SearchIndexClient
.get_search_client
inherits the API version.
- Fixed the issue that we missed ODATA header when using Entra ID auth.
- Added new models:
azure.search.documents.models.HybridCountAndFacetMode
azure.search.documents.models.HybridSearch
azure.search.documents.models.SearchScoreThreshold
azure.search.documents.models.VectorSimilarityThreshold
azure.search.documents.models.VectorThreshold
azure.search.documents.models.VectorThresholdKind
azure.search.documents.models.VectorizableImageBinaryQuery
azure.search.documents.models.VectorizableImageUrlQuery
azure.search.documents.indexes.models.AIServicesVisionParameters
azure.search.documents.indexes.models.AIServicesVisionVectorizer
azure.search.documents.indexes.models.AIStudioModelCatalogName
azure.search.documents.indexes.models.AzureMachineLearningParameters
azure.search.documents.indexes.models.AzureMachineLearningVectorizer
azure.search.documents.indexes.models.AzureOpenAIModelName
azure.search.documents.indexes.models.VectorEncodingFormat
azure.search.documents.indexes.models.VisionVectorizeSkill
- Added
hybrid_search
support forSearchClient.search
method. - Updated default API version to
2024-05-01-preview
.
- Fixed the bug that SearchClient failed when both answer count and answer threshold applied.
- Added
IndexerExecutionEnvironment
,IndexingMode
,LineEnding
,NativeBlobSoftDeleteDeletionDetectionPolicy
,ScalarQuantizationCompressionConfiguration
,ScalarQuantizationParameters
,SearchServiceCounters
,SearchServiceLimits
,SearchServiceStatistics
,VectorSearchCompressionConfiguration
&VectorSearchCompressionTargetDataType
. - Added
stored
inSearchField
.
SearchIndexerSkillset
,SearchField
,SearchIndex
,AnalyzeTextOptions
,SearchResourceEncryptionKey
,SynonymMap
,SearchIndexerDataSourceConnection
are no longer subclasses of_serialization.Model
.
- Fixed the issue that
SearchIndexerSkillset
,SearchField
,SearchIndex
,AnalyzeTextOptions
,SearchResourceEncryptionKey
,SynonymMap
,SearchIndexerDataSourceConnection
could not be serialized andas_dict
did not work. - Fixed the issue that
context
was missing forEntityRecognitionSkill
andSentimentSkill
. #34623
- Default to API version
V2024_03_01_PREVIEW
- Added back
semantic_query
forSearch
method. - Added back alias operations to
SearchIndexClient
. - Added back
AzureOpenAIEmbeddingSkill
,AzureOpenAIParameters
andAzureOpenAIVectorizer
. - Added back
query_language
,query_speller
,semantic_fields
anddebug
forSearch
method. - Added
send_request
method forSearchClient
&SearchIndexClient
to run a network request using the client's existing pipeline.
- Fixed the issue that we added unexpected
retrievable
property forSearchField
.
- Python 3.7 is no longer supported. Please use Python version 3.8 or later.
- Added new models:
VectorSearchAlgorithmMetric
IndexProjectionMode
SearchIndexerIndexProjections
SearchIndexerIndexProjectionSelector
SearchIndexerIndexProjectionsParameters
BlobIndexerDataToExtract
BlobIndexerImageAction
BlobIndexerParsingMode
CharFilterName
CustomEntity
CustomEntityAlias
DataChangeDetectionPolicy
DataDeletionDetectionPolicy
DefaultCognitiveServicesAccount
HighWaterMarkChangeDetectionPolicy
HnswAlgorithmConfiguration
IndexerExecutionResult
IndexingParameters
IndexingParametersConfiguration
IndexingSchedule
LexicalAnalyzerName
LexicalTokenizerName
PIIDetectionSkill
PIIDetectionSkillMaskingMode
ScoringProfile
SemanticSearch
- Added
index_projections
support forSearchIndexerSkillset
These changes do not impact the API of stable versions such as 11.3.0. Only code written against a beta version such as 11.4.0b11 may be affected.
