24],27 ['192.0.2.0', 24],28 ['192.88.99.0', 24],29 ['192.168.0.0', 16],30 ['198.18.0.0', 15],31 ['198.51.100.0', 24],32 ['203.0.113.0', 24],33 ['224.0.0.0', 3], // multicast 224/4 + reserved 240/4 + broadcast34])35 V4_BLOCKED.addSubnet(ip, bits, 'ipv4');36
37function v6ToBigInt(ip) {38 let s = ip.split('%')[0].toLowerCase();39 const v4 = /(\d+\.\d+\.\d+\.\d+)$/.exec(s);40 if (v4) {41 const p = v4[1].split('.').map(Number);42 s = s.slice(0, -v4[1].length) + ((p[0] << 8) | p[1]).toString(16) + ':' + ((p[2] << 8) | p[3]).toString(16);43 }44 const [head, tail] = s.split('::');45 const h = head ? head.split(':') : [];46 const t = tail ? tail.split(':') : [];47 const groups = s.includes('::') ? [...h, ...Array(8 - h.length - t.length).fill('0'), ...t] : h;48 return groups.reduce((n, g) => (n << 16n) | BigInt(parseInt(g || '0', 16)), 0n);49}50
51const v4FromBits = (n) => [24n, 16n, 8n, 0n].map((s) => (n >> s) & 255n).join('.');52const inPrefix = (n, prefix, bits) => n >> BigInt(128 - bits) === v6ToBigInt(prefix) >> BigInt(128 - bits);53
54// True only for public unicast addresses. IPv6 must be global unicast (2000::/3) outside the special blocks;55// IPv4-mapped, NAT64 and 6to4 addresses are judged by the IPv4 address they carry.56export function isPublicIp(ip) {57 const kind = net.isIP(ip.split('%')[0]);58 if (kind === 4) return !V4_BLOCKED.check(ip, 'ipv4');59 if (kind !== 6) return false;60 const n = v6ToBigInt(ip);61 if (inPrefix(n, '::ffff:0:0', 96) || inPrefix(n, '64:ff9b::', 96)) return isPublicIp(v4FromBits(n & 0xffffffffn));62 if (inPrefix(n, '2002::', 16)) return isPublicIp(v4FromBits((n >> 80n) & 0xffffffffn));63 if (!inPrefix(n, '2000::', 3)) return false; // loopback, unspecified, ULA, link-local, multicast, ...64 return !inPrefix(n, '2001::', 23) && !inPrefix(n, '2001:db8::', 32); // IETF/Teredo, documentation65}66
67const blocked = (host) => new Error(`${host} is not a public internet host`);68
69// dns.lookup replacement for sockets: resolves, then refuses unless every address is public.70const safeLookup = (isAllowed) => (host, opts, cb) =>71 dns.lookup(host, { ...opts, all: true }, (err, addrs) => {72 if (err) return cb(err);73 if (!addrs.length || !addrs.every((a) => isAllowed(a.address))) return cb(blocked(host));74 return opts.all ? cb(null, addrs) : cb(null, addrs[0].address, addrs[0].family);75 });76
77// One GET, no redirects. Resolves {status, location, body: Buffer} or rejects.78function once(u, isAllowed) {79 const host = u.hostname.replace(/^\[|\]$/g, '');80 if (net.isIP(host) && !isAllowed(host)) return Promise.reject(blocked(host));81 return new Promise((resolve, reject) => {82 const req = (u.protocol === 'https:' ? https : http).request(83 {84 protocol: u.protocol,85 hostname: host,86 port: u.port || undefined,87 path: u.pathname + u.search,88 headers: { 'user-agent': UA, accept: 'application/pdf,*/*;q=0.8' },89 lookup: safeLookup(isAllowed),90 agent: false,91 signal: AbortSignal.timeout(TIMEOUT_MS),92 },93 (res) => {94 const chunks = [];95 let bytes = 0;96 res.on('data', (c) => {97 bytes += c.length;98 if (bytes <= MAX_BYTES) return chunks.push(c);99 res.destroy();100 reject(new Error(`the file is larger than ${MAX_FILE_MB} MB`));101 });102 res.on('end', () => resolve({ status: res.statusCode, location: res.headers.location, body: Buffer.concat(chunks) }));103 res.on('error', reject);104 },105 );106 req.on('error', reject);107 req.end();108 });109}110
111// The bytes of a public URL. Every hop is checked (scheme, public IP). `isAllowed` exists only so tests can let one112// loopback test server through.113export async function fetchPublic(url, { isAllowed = isPublicIp } = {}) {114 let cur = url;115 for (let hop = 0; hop <= MAX_REDIRECTS; hop++) {116 let u;117 try {118 u = new URL(cur);119 } catch {120 throw new Error('the URL is not valid');121 }122 if (!/^https?:$/.test(u.protocol)) throw new Error('only http and https URLs work');123 const r = await once(u, isAllowed);124 if (r.status >= 300 && r.status < 400 && r.location) {125 cur = new URL(r.location, u).href;126 continue;127 }128 if (r.status !== 200) throw new Error(`HTTP ${r.status}`);129 return r.body;130 }131 throw new Error('too many redirects');132}133
134const KV_RECORD = /^\/v2\/key-value-stores\/([^/]+)\/records\/([^/]+)$/;135
136// {storeId, key, signature} for an https://api.apify.com key-value store record URL, else null.137export function apifyRecord(url) {138 let u;139 try {140 u = new URL(url);141 } catch {142 return null;143 }144 const m = u.protocol === 'https:' && u.hostname === 'api.apify.com' && !u.port ? KV_RECORD.exec(u.pathname) : null;145 return (146 m && { storeId: decodeURIComponent(m[1]), key: decodeURIComponent(m[2]), signature: u.searchParams.get('signature') ?? undefined }147 );148}149
150// Apify records (a file upload lands in one) go through the Apify client, so the run's token stays inside it.151export async function download(url, client) {152 const rec = apifyRecord(url);153 if (!rec) return fetchPublic(url);154 const record = await client155 .keyValueStore(rec.storeId)156 .getRecord(rec.key, { buffer: true, signature: rec.signature })157 .catch((e) => {158 if (e?.statusCode !== 403) throw e;159 throw new Error('this Apify store is private and the URL has no signature, so use a signed record URL');160 });161 if (!record) throw new Error('the key-value store record was not found');162 if (record.value.length > MAX_BYTES) throw new Error(`the file is larger than ${MAX_FILE_MB} MB`);163 return record.value;164}1// One statement in, one dataset item out. Rows leave only when every total of the statement matched, and only2// those statements are charged. No DOM, no network, no platform calls: main.js passes those in.3//4// Privacy (research/bank-actor-privacy-2026-10-30.md): the log and the status message hold counts and status words only.5// Never a row, an amount, a URL, or an error text from a file. Items go to the run's own dataset, in the user's account.6import { LAYOUT_PROBLEM, SCAN_PROBLEM } from './parsers/parse.js';7
8export const EVENT = 'statement-parsed';9export const MAX_FILE_MB = 20;10export const MAX_PAGES = 100;11export const BUDGET_NOTE =12 'Stopped: the run reached its maximum cost. The statements after this point were not read, and they cost nothing.';13export const MESSAGES = {14 scanned: 'Scanned, not supported. This PDF has no text layer, and this Actor does not run OCR. No charge.',15 unsupported:16 'Not a layout we support: Chase checking or credit card, Bank of America checking, Wells Fargo checking, Capital One credit card, or TD Bank checking. No charge.',17 password: 'This PDF has a password. Remove the password and run it again. No charge.',18 tooLong: `This PDF has more than ${MAX_PAGES} pages, so it is not one statement. No charge.`,19 unreadable: 'This file could not be read as a PDF. No charge.',20};21const NO_ROWS = ' No rows are returned, and there is no charge.';22
23// statementUrls + statementFile, trimmed, with repeats removed so one file is not charged twice.24export function statementUrls(input) {25 const list = Array.isArray(input.statementUrls) ? input.statementUrls : [];26 return [27 ...new Set(28 [...list, input.statementFile]29 .filter((u) => typeof u === 'string')30 .map((u) => u.trim())31 .filter(Boolean),32 ),33 ];34}35
36const item = (url, status, problem, extra = {}) => ({37 url,38 status,39 charged: false,40 layout: null,41 period: null,42 rowCount: 0,43 rows: [],44 totals: [],45 problem,46 ...extra,47});48
49// parsed: what parseStatement returned, which may hold rows even when a total failed.50export function toItem(url, { layout, period, rows, totals, ok, problem }) {51 if (ok) return item(url, 'passed', null, { charged: true, layout, period, rowCount: rows.length, rows, totals });52 if (problem === SCAN_PROBLEM) return item(url, 'scanned', MESSAGES.scanned);53 if (problem === LAYOUT_PROBLEM) return item(url, 'unsupported', MESSAGES.unsupported);54 const why = problem.replace(' Please check the rows before you use them.', '');55 return item(url, 'check_failed', why + NO_ROWS, { layout, period, totals });56}57
58async function readOne(url, { download, read, parse }) {59 let bytes;60 try {61 bytes = await download(url);62 } catch (e) {63 return item(url, 'download_failed', `Could not download this file: ${e.message}. No charge.`);64 }65 try {66 return toItem(url, parse((await read(bytes)).lines));67 } catch (e) {68 if (e?.name === 'PasswordException') return item(url, 'password', MESSAGES.password);69 if (e?.code === 'TOO_LONG') return item(url, 'too_long', MESSAGES.tooLong);70 return item(url, 'unreadable', MESSAGES.unreadable);71 }72}73
74// left(): how many statements the run can still pay for. A statement starts only if it could be paid for.75export async function processStatements({ urls, download, read, parse, push, charge, left, log }) {76 const s = { statements: urls.length, passed: 0, noCharge: 0, stopped: false };77 for (const [i, url] of urls.entries()) {78 if (left() < 1) {79 s.stopped = true;80 await push({ note: BUDGET_NOTE });81 break;82 }83 const it = await readOne(url, { download, read, parse });84 await push(it);85 if (it.charged) {86 await charge();87 s.passed++;88 } else s.noCharge++;89 log(`Statement ${i + 1} of ${urls.length}: ${it.status}`);90 }91 return s;92}93
94export const summary = (s) =>95 `${s.statements} statement${s.statements === 1 ? '' : 's'}: ${s.passed} passed and charged, ${s.noCharge} no charge.${s.stopped ? ' Stopped at the maximum cost.' : ''}`;1// Bank Statement Parser: digital statement PDFs to JSON rows, checked against the statement's own totals.2// The file stays in memory. Rows go only to the run's default dataset. See the privacy note at the top of lib.js.3import { Actor, log } from 'apify';4import { download } from './http.js';5import { EVENT, processStatements, statementUrls, summary } from './lib.js';6import { parseStatement } from './parsers/parse.js';7import { readPdf } from './pdf.js';8
9await Actor.init();10
