Format coverage
Tested across 41 formats in 7 countries
We don't claim “works on any document.” We list the exact formats we have fixtures for, with measured accuracy. New format requests ship in 48 hours.
Total formats
41
Banks
5
Payroll systems
16
Invoice + receipt
20
Banks
Bank statement formats
Bank statements are the highest-volume document we parse. Each bank below has a hand-written detail page with format quirks, sample JSON, and a working code snippet.
| Country | Format | Doc type | Fixtures | Accuracy | Last update | Notes |
|---|---|---|---|---|---|---|
| BE | KBC | bank_statement | 42 | 97% | 2026-04-22 | Belgium's largest retail bank — IBAN-formatted statements, Dutch + French + English layouts. |
| BE | Belfius | bank_statement | 38 | 96% | 2026-03-19 | Belgian retail and corporate bank — multi-language statements with split debit/credit columns. |
| DE | Sparkasse | bank_statement | 45 | 96% | 2026-03-04 | Germany's largest savings-bank network — DTAUS-style format with verwendungszweck details. |
| FR | BNP Paribas | bank_statement | 36 | 95% | 2026-04-19 | France's largest bank — Relevé de Compte format with operation-code prefixes. |
| NL | ING | bank_statement | 40 | 97% | 2026-03-26 | Dutch retail and SME bank — clean transaction-table format, English + Dutch variants. |
Payslip systems
Payroll-system formats
Tested against payslips from BE / DE / FR / NL / UK / US payroll providers. Country-specific deduction codes (RSZ, Lohnsteuer, CSG/CRDS, PAYE, FICA) extracted as structured fields.
| Country | Format | Doc type | Fixtures | Accuracy | Last update | Notes |
|---|---|---|---|---|---|---|
| BE | SD Worx | payslip | 38 | 96% | 2026-03-12 | Belgian-specific deduction codes (RSZ werknemer 13.07%, bedrijfsvoorheffing). Tri-language layout (NL / FR / EN); language auto-detected from header. |
| BE | Acerta | payslip | 32 | 95% | 2026-04-18 | Two-page layout: current period on page 1, YTD totals on page 2. Includes maaltijdcheques and ecocheques as separate line items. |
| BE | Securex | payslip | 28 | 94% | 2026-04-21 | Compact one-page layout with three-column deduction breakdown. NL and FR variants share template. |
| BE | Partena | payslip | 26 | 94% | 2026-04-23 | Partena Professional payslip — sectoral CCT references printed inline with bruto. Holiday pay (vakantiegeld) listed as separate line. |
| DE | DATEV | payslip | 44 | 97% | 2026-04-30 | German tax categorization (Lohnsteuer, Kirchensteuer, Solidaritätszuschlag). PDF metadata included with Steuerklasse and Sozialversicherungsnummer. |
| DE | SAP HR | payslip | 35 | 95% | 2026-04-20 | SAP HCM Entgeltabrechnung — heavy template variability across enterprise customers. Wage-type codes (Lohnarten) preserved verbatim alongside descriptions. |
| FR | Sage Paie | payslip | 33 | 95% | 2026-04-02 | French statutory deductions (CSG / CRDS / cotisations sociales). Two pages; YTD on page 2. Net imposable line called out separately from net à payer. |
| FR | ADP France | payslip | 30 | 94% | 2026-04-25 | ADP GlobalView fiche de paie — French statutory format with simplified bulletin de paie. URSSAF + AGIRC-ARRCO line items consolidated. |
| NL | Loket | payslip | 31 | 96% | 2026-04-22 | Dutch payroll provider — clean one-page layout. Loonheffing, ZVW, and pensioen split across separate columns. Cumulatief block at the foot of the page. |
| NL | Loonbedrijf | payslip | 27 | 93% | 2026-04-29 | Smaller-bureau Dutch payroll output — variable header positioning between bureaus. We auto-detect header anchor from the IBAN block. |
| GB | Sage Payroll | payslip | 34 | 96% | 2026-04-06 | UK Sage 50 / Sage Business Cloud Payroll — PAYE, NI, pension, and student-loan deductions itemized. Tax code printed in header. |
| GB | Xero Payroll | payslip | 36 | 96% | 2026-04-18 | Xero UK payslip — cloud-rendered HTML→PDF. Logo top-left always; YTD column on the right. Pension auto-enrolment line surfaces opt-in/opt-out status. |
| US | ADP | payslip | 41 | 96% | 2026-03-20 | ADP RUN / Workforce Now pay statement. Federal, state, and local taxes broken out. YTD column always present; check-stub voucher at the foot. |
| US | Paychex | payslip | 35 | 95% | 2026-04-21 | Paychex Flex pay statement — separate Earnings, Taxes, Deductions, and Net Pay sections. Pre-tax vs post-tax deductions clearly grouped. |
| US | Gusto | payslip | 39 | 97% | 2026-04-11 | Gusto pay stub — clean single-page HTML→PDF. Federal/FICA/Medicare/state tax rows always present; 401(k) and HSA contributions grouped under pre-tax. |
| US | Justworks | payslip | 28 | 94% | 2026-04-26 | Justworks PEO pay statement — co-employer model means employer-of-record is Justworks; client company surfaces as a separate header line. |
Invoice systems
Accounting-software invoice formats
Cloud accounting platforms render PDFs from HTML, which keeps layouts mostly stable. ERP-rendered invoices (NetSuite, SAP) carry the most variability — anchored on the line-item table.
