Fix multi index resolution on analytics engine

[finnegancarroll]

Description

RelNodeUtils.extractIndices() returned comma-delimited table names as a single array element (e.g. ["idx1,idx2"]). The security filter's IndexNameExpressionResolver tried to resolve this as one index name, was unable to, and resolved no concrete indices.

Due to the default setting of strict expand open on analytics query request index resolution errors are not fatal.

Fix

Split comma-delimited table names in collectIndices() using Strings.splitStringByCommaToArray() (same utility used by IndexResolution in the planner). Each index is now a separate entry in the String[] array, so security evaluates permissions on each independently.

[github-actions]

PR Reviewer Guide 🔍

Here are some key observations to aid the review process:

🧪 PR contains tests

🔒 No security concerns identified

✅ No TODO sections

🔀 No multiple PR themes

⚡ No major issues detected

[github-actions]

PR Code Suggestions ✨

Explore these optional code suggestions:

Category	Suggestion	Impact
Possible issue	Trim whitespace from split index tokens Strings.splitStringByCommaToArray does not trim whitespace, so a table name like "a, b" would yield " b" and bypass intended index matching in the security filter. Trim each token before adding (and re-check for empty after trim) to ensure each index is evaluated correctly. sandbox/plugins/analytics-engine/src/main/java/org/opensearch/analytics/planner/RelNodeUtils.java [234-242] String tableName = names.get(names.size() - 1); // PPL multi-source queries (source=a,b) produce a single TableScan with a // comma-delimited table name. Split so each index is evaluated independently // by the security filter — same logic as IndexResolution. for (String idx : Strings.splitStringByCommaToArray(tableName)) { - if (!idx.isEmpty()) { - indices.add(idx); + String trimmed = idx.trim(); + if (!trimmed.isEmpty()) { + indices.add(trimmed); } } Suggestion importance[1-10]: 4 __ Why: Trimming whitespace is a reasonable defensive measure, but PPL-generated table names from source=a,b syntax are unlikely to contain spaces, and the comment in the test (splitStringByCommaToArray trims and skips empty tokens) suggests the author believed trimming already happens. Minor robustness improvement.	Low

[github-actions]

✅ Gradle check result for 0982986: SUCCESS

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 73.32%. Comparing base (50accb8) to head (0982986).
⚠️ Report is 5 commits behind head on main.

Additional details and impacted files

@@             Coverage Diff              @@
##               main   #22314      +/-   ##
============================================
- Coverage     73.35%   73.32%   -0.03%     
+ Complexity    76086    76000      -86     
============================================
  Files          6075     6075              
  Lines        345371   345371              
  Branches      49718    49718              
============================================
- Hits         253348   253253      -95     
- Misses        71838    71850      +12     
- Partials      20185    20268      +83     

☔ View full report in Codecov by Harness.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.

[expani]

So, Vanilla allows incorrectly parsed names ?

[finnegancarroll]

some of these might be rejected up front by sql/ppl before making it into AE. I'm unsure. Adding SQL plugin integ tests for this PR here opensearch-project/sql#5581 and will cover these cases.