11try {12 const urls = statementUrls((await Actor.getInput()) ?? {});13 if (!urls.length) await Actor.fail('No statements were given. Add a statement URL or upload a PDF, then run again.');14 const charging = Actor.getChargingManager();15 const s = await processStatements({16 urls,17 download: (url) => download(url, Actor.apifyClient),18 read: readPdf,19 parse: parseStatement,20 push: (item) => Actor.pushData(item),21 charge: () => Actor.charge({ eventName: EVENT }),22 left: () => charging.calculateMaxEventChargeCountWithinLimit(EVENT),23 log: (line) => log.info(line),24 });25 const msg = summary(s);26 log.info(msg);27 await Actor.exit(msg);28} catch (err) {29 log.error(`The run stopped: ${err?.name ?? 'unknown error'}`); // the name only: a message could hold file text30 await Actor.fail('The run stopped on an internal error.');31}1// PDF bytes to lines with extractLines from brasspage.com/bank-statement-to-excel/ and the same pdf.js version as that page2// (the legacy build, which pdf.js asks for in Node).3import { MAX_PAGES } from './lib.js';4import { extractLines } from './parsers/parse.js';5
6const pdfjs = await import('pdfjs-dist/legacy/build/pdf.mjs');7pdfjs.GlobalWorkerOptions.workerSrc = import.meta.resolve('pdfjs-dist/legacy/build/pdf.worker.mjs');8
9const tooLong = () => Object.assign(new Error('too many pages'), { code: 'TOO_LONG' });10
11// extractLines takes the module, so this adds to its call: no pdf.js warnings in the log, and a page cap.12const capped = {13 getDocument(opts) {14 const task = pdfjs.getDocument({ ...opts, verbosity: 0 });15 return {16 destroy: () => task.destroy(),17 promise: task.promise18 .then((doc) => {19 if (doc.numPages > MAX_PAGES) throw tooLong();20 return doc;21 })22 .catch(async (e) => {23 await task.destroy();24 throw e;25 }),26 };27 },28};29
30export const readPdf = (bytes) => extractLines(capped, new Uint8Array(bytes));1// The download guard: only public hosts, on every hop, with a size cap. The one test server is on loopback and needs `isAllowed`.2import assert from 'node:assert/strict';3import http from 'node:http';4import { after, before, test } from 'node:test';5import { apifyRecord, download, fetchPublic, isPublicIp } from '../src/http.js';6
7let base;8let server;9const hits = [];10
11before(async () => {12 server = http.createServer((req, res) => {13 hits.push({ url: req.url, encoding: req.headers['accept-encoding'] });14 if (req.url === '/ok.pdf') return res.end('%PDF-file');15 if (req.url === '/to-ok') return res.writeHead(302, { location: '/ok.pdf' }).end();16 if (req.url === '/to-metadata') return res.writeHead(302, { location: 'http://169.254.169.254/latest/meta-data/' }).end();17 if (req.url === '/to-file') return res.writeHead(302, { location: 'file:///etc/passwd' }).end();18 if (req.url === '/loop') return res.writeHead(302, { location: '/loop' }).end();19 if (req.url === '/huge') {20 res.writeHead(200);21 const mb = Buffer.alloc(1024 * 1024);22 let n = 0;23 const send = () => {24 while (n++ < 21) if (!res.write(mb)) return res.once('drain', send);25 res.end();26 };27 return send();28 }29 return res.writeHead(404).end();30 });31 await new Promise((ok) => server.listen(0, '127.0.0.1', ok));32 base = `http://127.0.0.1:${server.address().port}`;33});34
35after(() => server.close());36
37const lo = { isAllowed: (ip) => ip === '127.0.0.1' || isPublicIp(ip) };38
39test('isPublicIp refuses private, metadata and mapped addresses', () => {40 for (const ip of [41 '127.0.0.1',42 '10.0.0.1',43 '169.254.169.254',44 '192.168.1.1',45 '100.64.0.1',46 '::1',47 'fe80::1',48 '::ffff:127.0.0.1',49 '64:ff9b::a9fe:a9fe',50 '2002:7f00:1::',51 ])52 assert.equal(isPublicIp(ip), false, ip);53 for (const ip of ['1.1.1.1', '2606:4700:4700::1111', '::ffff:1.1.1.1']) assert.equal(isPublicIp(ip), true, ip);54});55
56test('a public URL returns the file and asks for no compression', async () => {57 assert.equal((await fetchPublic(`${base}/ok.pdf`, lo)).toString(), '%PDF-file');58 assert.equal(hits.at(-1).encoding, undefined);59 assert.equal((await fetchPublic(`${base}/to-ok`, lo)).toString(), '%PDF-file');60});61
62test('private hosts are refused, on the first hop and after a redirect', async () => {63 await assert.rejects(fetchPublic(`${base}/ok.pdf`), /not a public internet host/); // no isAllowed: loopback64 await assert.rejects(fetchPublic('http://169.254.169.254/latest/meta-data/'), /not a public internet host/);65 await assert.rejects(fetchPublic('http://[::1]/x'), /not a public internet host/);66 await assert.rejects(fetchPublic(`${base}/to-metadata`, lo), /not a public internet host/);67});68
69test('other schemes, bad URLs, loops, errors and big files fail with a plain reason', async () => {70 await assert.rejects(fetchPublic('file:///etc/passwd'), /only http and https/);71 await assert.rejects(fetchPublic(`${base}/to-file`, lo), /only http and https/);72 await assert.rejects(fetchPublic('not a url'), /not valid/);73 await assert.rejects(fetchPublic(`${base}/loop`, lo), /too many redirects/);74 await assert.rejects(fetchPublic(`${base}/missing`, lo), /HTTP 404/);75 await assert.rejects(fetchPublic(`${base}/huge`, lo), /larger than 20 MB/);76});77
78test('apifyRecord matches only https api.apify.com record URLs', () => {79 assert.deepEqual(apifyRecord('https://api.apify.com/v2/key-value-stores/abc/records/my%20file?signature=s1'), {80 storeId: 'abc',81 key: 'my file',82 signature: 's1',83 });84 for (const u of [85 'http://api.apify.com/v2/key-value-stores/abc/records/k',86 'https://api.apify.com.evil.example/v2/key-value-stores/abc/records/k',87 'https://api.apify.com:8443/v2/key-value-stores/abc/records/k',88 'https://api.apify.com/v2/datasets/abc/items',89 'not a url',90 ])91 assert.equal(apifyRecord(u), null, u);92});93
94test('download sends Apify records through the client and everything else through the guard', async () => {95 const seen = [];96 const client = {97 keyValueStore: (id) => ({98 getRecord: async (key, opts) => (seen.push({ id, key, opts }), key === 'gone' ? undefined : { value: Buffer.from('%PDF') }),99 }),100 };101 const url = 'https://api.apify.com/v2/key-value-stores/abc/records/k?signature=s1';102 assert.equal((await download(url, client)).toString(), '%PDF');103 assert.deepEqual(seen[0], { id: 'abc', key: 'k', opts: { buffer: true, signature: 's1' } });104 await assert.rejects(download('https://api.apify.com/v2/key-value-stores/abc/records/gone', client), /not found/);105 await assert.rejects(download('http://127.0.0.1:1/x', client), /not a public internet host/);106 const denied = {107 keyValueStore: () => ({108 getRecord: async () => {109 throw Object.assign(new Error('Insufficient permissions'), { statusCode: 403 });110 },111 }),112 };113 await assert.rejects(download(url, denied), /use a signed record URL/);114 assert.equal(seen.length, 2);115});1// The whole statement path on real PDFs: the 7 made-up samples and the bad fixtures of the /bank-statement-to-excel/ tests.2import assert from 'node:assert/strict';3import fs from 'node:fs';4import path from 'node:path';5import { test } from 'node:test';6import { BUDGET_NOTE, MAX_PAGES, MESSAGES, processStatements, statementUrls, summary, toItem } from '../src/lib.js';7import { LAYOUT_PROBLEM, parseStatement, SCAN_PROBLEM } from '../src/parsers/parse.js';8import { readPdf } from '../src/pdf.js';9
10const REPO = path.resolve(import.meta.dirname, '../../..');11const SITE = path.join(REPO, 'site/bank-statement-to-excel');12const BAD = path.join(REPO, 'ops/fixtures/bank-csv');13const samples = fs.readdirSync(path.join(SITE, 'samples')).filter((f) => f.endsWith('.pdf'));14
15// Runs the path with file paths for URLs. left: how many statements the run can pay for.16async function run(urls, { left = Infinity, read = readPdf, download = (p) => fs.promises.readFile(p) } = {}) {17 const items = [];18 const lines = [];19 let charges = 0;20 const s = await processStatements({21 urls,22 download,23 read,24 parse: parseStatement,25 push: async (i) => items.push(i),26 charge: async () => charges++,27 left: () => left - charges,28 log: (l) => lines.push(l),29 });30 return { s, items, lines, charges };31}32
33test('every sample passes with rows, totals and one charge', async () => {34 assert.equal(samples.length, 7);35 const want = {36 'bofa-checking-sample.pdf': 'bofa-checking',37 'capone-card-sample.pdf': 'capone-card',38 'chase-card-sample.pdf': 'chase-card',39 'chase-checking-older-sample.pdf': 'chase-checking-sections',40 'chase-checking-sample.pdf': 'chase-checking',41 'td-checking-sample.pdf': 'td-checking',42 'wells-checking-sample.pdf': 'wells-checking',43 };44 for (const f of samples) {45 const { items, charges } = await run([path.join(SITE, 'samples', f)]);46 const [it] = items;47 assert.equal(it.status, 'passed', f);48 assert.equal(it.layout, want[f], f);49 assert.equal(it.charged, true, f);50 assert.equal(charges, 1, f);51 assert.ok(it.rows.length > 0 && it.rowCount === it.rows.length, f);52 assert.ok(it.totals.length > 0 && it.totals.every((t) => t.match), f);53 assert.match(it.period.start, /^\d{4}-\d{2}-\d{2}$/, f);54 assert.deepEqual(Object.keys(it.rows[0]), ['date', 'description', 'amount', 'balance', 'section'], f);55 assert.equal(it.problem, null, f);56 }57});58
59test('a scanned PDF says so, with no rows and no charge', async () => {60 const { items, charges } = await run([path.join(BAD, 'scan-no-text.pdf')]);61 assert.equal(items[0].status, 'scanned');62 assert.match(items[0].problem, /^Scanned, not supported/);63 assert.deepEqual(items[0].rows, []);64 assert.equal(items[0].charged, false);65 assert.equal(charges, 0);66});67
68test('a failed totals check returns no rows and no charge, and says which total failed', async () => {69 for (const f of ['chase-checking-off-by-a-cent.pdf', 'capone-total-off.pdf', 'td-subtotal-off.pdf', 'wells-daily-balance-off.pdf']) {70 const { items, charges } = await run([path.join(BAD, f)]);71 const [it] = items;72 assert.equal(it.status, 'check_failed', f);73 assert.deepEqual(it.rows, [], f);74 assert.equal(it.rowCount, 0, f);75 assert.equal(it.charged, false, f);76 assert.equal(charges, 0, f);77 assert.ok(78 it.totals.some((t) => !t.match),79 f,80 );81 assert.match(it.problem, /no rows are returned/i, f);82 assert.doesNotMatch(it.problem, /check the rows/, f);83 }84});85
