diff --git a/Extensions/Cosmos/Cosmos.DataTransfer.CosmosExtension.UnitTests/CosmosDataSinkExtensionTests.cs b/Extensions/Cosmos/Cosmos.DataTransfer.CosmosExtension.UnitTests/CosmosDataSinkExtensionTests.cs index a8c5da0..8eccfc1 100644 --- a/Extensions/Cosmos/Cosmos.DataTransfer.CosmosExtension.UnitTests/CosmosDataSinkExtensionTests.cs +++ b/Extensions/Cosmos/Cosmos.DataTransfer.CosmosExtension.UnitTests/CosmosDataSinkExtensionTests.cs @@ -1,5 +1,6 @@ using Cosmos.DataTransfer.Interfaces; using System.Dynamic; +using Microsoft.VisualStudio.TestTools.UnitTesting; namespace Cosmos.DataTransfer.CosmosExtension.UnitTests { @@ -56,6 +57,83 @@ public void BuildDynamicObjectTree_WithNestedArrays_WorksCorrectly() Assert.AreEqual("sub2-1", secondSubArray[0].id); } + [TestMethod] + public void BuildDynamicObjectTree_WithAnyCaseIds_UsesSourceIdValue() + { + var numeric = Random.Shared.Next(); + var lower = Guid.NewGuid().ToString(); + var upper = Guid.NewGuid().ToString(); + var mixed = Guid.NewGuid().ToString(); + var reversed = Guid.NewGuid().ToString(); + var item = new CosmosDictionaryDataItem(new Dictionary() + { + { "id", numeric }, + }); + + dynamic obj = item.BuildDynamicObjectTree(requireStringId: true, preserveMixedCaseIds: false)!; + Assert.AreEqual(numeric.ToString(), obj.id); + + item = new CosmosDictionaryDataItem(new Dictionary() + { + { "id", lower }, + }); + + obj = item.BuildDynamicObjectTree(requireStringId: true, preserveMixedCaseIds: false)!; + Assert.AreEqual(lower, obj.id); + + item = new CosmosDictionaryDataItem(new Dictionary() + { + { "ID", upper }, + }); + obj = item.BuildDynamicObjectTree(requireStringId: true, preserveMixedCaseIds: false)!; + Assert.AreEqual(upper, obj.id); + + item = new CosmosDictionaryDataItem(new Dictionary() + { + { "Id", mixed }, + }); + obj = item.BuildDynamicObjectTree(requireStringId: true, preserveMixedCaseIds: false)!; + Assert.AreEqual(mixed, obj.id); + + item = new CosmosDictionaryDataItem(new Dictionary() + { + { "iD", reversed }, + }); + obj = item.BuildDynamicObjectTree(requireStringId: true, preserveMixedCaseIds: false)!; + Assert.AreEqual(reversed, obj.id); + } + + [TestMethod] + public void BuildDynamicObjectTree_WithPreservedMixedCaseIds_PassesThroughSourceValues() + { + var id = Random.Shared.Next(); + var upper = Guid.NewGuid().ToString(); + var mixed = Guid.NewGuid().ToString(); + var item = new CosmosDictionaryDataItem(new Dictionary() + { + { "id", id }, + { "ID", upper }, + { "Id", mixed } + }); + + dynamic obj = item.BuildDynamicObjectTree(requireStringId: true, preserveMixedCaseIds: true)!; + Assert.AreEqual(id.ToString(), obj.id); + Assert.AreEqual(upper, obj.ID); + Assert.AreEqual(mixed, obj.Id); + + item = new CosmosDictionaryDataItem(new Dictionary() + { + { "ID", upper }, + { "Id", mixed } + }); + obj = item.BuildDynamicObjectTree(requireStringId: true, preserveMixedCaseIds: true)!; + Assert.AreEqual(upper, obj.ID); + Assert.AreEqual(mixed, obj.Id); + string? cosmosId = obj.id; + Assert.IsNotNull(cosmosId); + Assert.IsFalse(string.IsNullOrWhiteSpace(cosmosId)); + } + [TestMethod] public void BuildDynamicObjectTree_WithIgnoredNulls_ExcludesNullFields() { diff --git a/Extensions/Cosmos/Cosmos.DataTransfer.CosmosExtension/CosmosDataSinkExtension.cs b/Extensions/Cosmos/Cosmos.DataTransfer.CosmosExtension/CosmosDataSinkExtension.cs index b8ca140..30b8f7b 100644 --- a/Extensions/Cosmos/Cosmos.DataTransfer.CosmosExtension/CosmosDataSinkExtension.cs +++ b/Extensions/Cosmos/Cosmos.DataTransfer.CosmosExtension/CosmosDataSinkExtension.cs @@ -89,11 +89,14 @@ void ReportCount(int i) addedCount += i; if (addedCount % 500 == 0) { - logger.LogInformation("{AddedCount} records added after {TotalSeconds}s", addedCount, $"{timer.ElapsedMilliseconds / 1000.0:F2}"); + logger.LogInformation("{AddedCount} records added after {TotalSeconds}s ({AddRate} records/s)", addedCount, $"{timer.ElapsedMilliseconds / 1000.0:F2}", $"{(int)(addedCount / (timer.ElapsedMilliseconds / 1000.0))}"); } } - var convertedObjects = dataItems.Select(di => di.BuildDynamicObjectTree(requireStringId: true, ignoreNullValues: