Document Intelligence Releases

1.0 TBD

Feature

Initial release — IDocumentScanner over each platform’s native document camera: VisionKit on iOS and Mac Catalyst, ML Kit on Android (with optional PDF output), and Vision document segmentation plus perspective correction on macOS AppKit. No MAUI dependency; platform context is obtained natively.

Feature

IDocumentExtractor — on-device extraction of receipts, invoices, driver’s licenses (AAMVA PDF417), passports (ICAO 9303 MRZ with check-digit validation) and payment cards (Luhn-validated PAN, network, expiry, cardholder).

Feature

OCR geometry preserved end to end — recognized fragments carry normalized top-left bounds, are ordered into reading order and regrouped into visual rows before parsing, so keyword-anchored parsers see TOTAL and its amount on one line. Recognizers that report no geometry pass through unchanged.

Feature

IDataDetector — dates, addresses, phone numbers and links from recognized text via Apple’s NSDataDetector, applied as strictly additive enrichment so platforms without a detector produce a subset of the same result.

Feature

Replaceable ITextRecognizer and IBarcodeReader seams, with TextRecognitionOptions.Document / .Alphanumeric profiles for prose versus card numbers and MRZ lines.

Feature

Payment card results never expose a CVV and mask the PAN in ToString() — both enforced by tests.