86test('toItem never lets rows through when the parser says not ok', () => {87 const row = { date: '2026-01-02', description: 'x', amount: 1, balance: null, section: '' };88 const it = toItem('u', { layout: 'chase-card', period: null, rows: [row], totals: [], ok: false, problem: 'No ending balance.' });89 assert.equal(it.status, 'check_failed');90 assert.deepEqual(it.rows, []);91 assert.equal(it.charged, false);92 for (const [problem, status] of [93 [SCAN_PROBLEM, 'scanned'],94 [LAYOUT_PROBLEM, 'unsupported'],95 ])96 assert.equal(toItem('u', { layout: null, period: null, rows: [row], totals: [], ok: false, problem }).status, status);97});98
99test('a layout we do not know is unsupported and free', async () => {100 const { items, charges } = await run([path.join(BAD, 'not-chase.pdf')]);101 assert.equal(items[0].status, 'unsupported');102 assert.deepEqual(items[0].rows, []);103 assert.equal(charges, 0);104});105
106test('a Bank of America SafeBalance statement passes (it has no check section)', async () => {107 const { items, charges } = await run([path.join(BAD, 'bofa-safebalance.pdf')]);108 assert.equal(items[0].status, 'passed');109 assert.equal(charges, 1);110});111
112// A valid PDF of n empty pages, written by hand: pages tree, then a cross-reference table.113function emptyPdf(n) {114 const kids = Array.from({ length: n }, (_, i) => `${i + 3} 0 R`).join(' ');115 const objs = ['<</Type/Catalog/Pages 2 0 R>>', `<</Type/Pages/Kids[${kids}]/Count ${n}>>`];116 for (let i = 0; i < n; i++) objs.push('<</Type/Page/Parent 2 0 R/MediaBox[0 0 612 792]>>');117 let out = '%PDF-1.4\n';118 const at = objs.map((o, i) => {119 const offset = out.length;120 out += `${i + 1} 0 obj\n${o}\nendobj\n`;121 return offset;122 });123 const xref = out.length;124 out += `xref\n0 ${objs.length + 1}\n0000000000 65535 f \n`;125 for (const o of at) out += `${String(o).padStart(10, '0')} 00000 n \n`;126 return Buffer.from(`${out}trailer\n<</Size ${objs.length + 1}/Root 1 0 R>>\nstartxref\n${xref}\n%%EOF\n`, 'latin1');127}128
129test('real PDFs: a password PDF, one past the page cap, and an empty one at the cap', async () => {130 const pass = await run([path.join(import.meta.dirname, 'fixtures/password.pdf')]);131 assert.equal(pass.items[0].status, 'password');132 assert.equal(pass.charges, 0);133 const long = await run(['x'], { download: async () => emptyPdf(MAX_PAGES + 1) });134 assert.equal(long.items[0].status, 'too_long');135 assert.equal(long.charges, 0);136 const blank = await run(['x'], { download: async () => emptyPdf(MAX_PAGES) });137 assert.equal(blank.items[0].status, 'scanned'); // no text at all is what a scan looks like138 assert.equal(blank.charges, 0);139});140
141test('a file that is not a PDF, a password and a long PDF are free, and the run goes on', async () => {142 const pass = path.join(SITE, 'samples', samples[0]);143 const garbage = await run(['x', pass], { download: async (u) => (u === 'x' ? Buffer.from('not a pdf at all') : fs.readFileSync(u)) });144 assert.deepEqual(145 garbage.items.map((i) => i.status),146 ['unreadable', 'passed'],147 );148 assert.equal(garbage.charges, 1);149 for (const [name, code, status, message] of [150 ['PasswordException', undefined, 'password', MESSAGES.password],151 ['Error', 'TOO_LONG', 'too_long', MESSAGES.tooLong],152 ]) {153 const read = async () => {154 throw Object.assign(new Error('secret text'), { name, code });155 };156 const { items, charges } = await run(['x'], { read, download: async () => Buffer.alloc(1) });157 assert.equal(items[0].status, status);158 assert.equal(items[0].problem, message);159 assert.equal(charges, 0);160 }161 assert.match(MESSAGES.tooLong, new RegExp(String(MAX_PAGES)));162});163
164test('a download failure is one free item and the run goes on', async () => {165 const pass = path.join(SITE, 'samples', samples[0]);166 const download = async (u) => {167 if (u === 'bad') throw new Error('HTTP 404');168 return fs.readFileSync(u);169 };170 const { items, charges } = await run(['bad', pass], { download });171 assert.equal(items[0].status, 'download_failed');172 assert.match(items[0].problem, /HTTP 404/);173 assert.equal(items[1].status, 'passed');174 assert.equal(charges, 1);175});176
177test('the run stops before a statement it cannot pay for', async () => {178 const pass = path.join(SITE, 'samples', samples[0]);179 const other = path.join(SITE, 'samples', samples[1]);180 const read = [];181 const { s, items, charges } = await run([pass, other], {182 left: 1,183 download: async (u) => (read.push(u), fs.readFileSync(u)),184 });185 assert.equal(charges, 1);186 assert.deepEqual(read, [pass]);187 assert.equal(items.length, 2);188 assert.equal(items[1].note, BUDGET_NOTE);189 assert.equal(s.stopped, true);190 assert.match(summary(s), /Stopped/);191});192
193test('free statements do not stop at the maximum cost', async () => {194 const scan = path.join(BAD, 'scan-no-text.pdf');195 const { s, items } = await run([scan, path.join(BAD, 'not-chase.pdf')], { left: 1 });196 assert.equal(items.length, 2);197 assert.equal(s.noCharge, 2);198});199
200test('log lines and the summary hold no row, amount, URL or file text', async () => {201 const urls = samples.map((f) => path.join(SITE, 'samples', f));202 urls.push(path.join(BAD, 'chase-checking-off-by-a-cent.pdf'), 'https://secret.example/private-token/a.pdf');203 const { s, items, lines } = await run(urls, {204 download: async (u) => {205 if (u.startsWith('https')) throw new Error('HTTP 403');206 return fs.readFileSync(u);207 },208 });209 const text = [...lines, summary(s)].join('\n');210 for (const line of lines) assert.match(line, /^Statement \d+ of \d+: [a-z_]+$/);211 assert.doesNotMatch(text, /secret|\$|\/|\d\.\d\d/);212 for (const it of items)213 for (const r of it.rows ?? []) {214 assert.ok(!text.includes(r.description), 'a description reached the log');215 assert.ok(!text.includes(String(Math.abs(r.amount))) || Math.abs(r.amount) < 100, 'an amount reached the log');216 }217});218
219test('the summary counts one statement in the singular', () => {220 assert.equal(summary({ statements: 1, passed: 1, noCharge: 0, stopped: false }), '1 statement: 1 passed and charged, 0 no charge.');221});222
223test('statementUrls trims, drops junk and repeats, and adds the upload', () => {224 assert.deepEqual(statementUrls({ statementUrls: [' a ', 'a', '', 7, null], statementFile: 'b' }), ['a', 'b']);225 assert.deepEqual(statementUrls({ statementUrls: 'a' }), []);226 assert.deepEqual(statementUrls({}), []);227});228
229test('the parser copies are byte for byte the ones on brasspage.com', () => {230 for (const f of fs.readdirSync(path.join(import.meta.dirname, '../src/parsers'))) {231 const mine = fs.readFileSync(path.join(import.meta.dirname, '../src/parsers', f));232 assert.ok(mine.equals(fs.readFileSync(path.join(SITE, f))), `${f}: run cp from ${SITE}`);233 }234});235
236test('pdf.js is the version the page vendors', () => {237 const vendored = /pdfjsVersion = ([\d.]+)/.exec(fs.readFileSync(path.join(SITE, 'pdf.min.mjs'), 'utf8').slice(0, 2000))?.[1];238 const pkg = JSON.parse(fs.readFileSync(path.join(import.meta.dirname, '../package.json'), 'utf8'));239 assert.equal(pkg.dependencies['pdfjs-dist'], vendored);240});241
242test('the prices in the README, actor.json and pricing.json agree, and there is no start fee', () => {243 const root = path.join(import.meta.dirname, '..');244 const events = JSON.parse(fs.readFileSync(path.join(root, 'pricing.json'), 'utf8')).pricingInfos[0].pricingPerEvent.actorChargeEvents;245 assert.deepEqual(Object.keys(events), ['statement-parsed']);246 assert.equal(events['statement-parsed'].eventPriceUsd, 0.05);247 assert.match(JSON.parse(fs.readFileSync(path.join(root, '.actor/actor.json'), 'utf8')).description, /\$0\.05 per passed statement/);248 assert.match(fs.readFileSync(path.join(root, 'README.md'), 'utf8'), /\$0\.05/);249});1// Runs the whole Actor (src/main.js) on local storage. No public site is contacted: the URLs are loopback or .invalid.2import assert from 'node:assert/strict';3import { spawnSync } from 'node:child_process';4import fs from 'node:fs';5import os from 'node:os';6import path from 'node:path';7import { test } from 'node:test';8
9const root = path.resolve(import.meta.dirname, '..');10const storage = fs.mkdtempSync(path.join(os.tmpdir(), 'bankparser-'));11
12function run(input) {13 fs.rmSync(storage, { recursive: true, force: true });14 fs.mkdirSync(path.join(storage, 'key_value_stores/default'), { recursive: true });15 fs.writeFileSync(path.join(storage, 'key_value_stores/default/INPUT.json'), JSON.stringify(input));16 const r = spawnSync(process.execPath, ['src/main.js'], {17 cwd: root,18 env: { ...process.env, CRAWLEE_STORAGE_DIR: storage, CRAWLEE_PURGE_ON_START: '0' },19 encoding: 'utf8',20 });21 const dir = path.join(storage, 'datasets/default');22 const items = fs.existsSync(dir)23 ? fs24 .readdirSync(dir)25 .filter((f) => /^\d+\.json$/.test(f))26 .sort()27 .map((f) => JSON.parse(fs.readFileSync(path.join(dir, f), 'utf8')))28 : [];29 return { code: r.status, log: r.stderr + r.stdout, items };30}31
32test('no input fails the run with a clear message and no items', () => {33 const r = run({});34 assert.notEqual(r.code, 0, r.log);35 assert.match(r.log, /No statements were given/);36 assert.equal(r.items.length, 0);37});38
39test('blocked URLs give one free item each, the run succeeds, and the log holds no URL', () => {40 const urls = ['http://169.254.169.254/private-token/a.pdf', 'https://no-such-site.invalid/private-token/b.pdf'];41 const r = run({ statementUrls: urls });42 assert.equal(r.code, 0, r.log);43 assert.deepEqual(44 r.items.map((i) => [i.url, i.status, i.charged]),45 urls.map((u) => [u, 'download_failed', false]),46 );47 assert.match(r.items[0].problem, /not a public internet host/);48 assert.doesNotMatch(r.log, /private-token|169\.254|no-such-site/);49 assert.match(r.log, /2 statements: 0 passed and charged, 2 no charge/);50});1// Bank of America personal checking statements (the 2013 redesign, still in use in 2026), for parse.js.2// Layout: research/bank-csv-layout-bofa-2026-10-29.md. The tested statements are the made-up ones of3// ops/bank-csv-fixtures-banks.mjs. Credit card and business (credits/debits) statements are not read.4//5// An "Account summary" box, then one table per section (Deposits and other additions, Withdrawals and other6// subtractions, which some accounts split into ATM and debit card subtractions and Other subtractions, Checks,7// Service fees), each ending in a "Total ..." line. Rows print MM/DD/YY and a signed amount (money out has a8// minus), so the printed sign is the CSV sign. Checks print two across: Date, Check #, Amount, twice.9// Real text layers split words at odd places, so labels match with the spaces taken out.10import { cents, MONTHS, sum, total } from "./parse.js";11
