Conversation
- Implement comprehensive tests for the AsyncIterator streaming pattern in `tests/test_streaming_pattern.py`, covering various scenarios including basic pagination, sliding window, deduplication, batch size control, early termination, progress callback, invalid batch size, empty dataset, and large dataset simulation. - Introduce `tests/test_whale_block_pagination.py` to verify that the paging engine correctly detects and raises errors for whale blocks containing more transactions than the API's pagination limit, preventing silent data loss. - Ensure proper exception handling and telemetry recording for whale block detection.
|
You have run out of free Bugbot PR reviews for this billing cycle. This will reset on March 18. To receive reviews on all of your PRs, visit the Cursor dashboard to activate Pro and start your 14-day free trial. |
![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: 8c7a9f9b30
ℹ️ About Codex in GitHub
Your team has set up Codex to 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 👍.
Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".
| http=http_client, # type: ignore[arg-type] | ||
| endpoint_builder=self._network._url_builder, # type: ignore[arg-type] |
There was a problem hiding this comment.
This streaming call is passing incompatible dependencies into the paging layer: http comes from self._network._http2 (a boolean flag) and endpoint_builder is a UrlBuilder instance, while downstream functions expect an HttpClient (get/post) and an EndpointBuilder (open). In practice, non-V2 streaming paths fail at runtime with AttributeError instead of returning batches.
Useful? React with 👍 / 👎.
|
|
||
| # Fetch logs | ||
| try: | ||
| logs = await self.client.call(Method.EVENT_LOGS, **params) |
There was a problem hiding this comment.
iter_events() unconditionally calls Method.EVENT_LOGS, but the new v2 scanners used by default (from_config('etherscan', ...)) and by the recommended BlockScout v2 path do not provide EVENT_LOGS in their scanner specs. That means this method raises immediately in those common configurations, so the advertised event iteration API is unusable.
Useful? React with 👍 / 👎.
| resolver_result = await self.client.call( | ||
| Method.PROXY_ETH_CALL, | ||
| to=ENS_REGISTRY_ADDRESS, |
There was a problem hiding this comment.
Forward ENS resolution depends on Method.PROXY_ETH_CALL, but v2 scanners in this change do not expose that method, so client.call(...) fails and the broad exception handler returns None. This causes resolve_name() to silently miss for normal inputs on default/recommended scanner setups, making ENS support appear to work while dropping valid lookups.
Useful? React with 👍 / 👎.
- Add 15 new convenience methods to ChainscanClient covering ALL 28 Method enum values (except CONTRACT_VERIFY/VERIFY_STATUS workflows) - Fix get_transactions_df: use iter_transactions (full pagination) instead of single-page self.call() that silently truncated results - Add whale block warning in logs.py sliding-window mode - Add 38 new tests in test_client_convenience.py All 587 tests pass, mypy --strict clean, ruff clean.
- AGENTS.md: complete method reference (28 methods), paginated vs single-page table, updated test count (587+) - docs/skill.md: 30+ methods organized by category, common pitfalls table - docs/ROADMAP.md: mark streaming API as done, add v0.4.1 section, update release plan and priority matrix - README.md: replace client.call() examples with convenience methods, remove Moralis references, add streaming/DataFrame sections - Bump version to 0.4.1 - Auto-format test files (ruff format)
There was a problem hiding this comment.
💡 Codex Review
Here are some automated review suggestions for this pull request.
Reviewed commit: 29c69e33c1
ℹ️ About Codex in GitHub
Your team has set up Codex to 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 👍.
Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".
| page += 1 | ||
| return | ||
| # Get HTTP client from network | ||
| http_client = self._network._http2 |
There was a problem hiding this comment.
_http2 flag
iter_transactions_streaming() sets http_client = self._network._http2, but Network._http2 is a boolean configuration flag, not an HttpClient. The downstream streaming fetchers call http.get(...), so this path fails at runtime (e.g., when get_all_transactions() uses streaming) with an attribute error instead of fetching data; the same pattern is repeated in other new streaming helpers.
Useful? React with 👍 / 👎.
| params['topic2'] = topic2 | ||
| if topic3: | ||
| params['topic3'] = topic3 | ||
| result: list[dict[str, Any]] = await self.call(Method.EVENT_LOGS, **params) |
There was a problem hiding this comment.
EVENT_LOGS is unsupported by scanner
get_logs() unconditionally dispatches Method.EVENT_LOGS through self.call(...), but BlockScoutV2Scanner.SPECS in aiochainscan/scanners/blockscout_v2.py does not define EVENT_LOGS. With a blockscout_v2 client this now raises a "method not supported" error instead of returning logs, so single-page log retrieval is broken unless this method routes through the generic logs service fallback.
Useful? React with 👍 / 👎.
|
|
||
| # Fetch logs | ||
| try: | ||
| logs = await self.client.call(Method.EVENT_LOGS, **params) |
There was a problem hiding this comment.
SmartContract.iter_events() fetches logs with a single Method.EVENT_LOGS request and iterates that one response, so it only processes the first provider page (typically capped around ~1000 logs) and silently misses additional events for busy contracts. Since this API is exposed as an iterator for large histories, it should use paginated fetching (get_all_logs/streaming) to avoid truncating results.
Useful? React with 👍 / 👎.
- Replace vague all-methods-everywhere claim with explicit support matrix - blockscout_v2 supports ONLY 6 methods (balance, transactions, token portfolio, contract ABI, ENS reverse lookup x2) - Add scanner selection guide at top (choose scanner by task) - Document tx schema differences: blockscout_v2 uses nested from/to dicts - Add accurate network caveats (polygon 500, optimism 301 redirect) - Add per-scanner quick start examples and error handling patterns - Validated by real test agent: 5/7 tasks succeeded, 2 failed expectedly (get_block/gas_oracle on mainnet without etherscan key, as documented)
1 participant
No description provided.