fix: Import and Statistics fixes for Knowledge Bases

[erichare]

Summary

This PR fixes two correctness issues related to knowledge base ingestion and custom code execution.

For knowledge bases, component-based ingestion could leave embedding_metadata.json with stale metrics, especially when the file reported chunks = 0 even though Chroma data had already been written. This change makes new ingestions persist accurate chunk/text metrics and also allows existing stale knowledge bases to self-heal when metadata is read.

It also fixes dotted import handling in lfx.custom.validate so custom code follows normal Python import semantics for statements like import urllib.request and import urllib.request as request.

What changed

• Updated KnowledgeIngestionComponent to recompute and persist:
  • chunks
  • words
  • characters
  • avg_chunk_size
  • size
    after documents are written to Chroma.
• Updated KBAnalysisHelper.get_metadata(..., fast=True) to detect stale zero-chunk metadata when Chroma artifacts exist on disk, recount metrics from Chroma, and persist the corrected metadata.
• Fixed dotted import binding in lfx.custom.validate so non-aliased dotted imports bind the top-level package like normal Python imports, while aliased imports continue to work as expected.
• Regenerated component_index.json to keep the embedded component asset in sync with the ingestion changes.

Why

This closes the gap between what is actually stored in Chroma and what Langflow reports in knowledge base metadata. Without this fix, the KB modal/API could show misleading zeroed metrics after successful ingestion or partial failures.

It also makes custom code execution more predictable by aligning dotted import behavior with standard Python semantics.

Test coverage

• Added unit coverage for persisted KB metric updates after ingestion.
• Added backend API coverage for recounting stale zero-chunk metadata in fast metadata reads.
• Added unit coverage for dotted imports, aliased dotted imports, and global scope preparation in custom validation.

[coderabbitai]

Important

Review skipped

Auto reviews are disabled on base/target branches other than the default branch.

Please check the settings in the CodeRabbit UI or the .coderabbit.yaml file in this repository. To trigger a single review, invoke the @coderabbitai review command.

⚙️ Run configuration

Configuration used: Path: .coderabbit.yaml

Review profile: CHILL

Plan: Pro

Run ID: f062e7a5-161d-464c-b691-95ab8cf0c516

You can disable this status message by setting the reviews.review_status to false in the CodeRabbit configuration file.

Use the checkbox below for a quick retry:

• 🔍 Trigger review

✨ Finishing Touches

🧪 Generate unit tests (beta)

• Create PR with unit tests
• Commit unit tests in branch fix-kb-chroma-attribute

---

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.}

[codecov]

Codecov Report

❌ Patch coverage is 81.81818% with 4 lines in your changes missing coverage. Please review.
✅ Project coverage is 49.36%. Comparing base (b0a8662) to head (89cb662).
⚠️ Report is 1 commits behind head on release-1.9.0.

Files with missing lines	Patch %	Lines
src/backend/base/langflow/api/utils/kb_helpers.py	80.00%	2 Missing ⚠️
src/lfx/src/lfx/custom/validate.py	83.33%	1 Missing and 1 partial ⚠️

❌ Your project status has failed because the head coverage (48.00%) is below the target coverage (60.00%). You can increase the head coverage or adjust the target coverage.

Additional details and impacted files

@@                Coverage Diff                @@
##           release-1.9.0   #12446      +/-   ##
=================================================
+ Coverage          49.32%   49.36%   +0.04%     
=================================================
  Files               1924     1924              
  Lines             170395   170412      +17     
  Branches           24839    24841       +2     
=================================================
+ Hits               84043    84131      +88     
+ Misses             85348    85272      -76     
- Partials            1004     1009       +5     

Flag	Coverage Δ
backend	55.64% <80.00%> (+0.10%)	⬆️
frontend	48.23% <ø> (+<0.01%)	⬆️
lfx	48.00% <83.33%> (+0.13%)	⬆️

Flags with carried forward coverage won't be shown. Click here to find out more.

Files with missing lines	Coverage Δ
src/backend/base/langflow/api/utils/kb_helpers.py	77.44% <80.00%> (+6.69%)	⬆️
src/lfx/src/lfx/custom/validate.py	57.40% <83.33%> (+12.20%)	⬆️

... and 24 files with indirect coverage changes

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.
• 📦 JS Bundle Analysis: Save yourself from yourself by tracking and limiting bundle sizes in JS merges.

[github-actions]

Frontend Unit Test Coverage Report

Coverage Summary

Lines	Statements	Branches	Functions
	28.01% (29184/104184)	64.76% (3717/5739)	30.03% (689/2294)

Unit Test Results

Tests	Skipped	Failures	Errors	Time
3014	0 💤	0 ❌	0 🔥	4m 31s ⏱️

[Copilot]

Pull request overview

This PR improves correctness in two areas: (1) knowledge base metric accuracy when persisted metadata becomes stale relative to Chroma storage, and (2) runtime code execution/import handling so dotted imports bind the same names Python would normally bind.

Changes:

• Recount KB chunk/text metrics when embedding_metadata.json reports chunks=0 but Chroma data exists, and persist corrected metrics back to disk.
• Update KnowledgeIngestionComponent to recompute and persist chunk/word/character/size metrics immediately after ingestion.
• Fix dotted import binding behavior in custom code validation/execution and add unit tests for dotted/aliased imports.

Reviewed changes

Copilot reviewed 8 out of 8 changed files in this pull request and generated 2 comments.

Show a summary per file

File	Description
src/lfx/tests/unit/custom/component/test_validate.py	Adds tests covering dotted imports and aliased dotted imports for custom code execution.
src/lfx/src/lfx/custom/validate.py	Adjusts import binding to keep top-level package names for dotted imports (matching Python semantics).
src/lfx/src/lfx/components/files_and_knowledge/ingestion.py	Adds persisted metrics update after ingestion and returns Chroma from vector store creation for metric recounting.
src/lfx/src/lfx/_assets/component_index.json	Updates embedded component code hash/content to match ingestion changes.
src/backend/tests/unit/test_knowledge_bases_api.py	Adds coverage ensuring stale zero-chunk metadata triggers a recount in fast metadata mode.
src/backend/tests/unit/components/files_and_knowledge/test_ingestion.py	Adds coverage for persisting chunk/text metrics from a mocked Chroma collection.
.secrets.baseline	Updates baseline line numbers/timestamp due to test file changes.

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.