12const key = (s) => s.replace(/\s+/g, "").toLowerCase();13const DATE = /^(\d{2})\/(\d{2})\/(\d{2})$/;14const SECTIONS = [15 "Deposits and other additions",16 "Withdrawals and other subtractions",17 "ATM and debit card subtractions",18 "Other subtractions",19 "Checks",20 "Service fees",21];22const HEAD = new Map(SECTIONS.map((s) => [key(s), s]));23// Sub-sections roll up into the withdrawals line when the summary has no line of their own.24const PARTS = {25 "Withdrawals and other subtractions": [26 "Withdrawals and other subtractions",27 "ATM and debit card subtractions",28 "Other subtractions",29 ],30};31const PERIOD =32 /for ([A-Z][a-z]+) (\d{1,2}), (\d{4}) to ([A-Z][a-z]+) (\d{1,2}), (\d{4})/;33const pad = (n) => String(n).padStart(2, "0");34
35// Business statements say "credits" and "debits"; they are not tested, so they get the plain layout message.36export const bofa = {37 detect: (has) =>38 has(/Bank of America/i) &&39 has(/^Account summary\b/) &&40 has(/^Beginning balance on [A-Z][a-z]+ \d{1,2}, \d{4}/) &&41 !has(/^(Deposits and other credits|Withdrawals and other debits)/i)42 ? "bofa-checking"43 : null,44 parse,45};46
47function findPeriod(L) {48 for (const { text } of L) {49 const m = PERIOD.exec(text);50 if (m && MONTHS.includes(m[1]) && MONTHS.includes(m[4]))51 return {52 start: `${m[3]}-${pad(MONTHS.indexOf(m[1]) + 1)}-${pad(m[2])}`,53 end: `${m[6]}-${pad(MONTHS.indexOf(m[4]) + 1)}-${pad(m[5])}`,54 };55 }56 return null;57}58
59const iso = (m) => `20${m[3]}-${m[1]}-${m[2]}`;60
61function parse(L) {62 const summary = new Map();63 const totals = []; // the "Total ..." lines: { label, cents, section }64 const rows = [];65 let mode = null; // summary | rows | null66 let done = false; // the summary box is read once ("Account summary - continued" is interest text)67 let section = null;68 let descX = 0;69 let last = null;70 let page = 0;71 for (const l of L) {72 if (l.page !== page) [page, last] = [l.page, null];73 const { cells } = l;74 const k = key(l.text);75 // Page furniture: page numbers, the running header (NAME ! Account # 0000 0000 0000 ! period, a 12 digit76 // number, so a description that says "Account # 1234" stays), the page 1 mail codes.77 if (78 /page\d+of\d+$/.test(k) ||79 (/account#\d{12}(?!\d)/.test(k) && !DATE.test(cells[0].str)) ||80 /^pull:/.test(k) ||81 k === "continuedonthenextpage"82 )83 continue;84 const amount = cents(cells.at(-1).str);85 const label = key(86 cells87 .slice(0, amount === null ? cells.length : -1)88 .map((c) => c.str)89 .join(""),90 );91 if (!done && /^accountsummary/.test(k) && amount === null) {92 mode = "summary";93 continue;94 }95 if (mode === "summary") {96 // Label, then its amount; the right column (bank address, notes) can share the line after it.97 const at = cells.findIndex((c) => cents(c.str) !== null);98 if (at < 1) continue;99 const name = key(100 cells101 .slice(0, at)102 .map((c) => c.str)103 .join(""),104 );105 const v = cents(cells[at].str);106 if (name.startsWith("beginningbalanceon")) summary.set("begin", v);107 else if (name.startsWith("endingbalanceon")) {108 summary.set("end", v);109 [mode, done] = [null, true];110 } else if (HEAD.has(name)) summary.set(HEAD.get(name), v);111 continue;112 }113 const head = HEAD.get(k.replace(/-continued$/, ""));114 if (head && amount === null) {115 [mode, section, last] = ["rows", head, null];116 continue;117 }118 if (/^(dailyledgerbalances|importantinformation)/.test(k)) {119 [mode, last] = [null, null];120 continue;121 }122 if (mode !== "rows") continue;123 if (cells[0].str === "Date") {124 descX = cells[1]?.x ?? 0;125 continue;126 }127 if (label.startsWith("total")) {128 // Only a section's own total counts. "Total # of checks 2" and the overdraft box ("Total Overdraft129 // fees $0.00 $0.00", "Total NSF: Returned Item fees ...") are not money for these rows.130 last = null;131 const named = SECTIONS.find((s) => label === "total" + key(s));132 if (named && amount !== null)133 totals.push({134 label: l.text.slice(0, l.text.lastIndexOf(" ")),135 cents: amount,136 section: named,137 });138 continue;139 }140 if (section === "Checks") {141 // Two checks across: Date, Check #, Amount, then the same again. A "*" marks a gap in the numbers;142 // it can be its own item.143 const cs = cells.filter((c) => c.str !== "*");144 for (let i = 0; i + 2 < cs.length; i++) {145 const d = DATE.exec(cs[i].str);146 const v = cents(cs[i + 2].str);147 const n = /^(\d+)\s*\*?$/.exec(cs[i + 1].str);148 if (!d || v === null || !n) continue;149 rows.push({150 date: iso(d),151 description: `Check #${n[1]}`,152 cents: v,153 balance: null,154 section,155 });156 i += 2;157 }158 last = null;159 continue;160 }161 const d = DATE.exec(cells[0].str);162 if (d) {163 const desc = cells.slice(1, amount === null ? cells.length : -1);164 last = {165 date: iso(d),166 description: desc.map((c) => c.str).join(" "),167 cents: amount, // null: a long description whose amount sits on a later line168 balance: null,169 section,170 y: l.y,171 };172 rows.push(last);173 continue;174 }175 // A wrapped description: no date, in the description column, right under its row.176 if (last && cells[0].x >= descX - 5 && last.y - l.y <= 12) {177 const text = cells.slice(0, amount === null ? cells.length : -1);178 if (text.length)179 last.description += " " + text.map((c) => c.str).join(" ");180 if (amount !== null && last.cents === null) last.cents = amount;181 last.y = l.y;182 } else last = null;183 }184
185 // A row whose amount never showed up is left out, so its section total says "Does not match".186 const kept = rows.filter((r) => r.cents !== null);187 const raw = [];188 for (const [name, v] of summary) {189 if (name === "begin" || name === "end") continue;190 const parts = PARTS[name]?.filter((p) => p === name || !summary.has(p)) || [191 name,192 ];193 raw.push(194 total(name, v, sum(kept.filter((r) => parts.includes(r.section)))),195 );196 }197 // "Total withdrawals and other subtractions" on a split statement covers both sub-sections.198 for (const t of totals) {199 const parts = PARTS[t.section] || [t.section];200 raw.push(201 total(202 t.label,203 t.cents,204 sum(kept.filter((r) => parts.includes(r.section))),205 ),206 );207 }208 if (summary.has("begin") && summary.has("end"))209 raw.push(210 total(211 "Ending balance (Beginning balance + all rows)",212 summary.get("end"),213 summary.get("begin") + sum(kept),214 true,215 ),216 );217 return { period: findPeriod(L), rows: kept, raw };218}1// Capital One credit card statements (the layout in use from mid 2021 to 2026), for parse.js.2// Layout: research/bank-csv-layout-capitalone-2026-10-30.md sections 2 to 10. The tested statements are the3// made-up ones of ops/bank-csv-fixtures-capone.mjs. 360 Checking and the 2019 to 2021 card layout (one date4// column, the amount alone under a wrapped description) are not read.5//6// Page 1: an Account Summary box whose amounts carry a printed sign ("- $", "+ $", "= $"; the sign can be its7// own item). Then Transactions: per cardholder "<NAME> #NNNN: Payments, Credits and Adjustments" and8// "<NAME> #NNNN: Transactions" blocks of "Mon D Mon D DESCRIPTION amount" rows (trans date, post date, no9// year), Fees, and Interest Charged lines that have no date. Payments and credits print "- $350.00" and10// purchases "$60.00", so amount = -printed, as on Chase cards. Spanish statements repeat every page, Spanish11// first ("Página 1 de 6"): those pages are skipped. Real text layers split words at odd places, so labels match12// with the spaces taken out.13import { cents, MONTHS, rowDate, sum, total } from "./parse.js";14
15const tight = (s) => s.replace(/\s+/g, "");16// A phrase that still matches when the PDF splits its words anywhere.17const loose = (phrase) =>18 new RegExp(19 [...tight(phrase)]20 .map((ch) => ch.replace(/[.*+?^${}()|[\]\\/]/g, "\\$&"))21 .join("\\s*"),22 "i",23 );24const HEADER = loose("Trans Date Post Date Description Amount");25const AMT = "(\\d{1,3}(?:,\\d{3})*\\.\\d{2})";26// "Dec 14, 2025 - Jan 13, 2026 | 31 days in Billing Cycle", on the spaceless text. "Jun. 05", en dash, U+2212.27const PERIOD = new RegExp(28 `([A-Z][a-z]{2})\\.?(\\d{1,2}),(\\d{4})[-\u2013\u2212]([A-Z][a-z]{2})\\.?(\\d{1,2}),(\\d{4})\\|?\\d+daysinBillingCycle`,29);30const ROW = new RegExp(31 `^([A-Z][a-z]{2})\\.?\\s*(\\d{1,2})\\s+([A-Z][a-z]{2})\\.?\\s*(\\d{1,2})\\s+(.*?)\\s*(-)?\\s*\\$${AMT}$`,32);33const TAIL = new RegExp(`(-?)\\$${AMT}$`); // a line's last amount, on the spaceless text34const AUTH = /\s*AuthDate\s*\d{1,2}-[A-Za-z]{3}$/; // "CAPITAL ONE MOBILE PYMTAuthDate 22-May"35// Summary labels and the sign each must print. "+" and "=" always print, so "Available Credit for Cash36// Advances $0.00" and the coupon's "New Balance $X" (no sign) are not read as the summary line.37const SUMMARY = [38 ["Previous Balance", "-?"],39 ["Payments", "-?"],40 ["Other Credits", "-?"],41 ["Transactions", "\\+"],42 ["Cash Advances", "\\+"],43 ["Fees Charged", "\\+"],44 ["Interest Charged", "\\+"],45 ["New Balance", "="],46].map(([label, sign]) => [47 label,48 new RegExp(`${tight(label)}(${sign})\\$${AMT}`, "i"),49]);50// Printed totals: pattern, label, and the start of the section names of the rows each covers.51const TOTALS = [52 [53 /#(\d{4}):TotalTransactions-?\$/i,54 "Total Transactions #$1",55 "Transactions #$1",56 ],57 [58 /^TotalTransactionsforThisPeriod-?\$/i,59 "Total Transactions for This Period",60 "Transactions #",61 ],62 [/^TotalFeesforThisPeriod-?\$/i, "Total Fees for This Period", "Fees"],63 [64 /^TotalInterestforThisPeriod-?\$/i,65 "Total Interest for This Period",66 "Interest Charged",67 ],68];69// Summary line, and the start of the section names of the rows it covers.70const COVERS = {71 Transactions: "Transactions #",72 "Fees Charged": "Fees",73 "Interest Charged": "Interest Charged",74};75const pad = (n) => String(n).padStart(2, "0");76const month = (abbr) => MONTHS.findIndex((m) => m.startsWith(abbr)) + 1; // 0: not a month77