| Country | Format | Doc type | Fixtures | Accuracy | Last update | Notes |
|---|---|---|---|---|---|---|
| GLOBAL | Xero | invoice | 43 | 97% | 2026-03-08 | Cloud-only — PDF rendering uses their HTML→PDF pipeline. Logo top-left always. Tax-rate column shows percentage; supports inclusive/exclusive tax modes per line. |
| GLOBAL | QuickBooks | invoice | 41 | 96% | 2026-04-09 | QuickBooks Online + Desktop variants share core layout but differ in tax sub-totals. Custom-template variability is the main quirk; we anchor on the line-item table header. |
| GLOBAL | FreshBooks | invoice | 32 | 95% | 2026-04-19 | FreshBooks classic + new — single-page invoices with bold totals card. Time-tracking line items include hours and hourly rate as separate columns. |
| GLOBAL | Sage | invoice | 36 | 95% | 2026-04-20 | Sage Business Cloud Accounting + Sage 50 — high template variability across regions. We auto-detect VAT/GST/sales-tax label and route to the right tax schema. |
| GLOBAL | NetSuite | invoice | 30 | 93% | 2026-04-25 | NetSuite SuiteScript-rendered invoices — heavy customer customization. Subsidiary block at the top identifies the legal entity; multi-currency conversions inline. |
| GLOBAL | SAP | invoice | 27 | 92% | 2026-04-28 | SAP S/4HANA + ECC SmartForms output — extreme template variability. We extract via line-item-table anchor; header fields are best-effort across SAP variants. |
| DE | DATEV | invoice | 38 | 96% | 2026-03-12 | German VAT categorization. UStG §14 mandatory fields enforced (Steuernummer or USt-IdNr, Rechnungsdatum, fortlaufende Nummer). Reverse-charge flagged when present. |
| BE | Yuki | invoice | 29 | 95% | 2026-04-22 | Belgian/Dutch SME accounting platform — clean PDF output. Structured communication block (+++123/4567/89001+++) always present and extracted as a first-class field. |
| NL | Exact | invoice | 33 | 95% | 2026-04-19 | Exact Online — Dutch + EN variants. BTW (VAT) breakdown by rate (0% / 9% / 21%) always itemized. KvK + BTW number printed in footer. |
| GLOBAL | Wave | invoice | 26 | 94% | 2026-04-30 | Wave Accounting — free-tier invoicing tool, common with US/CA freelancers. Single-template HTML→PDF; layout is highly stable but limited tax-mode flexibility. |
Receipt categories
Receipt formats by merchant type
Per-merchant receipt fixtures would be infeasible. We group by merchant category and validate against a representative sample within each — variability is highest for retail and grocery.
| Country | Format | Doc type | Fixtures | Accuracy | Last update | Notes |
|---|---|---|---|---|---|---|
| GLOBAL | Restaurants | receipt | 47 | 94% | 2026-04-04 | Variable layouts — table-service prints itemized; fast-food prints summary. Tip line where applicable; service-charge vs tip distinguished by label. |
| GLOBAL | Gas stations | receipt | 38 | 95% | 2026-04-16 | Thermal-paper receipts — pump number, fuel grade, price-per-litre/gallon, and total. Multi-product receipts (fuel + shop) handled with line-item separation. |
| GLOBAL | Retail (general) | receipt | 44 | 93% | 2026-04-17 | Long-tail merchant variability — POS systems range from Square / Shopify to bespoke registers. We anchor on the line-item table and infer tax schema from currency + locale. |
| GLOBAL | Grocery | receipt | 41 | 93% | 2026-04-12 | Long item lists (50+ lines common). Loyalty-card discounts itemized as negative lines; bottle-deposit (statiegeld / Pfand) recognized as a separate line type. |
| GLOBAL | Parking | receipt | 28 | 95% | 2026-04-22 | Compact thermal receipts — entry/exit timestamps, duration, rate, total. App-based parking (Parkmobile, EasyPark) emails handled separately as PDF/HTML. |
| GLOBAL | Hotels | receipt | 33 | 94% | 2026-04-19 | Folio-style invoices with per-night breakdown, city tax / tourist tax as separate line, and incidentals (minibar, restaurant) grouped at the foot. VAT-inclusive in EU. |
| GLOBAL | Rideshare (Uber / Lyft) | receipt | 36 | 96% | 2026-03-30 | HTML email receipts converted to PDF. Pickup + dropoff addresses, distance, duration, base fare, tip. Surge multiplier surfaced when present. Uber Eats receipts share the template but flag delivery instead of trip. |
| GLOBAL | Travel / airline | receipt | 31 | 93% | 2026-04-21 | E-ticket itineraries vary wildly by airline + GDS (Sabre / Amadeus / Travelport). PNR, fare basis, and tax breakdown (YQ, YR, etc.) extracted as a structured taxes array. |
| GLOBAL | Subscription services | receipt | 35 | 96% | 2026-04-08 | SaaS / streaming invoices — Stripe-rendered or vendor-rendered HTML→PDF. Billing period, plan name, prorations, and tax (VAT MOSS / US sales tax / GST) extracted as structured fields. |
| GLOBAL | Healthcare / pharmacy | receipt | 29 | 92% | 2026-04-25 | Pharmacy and clinic receipts — prescription codes (RX numbers), insurance-paid vs patient-paid split, and reimbursable-amount fields where the locale provides them (e.g. NL ZVW, BE mutualiteit). |
Don't see your format? 48 hours.
Email us a sample (anonymized is fine). If it's a bank, payroll, invoice, or receipt format we don't cover, we ship test fixtures and parser support within 48 hours — typically same-day for European banks. No procurement cycle, no “raise a ticket with our solutions team.” Just an email and a sample PDF.
Request a format →Found inaccuracy? Tell us.
Coverage numbers above are measured, not aspirational, but measurement is hard. If a number on this page doesn't match what you're seeing in production, we want to know — and we'll publish the correction.
support@tryparsr.dev →