mirror of
https://github.com/storytold/pdfcraft.git
synced 2026-11-11 02:41:31 -06:00
f9c150c17c
* Add IBAN matching to Search and Redact patterns
The Pattern/PATTERNS/from_id interface is consumed generically by redact_mark and the UI pattern combobox through the engine re-exports, so one new variant and matcher exposes IBAN in Search and Redact everywhere. A unit test shows valid compact and spaced IBANs are marked while wrong check digits, short runs, unknown countries, letter check digits and token-glued IBANs are not.
Add an Iban variant to Pattern, list it in PATTERNS, handle it in label, id and find, and add an iban matcher with a country-code length table, compact or group-of-four-spaced acceptance, alphanumeric start and end boundaries and ISO 7064 mod-97 check-digit validation.
Test: iban_matches_compact_spaced_and_rejects_near_matches failed before the change and passes after.
* Address review for pull request 4332
* Address review for pull request 4332
* review: address CodeRabbit findings
- patterns.rs: fix SWIFT registry IBAN lengths EG 27->29, LY 27->25, SD 24->18 and add NI 28, BI 27, DJ 27, OM 23 (crates/redact/src/patterns.rs:545,549,551,558-561)
- patterns.rs: state table source as SWIFT IBAN registry without claiming full ISO 13616 coverage (crates/redact/src/patterns.rs:475)
- patterns.rs: regression test with valid registry examples for EG, LY, SD, NI, BI, DJ, OM and a wrong-length EG rejection (crates/redact/src/patterns.rs:892)
* Complete the IBAN country table and accept spaces only between groups
The country table now lists all 89 countries of the SWIFT IBAN registry
(release 99) as one sorted constant. Honduras, Russia, Yemen, Falkland
Islands, Mongolia and Somalia were missing.
A space is accepted only after a full group of four characters, so a code
followed by ordinary words ("BA37 leaves at noon") ends the run instead of
being tried against the check digits.
Tests: the registry example of every country matches compact and printed,
changes to one digit are rejected, and the table and the examples agree.
* Advertise the IBAN pattern in the redact_mark tool schema
The pattern enum in the advertised schema was kept by hand and still
listed phone, email, credit-card, ssn and date, so clients that obey the
schema could not select iban, which the runtime already accepted. The
enum is now built from the engine's pattern registry, so it cannot drift
from it again, and the description names iban.
Tests: tools/list advertises exactly the registry's pattern ids and
names each in the description, and an iban mark made through the tool
covers the spaced German sample and nothing else.
---------
Co-authored-by: eliets2 <287273517+eliets2@users.noreply.github.com>
Co-authored-by: FredApps <153173854+FredApps@users.noreply.github.com>