78export const capone = {79 detect: (has) =>80 has(/capital\s*one/i) &&81 has(/ending\s*in\s*\d{4}/i) &&82 has(loose("days in Billing Cycle")) &&83 has(HEADER)84 ? "capone-card"85 : null,86 parse,87};88
89function parse(L) {90 // A page that says "Página N de M" is the Spanish copy; an unnumbered page (terms) goes with the page before.91 const lang = {};92 for (const l of L) {93 const t = tight(l.text);94 if (/P\S{1,2}gina\d+de\d+/.test(t)) lang[l.page] = "es";95 else if (/Page\d+of\d+/.test(t)) lang[l.page] ??= "en";96 }97 const summary = new Map();98 const printed = []; // the "Total ..." lines: { label, cents, covers }99 const rows = [];100 let period = null;101 let mode = null; // summary | rows102 let section = "";103 let descX = 0;104 let last = null;105 let page = 0;106 let spanish = false;107 for (const l of L) {108 if (l.page !== page) {109 [page, last] = [l.page, null];110 if (lang[page]) spanish = lang[page] === "es";111 }112 if (spanish) continue;113 const t = tight(l.text);114 let m = PERIOD.exec(t);115 if (m && !period && month(m[1]) && month(m[4]))116 period = {117 start: `${m[3]}-${pad(month(m[1]))}-${pad(m[2])}`,118 end: `${m[6]}-${pad(month(m[4]))}-${pad(m[5])}`,119 };120 // The dark bar: "Transactions", or "Transactions (Continued)" on the next page (the open block carries on).121 if (/^Transactions(\(Continued\))?$/i.test(t)) {122 [mode, last] = ["rows", null];123 continue;124 }125 if (!mode && /AccountSummary/i.test(t)) {126 mode = "summary";127 continue;128 }129 if (mode === "summary") {130 // The Payment Information and Rewards boxes share these lines; only label + sign + $amount counts.131 for (const [label, re] of SUMMARY)132 if (!summary.has(label) && (m = re.exec(t)))133 // Payments and Other Credits print "-" as a word for minus: kept unsigned. A "- $" Previous134 // Balance is a credit balance, so it keeps its sign.135 summary.set(136 label,137 (label === "Previous Balance" && m[1] ? -1 : 1) * cents(m[2]),138 );139 continue;140 }141 if (mode !== "rows") continue;142 if (/^TotalsYear-to-Date/i.test(t)) break; // then the APR table: amounts that are not rows143 if ((m = /#(\d{4}):Payments,?Credits?andAdjustments?$/i.exec(t))) {144 [section, last] = [`Payments, Credits and Adjustments #${m[1]}`, null];145 continue;146 }147 if ((m = /#(\d{4}):Transactions$/i.exec(t))) {148 [section, last] = [`Transactions #${m[1]}`, null];149 continue;150 }151 if (/^(Fees|InterestCharged)$/i.test(t)) {152 [section, last] = [/^fees$/i.test(t) ? "Fees" : "Interest Charged", null];153 continue;154 }155 if (HEADER.test(l.text)) {156 // pdf.js can join the header into one item: estimate where "Description" starts in it.157 const d = l.cells.find((c) => /desc/i.test(c.str));158 descX = d159 ? d.x + (d.w * d.str.toLowerCase().indexOf("desc")) / d.str.length160 : 0;161 continue;162 }163 const tail = TAIL.exec(t);164 const v = tail && cents(tail[1] + tail[2]);165 const named = TOTALS.find(([re]) => re.test(t));166 if (named && tail) {167 const n = named[0].exec(t)[1];168 const [label, covers] = named.slice(1).map((s) => s.replace("$1", n));169 printed.push({ label, cents: v, covers });170 last = null;171 continue;172 }173 // Interest has no date: one row per non-zero line, dated the closing date.174 m = /^InterestChargeon(Purchases?|CashAdvances|OtherBalances)-?\$/i.exec(t);175 if (m && tail && section === "Interest Charged") {176 if (v)177 rows.push({178 description:179 "Interest Charge on " + m[1].replace(/([a-z])([A-Z])/g, "$1 $2"),180 printed: v,181 section,182 });183 last = null;184 continue;185 }186 const r = ROW.exec(l.text);187 if (r && month(r[1]) && month(r[3])) {188 last = {189 mm: pad(month(r[1])),190 dd: pad(r[2]),191 description: r[5].replace(AUTH, ""),192 printed: (r[6] ? -1 : 1) * cents(r[7]),193 section,194 y: l.y,195 };196 rows.push(last);197 continue;198 }199 // A foreign purchase's extra lines ("$40.00", "EUR", "0.920000000 Exchange Rate") are not rows or text.200 if (tail || /ExchangeRate$/i.test(t)) {201 last = null;202 continue;203 }204 // A wrapped description: no date, no amount, in the description column, right under its row.205 if (last && l.cells[0].x >= descX - 5 && last.y - l.y <= 11) {206 last.description += " " + l.text;207 last.y = l.y;208 } else last = null;209 }210
211 if (!period) return { period: null, rows: [], raw: [] };212 const out = rows.map((r) => ({213 date: r.mm ? rowDate(r.mm, r.dd, period) : period.end,214 description: r.description,215 cents: -r.printed,216 balance: null,217 section: r.section,218 }));219 const of = (start) => sum(out.filter((r) => r.section.startsWith(start)));220 // Every total in the CSV's signs: purchases, fees and interest negative, payments and credits positive.221 const raw = printed.map((p) => total(p.label, -p.cents, of(p.covers)));222 if (summary.has("Payments") || summary.has("Other Credits"))223 raw.push(224 total(225 "Payments + Other Credits",226 (summary.get("Payments") ?? 0) + (summary.get("Other Credits") ?? 0),227 of("Payments, Credits and Adjustments #"),228 ),229 );230 // A non-zero Cash Advances has no rows we read (not confirmed where they print), so a check below fails.231 for (const [label, start] of Object.entries(COVERS))232 if (summary.has(label))233 raw.push(total(label, -summary.get(label), of(start)));234 if (summary.has("Previous Balance") && summary.has("New Balance"))235 raw.push(236 total(237 "New Balance (Previous Balance + all activity)",238 summary.get("New Balance"),239 summary.get("Previous Balance") - sum(out),240 true,241 ),242 );243 return { period, rows: out, raw };244}1// TD Bank (US) checking statements (the "STATEMENT OF ACCOUNT" layout, seen on real PDFs from 2017 to 2025), for2// parse.js. Layout: research/bank-csv-layout-tdbank-2026-10-30.md. The tested statements are the made-up ones of3// ops/bank-csv-fixtures-td.mjs. The municipal "Transactions by Date" layout and TD Canada Trust are not read.4//5// An ACCOUNT SUMMARY box (left column: the money lines; right column: interest and days, not money we check),6// then DAILY ACCOUNT ACTIVITY: one table per section (Deposits, Electronic Deposits, Other Credits, Checks Paid,7// Electronic Payments, Other Withdrawals, Service Charges), each ending in "Subtotal:". Rows print MM/DD (M/D on8// the older render) and an unsigned amount: the section says which way the money went. Checks print two across,9// left column first. Page 2 is a balancing worksheet. A DAILY BALANCE SUMMARY ends the statement.10// Real text layers drop the spaces inside words ("ElectronicDeposits"), so labels match with the spaces taken out.11import { cents, MONTHS, rowDate, sum, total } from "./parse.js";12
13const tight = (s) => s.replace(/\s+/g, "");14// A phrase that still matches when the PDF splits its words anywhere.15const loose = (phrase) =>16 new RegExp(17 [...tight(phrase)]18 .map((ch) => ch.replace(/[.*+?^${}()|[\]\\/]/g, "\\$&"))19 .join("\\s*"),20 "i",21 );22const MD = /^(\d{1,2})\/(\d{1,2})$/;23const PERIOD =24 /StatementPeriod:([A-Z][a-z]{2})(\d{1,2})(\d{4})-([A-Z][a-z]{2})(\d{1,2})(\d{4})/;25// Each section and its sign: money in (+) or money out (-). The summary box uses the same names.26const SECTIONS = {27 Deposits: 1,28 "Electronic Deposits": 1,29 "Other Credits": 1,30 "Checks Paid": -1,31 "Electronic Payments": -1,32 "Other Withdrawals": -1,33 "Service Charges": -1,34};35// Headings are Title Case, so a wrapped description that says "DEPOSITS" is not one.36const HEAD = new Map(Object.keys(SECTIONS).map((s) => [tight(s), s]));37const LABEL = new Map(38 Object.keys(SECTIONS).map((s) => [tight(s).toLowerCase(), s]),39);40const pad = (n) => String(n).padStart(2, "0");41// A debit card row's last line is the full card number: keep only its last 4 digits, however it is spaced.42const CARD = /\b\d{4}[ -]?\d{4}[ -]?\d{4}[ -]?\d{4}\b/g;43const mask = (s) => s.replace(CARD, (n) => `CARD ENDING ${n.slice(-4)}`);44const month = (abbr) => MONTHS.findIndex((m) => m.startsWith(abbr)) + 1;45
46export const td = {47 detect: (has) =>48 has(/TD Bank, N\.A\.|tdbank\.com/i) &&49 has(loose("STATEMENT OF ACCOUNT")) &&50 has(loose("Statement Period:")) &&51 has(loose("Primary Account #:")) &&52 has(loose("ACCOUNT SUMMARY")) &&53 (has(loose("DAILY ACCOUNT ACTIVITY")) ||54 has(loose("No Transactions this Statement Period"))) &&55 !has(loose("Transactions by Date")) &&56 !has(/^TD\s.*Savings$/i) // savings is not tested, and its OFX would say CHECKING57 ? "td-checking"58 : null,59 parse,60};61
62// "Statement Period: Dec 06 2025-Jan 05 2026", on the page 1 header and every continuation page.63function findPeriod(L) {64 for (const l of L) {65 const m = PERIOD.exec(tight(l.text));66 if (!m || !month(m[1]) || !month(m[4])) continue;67 return {68 start: `${m[3]}-${pad(month(m[1]))}-${pad(m[2])}`,69 end: `${m[6]}-${pad(month(m[4]))}-${pad(m[5])}`,70 };71 }72 return null;73}74
