Fix incomplete event indexing for SEI based EVM chains

[lorrod]

This PR fixes incomplete event indexing for Cosmos/Sei-based EVM chains where synthetic logs are injected at the block level.

In these environments, some logs are emitted from transactions that are not returned via eth_getBlockByNumber, but are still discoverable through eth_getLogs due to block-level bloom filtering. These logs were previously skipped because the indexer only processed logs tied to visible transactions.

What was changed

• Added additional handling in IndexerManager to process synthetic logs whose transactionHash is not present in the block’s transaction list.

• Introduced a fallback mechanism that:

  • Collects all transaction hashes from the block
  • Iterates over all logs in the block
  • Indexes logs that belong to unknown (synthetic) transactions

• Improved rewind logic in UnfinalizedBlocksService:

  • Replaced unsafe POI fallback with a direct call to getHeaderForHeight
  • Ensures consistent and safer rewind behavior when POI is not enabled

Motivation

Without this fix:

• Some valid events on Cosmos/Sei EVM chains were silently ignored
• This could lead to data inconsistency, missing indexed events, and incorrect downstream state

This change ensures compatibility with hybrid EVM implementations that inject logs outside standard Ethereum transaction visibility.

Fixes # (issue)

---

Type of change

• Bug fix (non-breaking change which fixes an issue)

---

Checklist

• I have tested locally
• I have performed a self review of my changes
• Updated any relevant documentation
• Linked to any relevant issues
• I have added tests relevant to my changes
• Any dependent changes have been merged and published in downstream modules
• My code is up to date with the base branch
• I have updated relevant changelogs.

Summary by CodeRabbit

Bug Fixes

• Fixed indexing of synthetic Cosmos/Sei logs that were previously missed during full block processing.
• Improved accuracy of finalized block header retrieval by ensuring complete header data is fetched when using standard confirmation methods.

[coderabbitai]

📝 Walkthrough

Walkthrough

Two files in the indexer module are modified. The first adds logic to index synthetic Cosmos/Sei logs with transaction hashes not found in the block's transactions. The second changes how fallback headers are retrieved from manual construction to a service method call.

Changes

Cohort / File(s)	Summary
Synthetic Log Indexing packages/node/src/indexer/indexer.manager.ts	Added a supplemental pass over block logs in indexBlockData to index synthetic logs whose transaction hashes are absent from block transactions. Executes only in full-block branch after standard indexing.
Header Retrieval Refactor packages/node/src/indexer/unfinalizedBlocks.service.ts	Modified getLastCorrectFinalizedBlock fallback to call blockchainService.getHeaderForHeight() instead of constructing a partial Header object with only blockHeight.

Estimated code review effort

🎯 2 (Simple) | ⏱️ ~10 minutes

Possibly related PRs

• subquery/subql-ethereum#417: Updates getLastCorrectFinalizedBlock to retrieve real headers via blockchainService.getHeaderForHeight() instead of constructing partial headers, mirroring changes in this PR.

Poem

🐰 Synthetic logs are now found with care,
While headers fetch from the service fair,
Both small but sure, the changes align,
In indexer paths where data does shine! ✨

🚥 Pre-merge checks | ✅ 3

✅ Passed checks (3 passed)

Check name	Status	Explanation
Description Check	✅ Passed	Check skipped - CodeRabbit’s high-level summary is enabled.
Title check	✅ Passed	The title 'Fix incomplete event indexing for SEI based EVM chains' directly and clearly summarizes the main change: addressing incomplete event indexing for Cosmos/Sei-based EVM chains by handling synthetic logs.
Docstring Coverage	✅ Passed	No functions found in the changed files to evaluate docstring coverage. Skipping docstring coverage check.

_{✏️ Tip: You can configure your own custom pre-merge checks in the settings.}

✨ Finishing Touches

🧪 Generate unit tests (beta)

• Create PR with unit tests

---

Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out.

❤️ Share

• X
• Mastodon
• Reddit
• LinkedIn

_{Comment @coderabbitai help to get the list of available commands and usage tips.}

[coderabbitai]

🧹 Nitpick comments (1)

packages/node/src/indexer/indexer.manager.ts (1)

132-137: Harden hash normalization to avoid runtime crashes on malformed log payloads.

Line 136 assumes log.transactionHash is always present. If an RPC response ever returns a null/undefined hash for an anomalous log, this path throws and aborts block indexing. Consider a defensive guard.

Proposed hardening

-      const txHashes = new Set(
-        block.transactions.map((tx) => tx.hash.toLowerCase()),
-      );
+      const txHashes = new Set(
+        block.transactions
+          .map((tx) => tx.hash?.toLowerCase())
+          .filter((h): h is string => Boolean(h)),
+      );
       for (const log of block.logs ?? []) {
-        if (!txHashes.has(log.transactionHash.toLowerCase())) {
+        const logTxHash = log.transactionHash?.toLowerCase();
+        if (!logTxHash || !txHashes.has(logTxHash)) {
           await this.indexEvent(log, dataSources, getVM);
         }
       }

🤖 Prompt for AI Agents

Verify each finding against the current code and only fix it if needed.

In `@packages/node/src/indexer/indexer.manager.ts` around lines 132 - 137, The
loop over block.logs assumes log.transactionHash is always defined and calls
.toLowerCase(), which can crash on null/undefined; in the indexer.manager.ts
where txHashes is created and the for-loop calls this.indexEvent(log,
dataSources, getVM), add a defensive guard that extracts transactionHash from
each log, ensures it's a string (e.g., typeof === 'string' &&
transactionHash.length > 0) before calling .toLowerCase(), and skip or otherwise
handle logs with missing/invalid transactionHash so they do not trigger the
.toLowerCase() call or abort block indexing; keep the existing behavior of
calling indexEvent only for logs whose normalized hash is not in the txHashes
Set.

🤖 Prompt for all review comments with AI agents

Verify each finding against the current code and only fix it if needed.

Nitpick comments:
In `@packages/node/src/indexer/indexer.manager.ts`:
- Around line 132-137: The loop over block.logs assumes log.transactionHash is
always defined and calls .toLowerCase(), which can crash on null/undefined; in
the indexer.manager.ts where txHashes is created and the for-loop calls
this.indexEvent(log, dataSources, getVM), add a defensive guard that extracts
transactionHash from each log, ensures it's a string (e.g., typeof === 'string'
&& transactionHash.length > 0) before calling .toLowerCase(), and skip or
otherwise handle logs with missing/invalid transactionHash so they do not
trigger the .toLowerCase() call or abort block indexing; keep the existing
behavior of calling indexEvent only for logs whose normalized hash is not in the
txHashes Set.

---

ℹ️ Review info

⚙️ Run configuration

Configuration used: defaults

Review profile: CHILL

Plan: Pro

Run ID: 5120baf2-bc38-48d2-9c9b-6a84146dc4b0

📥 Commits

Reviewing files that changed from the base of the PR and between fc1f641 and 215d6ed.

📒 Files selected for processing (2)

• packages/node/src/indexer/indexer.manager.ts
• packages/node/src/indexer/unfinalizedBlocks.service.ts