archetech / archon / 34228550267

87%

push

github

feat(python-cli): Open the same SQLite wallet the TypeScript CLI does (#1076)

* feat(python-cli): Open the same SQLite wallet the TypeScript CLI does

The TypeScript CLI reads ARCHON_WALLET_TYPE and opens the matching
store. The Python CLI never read it and always built a JsonWalletStore,
so on a machine with sqlite configured the two disagreed about which
file holds the identity: `keymaster` opened the database, the Python CLI
found no JSON wallet where it looked and offered to create one.

SqliteWalletStore issues the SQL WalletSQLite issues -- the same table,
the same delete-then-insert, the same LIMIT 1 reads -- and writes
JSON.stringify's separators, so a wallet either CLI writes is one the
other opens. Verified both directions against a local node.

The location logic that #1072 kept JSON-only, because backend-aware
resolution without a backend finds a file it then opens as the wrong
format, is now backend-aware: the .db default, the data/ candidates a
wallet may still be sitting in, and the refusal to provision over one
stranded there.

Neither CI suite can run the other language, so a guard compares the SQL
literals in both sources instead.

Closes #1075

Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01GYGMfX3foBcVqP7AQNXjWj

* fix(cli): Refuse an ARCHON_WALLET_TYPE neither CLI can honour

Anything that was not `sqlite` selected JSON, so a typo opened a JSON
wallet for someone who has a SQLite one and offered them a new identity
-- the second wallet these rules exist to prevent. Both CLIs did it, and
refusing it in the Python port alone would leave the two disagreeing
about which configurations are legal.

walletBackend() is the one rule now: both CLIs read it, and MCP reads it
in place of the copy it was carrying.

Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01GYGMfX3foBcVqP7AQNXjWj

* test(cli): Compare the wallet location rules a... (continued)

3977 of 4566 branches covered (87.1%)

Branch coverage included in aggregate %.

5 of 5 new or added lines in 2 files covered. (100.0%)

8625 of 9051 relevant lines covered (95.29%)

717.35 hits per line