75function parse(L) {76 // Page 2 (How to Balance your Account) repeats the ending balance and says "Deposits": not rows.77 const worksheet = new Set(78 L.filter((l) => /^HowtoBalanceyourAccount/i.test(tight(l.text))).map(79 (l) => l.page,80 ),81 );82 const summary = new Map(); // "begin", "end" and section names: the printed amount83 const subtotals = []; // { section, cents } with the CSV sign84 const daily = []; // [mm, dd, balance cents]85 const rows = [];86 let right = []; // this page's right-hand check column, listed after the left one87 let mode = null; // summary | rows | daily88 let section = null;89 let amtX = 0; // left edge of the AMOUNT header: amounts end right of it90 let descX = 0;91 let last = null;92 let page = 0;93 const flush = () => {94 rows.push(...right);95 right = [];96 };97 for (const l of L) {98 if (worksheet.has(l.page)) continue;99 if (l.page !== page) {100 flush();101 [page, last] = [l.page, null];102 }103 const { cells } = l;104 const t = tight(l.text);105 if (/^ACCOUNTSUMMARY/i.test(t)) {106 mode = "summary";107 continue;108 }109 if (/^DAILYACCOUNTACTIVITY$/i.test(t)) {110 [mode, last] = ["rows", null];111 continue;112 }113 if (/^DAILYBALANCESUMMARY$/i.test(t)) {114 flush();115 [mode, last] = ["daily", null];116 continue;117 }118 if (mode === "summary") {119 // Label and amount pairs, left to right. Only the left column's labels are known, so the right column120 // (Average Collected Balance, Interest Earned This Period, ...) is skipped. Zero lines are not printed.121 let label = "";122 for (const c of cells) {123 const v = cents(c.str);124 if (v === null) {125 label += tight(c.str).toLowerCase();126 continue;127 }128 if (label === "beginningbalance") summary.set("begin", v);129 else if (label === "endingbalance") {130 summary.set("end", v);131 mode = null;132 } else if (LABEL.has(label)) summary.set(LABEL.get(label), v);133 label = "";134 }135 continue;136 }137 if (mode === "daily") {138 // Date and balance pairs, two across.139 for (let i = 0; i + 1 < cells.length; i++) {140 const d = MD.exec(cells[i].str);141 const v = cents(cells[i + 1].str);142 if (!d || v === null) continue;143 daily.push([d[1], d[2], v]);144 i++;145 }146 continue;147 }148 if (mode !== "rows") continue;149 // "Electronic Payments (continued)" on a continuation page; Checks Paid can carry "No. Checks: 3" on its line.150 const head = HEAD.get(151 t.replace(/No\.Checks:\d+$/i, "").replace(/\(continued\)$/i, ""),152 );153 if (head) {154 flush();155 [section, last] = [head, null];156 continue;157 }158 if (/^POSTINGDATE/i.test(t)) {159 amtX = cells.find((c) => /^AMOUNT$/i.test(c.str))?.x ?? amtX;160 descX = cells.find((c) => /^DESCRIPTION$/i.test(c.str))?.x ?? descX;161 last = null;162 continue;163 }164 if (/^Subtotal:/i.test(t)) {165 flush();166 const v = cents(cells.at(-1).str);167 if (section && v !== null)168 subtotals.push({ section, cents: SECTIONS[section] * v });169 last = null;170 continue;171 }172 if (!section) continue;173 const sign = SECTIONS[section];174 const lead = MD.exec(cells[0].str);175 if (section === "Checks Paid" && lead) {176 // Date, serial number, amount, then the same again. A "*" after the number marks a gap in the numbers;177 // it can be its own item. The header, "No. Checks:" and the "* Indicates ..." note hold no such triple.178 const cs = cells.filter((c) => c.str !== "*");179 let found = 0;180 for (let i = 0; i + 2 < cs.length; i++) {181 const d = MD.exec(cs[i].str);182 const n = /^(\d+)\s*\*?$/.exec(cs[i + 1].str);183 const v = cents(cs[i + 2].str);184 if (!d || !n || v === null) continue;185 // ponytail: left or right column by the page middle; read the header's second DATE x if a layout moves it186 (cs[i].x > 306 ? right : rows).push({187 date: [d[1], d[2]],188 description: `Check #${n[1]}`,189 cents: sign * v,190 balance: null,191 section,192 });193 found++;194 i += 2;195 }196 last = null;197 if (found) continue; // else a check printed like any other row (not seen yet): read it below198 }199 // The amount comes only from the amount column: a wrapped line can hold a number that looks like one.200 const inCol = (c) => cents(c.str) !== null && c.x + c.w >= amtX;201 const amount = inCol(cells.at(-1)) ? cents(cells.at(-1).str) : null;202 if (lead && amount !== null) {203 last = {204 date: [lead[1], lead[2]],205 description: cells206 .slice(1, -1)207 .map((c) => c.str)208 .join(" "),209 cents: sign * amount,210 balance: null,211 section,212 y: l.y,213 };214 rows.push(last);215 continue;216 }217 // A wrapped description: no date, no amount, in the description column, right under its row.218 if (219 last &&220 !lead &&221 !cells.some(inCol) &&222 cells[0].x >= descX - 5 &&223 last.y - l.y <= 14224 ) {225 last.description += " " + l.text;226 last.y = l.y;227 } else last = null;228 }229 flush();230
231 const period = findPeriod(L);232 if (!period) return { period: null, rows: [], raw: [] };233 const iso = ([mm, dd]) => rowDate(pad(mm), pad(dd), period);234 const out = rows.map((r) => ({235 ...r,236 date: iso(r.date),237 description: mask(r.description),238 }));239 const of = (s) => sum(out.filter((r) => r.section === s));240 const raw = [];241 for (const [name, v] of summary)242 if (SECTIONS[name]) raw.push(total(name, SECTIONS[name] * v, of(name)));243 for (const s of subtotals)244 raw.push(total(`${s.section} Subtotal`, s.cents, of(s.section)));245 const begin = summary.get("begin");246 const end = summary.get("end");247 if (begin === undefined) return { period, rows: out, raw };248 if (daily.length) {249 // Every printed daily balance against Beginning Balance + the rows up to that date, added up into one check.250 // ponytail: rows x days scan, fine for one statement251 let printed = 0;252 let ran = 0;253 for (const d of daily) {254 const day = iso(d);255 printed += d[2];256 ran += begin + sum(out.filter((r) => r.date <= day));257 }258 const n = daily.length;259 raw.push(260 total(261 `Daily Balance Summary (${n} ${n === 1 ? "day" : "days"} added up)`,262 printed,263 ran,264 ),265 );266 }267 if (end !== undefined)268 raw.push(269 total(270 "Ending Balance (Beginning Balance + all rows)",271 end,272 begin + sum(out),273 true,274 ),275 );276 return { period, rows: out, raw };277}1// Wells Fargo personal checking statements (the layout in use from at least 2018 to 2026), for parse.js.2// Layout: research/bank-csv-layout-wellsfargo-2026-10-29.md. The tested statements are the made-up ones of3// ops/bank-csv-fixtures-banks.mjs. Business checking (Credits/Debits) and the 2002 layout are not read.4//5// One "Transaction history" table: Date (M/D, no year), Check Number, Description, Deposits/Additions,6// Withdrawals/Subtractions, Ending daily balance (only on the last row of each day). Amounts carry no sign:7// the column says which way the money went, so each amount goes to the header whose right edge is nearest8// (amounts and these headers are right aligned on the same edge).9// Real text layers split words at odd places ("Transaction hi story"), so labels match with the spaces taken out.10import { cents, MONTHS, rowDate, sum, total } from "./parse.js";11
12const tight = (s) => s.replace(/\s+/g, "");13// A phrase that still matches when the PDF splits its words anywhere.14const loose = (phrase) =>15 new RegExp(16 [...tight(phrase)]17 .map((ch) => ch.replace(/[.*+?^${}()|[\]\\/]/g, "\\$&"))18 .join("\\s*"),19 "i",20 );21const SUMMARY = loose("Statement period activity summary");22const MD = /^(\d{1,2})\/(\d{1,2})$/;23const HEADER_DATE = /([A-Z][a-z]+)(\d{1,2}),(\d{4})■?Page\d+of\d+/;24const pad = (n) => String(n).padStart(2, "0");25
26export const wells = {27 detect: (has) =>28 has(/Wells\s*Far\s*go/i) &&29 has(SUMMARY) &&30 has(loose("Transaction history"))31 ? "wells-checking"32 : null,33 parse,34};35
36function findPeriod(L, begin) {37 for (const l of L) {38 const m = HEADER_DATE.exec(tight(l.text));39 if (!m || !MONTHS.includes(m[1]) || !begin) continue;40 const end = `${m[3]}-${pad(MONTHS.indexOf(m[1]) + 1)}-${pad(m[2])}`;41 let start = `${m[3]}-${pad(begin.mm)}-${pad(begin.dd)}`;42 if (start > end) start = `${Number(m[3]) - 1}${start.slice(4)}`;43 return { start, end };44 }45 return null;46}47
48function parse(L) {49 const summary = {};50 const table = {};51 const rows = [];52 let mode = null;53 let cols = null; // right edges of the Additions, Subtractions and balance headers on this page54 let descX = 0;55 let last = null;56 let page = 0;57 for (const l of L) {58 if (l.page !== page) [page, cols, last] = [l.page, null, null];59 const t = tight(l.text);60 if (/Page\d+of\d+$/.test(t)) continue;61 if (SUMMARY.test(l.text) && !summary.done) {62 mode = "summary";63 continue;64 }65 if (mode === "summary") {66 // Label, then its amount. "- 228.39" can be one item or two; the account box can follow on the line.67 const at = l.cells.findIndex(68 (c) => cents(c.str.replace(/^-\s+/, "-")) !== null,69 );70 if (at < 1) continue;71 let label = tight(72 l.cells73 .slice(0, at)74 .map((c) => c.str)75 .join(""),76 );77 let v = cents(l.cells[at].str.replace(/^-\s+/, "-"));78 if (label.endsWith("-")) [label, v] = [label.slice(0, -1), -Math.abs(v)];79 let m;80 if ((m = /^Beginningbalanceon(\d{1,2})\/(\d{1,2})$/i.exec(label)))81 [summary.beginDay, summary.begin] = [{ mm: m[1], dd: m[2] }, v];82 else if (/^Deposits\/Additions$/i.test(label)) summary.deposits = v;83 else if (/^Withdrawals\/Subtractions$/i.test(label))84 summary.withdrawals = -Math.abs(v);85 else if (/^Endingbalanceon\d{1,2}\/\d{1,2}$/i.test(label)) {86 summary.end = v;87 [summary.done, mode] = [true, null];88 }89 continue;90 }91 if (/^Transactionhistory/i.test(t)) {92 [mode, last] = ["rows", null];93 continue;94 }95 if (mode !== "rows") continue;96 if (/^(Summaryofcheckswritten|Monthlyservicefeesummary)/i.test(t)) {97 mode = null;98 continue;99 }100 const right = (c) => c.x + c.w;101 const find = (word) =>102 l.cells.find((c) => tight(c.str).toLowerCase().endsWith(word));103 if (/^DateNumberDescription/i.test(t)) {104 const heads = ["additions", "subtractions", "balance"].map(find);105 if (heads.every(Boolean)) cols = heads.map(right);106 // pdf.js can join "Number Description" into one item: estimate where "Description" starts in it.107 const d = find("description");108 descX = d109 ? d.x +110 (d.w * d.str.toLowerCase().lastIndexOf("description")) / d.str.length111 : 0;112 continue;113 }114 if (!cols) continue;115 // Each money cell goes to the nearest amount column; one far from all three is part of the text.116 const col = (c) => {117 if (cents(c.str) === null) return -1;118 const d = cols.map((e) => Math.abs(e - right(c)));119 const i = d.indexOf(Math.min(...d));120 return d[i] <= 20 ? i : -1;121 };122 const amounts = [[], [], []];123 const text = [];124 for (const c of l.cells.slice(1)) {125 const i = col(c);126 if (i < 0) text.push(c);127 else amounts[i].push(cents(c.str));128 }129 if (/^Endingbalanceon\d/i.test(t)) {130 table.end = amounts[2][0];131 last = null;132 continue;133 }134 if (/^Totals/i.test(t)) {135 [table.deposits, table.withdrawals] = [amounts[0][0], amounts[1][0]];136 [mode, last] = [null, null];137 continue;138 }139 const lead = MD.exec(l.cells[0].str);140 if (lead) {141 // Exactly one of the two money columns; anything else is not a row we can trust.142 if (143 amounts[0].length + amounts[1].length !== 1 ||144 amounts[2].length > 1145 ) {146 last = null;147 continue;148 }149 // The check number ends 5 pt left of the description, so pdf.js often joins the two into one item.150 let check = "";151 const m =152 text[0]?.x < descX - 1 && /^(\d+)(?:\s+(.*))?$/.exec(text[0].str);153 if (m) {154 check = m[1];155 text.shift();156 if (m[2]) text.unshift({ x: descX, w: 0, str: m[2] });157 }158 const description = text.map((c) => c.str).join(" ");159 last = {160 date: [lead[1], lead[2]],161 description: check162 ? `${description || "Check"} #${check}`163 : description,164 cents: amounts[0].length165 ? Math.abs(amounts[0][0])166 : -Math.abs(amounts[1][0]),167 balance: amounts[2].length ? amounts[2][0] : null,168 section: "",169 y: l.y,170 };171 rows.push(last);172 continue;173 }174 // A wrapped description: no date, no amount, in the description column, right under its row.175 if (176 last &&177 !amounts.flat().length &&178 l.cells[0].x >= descX - 5 &&179 last.y - l.y <= 11180 ) {181 last.description += " " + l.text;182 last.y = l.y;183 } else last = null;184 }185
