fix ai search (#119)

* fix ai search

* allow names to be specified for indexer, skillset, and datasource; set defaults based on index name

src/ai/.x/help/search.index.update

@@ -18,6 +18,8 @@ USAGE: ai index update [...]
   AZURE SEARCH DATA SOURCE
     --data-source-connection NAME           (see: ai help search data source connection)
     --blob-container ENDPOINT/NAME          (see: ai help search data source blob container)
+    --indexer-name NAME                     (see: ai help search indexer name)
+    --skillset-name NAME                    (see: ai help search skillset name)
     --id-field NAME                         (see: ai help search id field name)
     --content-field NAME                    (see: ai help search content field name)
     --vector-field NAME                     (see: ai help search vector field name)

src/ai/commands/parsers/search_command_parser.cs

@@ -103,6 +103,7 @@ public CommonSearchNamedValueTokenParsers() : base(
 
             BlobContainerToken.Parser(),
             SearchIndexerDataSourceConnectionNameToken.Parser(),
+            SearchIndexerSkillsetNameToken.Parser(),
             IndexIdFieldNameToken.Parser(),
             IndexContentFieldNameToken.Parser(),
             IndexVectorFieldNameToken.Parser(),

src/ai/commands/search_command.cs

@@ -108,13 +108,15 @@ private void DoIndexUpdate()
                 var embeddingsEndpoint = DemandEmbeddingsEndpointUri(action, command);
                 var embeddingsApiKey = DemandEmbeddingsApiKey(action, command);
                 var embeddingModelDeployment = SearchEmbeddingModelDeploymentNameToken.Data().Demand(_values, action, command, checkConfig: "embedding.model.deployment.name");
-                var dataSourceConnectionName = SearchIndexerDataSourceConnectionNameToken.Data().GetOrDefault(_values, "datasource");
+                var dataSourceConnectionName = SearchIndexerDataSourceConnectionNameToken.Data().GetOrDefault(_values, $"{searchIndexName}-datasource");
+                var skillsetName = SearchIndexerSkillsetNameToken.Data().GetOrDefault(_values, $"{searchIndexName}-skillset");
+                var indexerName = SearchIndexerNameToken.Data().GetOrDefault(_values, $"{searchIndexName}-indexer");
 
                 var idFieldName = IndexIdFieldNameToken.Data().GetOrDefault(_values, "id");
                 var contentFieldName = IndexContentFieldNameToken.Data().GetOrDefault(_values, "content");
                 var vectorFieldName = IndexVectorFieldNameToken.Data().GetOrDefault(_values, "contentVector");
 
-                output = DoIndexUpdateWithAISearch(aiServicesApiKey, searchEndpoint, searchApiKey, embeddingsEndpoint, embeddingModelDeployment, embeddingsApiKey, searchIndexName, dataSourceConnectionName, blobContainer, pattern, idFieldName, contentFieldName, vectorFieldName).Result;
+                output = DoIndexUpdateWithAISearch(aiServicesApiKey, searchEndpoint, searchApiKey, embeddingsEndpoint, embeddingModelDeployment, embeddingsApiKey, searchIndexName, dataSourceConnectionName, blobContainer, pattern, skillsetName, indexerName, idFieldName, contentFieldName, vectorFieldName).Result;
             }
             else if (useSK)
             {
@@ -162,15 +164,15 @@ private string DoIndexUpdateWithGenAi(string subscription, string groupName, str
             return PythonSDKWrapper.UpdateMLIndex(_values, subscription, groupName, projectName, indexName, embeddingModelDeployment, embeddingModelName, dataFiles, externalSourceUrl);
         }
 
-        private async Task<string> DoIndexUpdateWithAISearch(string aiServicesApiKey, string searchEndpoint, string searchApiKey, string embeddingsEndpoint, string embeddingsDeployment, string embeddingsApiKey, string searchIndexName, string dataSourceConnectionName, string blobContainer, string pattern, string idFieldName, string contentFieldName, string vectorFieldName)
+        private async Task<string> DoIndexUpdateWithAISearch(string aiServicesApiKey, string searchEndpoint, string searchApiKey, string embeddingsEndpoint, string embeddingsDeployment, string embeddingsApiKey, string searchIndexName, string dataSourceConnectionName, string blobContainer, string pattern, string skillsetName, string indexerName, string idFieldName, string contentFieldName, string vectorFieldName)
         {
             var (connectionString, containerName) = await UploadFilesToBlobContainer(blobContainer, pattern);
 
             Console.WriteLine("Connecting to Search ...");
             var datasourceIndex = PrepGetSearchIndex(embeddingsEndpoint, embeddingsDeployment, embeddingsApiKey, searchIndexName, idFieldName, contentFieldName, vectorFieldName);
             var dataSource = PrepGetDataSourceConnection(dataSourceConnectionName, connectionString, containerName);
-            var skillset = PrepGetSkillset(aiServicesApiKey, embeddingsEndpoint, embeddingsDeployment, embeddingsApiKey, idFieldName, contentFieldName, vectorFieldName, datasourceIndex);
-            var indexer = PrepGetIndexer(datasourceIndex, dataSource, skillset);
+            var skillset = PrepGetSkillset(skillsetName, aiServicesApiKey, embeddingsEndpoint, embeddingsDeployment, embeddingsApiKey, idFieldName, contentFieldName, vectorFieldName, datasourceIndex);
+            var indexer = PrepGetIndexer(indexerName, datasourceIndex, dataSource, skillset);
 
