Fixes the indexer crash on chipnet: bitcoincash 0.32.1's
Opcode::classify panicked on BCH re-enabled opcodes (e.g. OP_SPLIT),
which riftenlabs-defi hit while parsing input scriptSigs in
parse_cauldrons_from_tx. bitcoincash 0.32.2 makes classify total and
panic-free (and adds the May 2026 upgrade opcodes); riftenlabs-defi
0.4.1 additionally hardens read_push_from_script to not classify
non-push opcodes at all.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Migrate off the deprecated bitcoincash 0.29 API (Amount/Version types,
FromHex -> FromStr/parse, non-exhaustive Network) across the indexer and
tests.
Add indexing of native token-A <-> token-B (TokenToken) AMM pools:
- tokentoken_pool / tokentoken_pool_history_entry tables in cauldron.db,
created via an always-run idempotent migration (no DB_VERSION bump, so
existing databases upgrade in place)
- block-path indexing sharing the cauldron write transaction and reorg
undo, with creation/swap/withdrawal state tracking and reserve deltas
- mempool indexing: electrum mempool.get filters on the tokentoken
contract code (spends) and the CONJURE op_return hint (creations);
first_seen_timestamp reconciles with mtp on confirmation
- RPC endpoints /tokentoken/pool/active (pair lookup, order-insensitive)
and /tokentoken/tokens
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
This makes the application fully async; freeing web server threads to
handle new connections when waiting on SQL queries.
Additionally sqlx will allow easier move to a different database if
needed in the future.
Includes some SQL optimizations as well (slow queries more easliy identifiable
with sqlx).