186 const period = findPeriod(L, summary.beginDay);187 if (!period) return { period: null, rows: [], raw: [] };188 const out = rows.map((r) => ({189 ...r,190 date: rowDate(pad(r.date[0]), pad(r.date[1]), period),191 }));192 const raw = [];193 const ins = sum(out.filter((r) => r.cents > 0));194 const outs = sum(out.filter((r) => r.cents < 0));195 if (summary.deposits !== undefined)196 raw.push(total("Deposits/Additions", summary.deposits, ins));197 if (summary.withdrawals !== undefined)198 raw.push(total("Withdrawals/Subtractions", summary.withdrawals, outs));199 if (table.deposits !== undefined)200 raw.push(total("Totals line, Deposits/Additions", table.deposits, ins));201 if (table.withdrawals !== undefined)202 raw.push(203 total(204 "Totals line, Withdrawals/Subtractions",205 -Math.abs(table.withdrawals),206 outs,207 ),208 );209 if (summary.begin !== undefined) {210 // Every printed ending daily balance against the running balance on that row, added up into one check.211 let run = summary.begin;212 let printed = 0;213 let ran = 0;214 let days = 0;215 for (const r of out) {216 run += r.cents;217 if (r.balance === null) continue;218 [printed, ran, days] = [printed + r.balance, ran + run, days + 1];219 }220 if (days)221 raw.push(222 total(223 `Ending daily balances (${days} ${days === 1 ? "day" : "days"} added up)`,224 printed,225 ran,226 ),227 );228 if (summary.end !== undefined) {229 if (table.end !== undefined)230 raw.push(231 total("Ending balance (summary and table)", summary.end, table.end),232 );233 raw.push(234 total(235 "Ending balance (Beginning balance + all rows)",236 summary.end,237 summary.begin + sum(out),238 true,239 ),240 );241 }242 }243 return { period, rows: out, raw };244}1// Bank statement PDF to rows and CSV, in the browser. No DOM, no network.2// The caller loads the vendored pdf.js (./pdf.min.mjs), sets GlobalWorkerOptions.workerSrc to3// ./pdf.worker.min.mjs, and passes the module in. The layouts it knows are the made-up statements of4// ops/bank-csv-fixtures.mjs, ops/bank-csv-fixtures-banks.mjs, -capone.mjs and -td.mjs, which copy each bank's public5// statement layout. Chase is read here; Bank of America (parse-bofa.js), Wells Fargo (parse-wells.js), Capital One6// cards (parse-capone.js) and TD Bank (parse-td.js) in their own files.7//8// Line shape (extractLines output, parseStatement input), in reading order (page, then top to bottom):9// { page: 1-based page number,10// y: text baseline in PDF points from the page bottom (bigger is higher),11// text: the cell strings joined by one space, left to right,12// cells: [{ x: left edge in PDF points, w: width in PDF points, str: trimmed text item }] sorted by x }13// One cell is one pdf.js text item, so a right-edge amount column stays its own cell.14//15// Every sum and comparison uses integer cents. Row amounts: money in positive, money out negative.16
17import { bofa } from "./parse-bofa.js";18import { wells } from "./parse-wells.js";19import { capone } from "./parse-capone.js";20import { td } from "./parse-td.js";21
22export const SCAN_PROBLEM =23 "This PDF has no text layer, so it is a scan. Scanned statements are not supported.";24export const LAYOUT_PROBLEM =25 "This does not look like a statement layout we have tested (Chase, Bank of America, Wells Fargo, Capital One or TD Bank).";26
27export async function extractLines(pdfjsLib, data) {28 const task = pdfjsLib.getDocument({29 data,30 isEvalSupported: false,31 disableFontFace: true,32 });33 const doc = await task.promise;34 const lines = [];35 let chars = 0;36 try {37 for (let p = 1; p <= doc.numPages; p++) {38 const page = await doc.getPage(p);39 const { items } = await page.getTextContent();40 const rows = [];41 for (const it of items) {42 const str = (it.str || "").trim();43 if (!str) continue;44 chars += str.replace(/\s/g, "").length;45 const x = Math.round(it.transform[4] * 10) / 10;46 const y = Math.round(it.transform[5] * 10) / 10;47 const w = Math.round((it.width || 0) * 10) / 10;48 // ponytail: linear scan per item, fine for a few hundred items per page49 let row = rows.find((r) => Math.abs(r.y - y) <= 2);50 if (!row) rows.push((row = { page: p, y, text: "", cells: [] }));51 row.cells.push({ x, w, str });52 }53 rows.sort((a, b) => b.y - a.y);54 for (const r of rows) {55 r.cells.sort((a, b) => a.x - b.x);56 r.text = r.cells.map((c) => c.str).join(" ");57 }58 lines.push(...rows);59 page.cleanup();60 }61 return { pages: doc.numPages, lines, chars };62 } finally {63 await task.destroy();64 }65}66
67export const DATE = /^(\d{2})\/(\d{2})$/;68const MONEY = /^([-+])?\$?(-)?(\d{1,3}(?:,\d{3})*|\d+)\.(\d{2})$/;69export const MONTHS =70 "January February March April May June July August September October November December".split(71 " ",72 );73const NOISE = /(^|\s)Page \d+ of \d+$|^(Primary )?Account Number:/i;74
75export function cents(s) {76 const m = MONEY.exec(s);77 if (!m) return null;78 const v = Number(m[3].replace(/,/g, "")) * 100 + Number(m[4]);79 return m[1] === "-" || m[2] ? -v : v;80}81
82export function money(c) {83 const a = Math.abs(c);84 const whole = String(Math.floor(a / 100)).replace(/\B(?=(\d{3})+$)/g, ",");85 return (c < 0 ? "-" : "") + whole + "." + String(a % 100).padStart(2, "0");86}87
88// Label and amount pairs on one line, left to right. Works on two-column lines too89// ("Previous Balance $1.00 ... New Balance $2.00"). Bare integers (the INSTANCES column) are skipped.90function pairs(cells) {91 const out = [];92 let label = [];93 for (const c of cells) {94 const v = cents(c.str);95 if (v !== null) {96 if (label.length) out.push({ label: label.join(" "), cents: v });97 label = [];98 } else if (!/^\d+$/.test(c.str)) label.push(c.str);99 }100 return out;101}102
103// Headings, matched case sensitive (Chase prints them in capitals; summary labels are Title Case).104// kind: summary | rows | stop | activity (keep the open section, for "ACCOUNT ACTIVITY (CONTINUED)").105// final: nothing after it is a row. sign maps the printed amount to money in (+) or money out (-).106// sum: the matching label in the summary box.107const CHECKING_HEADS = {108 "CHECKING SUMMARY": { kind: "summary", prefix: true },109 "TRANSACTION DETAIL": { kind: "rows", prefix: true, section: "", sign: 1 },110 "DEPOSITS AND ADDITIONS": {111 kind: "rows",112 section: "Deposits and Additions",113 sign: 1,114 },115 "CHECKS PAID": { kind: "rows", section: "Checks Paid", sign: -1 },116 "ATM & DEBIT CARD WITHDRAWALS": {117 kind: "rows",118 section: "ATM & Debit Card Withdrawals",119 sign: -1,120 },121 "ELECTRONIC WITHDRAWALS": {122 kind: "rows",123 section: "Electronic Withdrawals",124 sign: -1,125 },126 "OTHER WITHDRAWALS": { kind: "rows", section: "Other Withdrawals", sign: -1 },127 FEES: { kind: "rows", section: "Fees", sign: -1 },128 "DAILY ENDING BALANCE": { kind: "stop" },129 "SERVICE CHARGE SUMMARY": { kind: "stop" },130 "IN CASE OF ERRORS": { kind: "stop", prefix: true },131};132// Card statements print purchases, fees and interest positive (they add to what you owe) and payments133// and credits negative. Money out is a purchase, so every card row gets amount = -printed.134const CARD_HEADS = {135 "ACCOUNT ACTIVITY": { kind: "activity" },136 "PAYMENTS AND OTHER CREDITS": {137 kind: "rows",138 section: "Payments and Other Credits",139 sum: /^Payments?, Credits$/,140 sign: -1,141 },142 PURCHASE: { kind: "rows", section: "Purchase", sum: /^Purchases$/, sign: -1 },143 PURCHASES: {144 kind: "rows",145 section: "Purchase",146 sum: /^Purchases$/,147 sign: -1,148 },149 "CASH ADVANCE": {150 kind: "rows",151 section: "Cash Advances",152 sum: /^Cash Advances$/,153 sign: -1,154 },155 "CASH ADVANCES": {156 kind: "rows",157 section: "Cash Advances",158 sum: /^Cash Advances$/,159 sign: -1,160 },161 "BALANCE TRANSFER": {162 kind: "rows",163 section: "Balance Transfers",164 sum: /^Balance Transfers$/,165 sign: -1,166 },167 "BALANCE TRANSFERS": {168 kind: "rows",169 section: "Balance Transfers",170 sum: /^Balance Transfers$/,171 sign: -1,172 },173 "FEES CHARGED": {174 kind: "rows",175 section: "Fees Charged",176 sum: /^Fees Charged$/,177 sign: -1,178 },179 "INTEREST CHARGED": {180 kind: "rows",181 section: "Interest Charged",182 sum: /^Interest Charged$/,183 sign: -1,184 },185 "INTEREST CHARGES": { kind: "stop", final: true },186};187
188function heading(text, heads) {189 const t = text.replace(/\s*\(continued\)$/i, "");190 for (const [k, h] of Object.entries(heads))191 if (t === k || (h.prefix && t.startsWith(k))) return h;192 if (heads === CARD_HEADS && /Totals Year-to-Date/i.test(t))193 return { kind: "stop", final: true };194 return null;195}196
197function findPeriod(L) {198 const iso = (y, m, d) =>199 `${y}-${String(m).padStart(2, "0")}-${String(d).padStart(2, "0")}`;200 for (const { text } of L) {201 let m =202 /([A-Z][a-z]+) (\d{1,2}), (\d{4}) through ([A-Z][a-z]+) (\d{1,2}), (\d{4})/.exec(203 text,204 );205 if (m && MONTHS.includes(m[1]) && MONTHS.includes(m[4]))206 return {207 start: iso(m[3], MONTHS.indexOf(m[1]) + 1, m[2]),208 end: iso(m[6], MONTHS.indexOf(m[4]) + 1, m[5]),209 };210 m =211 /Opening\/Closing Date:?\s*(\d{2})\/(\d{2})\/(\d{2})\s*-\s*(\d{2})\/(\d{2})\/(\d{2})/.exec(212 text,213 );214 if (m)215 return {216 start: iso("20" + m[3], m[1], m[2]),217 end: iso("20" + m[6], m[4], m[5]),218 };219 }220 return null;221}222
