- Added support for vector search
- Added support for semantic search and reranking
- New
SearchIndexerSkill
sEntityLinkingSkill
EntityRecognitionSkillV3
PIIDetectionSkill
SentimentSkillV3
- Minor type changes
- All nullable optional properties can no longer be assigned
null
. Useundefined
instead. - Updated specific property types for enhanced precision, replacing
any
with more defined types.
- All nullable optional properties can no longer be assigned
- Narrowed some enum property types to the respective string literal union
BlobIndexerDataToExtract
BlobIndexerImageAction
BlobIndexerParsingMode
BlobIndexerPDFTextRotationAlgorithm
CustomEntityLookupSkillLanguage
EntityCategory
EntityRecognitionSkillLanguage
ImageAnalysisSkillLanguage
ImageDetail
IndexerExecutionEnvironment
KeyPhraseExtractionSkillLanguage
OcrSkillLanguage
RegexFlags
SearchIndexerDataSourceType
SentimentSkillLanguage
SplitSkillLanguage
TextSplitMode
TextTranslationSkillLanguage
VisualFeature
- Removed
KnownLexicalAnalyzerName
as a duplicate ofKnownAnalyzerNames
- Removed
KnownCharFilterName
as a duplicate ofKnownCharFilterNames
- Removed
KnownTokenFilterName
as a duplicate ofKnownTokenFilterNames
- Removed
SearchRequest
as a duplicate ofSearchRequestOptions
- Added type narrowing for response types based on
select
ed fields - For types that are generic on a model type, the model type is now constrained to extend
object
- Deprecated
apiVersion
in favor ofserviceVersion
in clients:SearchClient
SearchIndexClient
SearchIndexerClient
- This GA release introduces bug fixes, convenience improvements, and select features from prior beta releases. Except for the features listed in these release notes, no preview features are being pulled into this release.
- Fix some
SearchIndex
fields being improperly serialized.#27238
- Fix ISO8601 deserialization. #25802
- Fix all clients adding one or more duplicate user agents. #26298
- Fix
serializerOptions
andonResponse
options for SearchClient methods. #26327 - Fix discarded user-defined
onResponse
callback. #24479 - Fix type error on
select
statement with nested fields. #26407
- Add deprecation warning to
EntityRecognitionSkill
andSentimentSkill
#25686
- Addressed an issue where
similarity
settings on indexes wouldn't populate correctly. See #23811
- Enabled national cloud support for Azure Search SDK. Please refer #22887 for further details.
- Support for TokenCredential has been added. With this addition, the Search SDK supports authentication via AAD.
- Converted the complex fields correctly within the Search Fields. Please refer #16489 for more details.
- Fixed the typos
anayzerName
toanalyzerName
inconvertFieldsToPublic
method ofserviceUtils.ts
. - Fixed the issue with the presence of recursive structure while uploading documents. Please refer #15656 for further details.
-
This GA release includes AAD with multi-cloud support and all the bug fixes since the last 11.2.0 GA release. Other preview features and breaking changes from the 11.3.0-beta.1 to 11.3.0-beta.8 releases are not included in this GA release.
-
Add deprecation warning to
EntityRecognitionSkill
andSentimentSkill
#25686
- Added
core-http-compat
dependency - Enabled national cloud support for Azure Search SDK. Please refer #22887 for further details.
- Added new APIs
createAlias
,createOrUpdateAlias
,deleteAlias
,getAlias
&listAliases
operations to theSearchIndexClient
.
- Added new type of SearchIndexer skill -
AzureMachineLearningSkill
. Please refer #20183 for further details.
- Deprecated
SearchClientOptions.apiVersion
in favor ofSearchClientOptions.serviceVersion
.apiVersion
will continue to be supported in version 11.x; however, customers are encouraged to migrate toserviceVersion
instead which is more consistent with the rest of our client libraries.
- Added
semanticConfiguration
property toSearchRequest
object. - Several new languages are added to the
KnownQueryLanguage
object. - Added
semanticSettings
property toSearchIndex
object.
- Renamed
IndexerCurrentState
property toIndexerState
property. - Renamed
AllDocsInitialChangeTrackingState
property toAllDocumentsInitialChangeTrackingState
property. - Renamed
AllDocsFinalChangeTrackingState
property toAllDocumentsFinalChangeTrackingState
property. - Renamed
ResetDocsInitialChangeTrackingState
property toResetDocumentsInitialChangeTrackingState
property. - Renamed
ResetDocsFinalChangeTrackingState
property toResetDocumentsFinalChangeTrackingState
property. - Renamed
SkillNames
property toResetSkillsOptions
(with aSkillNames
property) - Renamed
ResetDocs
method toResetDocuments
in the SDK client.