- Renamed
AnswerResult
toQueryAnswerResult
andCaptionResult
toQueryCaptionResult
. - Renamed
SemanticErrorHandling
toSemanticErrorMode
. - Renamed
RawVectorQuery
toVectorizedQuery
. - Renamed
ExhaustiveKnnVectorSearchAlgorithmConfiguration
toExhaustiveKnnAlgorithmConfiguration
. - Renamed
PrioritizedFields
toSemanticPrioritizedFields
. - Renamed
query_caption_highlight
toquery_caption_highlight_enabled
. query_language
andquery_speller
are not available forSearch
method in this stable release.alias
operations are not available in this stable release.AzureOpenAIEmbeddingSkill
,AzureOpenAIParameters
andAzureOpenAIVectorizer
are not available in 11.4.0.- Renamed
vector_search_profile
tovector_search_profile_name
inSearchField
. - Renamed
SemanticSettings
toSemanticSearch
.
- Used API version "2023-11-01".
- Added
vector_filter_mode
support forSearch
method. - Exposed
VectorizableTextQuery
inazure.search.document.models
.
These changes do not impact the API of stable versions such as 11.3.0. Only code written against a beta version such as 11.4.0b6 may be affected.
- Renamed
vector_search_configuration
tovector_search_profile
inSearchField
. - Renamed
vectors
tovector_queries
inSearch
method. - Renamed
azure.search.documents.models.Vector
toazure.search.documents.models.VectorQuery
. - Stopped supporting api version
V2023_07_01_PREVIEW
anymore.
- Default to use API version
V2023_10_01_PREVIEW
- Fixed the bug that list type of
order_by
was not correctly handled. #31837
- Exposed
HnswVectorSearchAlgorithmConfiguration
These changes do not impact the API of stable versions such as 11.3.0. Only code written against a beta version such as 11.4.0b6 may be affected.
- Instead of using
VectorSearchAlgorithmConfiguration
, now you need to use concrete types likeHnswVectorSearchAlgorithmConfiguration
.
- Added multi-vector search support. Now instead of passing in
vector
,top_k
andvector_fields
, search method acceptsvectors
which is a list ofVector
object.
These changes do not impact the API of stable versions such as 11.3.0. Only code written against a beta version such as 11.4.0b6 may be affected.
- Stopped supporting
vector
,top_k
andvector_fields
inSearchClient.search
method.
- Added
top_k
support forVectorSearch
.
- Exposed
azure.search.documents.models.Vector
.
- Added
VectorSearch
support.
- Deprecated
SentimentSkillV1
andEntityRecognitionSkillV1
.
- Added the semantic reranker score and captions on
SearchResult
.(thanks to @LucasVascovici for the contribution)
- Enabled
OcrSkill
andImageAnalysisSkill
- Added Python 3.11 support.
- Added support to create, update and delete aliases via the
SearchIndexClient
.
- Some of the features that were available in the
11.3.0b8
version are not available in this GA. They would be available in the upcoming beta release.
- Added support for other national clouds.
- Added support for TokenCredential
- Fixed issue where async
search
call would fail with a 403 error when retrieving large number of documents.
- Python 3.6 is no longer supported. Please use Python version 3.7 or later.
- Fixes a bug allowing users to set keys for cognitive service skills using the API. Exposes
DefaultCognitiveServicesAccount
andCognitiveServicesAccountKey
- Added support to create, update and delete aliases via the
SearchIndexClient
.
- Support for
AzureMachineLearningSkill
. The AML skill allows you to extend AI enrichment with a custom Azure Machine Learning (AML) model. Once an AML model is trained and deployed, an AML skill integrates it into AI enrichment.
- Python 2.7 is no longer supported. Please use Python version 3.6 or later.
Minor updates.
