Wire analytics-engine as extendedPlugins dependency

[ahkcs]

Summary

• Wire analytics-engine as extendedPlugins parent classloader for shared Calcite classes
• Add bundlePlugin excludes to prevent jar hell between parent/child classloaders
• Add analytics-engine plugin to all test clusters (integTest, yamlRestTest, security, JDBC, doctest)
• Add commons-text to analytics-engine ZIP for Calcite fuzzy matching support
• Fix isAnalyticsIndex to use lightweight parsing context (no cluster state needed)

Test plan

• All existing integration tests pass with analytics-engine loaded
• RestUnifiedQueryActionTest passes (routing logic)
• Doctests pass (no ClassNotFoundException for LevenshteinDistance)
• No jar hell errors on plugin startup

[github-actions]

PR Code Analyzer ❗

AI-powered 'Code-Diff-Analyzer' found issues on commit 2c4c401.

Path	Line	Severity	Description
libs/analytics-engine-3.6.0-SNAPSHOT.jar	1	high	Unverifiable binary JAR committed directly to the repository. SNAPSHOT artifacts bypass standard artifact signing and reproducibility guarantees. Cannot audit for malicious bytecode without decompilation.
libs/analytics-framework-3.6.0-SNAPSHOT.jar	1	high	Unverifiable binary JAR committed directly to the repository. SNAPSHOT artifacts bypass standard artifact signing and reproducibility guarantees. Cannot audit for malicious bytecode without decompilation.
libs/analytics-engine-3.6.0-SNAPSHOT.zip	1	high	Unverifiable binary ZIP (plugin bundle) committed directly to the repository. Contents cannot be audited and are deployed as a live OpenSearch plugin in all test clusters.
core/build.gradle	66	high	Dependency added from a local binary file outside of a package registry (files(...)) referencing a SNAPSHOT JAR in libs/. Bypasses normal dependency verification, checksums, and provenance tracking.
plugin/build.gradle	58	high	extendedPlugins now includes 'analytics-engine', making the analytics-engine classloader a trusted parent of the SQL plugin classloader. This grants analytics-engine code access to SQL plugin internals and elevates its privilege in the OpenSearch plugin security model. Combined with the unaudited binary ZIP, this is a significant trust escalation.
core/src/main/java/org/opensearch/sql/calcite/utils/CalciteToolsHelper.java	399	medium	Uses reflection with setAccessible(true) to access the private internalParameters field of CalcitePreparingStmt. Bypasses Java access controls to read/write internal Calcite state; if the field mapping is wrong or manipulated, it could allow injection of arbitrary execution parameters.
core/src/main/java/org/opensearch/sql/calcite/utils/CalciteToolsHelper.java	423	medium	Dynamically compiles and instantiates arbitrary Java source code (generated from Calcite query plans) using a reflection-invoked Janino compiler with a custom classloader. The compiled class is executed directly as a Bindable. Malicious input to the query plan could lead to code execution if Calcite code generation is compromised.
opensearch/src/main/java/org/opensearch/sql/opensearch/storage/script/CalciteScriptEngine.java	154	medium	Dynamically compiles arbitrary Java source (Rex expressions) via reflection-loaded Janino ClassBodyEvaluator with a custom classloader, then instantiates and executes the resulting class. Same risk profile as CalciteToolsHelper: if Rex code generation can be influenced by attacker-controlled input, arbitrary code execution is possible.
plugin/build.gradle	201	medium	Multiple runtime JARs explicitly excluded from the plugin bundle (calcite-core, guava, janino, jackson, etc.) because they are expected to be provided by the analytics-engine plugin. This creates a hard runtime dependency on the unaudited analytics-engine binary; if that binary is replaced or tampered with, the SQL plugin will silently use the attacker-supplied versions of these critical libraries.
plugin/src/main/java/org/opensearch/sql/plugin/rest/RestUnifiedQueryAction.java	171	low	Schema is now built from live cluster state (clusterService.state()) via OpenSearchSchemaBuilder from the unaudited analytics-framework JAR. If that library exfiltrates or logs cluster metadata (index names, mappings, settings), it would have access to full cluster state on every query context construction.

The table above displays the top 10 most important findings.

Total: 10 | Critical: 0 | High: 5 | Medium: 4 | Low: 1

---

Pull Requests Author(s): Please update your Pull Request according to the report above.

Repository Maintainer(s): You can bypass diff analyzer by adding label skip-diff-analyzer after reviewing the changes carefully, then re-run failed actions. To re-enable the analyzer, remove the label, then re-run all actions.

---

⚠️ Note: The Code-Diff-Analyzer helps protect against potentially harmful code patterns. Please ensure you have thoroughly reviewed the changes beforehand.

Thanks.