Migrate from databricks-vectorsearch to databricks-ai-search#437
Merged
Conversation
`databricks-vectorsearch` is deprecated and has been renamed to `databricks-ai-search`. Per upstream's deprecation warning: databricks-vectorsearch is deprecated and has been renamed to databricks-ai-search. Imports under 'databricks.vector_search.*' will continue to work as a thin re-export of 'databricks.ai_search.*', but new code should switch to 'pip install databricks-ai-search' and 'from databricks.ai_search.* import ...'. Follow that recommendation literally: * pyproject.toml: replace `databricks-vectorsearch>=*` with `databricks-ai-search>=0.73` (root, openai, langchain, llamaindex). * All `from databricks.vector_search.* import ...` -> ai_search. * Mock-patch targets and docstring references updated to match. Symbol names are unchanged because `databricks-ai-search==0.73` exposes back-compat aliases: - databricks/ai_search/client.py:1090 VectorSearchClient = AISearchClient - databricks/ai_search/index.py:682 VectorSearchIndex = AISearchIndex - databricks/ai_search/exceptions.py:47 VectorSearchException = AISearchException The reranker / utils symbols (`Reranker`, `DatabricksReranker`, `CredentialStrategy`) were already used as-is and exist at the same relative paths in ai_search. Validated: * Fresh venv: only databricks-ai-search==0.73 installed (no vectorsearch). * 53/53 openai unit tests pass. * 258/258 langchain unit tests pass (4 skipped as before). * Smoke: `from databricks_openai.vector_search_retriever_tool import VectorSearchRetrieverTool` resolves `VectorSearchIndex` to `databricks.ai_search.index` even with the `databricks.vector_search` package completely absent.
bbqiu
approved these changes
Jun 9, 2026
The previous commit's bulk sed also rewrote the Sphinx doc links in six docstrings to point at `databricks.ai_search.html`. That page currently 403s on api-docs.databricks.com — the docs site has not yet republished under the `databricks-ai-search` name and there is no redirect from the old URL. Until the docs team ships docs under the new package name (and adds redirects from the old), point users at the URL that actually resolves today. The page that opens at `databricks.vector_search.html` documents the same `VectorSearchIndex.similarity_search` API the bridge calls, so the link target is still correct content-wise. This reverts only the URL strings; the actual import paths (`from databricks.ai_search.* import ...`) are unchanged.
…ot `.client`) The original docstrings said `databricks.vector_search.client. VectorSearchIndex.similarity_search` — the same wrong-path incidental-re-export bug that PR #435 fixed in the actual code. The earlier bulk sed in this PR carried the bug forward into the new package name, producing `databricks.ai_search.client.VectorSearchIndex` which does not even resolve (the new `databricks.ai_search.client` imports `AISearchIndex`, not `VectorSearchIndex`; only the alias on `.index` exists). Fix the docstrings to point at the real home of the class: databricks.ai_search.client.VectorSearchIndex → databricks.ai_search.index.VectorSearchIndex
3 tasks
jennsun
added a commit
that referenced
this pull request
Jun 10, 2026
Both jobs ran current CI against the frozen `integrations/openai` / `integrations/langchain` source from historical tags (databricks-ai- v0.2.0 through v0.5.0). They worked while PyPI dep evolution was backward-compatible. That contract broke with the databricks-vectorsearch → databricks-ai- search rename. The frozen historical sources import VectorSearchIndex from `databricks.vector_search.client` (the path that was removed in the rename), and there's no PR to HEAD that can update that frozen source. The matrix is permanently red regardless of what HEAD does. Per Ann: "we can restart them from the version we are deploying." Remove both jobs now; once we ship a release containing the ai-search migration (#437), the cross-version matrix can be re-introduced with that release as the new stable baseline. The bridge's own unit tests + integration tests still cover regression risk. This removes a non-Required CI signal that's now noise rather than information.
kunyuchen
added a commit
that referenced
this pull request
Jun 10, 2026
## databricks-ai-bridge 0.20.0 databricks-langchain 0.20.0 databricks-openai 0.16.0 (2026-06-10) ### New Features - databricks-ai-bridge: Durable, crash-resumable agent execution for MLflow ResponsesAgent handlers (#425) ### Improvements - databricks-ai-bridge / databricks-langchain / databricks-openai: migrate databricks-vectorsearch to databricks-ai-search (#437) ### Bug Fixes - databricks-openai: forward extra `kwargs` to maintain compatibility with newer versions of openai-agents (#426, #429)
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Summary
databricks-vectorsearchis deprecated and has been renamed todatabricks-ai-search. Follow-up to #435.When tests run,
databricks-vectorsearchnow emits:This PR follows that recommendation literally.
Changes
pyproject.toml× 4 (root,integrations/openai,integrations/langchain,integrations/llamaindex): replacedatabricks-vectorsearch>=*withdatabricks-ai-search>=0.73.src/,integrations/*/src/,integrations/*/tests/, andtests/: everyfrom databricks.vector_search.* import ...→from databricks.ai_search.* import .... Includesmock.patchtarget strings and theImportErrorfallback message indatabricks_langchain/vectorstores.py.databricks-ai-search==0.73exposes back-compat aliases for the names the bridge uses:databricks/ai_search/client.py:1090:VectorSearchClient = AISearchClientdatabricks/ai_search/index.py:682:VectorSearchIndex = AISearchIndexdatabricks/ai_search/exceptions.py:47:VectorSearchException = AISearchExceptionReranker,DatabricksReranker,CredentialStrategyexist under the same submodule paths.Net diff: +54 / −54 across 16 files (pure rename).
Validation
databricks-ai-search==0.73is present —databricks-vectorsearchis no longer pulled at all.from databricks_openai.vector_search_retriever_tool import VectorSearchRetrieverToolresolvesVectorSearchIndextodatabricks.ai_search.indexeven with thedatabricks.vector_searchpackage completely absent.ruff check+ruff format --checkclean.Companion notes
VectorSearchIndexcanonical-path fix). With both changes shipped, the bridge no longer touches thedatabricks.vector_search.*namespace at all and stops triggering the deprecation warning.databricks-openai = { path = "../openai", editable = true }source pin (added in Import VectorSearchIndex from its canonical module #435) is preserved — still needed until the nextdatabricks-openairelease containing the canonical-path fix ships to PyPI; can be removed in a follow-up cleanup after that release.openai_test (3.10, v0.x)/langchain_test (3.10, v0.x)) still tests historical tags whose frozen source importsdatabricks.vector_search.client.VectorSearchIndex. Those tags cannot be retroactively migrated — this PR doesn't try to. Tracked separately for Dhruv.Test plan
databricks_ai_bridge,databricks_langchain,databricks_openai, ordatabricks_llamaindexno longer see the upstreamDeprecationWarningaboutdatabricks-vectorsearch