OpenMetadata Service New-Code Coverage

❌ FAIL. Required changed-line coverage: 90.00% overall and per touched production file.

• Overall executable changed lines: 2/28 covered (7.14%)
• Missed executable changed lines: 26
• Non-executable changed lines ignored by JaCoCo: 35
• Changed production files: 4

Files below threshold:

• openmetadata-service/src/main/java/org/openmetadata/service/jdbi3/GlossaryTermRepository.java: 0/13 covered (0.00%), uncovered lines 177, 178, 179, 182, 183, 184, 190, 200, 212, 213, 214, 215, +1 more
• openmetadata-service/src/main/java/org/openmetadata/service/search/InheritedFieldEntitySearch.java: 0/9 covered (0.00%), uncovered lines 249, 250, 251, 252, 253, 254, 255, 256, 257
• openmetadata-service/src/main/java/org/openmetadata/service/resources/glossary/GlossaryTermResource.java: 0/3 covered (0.00%), uncovered lines 447, 883, 931
• openmetadata-service/src/main/java/org/openmetadata/service/search/QueryFilterBuilder.java: 2/3 covered (66.67%), uncovered lines 139

Only changed executable lines under openmetadata-service/src/main/java are counted. Test files, comments, imports, and non-executable lines are excluded.

🟡 Playwright Results — all passed (27 flaky)

✅ 3434 passed · ❌ 0 failed · 🟡 27 flaky · ⏭️ 223 skipped

📦 Download artifacts

# Download playwright-test-results-<shard> artifact and unzip
npx playwright show-trace path/to/trace.zip    # view trace

// Option: add a thread-local or pass certClassification as parameter
// For example, in getTags():
private List<TagLabel> filterCertTags(List<TagLabel> tags, String certClassification) {
  if (certClassification != null && tags != null) {
    tags = new ArrayList<>(tags);
    tags.removeIf(tag -> certClassification.equals(
        FullyQualifiedName.getParentFQN(tag.getTagFQN())));
  }
  return tags;
}

Code Review: Adds parent-scoped asset counts and assets listing for glossary terms with performance optimizations including Elasticsearch query pushdown and reduced deserialization. Consider consolidating the redundant getCertificationClassification() call to avoid duplicate processing per entity read.

1. 💡 Performance: getCertificationClassification() called redundantly per entity read

   The `getCertificationClassification()` method is called separately from `getCertification()`, `getTags()`, and `getTagsByPrefix()` during a single entity read operation. While `SettingsCache` is memory-cached (3-min TTL), the repeated cache lookups, string operations, and `AssetCertificationSettings` deserialization are unnecessary overhead that compounds in bulk/list operations. Consider caching the result in a local variable at the operation level or passing it as a parameter.

   Suggested fix:
   // Option: add a thread-local or pass certClassification as parameter
   // For example, in getTags():
   private List<TagLabel> filterCertTags(List<TagLabel> tags, String certClassification) {
     if (certClassification != null && tags != null) {
       tags = new ArrayList<>(tags);
       tags.removeIf(tag -> certClassification.equals(
           FullyQualifiedName.getParentFQN(tag.getTagFQN())));
     }
     return tags;
   }

gitar display:verbose         

_{Was this helpful? React with 👍 / 👎 | Gitar}

@gitar-bot can you review the backend changes again #26801 (comment)

@gitar-bot can you review the backend changes again #26801 (comment)

Here's my review of the latest backend changes:

GlossaryTermRepository.java — getAllGlossaryTermsWithAssetsCount(String parent)

The refactor correctly addresses the previous concerns:

• When parent is provided, getNestedTerms(parent) is called and only the fullyQualifiedName field is extracted via JsonUtils.readTree(json).path("fullyQualifiedName").asText(). This avoids full GlossaryTerm deserialization — good.
• When parent is null/empty, listAll(getFields("fullyQualifiedName"), ...) is used — consistent with the previous unfiltered path.
• The loop now operates on String fqn instead of GlossaryTerm objects — clean and correct.

One remaining concern:

getNestedTerms may return empty strings silently. If JsonUtils.readTree(json).path("fullyQualifiedName") is missing or null in a record, asText() returns "" (empty string). The resulting empty key in glossaryTermAssetCounts could produce a spurious Elasticsearch query. Consider filtering with .filter(fqn -> !fqn.isEmpty()) after the .map(...) call.

GlossaryTermRepository.java — getGlossaryTermAssets

The change here is a minor cosmetic refactor (extract fqn variable to reduce repetition in the log call). No behavioral change. Looks fine.

Note: the earlier finding about the missing parent-filter guard in getGlossaryTermAssets (where an out-of-scope term would return assets instead of an empty list) was addressed in a prior iteration via Elasticsearch-side filtering, so that concern is resolved.

GlossaryTermResource.java

Clean — @QueryParam("parent") is correctly wired through with a descriptive @Parameter annotation. No issues.

DefaultInheritedFieldEntitySearchTest.java

The added assertion assertFalse(glossary.isSupportsHierarchy()) is valid and correctly documents the expected behavior. No issues.

GlossaryTermResourceIT.java

Test coverage is solid overall. A few observations:

1. get_assetsCountsWithParentFilter verifies that filtering by glossary FQN includes only that glossary's terms — good. However, it only asserts that keys exist in the map, not their values. This is acceptable for an integration smoke test, but be aware it won't catch count miscalculations.

2. get_assetsCountsWithNoParent only asserts assertNotNull — effectively just a "doesn't crash" test. Low value but not harmful.

3. getAssetCounts, getTermAssetsById, and getTermAssetsByName helpers directly call raw HTTP endpoints. This is fine for integration tests where the typed SDK doesn't expose these endpoints yet.

Summary: The changes are in good shape. The only actionable item is adding a .filter(fqn -> !fqn.isEmpty()) guard after the .map(...) in the parent-scoped branch of getAllGlossaryTermsWithAssetsCount to handle potential missing FQN fields in the raw JSON.

Quality Gate passed for 'open-metadata-ingestion'

Issues
0 New issues
0 Accepted issues

Measures
0 Security Hotspots
0.0% Coverage on New Code
0.0% Duplication on New Code

See analysis details on SonarQube Cloud