223// MM/DD gets the year (the start's, one before or one after) that puts it closest to the period. A card224// statement can list a purchase from before its period (12/31 on a Jan 4 to Feb 3 statement is last year).225export function rowDate(mm, dd, period) {226 const sy = Number(period.start.slice(0, 4));227 const [s, e] = [period.start, period.end].map(Date.parse);228 const away = (y) => {229 const t = Date.UTC(y, mm - 1, dd);230 return Math.max(s - t, t - e, 0);231 };232 const y = [sy - 1, sy + 1].reduce((a, b) => (away(b) < away(a) ? b : a), sy);233 return `${y}-${mm}-${dd}`;234}235
236// Walk the lines once: summary pairs, rows, section totals, and the table's own balance lines.237function walk(L, heads, card) {238 const rows = [];239 const summary = new Map();240 const sectionTotals = [];241 const table = {};242 let mode = card ? "summary" : null;243 let head = null;244 let last = null;245 let amtX = 0;246 let page = 0;247 for (const l of L) {248 if (l.page !== page) [page, amtX, last] = [l.page, 0, null];249 if (NOISE.test(l.text)) continue;250 const { cells } = l;251 const lead = DATE.exec(cells[0].str);252 const h =253 !lead && cells.every((c) => cents(c.str) === null)254 ? heading(l.text, heads)255 : null;256 if (h) {257 if (h.final) break;258 [mode, last] = [259 h.kind === "activity" ? (head ? "rows" : null) : h.kind,260 null,261 ];262 if (h.kind === "rows") head = h;263 continue;264 }265 const amtHead = cells.find((c) => /^(\$ )?amount$/i.test(c.str));266 if (amtHead && cells.every((c) => cents(c.str) === null)) {267 amtX = amtHead.x - 80; // amounts are right aligned under the header; descriptions sit far left268 continue;269 }270 if (mode === "summary") {271 for (const p of pairs(cells))272 if (!summary.has(p.label)) summary.set(p.label, p.cents);273 if (!card && summary.has("Ending Balance")) mode = null;274 continue;275 }276 if (mode !== "rows") continue;277 const tail = [];278 for (let i = cells.length - 1; i > 0 && tail.length < 2; i--) {279 const v = cents(cells[i].str);280 if (v === null || cells[i].x < amtX) break;281 tail.unshift({ i, v });282 }283 const check =284 !lead &&285 head.section === "Checks Paid" &&286 /^\d+$/.test(cells[0].str) &&287 tail.length;288 if ((lead || check) && tail.length) {289 let date = lead;290 let desc = cells.slice(1, tail[0].i);291 if (check) {292 const at = desc.findIndex((c) => DATE.test(c.str));293 if (at < 0) continue;294 date = DATE.exec(desc[at].str);295 desc = [296 { str: "Check #" + cells[0].str },297 ...desc.filter((c, i) => i !== at && c.str !== "^"),298 ];299 } else if (card && desc[0] && DATE.test(desc[0].str))300 desc = desc.slice(1); // a post date column301 const twoCol = !card && head.section === "" && tail.length === 2;302 const printed = twoCol ? tail[0].v : tail[tail.length - 1].v;303 last = {304 mm: date[1],305 dd: date[2],306 description: desc.map((c) => c.str).join(" "),307 printed,308 cents: card ? -printed : head.sign * printed,309 balance: twoCol ? tail[1].v : null,310 section: head.section,311 y: l.y,312 };313 rows.push(last);314 continue;315 }316 const v = cents(cells[cells.length - 1].str);317 if (v === null) {318 // ponytail: a continuation line sits within 11 pt under its row (real Chase: 9 to 10.4 pt);319 // measure more statements if descriptions get cut short320 if (last && last.y - l.y <= 11) {321 last.description += " " + l.text;322 last.y = l.y;323 }324 continue;325 }326 last = null;327 const label = pairs(cells)[0]?.label || "";328 if (/^Beginning Balance$/i.test(label)) table.begin = v;329 else if (/^Ending Balance$/i.test(label)) [table.end, mode] = [v, null];330 else if (/^Total /i.test(label))331 sectionTotals.push({ label, cents: v, head });332 }333 return { rows, summary, sectionTotals, table };334}335
336// every: this check covers every row (a balance check). A CSV is sold only when one matches.337export const total = (label, statement, parsed, every = false) => ({338 label,339 statement,340 parsed,341 every,342});343export const sum = (rows, f = (r) => r.cents) =>344 rows.reduce((n, r) => n + f(r), 0);345
346function checkingTotals(w, layout) {347 const out = [];348 const { rows, summary, table } = w;349 const begin = summary.get("Beginning Balance") ?? table.begin;350 const end = summary.get("Ending Balance") ?? table.end;351 const cats = [...summary].filter(352 ([k]) => !/^(Beginning|Ending) Balance$/.test(k),353 );354 if (layout === "chase-checking") {355 const dep = summary.get("Deposits and Additions");356 if (dep !== undefined)357 out.push(358 total(359 "Deposits and Additions",360 dep,361 sum(rows.filter((r) => r.cents > 0)),362 ),363 );364 const wd = cats.filter(([k]) => k !== "Deposits and Additions");365 if (wd.length)366 out.push(367 total(368 wd.map(([k]) => k).join(" + "),369 wd.reduce((n, [, v]) => n - Math.abs(v), 0),370 sum(rows.filter((r) => r.cents < 0)),371 ),372 );373 } else {374 for (const [k, v] of cats) {375 const sec = rows.filter(376 (r) => r.section.toLowerCase() === k.toLowerCase(),377 );378 out.push(total(k, v, sum(sec)));379 }380 for (const t of w.sectionTotals) {381 const sec = rows.filter((r) => r.section === t.head?.section);382 out.push(383 total(384 t.label,385 t.cents,386 sum(sec, (r) => r.printed),387 ),388 );389 }390 }391 if (table.begin !== undefined && summary.has("Beginning Balance"))392 out.push(393 total(394 "Beginning Balance (summary and table)",395 summary.get("Beginning Balance"),396 table.begin,397 ),398 );399 if (begin !== undefined && end !== undefined)400 out.push(401 total(402 "Ending Balance (Beginning Balance + all rows)",403 end,404 begin + sum(rows),405 true,406 ),407 );408 return out;409}410
411function cardTotals(w) {412 const out = [];413 const { rows, summary } = w;414 const heads = [...new Set(Object.values(CARD_HEADS).filter((h) => h.sum))];415 const seen = new Set();416 for (const [k, v] of summary) {417 const h = heads.find((x) => x.sum.test(k));418 if (!h || seen.has(h.section)) continue;419 seen.add(h.section);420 const sec = rows.filter((r) => r.section === h.section);421 out.push(422 total(423 k,424 v,425 sum(sec, (r) => r.printed),426 ),427 );428 }429 for (const t of w.sectionTotals) {430 const sec = rows.filter((r) => r.section === t.head?.section);431 out.push(432 total(433 t.label,434 t.cents,435 sum(sec, (r) => r.printed),436 ),437 );438 }439 const prev = summary.get("Previous Balance");440 const next = summary.get("New Balance");441 if (prev !== undefined && next !== undefined)442 out.push(443 total(444 "New Balance (Previous Balance + all activity)",445 next,446 prev + sum(rows, (r) => r.printed),447 true,448 ),449 );450 return out;451}452
453const result = (layout, problem, period = null, rows = [], totals = []) => ({454 layout,455 period,456 rows,457 totals,458 ok: !problem,459 problem,460});461
462// rows: [{ date: "YYYY-MM-DD", description, cents, balance: cents or null, section }]. raw: total() checks.463function finish(layout, period, rows, raw) {464 if (!period)465 return result(466 layout,467 "We could not find the statement period, so we cannot tell the year of each date.",468 );469 const out = rows.map((r) => ({470 date: r.date,471 description: r.description,472 amount: r.cents / 100,473 balance: r.balance === null ? null : r.balance / 100,474 section: r.section,475 }));476 const totals = raw.map((t) => ({477 label: t.label,478 statement: t.statement / 100,479 parsed: t.parsed / 100,480 match: t.statement === t.parsed,481 }));482 let problem = null;483 if (!rows.length) problem = "We found no transactions in this statement.";484 else if (!totals.length)485 problem =486 "We could not find the statement's own totals, so we cannot check the rows.";487 else {488 const bad = raw.filter((t) => t.statement !== t.parsed);489 if (bad.length)490 problem =491 "These totals do not match the statement: " +492 bad493 .map(494 (t) =>495 `${t.label} (statement ${money(t.statement)}, rows ${money(t.parsed)})`,496 )497 .join("; ") +498 ". Please check the rows before you use them.";499 // Section totals can match while rows outside them go unchecked: only the balance check sees every row.500 else if (!raw.some((t) => t.every))501 problem =502 "We could not find the statement's ending balance, so we cannot check every row.";503 }504 return result(layout, problem, period, out, totals);505}506
507export function parseStatement(lines) {508 const L = [];509 for (const l of lines || []) {510 const cells = (l.cells || [])511 .map((c) => ({ x: c.x, w: c.w || 0, str: String(c.str).trim() }))512 .filter((c) => c.str && !/^\*(start|end)\*/.test(c.str)); // hidden section markers in real Chase PDFs513 if (cells.length)514 L.push({515 page: l.page,516 y: l.y,517 cells,518 text: cells.map((c) => c.str).join(" "),519 });520 }521 if (L.reduce((n, l) => n + l.text.replace(/\s/g, "").length, 0) < 20)522 return result(null, SCAN_PROBLEM);523 const has = (re) => L.some((l) => re.test(l.text));524 for (const bank of [bofa, wells, capone, td]) {525 const layout = bank.detect(has);526 if (!layout) continue;527 const { period, rows, raw } = bank.parse(L, layout);528 return finish(layout, period, rows, raw);529 }530 let layout = null;531 if (has(/\bchase\b/i) && has(/^CHECKING SUMMARY/))532 layout = has(/^TRANSACTION DETAIL/)533 ? "chase-checking"534 : "chase-checking-sections";535 else if (536 has(/\bchase\b/i) &&537 has(/^ACCOUNT ACTIVITY/) &&538 has(/Opening\/Closing Date/)539 )540 layout = "chase-card";541 if (!layout) return result(null, LAYOUT_PROBLEM);542 const period = findPeriod(L);543 if (!period) return finish(layout, null);544 const card = layout === "chase-card";545 const w = walk(L, card ? CARD_HEADS : CHECKING_HEADS, card);546 const rows = w.rows.map((r) => ({547 ...r,548 date: rowDate(r.mm, r.dd, period),549 }));550 return finish(551 layout,552 period,553 rows,554 card ? cardTotals(w) : checkingTotals(w, layout),555 );556}557
558const cell = (s) => {559 let v = String(s ?? "");560 if (/^[=+\-@\t\r]/.test(v)) v = "'" + v; // CSV injection guard for spreadsheet apps561 return /[",\r\n]/.test(v) ? `"${v.replace(/"/g, '""')}"` : v;562};563const num = (n) =>564 n === null || n === undefined565 ? ""566 : money(Math.round(n * 100)).replace(/,/g, "");567
568export function toCsv(rows) {569 const out = ["Date,Description,Amount,Balance,Section"];570 for (const r of rows)571 out.push(572 [573 r.date,574 cell(r.description),575 num(r.amount),576 num(r.balance),577 cell(r.section),578 ].join(","),579 );580 return out.join("\r\n") + "\r\n";581}