settings.IgnoreNullValues)).Where(o => o != null).OfType(); + var convertedObjects = dataItems + .Select(di => di.BuildDynamicObjectTree(requireStringId: true, ignoreNullValues: settings.IgnoreNullValues, preserveMixedCaseIds: settings.PreserveMixedCaseIds)) + .Where(o => o != null) + .OfType(); var batches = convertedObjects.Buffer(settings.BatchSize); var retry = GetRetryPolicy(settings.MaxRetryCount, settings.InitialRetryDurationMs); await foreach (var batch in batches.WithCancellation(cancellationToken)) @@ -111,7 +114,7 @@ void ReportCount(int i) throw new Exception($"Only {addedCount} of {inputCount} records were added to Cosmos"); } - logger.LogInformation("Added {AddedCount} total records in {TotalSeconds}s", addedCount, $"{timer.ElapsedMilliseconds / 1000.0:F2}"); + logger.LogInformation("Added {AddedCount} total records in {TotalSeconds}s ({AddRate} records/s)", addedCount, $"{timer.ElapsedMilliseconds / 1000.0:F2}", $"{(int)(addedCount / (timer.ElapsedMilliseconds / 1000.0))}"); } private static AsyncRetryPolicy GetRetryPolicy(int maxRetryCount, int initialRetryDuration) diff --git a/Extensions/Cosmos/Cosmos.DataTransfer.CosmosExtension/CosmosSinkSettings.cs b/Extensions/Cosmos/Cosmos.DataTransfer.CosmosExtension/CosmosSinkSettings.cs index a17a73d..7fbe318 100644 --- a/Extensions/Cosmos/Cosmos.DataTransfer.CosmosExtension/CosmosSinkSettings.cs +++ b/Extensions/Cosmos/Cosmos.DataTransfer.CosmosExtension/CosmosSinkSettings.cs @@ -14,6 +14,7 @@ public class CosmosSinkSettings : CosmosSettingsBase, IDataExtensionSettings public bool UseAutoscaleForCreatedContainer { get; set; } = true; public bool IsServerlessAccount { get; set; } = false; public bool UseSharedThroughput { get; set; } = false; + public bool PreserveMixedCaseIds { get; set; } = false; public DataWriteMode WriteMode { get; set; } = DataWriteMode.Insert; public bool IgnoreNullValues { get; set; } = false; public List? PartitionKeyPaths { get; set; } diff --git a/Extensions/Cosmos/README.md b/Extensions/Cosmos/README.md index 2c8883d..b3b9de2 100644 --- a/Extensions/Cosmos/README.md +++ b/Extensions/Cosmos/README.md @@ -44,7 +44,7 @@ Or with RBAC: } ``` -Sink requires an additional `PartitionKeyPath` parameter which is used when creating the container if it does not exist. To use hierarchical partition keys, instead use the `PartitionKeyPaths` setting to supply an array of up to 3 paths. It also supports an optional `RecreateContainer` parameter (`false` by default) to delete and then recreate the container to ensure only newly imported data is present. The optional `BatchSize` parameter (100 by default) sets the number of items to accumulate before inserting. `ConnectionMode` can be set to either `Gateway` (default) or `Direct` to control how the client connects to the CosmosDB service. For situations where a container is created as part of the transfer operation `CreatedContainerMaxThroughput` (in RUs) and `UseAutoscaleForCreatedContainer` provide the initial throughput settings which will be in effect when executing the transfer. To instead use shared throughput that has been provisioned at the database level, set the `UseSharedThroughput` parameter to `true`. The optional `WriteMode` parameter specifies the type of data write to use: `InsertStream`, `Insert`, `UpsertStream`, or `Upsert`. The `IsServerlessAccount` parameter specifies whether the target account uses Serverless instead of Provisioned throughput, which affects the way containers are created. Additional parameters allow changing the behavior of the Cosmos client appropriate to your environment. The `IgnoreNullValues` parameter allows for excluding fields with null values when writing to Cosmos DB. +Sink requires an additional `PartitionKeyPath` parameter which is used when creating the container if it does not exist. To use hierarchical partition keys, instead use the `PartitionKeyPaths` setting to supply an array of up to 3 paths. It also supports an optional `RecreateContainer` parameter (`false` by default) to delete and then recreate the container to ensure only newly imported data is present. The