pypi / azure-search-documents
azure-search-documents
What changed in azure-search-documents 12, from its official release notes, upgrade guides and advisories. Sources checked 2026-10-28.
5 breaking1 deprecations13 new APIs and options2 behavior and defaults1 fixes and other
See which of these apply to your code, ranked, in your agent:
npx @unstale/cli initazure-search-documents 1222 changes
Breaking changes 5
Model deserialize() methods are removed; use the model constructor to deserialize.
Replaces PathHierarchyTokenizer under its renamed model name.
The MYSQL enum value replaces MY_SQL.
The ONELAKE enum value replaces ONE_LAKE.
Model serialize() methods are removed; use as_dict() to serialize.
Deprecations 1
retrievable is now the native property instead of hidden; hidden remains as a backward-compatible inverse getter/setter.
New APIs and options 13
Adds a model for Azure Blob knowledge sources.
Adds native Markdown file parsing for blob indexers.
Adds a model for chat completion skills.
Adds a model for content understanding skills.
Adds a model for indexed OneLake knowledge sources.
Adds a setting for Markdown sectioning header depth, from h1 through h6.
Adds a setting to control Markdown document splitting with oneToOne or oneToMany.
Adds the KnowledgeBase model for knowledge base support.
Adds a client for knowledge base retrieval.
Adds managed identity support for data source connections.
Adds managed identity support for knowledge store projections.
Adds a model for search index knowledge sources.
Adds a model for web knowledge sources.
Behavior and default changes 2
The default API version is updated to 2026-04-01.
Enum values are now UPPER_CASE; PascalCase aliases remain available at runtime.
Fixes and other changes 1
The key_version property is now optional, aligning with service behavior.