- Fixed the issue with the presence of recursive structure while uploading documents. Please refer #15656 for further details.
- Added properties
currentState
&statusDetail
to theIndexerExecutionResult
object. - Added operations
resetDocs
&resetSkills
to theSearchIndexerClient
class.
- Changed property
ignoreResetRequirements
toskipIndexerResetRequirementForCache
inCreateorUpdateDataSourceConnectionOptions
,CreateorUpdateIndexerOptions
&CreateOrUpdateSkillsetOptions
objects. - Changed the type of
answers
property fromAnswers
toQueryAnswerType
, in theSearchRequest
object. - Changed the type of
captions
property fromCaptions
toQueryCaptionType
, in theSearchRequest
object. - Changed the type of
speller
property fromSpeller
toQuerySpellerType
, in theSearchRequest
object. - Changed the underlying dependency of the SDK from
core-http
tocore-rest-pipeline
package.
- Added a new property
normalizerName
to theAnalyzeRequest
object. - Added 2 new properties
captions
&semanticFields
to theSearchRequestOptions
object.
- Added a new parameter
ignoreResetRequirements
for thecreateOrUpdate
operation of Data Sources. - Added new parameters
ignoreResetRequirements
&disableCacheReprocessingChangeDetection
for thecreateOrUpdate
operation of Indexers and Skillsets.
- Converted the complex fields correctly within the Search Fields. Please refer #16489 for more details.
- Fixed the typos
anayzerName
toanalyzerName
andnormalizerNames
tonormalizerName
inconvertFieldsToPublic
method ofserviceUtils.ts
.
- Removed the
| null
from the definitions of the optional values. Please refer #16694 for more details.
-
With the dropping of support for Node.js versions that are no longer in LTS, the dependency on
@types/node
has been updated to version 12. Read our support policy for more details. -
Updated our internal core package dependencies to their latest versions in order to add support for Opentelemetry 1.0.0 which is compatible with the latest versions of our other client libraries.
-
Changed TS compilation target to ES2017 in order to produce smaller bundles and use more native platform features
-
Regenerated the search SDK with the latest swaggers that includes the following changes:
- Support for
TokenCredential
has been added. With this addition, the Search SDK supports authentication via AAD. - Identity types -
SearchIndexerDataNoneIdentity
&SearchIndexerDataUserAssignedIdentity
have been added. - The following new skills have been added:
- SentimentSkill(V3)
- EntityLinkingSkill(V3)
- EntityRecognitionSkill(V3)
- PIIDetectionSkill
- A new property
lineEnding
has been added to the skillOcrSkill
.
- Support for
The list of changes in 11.2.0 since 11.1.0 & 11.2.0-beta.2 are provided below:
Changes since 11.1.0
- Added support for Knowledge Store feature through the new
SearchIndexerKnowledgeStore
in theSearchIndexerSkillset
object. - The
skillsetCounter
property inServiceCounters
object has been made optional. - Added Support for new datasource
adlsgen2
. Please refer #14620 for further details. - Added Support for new skills such as
CustomEntityLookupSkill
,DocumentExtractionSkill
, etc. Please refer #14620 for further details.
Changes since 11.2.0-beta.2
- Removed Support for Semantic Search and introduced new properties in
SearchOptions
,SearchRequest
,SearchResult
andSearchDocumentsResult
objects. - Removed Support for normalizers
LexicalNormalizer
&CustomNormalizer
. Please refer #14620 for further details.
- Added Support for Semantic Search and introduced new properties in
SearchOptions
,SearchRequest
,SearchResult
andSearchDocumentsResult
objects. - Added support for Knowledge Store feature through the new
SearchIndexerKnowledgeStore
in theSearchIndexerSkillset
object. - The
skillsetCounter
property inServiceCounters
object has been made optional.
- Added Support for new skills such as
CustomEntityLookupSkill
,DocumentExtractionSkill
, etc. Please refer #14620 for further details. - Added Support for new datasource
adlsgen2
. Please refer #14620 for further details. - Added Support for normalizers
LexicalNormalizer
&CustomNormalizer
. Please refer #14620 for further details.
- The list of changes in 11.1.0 since 11.0.3 & 11.1.0-beta.2 are provided below:
Changes since 11.0.3
- Added Batching ability to the search SDK. The
SearchIndexingBufferedSender
class enables the user to perform indexing documents in batch mode. There are several user configurable properties such asautoFlush
,flushWindowInMs
,throttlingDelayInMs
, etc. - The types
BlobIndexerDataToExtract
,BlobIndexerImageAction
,BlobIndexerParsingMode
,BlobIndexerPDFTextRotationAlgorithm
have changed from union of string literals to string. This is to support the service definition for these types to be extensible enums. The documentation on methods that use these have been updated with known values that can be used for these types.
