feat: pulls out btc package #696
Conversation
…utilities Adds a new `btc` namespace to the SDK with the following modules: - btc.slip132: SLIP-132 xpub format conversion (normalize, format, getPrefix, inferPurpose) - btc.network: Network inference from xpub (inferFromXpub, getCoinType, isTestnet) - btc.getXpub/getXpubs/getAllXpubs: Flexible xpub fetching with purpose/coinType support - btc.tx: Transaction building (buildTxReq, estimateFee) with UTXO selection - btc.wallet: Wallet utilities (getSummary, getSnapshot) for balance/UTXO aggregation - btc.provider: Blockbook provider abstraction for chain data Key design decisions: - No breaking changes: existing fetchBtcXpub/signBtcTx APIs unchanged - Uses Uint8Array/DataView instead of Node.js Buffer for browser compatibility - Provider interface allows swapping Blockbook for other backends - Types reuse existing PreviousOutput for Lattice signing compatibility Includes 41 new tests across unit, integration, and e2e test suites.
Move BTC utilities from packages/sdk/src/btc/ to new packages/btc/ package for independent versioning and reuse. Changes: - Create @gridplus/btc package with slip132, network, tx, wallet utilities - SDK now depends on @gridplus/btc via workspace:* - SDK re-exports @gridplus/btc for backward compatibility - Keep xpub.ts in SDK (has Lattice-specific dependencies) - Add type guards and safeParseInt for robustness - Use Uint8Array instead of Buffer for browser compatibility
![chatgpt-codex-connector[bot]](https://avatars.githubusercontent.com/in/1144995?s=60&v=4){kind=small}
There was a problem hiding this comment.
💡 Codex Review
Here are some automated review suggestions for this pull request.
Reviewed commit: e309972391
ℹ️ About Codex in GitHub
Codex has been enabled to automatically review pull requests in this repo. Reviews are triggered when you
- Open a pull request for review
- Mark a draft as ready
- Comment "@codex review".
If Codex has suggestions, it will comment; otherwise it will react with 👍.
When you sign up for Codex through ChatGPT, Codex can also answer questions or update the PR, like "@codex address that feedback".
| return [ | ||
| purpose + HARDENED_OFFSET, | ||
| coinType + HARDENED_OFFSET, | ||
| HARDENED_OFFSET, | ||
| 1, |
There was a problem hiding this comment.
The change derivation path hard-codes the account component to 0' (HARDENED_OFFSET) instead of using the account that the UTXOs/xpub belong to. If a caller builds a transaction for a non‑zero account (e.g., they fetched an account-1 xpub via getXpub({ account: 1 }) and pass its UTXOs here), the change output will be sent to account 0, which is not covered by the account-1 xpub and will appear “missing” to the wallet. This is a functional regression for multi‑account use; the change path should be derived from the UTXO path or be provided explicitly.
Useful? React with 👍 / 👎.
| const hasSegwit = inputType !== 'p2pkh'; | ||
| const segwitOverhead = hasSegwit ? VBYTE_SIZES.SEGWIT_MARKER : 0; | ||
|
|
||
| const outputSize = VBYTE_SIZES.P2WPKH_OUTPUT; | ||
|
|
There was a problem hiding this comment.
Fee estimation always treats outputs as P2WPKH (outputSize = P2WPKH_OUTPUT), even when the wallet purpose is legacy or wrapped segwit. For purpose 44/49, the change output is P2PKH or P2SH‑P2WPKH, which are larger; this underestimates vbytes and can select too few inputs or produce too-low fees for non‑segwit outputs. This shows up when building legacy/wrapped transactions, where fee and change calculations rely on this estimate.
Useful? React with 👍 / 👎.
The cleanup steps (show logs, stop simulator) used working-directory which fails if the simulator checkout didn't happen. Now checks if directory exists before accessing files.
The SDK depends on @gridplus/btc via workspace:*, so btc must be built first. Updated root package.json scripts to build/test btc package before SDK.
commit: |
📝 Summary
🔧 Context / Implementation
🧪 Test Plan
🖼️ Screenshots (if applicable)