Claude c5b1d0051a refactor(cashu): move out-of-band token parser + model to quartz ...

Follow-up to the V4Encoder move. Quartz could encode a cashuB string but
could not parse cashuA/cashuB back, and the parsing lived in amethyst even
though it is pure NUT-00 wire-format protocol. Worse, commons.RichTextParser
already detects cashuA/cashuB words while the parser sat up in the app, so
Desktop (its own rich-text viewer) could not parse a received token at all.

Consolidate the legacy out-of-band redeem stack onto quartz:

- new quartz CashuTokenB64Parser parses cashuA (standard-Base64 JSON, rewritten
  off Jackson onto kotlinx.serialization to satisfy quartz's no-Jackson rule)
  and cashuB (Base64URL CBOR, reusing the V4Token models), returning quartz
  types. It is the inverse of V4Encoder.
- move the CashuToken container model from commons to quartz, switching its
  proofs from the duplicate commons Proof (field C, amount Int) onto the
  canonical quartz CashuProof (field c, amount Long). The duplicate Proof
  type is deleted.
- delete amethyst V3Parser/V3Token/V4Parser; CashuParser/CachedCashuParser
  stay as thin amethyst adapters (off-main-thread guard + GenericLoadable +
  LruCache) over the quartz parser.
- this removes the manual Proof -> CashuProof conversion shims that
  MeltProcessor and CashuWalletViewModel previously carried.

Tests: full cashuA + cashuB vector coverage moves to quartz commonTest
(CashuTokenB64ParserTest, runs on JVM) plus an encode/parse round-trip; the
superseded amethyst CashuV4ParserTest is removed and CashuBTest stays as the
adapter integration test.

Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_013UMNKix4qEfiAPP9s2a4gB

2026-06-19 15:32:48 +00:00

..

composeResources

fix: remove literal backslash in DM relay-incomplete label

2026-06-18 22:25:34 +02:00

kotlin/com/vitorpamplona/amethyst/commons

refactor(cashu): move out-of-band token parser + model to quartz

2026-06-19 15:32:48 +00:00