             Uri endpoint = new Uri(searchEndpoint);
             AzureKeyCredential credential = new AzureKeyCredential(searchApiKey);
@@ -365,7 +367,7 @@ private static SearchIndexerDataSourceConnection PrepGetDataSourceConnection(str
                     new SearchIndexerDataContainer(dataSourceContainerName));
         }
 
-        private static SearchIndexerSkillset PrepGetSkillset(string aiServicesApiKey, string embeddingsEndpoint, string embeddingsDeployment, string embeddingsApiKey, string idFieldName, string contentFieldName, string vectorFieldName, SearchIndex datasourceIndex)
+        private static SearchIndexerSkillset PrepGetSkillset(string skillsetName, string aiServicesApiKey, string embeddingsEndpoint, string embeddingsDeployment, string embeddingsApiKey, string idFieldName, string contentFieldName, string vectorFieldName, SearchIndex datasourceIndex)
         {
             var useOcr = true;
 
@@ -394,11 +396,9 @@ private static SearchIndexerSkillset PrepGetSkillset(string aiServicesApiKey, st
                     InsertPostTag = " "
                 };
 
-
             var splitSkill = new SplitSkill(
                 new List<InputFieldMappingEntry> {
-                    new InputFieldMappingEntry("text") { Source = useOcr ? "/document/mergedText" : "/document/content" },
-                    new InputFieldMappingEntry("languageCode") { Source = "/document/language" }
+                    new InputFieldMappingEntry("text") { Source = useOcr ? "/document/mergedText" : "/document/content" }
                 },
                 new List<OutputFieldMappingEntry> {
                     new OutputFieldMappingEntry("textItems") { TargetName = "pages"}
@@ -415,7 +415,7 @@ private static SearchIndexerSkillset PrepGetSkillset(string aiServicesApiKey, st
                     new InputFieldMappingEntry("text") { Source = "/document/pages/*" }
                 },
                 new List<OutputFieldMappingEntry> {
-                    new OutputFieldMappingEntry("embedding") { TargetName = vectorFieldName}
+                    new OutputFieldMappingEntry("embedding") { TargetName = "vector" }
                 }) {
                     Context = "/document/pages/*",
                     ResourceUri = new Uri(embeddingsEndpoint),
@@ -442,20 +442,20 @@ private static SearchIndexerSkillset PrepGetSkillset(string aiServicesApiKey, st
                 };
 
             var skillset = !string.IsNullOrEmpty(aiServicesApiKey)
-                ? new SearchIndexerSkillset("datasourceskillset", skills) {
+                ? new SearchIndexerSkillset(skillsetName, skills) {
                     CognitiveServicesAccount = new CognitiveServicesAccountKey(aiServicesApiKey),
                     IndexProjections = indexProjections
                 }
-                : new SearchIndexerSkillset("datasourceskillset", skills) {
+                : new SearchIndexerSkillset(skillsetName, skills) {
                     IndexProjections = indexProjections
                 };
 
             return skillset;
         }
 
-        private static SearchIndexer PrepGetIndexer(SearchIndex datasourceIndex, SearchIndexerDataSourceConnection dataSource, SearchIndexerSkillset skillset)
+        private static SearchIndexer PrepGetIndexer(string indexerName, SearchIndex datasourceIndex, SearchIndexerDataSourceConnection dataSource, SearchIndexerSkillset skillset)
         {
-            return new SearchIndexer("datasourceindexer", dataSource.Name, datasourceIndex.Name)
+            return new SearchIndexer(indexerName, dataSource.Name, datasourceIndex.Name)
             {
                 Description = "Data indexer",
                 Schedule = new IndexingSchedule(TimeSpan.FromDays(1))

src/common/details/named_values/tokens/search_indexer_name_token.cs

@@ -0,0 +1,18 @@
+//
+// Copyright (c) Microsoft. All rights reserved.
+// Licensed under the MIT license. See LICENSE.md file in the project root for full license information.
+//
+
+namespace Azure.AI.Details.Common.CLI
+{
+    public class SearchIndexerNameToken
+    {
+        public static NamedValueTokenData Data() => new NamedValueTokenData(_optionName, _fullName, _optionExample, _requiredDisplayName);
+        public static INamedValueTokenParser Parser() => new NamedValueTokenParser(_optionName, _fullName, "011;01", "1");
+
+        private const string _requiredDisplayName = "indexer name";
+        private const string _optionName = "--indexer-name";
+        private const string _optionExample = "NAME";
+        private const string _fullName = "search.indexer.name";
+    }
+}

src/common/details/named_values/tokens/search_indexer_skillset_name_token.cs

@@ -0,0 +1,18 @@
+//
+// Copyright (c) Microsoft. All rights reserved.
+// Licensed under the MIT license. See LICENSE.md file in the project root for full license information.
+//
+
+namespace Azure.AI.Details.Common.CLI
+{
+    public class SearchIndexerSkillsetNameToken
+    {
+        public static NamedValueTokenData Data() => new NamedValueTokenData(_optionName, _fullName, _optionExample, _requiredDisplayName);
+        public static INamedValueTokenParser Parser() => new NamedValueTokenParser(_optionName, _fullName, "0011;0010", "1");
+
+        private const string _requiredDisplayName = "skillset name";
+        private const string _optionName = "--skillset-name";
+        private const string _optionExample = "NAME";
+        private const string _fullName = "search.indexer.skillset.name";
+    }
+}