- Added properties to
SearchClient.search
:semantic_configuration_name
- Added properties to
SearchIndex
:semantic_settings
- Added models:
PrioritizedFields
,SemanticConfiguration
,SemanticField
,SemanticSettings
- Added new values to model
QueryLanguage
- Added properties to
SearchClient.search
:session_id
,scoring_statistics
. - Added properties to
SearchIndexerDataSourceConnection
:identity
,encryption_key
. - Added
select
property to the followingSearchIndexClient
operations:get_synonym_maps
,list_indexes
. - Added
select
property to the followingSearchIndexersClient
operations:get_data_source_connections
,get_indexers
,get_skillsets
. - Added operations to
SearchIndexerClient
:reset_skills
,reset_documents
. - Added model:
DocumentKeysOrIds
- Added properties to
SearchClient
:query_answer
,query_answer_count
,query_caption
,query_caption_highlight
andsemantic_fields
.
- Renamed
SearchClient.speller
toSearchClient.query_speller
. - Renamed model
Speller
toQuerySpellerType
. - Renamed model
Answers
toQueryAnswerType
. - Removed keyword arguments from
SearchClient
:answers
andcaptions
. SentimentSkill
,EntityRecognitionSkill
: added client-side validation to prevent sending unsupported parameters.- Renamed property
ignore_reset_requirements
toskip_indexer_reset_requirement_for_cache
.
- Added new models:
azure.search.documents.models.QueryCaptionType
azure.search.documents.models.CaptionResult
azure.search.documents.indexes.models.CustomEntityLookupSkillLanguage
azure.search.documents.indexes.models.EntityRecognitionSkillVersion
azure.search.documents.indexes.models.LexicalNormalizerName
azure.search.documents.indexes.models.PIIDetectionSkill
azure.search.documents.indexes.models.PIIDetectionSkillMaskingMode
azure.search.documents.indexes.models.SearchIndexerCache
azure.search.documents.indexes.models.SearchIndexerDataIdentity
azure.search.documents.indexes.models.SearchIndexerDataNoneIdentity
azure.search.documents.indexes.models.SearchIndexerDataUserAssignedIdentity
azure.search.documents.indexes.models.SentimentSkillVersion
- Added
normalizer_name
property toAnalyzeTextOptions
model.
- Removed:
azure.search.documents.indexes.models.SentimentSkillV3
azure.search.documents.indexes.models.EntityRecognitionSkillV3
- Renamed:
SearchField.normalizer
renamed toSearchField.normalizer_name
.
SentimentSkill
andEntityRecognitionSkill
can now be created by specifying theskill_version
keyword argument with aSentimentSkillVersion
orEntityRecognitionSkillVersion
, respectively. The default behavior ifskill_version
is not specified is to create a version 1 skill.
- Added new skills:
SentimentSkillV3
,EntityLinkingSkill
,EntityRecognitionSkillV3
- Added AAD support
- Added support for semantic search
- Added normalizer support
This version will be the last version to officially support Python 3.5, future versions will require Python 2.7 or Python 3.6+.