Changes since 11.1.0-beta.2
- [Breaking] Hidden the constructor of
SearchIndexingBufferedSender
and made it to an interface. Please refer #11785 for further details. - Added
encryptionKey
property toSearchIndexerDataSource
,SearchIndexer
andSearchIndexerSkillSet
objects. Please refer #12275 for further details. - Added user configurable properties such as
flushWindowInMs
,initialBatchActionCount
,maxRetries
, etc to theSearchIndexingBufferedSenderOptions
object. Please refer #12297 for further details.
- Added Batching ability to the search SDK. Please refer #11544 for further details.
- Regenerated the search SDK using the latest swaggers. Please refer #11533 for further details.
- Fixed the testcases to ensure success in CI. Please refer #11518 for further details.
- Added more test cases for odata scenario. Please refer #11321 for further details.
- Fixed the parented tracing span in the searchClient.search API. Please refer #10442 for further details.
- Fixed the samples in the Readme File. Please refer #10383 for further details.
- [BugFix] Set the correct continuation token in the search documrnts API. Please refer #10146 for further details.
- Set
ConnectionString
value to<unchanged>
inSearchIndexerDataSourceConnection
, if the value is not set by the user. - [Breaking] In Suggest API & Search API return values, a new property called
document
is introduced. All user-defined fields are moved inside thisdocument
property. - [Breaking] In
analyzeText
API, thetext
parameter is moved from method level to insideoptions
bag. - [Breaking] In
search
API,includeTotalResultCount
property is renamed toincludeTotalCount
. - [Breaking] In
ServiceCounters
, theskillsetCounter
property has been removed. - [Breaking] Modified the names of several properties. Please refer #9321 for a detailed list of renames.
- [Breaking] Changed
ListIndexes
operation to a pageable operation. - [Breaking] Added
onlyIfUnchanged
parameter for CreateOrUpdate and Delete operations. - [Breaking] Removed
$select
property for the List operations. - [Breaking] Refactored
SearchServiceClient
and split it toSearchIndexClient
andSearchIndexerClient
and changedSearchIndexClient
class toSearchClient
class. - [Breaking] Moved non-optional parameters from options bag.
- [Breaking] Renamed
countDocuments
method togetDocumentsCount
method. - [Breaking] In
search
method, moved thesearchText
parameter from the options bag to method parameter. - [Breaking] In
indexDocuments
method, the options parameter is renamed toIndexDocumentsOptions
. - [Breaking] Modified
deleteDocuments
method to get documents as a parameter. - [Breaking] In
getIndexStatistics
method, renamed the return type fromGetIndexStatisticsResult
toSearchIndexStatistics
. - [Breaking] In
getServiceStatistics
method, renamed the return type fromServiceStatistics
toSearchServiceStatistics
. - [Breaking] Modified
DataSource
model name toDataSourceConnection
. Changed all references in the method names, parameters, etc. - [Breaking] Renamed
SimpleDataType
model toSearchFieldDataType
model. - [Breaking] Modified the names of several models & parameters. Please refer #8984, #9037 and #8383 for a detailed list of renames.
- Added separate methods for getting just names such as
listIndexesNames
,listSynonymMapsNames
, etc. - Added
getSearchClient
method to theSearchIndexClient
class.
- Added support for Indexers API (Create, Get, List, etc.)
- Added support for Datasources API.(Create, Get, List, etc.)
- Fixed a bug where GeographyPoints were serialized incorrectly causing latitude and longitude to be flipped.
- [Breaking] Package renamed to
@azure/search-documents
and version number reset to1.0.0-preview.2
. - Support for index management operations using the
SearchServiceClient
. - [Breaking]
indexDocuments
onSearchIndexClient
now takes anIndexDocumentsBatch
object instead of a raw action array. This new type helps compose an array of actions to be performed on the index. - [Breaking] In
SearchIndexClient
, removed optionsmergeIfExists
anduploadIfNotExists
onuploadDocuments
andmergeDocuments
and replaced them with new helpermergeOrUploadDocuments
. - The type
IndexAction
was renamed toIndexDocumentsAction
. - [Breaking] Removed
SearchApiKeyCredential
and replaced withAzureKeyCredential
. - [Breaking] Search results accessed
byPage
now have an opaquecontinuationToken
in place ofnextLink
andnextPageParameters
.
- Initial implementation of the data-plane Cognitive Search Client. The version number starts at 11 to align with client libraries in other languages.
- This first preview has support for document operations on an index, such as querying and document management.