New features
- Added support for knowledge store #18461
- Added new data source type ADLS gen2 #16852
Breaking Changes
IndexDocumentsBatch
does not supportenqueue_action
any longer.enqueue_actions
takes a single action too.max_retries
ofSearchIndexingBufferedSender
is renamed tomax_retries_per_action
SearchClient
does not supportget_search_indexing_buffered_sender
Features
- Added
get_search_indexing_buffered_sender
support forSearchClient
- Added
initial_batch_action_count
support forSearchIndexingBufferedSender
- Added
max_retries
support forSearchIndexingBufferedSender
Breaking Changes
- Renamed
SearchIndexDocumentBatchingClient
toSearchIndexingBufferedSender
- Renamed
SearchIndexDocumentBatchingClient.add_upload_actions
toSearchIndexingBufferedSender.upload_documents
- Renamed
SearchIndexDocumentBatchingClient.add_delete_actions
toSearchIndexingBufferedSender.delete_documents
- Renamed
SearchIndexDocumentBatchingClient.add_merge_actions
toSearchIndexingBufferedSender.merge_documents
- Renamed
SearchIndexDocumentBatchingClient.add_merge_or_upload_actions
toSearchIndexingBufferedSender.merge_or_upload_documents
- Stopped supporting
window
kwargs forSearchIndexingBufferedSender
- Splitted kwarg
hook
intoon_new
,on_progress
,on_error
,on_remove
forSearchIndexingBufferedSender
Features
- Added
auto_flush_interval
support forSearchIndexingBufferedSender
Features
- Added
azure.search.documents.RequestEntityTooLargeError
Flush
method inBatchClient
now will not return until all actions are done
Breaking Changes
- Removed
succeeded_actions
&failed_actions
fromBatchClient
- Removed
get_index_document_batching_client
fromSearchClient
Features
- new
SearchIndexDocumentBatchingClient
SearchIndexDocumentBatchingClient
supports handling document indexing actions in an automatic way. It can trigger the flush method automatically based on pending tasks and idle time.
- Doc & Sample fixes
Features
-
Exposed more models:
- BM25SimilarityAlgorithm
- ClassicSimilarityAlgorithm
- EdgeNGramTokenFilterSide
- EntityCategory
- EntityRecognitionSkillLanguage
- FieldMapping
- FieldMappingFunction
- ImageAnalysisSkillLanguage
- ImageDetail
- IndexerExecutionStatus
- IndexerStatus
- KeyPhraseExtractionSkillLanguage
- MicrosoftStemmingTokenizerLanguage
- MicrosoftTokenizerLanguage
- OcrSkillLanguage
- PhoneticEncoder
- ScoringFunctionAggregation
- ScoringFunctionInterpolation
Breaking Changes
-
Reorganized
SearchServiceClient
intoSearchIndexClient
&SearchIndexerClient
#11507 -
Split searchindex.json and searchservice.json models and operations into separate namespaces #11508
-
Renamed
edm
toSearchFieldDataType
#11511 -
Now Search Synonym Map creation/update returns a model #11514
-
Renaming #11565
- SearchIndexerDataSource -> SearchIndexerDataSourceConnection
- SearchField.SynonymMaps -> SearchField.SynonymMapNames
- SearchField.Analyzer -> SearchField.AnalyzerName
- SearchField.IndexAnalyzer -> SearchField.IndexAnalyzerName
- SearchField.SearchAnalyzer -> SearchField.SearchAnalyzerName
- SearchableField.SynonymMaps -> SearchableField.SynonymMapNames
- SearchableField.Analyzer -> SearchableField.AnalyzerName
- SearchableField.IndexAnalyzer -> SearchableField.IndexAnalyzerName
- SearchableField.SearchAnalyzer -> SearchableField.SearchAnalyzerName
- Similarity -> SimilarityAlgorithm
- Suggester -> SearchSuggester
- PathHierarchyTokenizerV2 -> PathHierarchyTokenizer
-
Renamed DataSource methods to DataSourceConnection #11693
-
Autocomplete & suggest methods now takes arguments search_text & suggester_name rather than query objects #11747
-
Create_or_updates methods does not support partial updates #11800
-
Renamed AnalyzeRequest to AnalyzeTextOptions #11800
-
Renamed Batch methods #11800
Features
- Add support for synonym maps operations #10830
- Add support for skillset operations #10832
- Add support of indexers operation #10836
- Add helpers for defining searchindex fields #10833
Breaking Changes
SearchIndexClient
renamed toSearchClient
Features
- Added index service client #10324
- Accepted an array of RegexFlags for PatternAnalyzer and PatternTokenizer #10409
Breaking Changes
- Removed
SearchApiKeyCredential
and now usingAzureKeyCredential
from azure.core.credentials as key credential
First release of Azure Search SDK for Python