optional `BatchSize` parameter (100 by default) sets the number of items to accumulate before inserting. `ConnectionMode` can be set to either `Gateway` (default) or `Direct` to control how the client connects to the CosmosDB service. For situations where a container is created as part of the transfer operation `CreatedContainerMaxThroughput` (in RUs) and `UseAutoscaleForCreatedContainer` provide the initial throughput settings which will be in effect when executing the transfer. To instead use shared throughput that has been provisioned at the database level, set the `UseSharedThroughput` parameter to `true`. The optional `WriteMode` parameter specifies the type of data write to use: `InsertStream`, `Insert`, `UpsertStream`, or `Upsert`. The `IsServerlessAccount` parameter specifies whether the target account uses Serverless instead of Provisioned throughput, which affects the way containers are created. Additional parameters allow changing the behavior of the Cosmos client appropriate to your environment. The `PreserveMixedCaseIds` parameter (`false` by default) ignores differently cased `id` fields and writes them through without modification, while generating a separate lowercased `id` field as required by Cosmos. The `IgnoreNullValues` parameter allows for excluding fields with null values when writing to Cosmos DB. ### Sink @@ -62,6 +62,7 @@ Sink requires an additional `PartitionKeyPath` parameter which is used when crea "CreatedContainerMaxThroughput": 1000, "UseAutoscaleForCreatedContainer": true, "WriteMode": "InsertStream", + "PreserveMixedCaseIds": false, "IgnoreNullValues": false, "IsServerlessAccount": false, "UseSharedThroughput": false diff --git a/Interfaces/Cosmos.DataTransfer.Interfaces/DataItemExtensions.cs b/Interfaces/Cosmos.DataTransfer.Interfaces/DataItemExtensions.cs index 90c87ad..7e0f4b9 100644 --- a/Interfaces/Cosmos.DataTransfer.Interfaces/DataItemExtensions.cs +++ b/Interfaces/Cosmos.DataTransfer.Interfaces/DataItemExtensions.cs @@ -10,9 +10,10 @@ public static class DataItemExtensions /// /// If true, adds a new GUID "id" field to any top level items where one is not already present. /// If true, excludes fields containing null values from output. + /// If true, disregards differently cased "id" fields for purposes of required "id" and passes them through. /// A dynamic object containing the entire data structure. /// The returned ExpandoObject can be used directly as an IDictionary. - public static ExpandoObject? BuildDynamicObjectTree(this IDataItem? source, bool requireStringId = false, bool ignoreNullValues = false) + public static ExpandoObject? BuildDynamicObjectTree(this IDataItem? source, bool requireStringId = false, bool ignoreNullValues = false, bool preserveMixedCaseIds = false) { if (source == null) { @@ -21,20 +22,25 @@ public static class DataItemExtensions var fields = source.GetFieldNames().ToList(); var item = new ExpandoObject(); - + /* * If the item contains a lowercase id field, we can take it as is. - * If we have an uppercase Id or ID field, but no lowercase id, we will rename it to id. + * If we have an uppercase Id or ID field, but no lowercase id, we will rename it to id, unless `preserveMixedCaseIds` is set to true. + * If `preserveMixedCaseIds` is set to true, any differently cased "id" fields will be passed through as normal properties with no casing change and a separate "id" will be generated. * Then it can be used i.e. as CosmosDB primary key, when `requireStringId` is set to true. */ var containsLowercaseIdField = fields.Contains("id", StringComparer.CurrentCulture); var containsAnyIdField = fields.Contains("id", StringComparer.CurrentCultureIgnoreCase); - - if (requireStringId && !containsAnyIdField) + + if (requireStringId) { - item.TryAdd("id", Guid.NewGuid().ToString()); + bool mismatchedIdCasing = preserveMixedCaseIds && !containsLowercaseIdField; + if (!containsAnyIdField || mismatchedIdCasing) + { + item.TryAdd("id", Guid.NewGuid().ToString()); + } } - + foreach (string field in fields) { object? value = source.GetValue(field); @@ -44,10 +50,24 @@ public static class DataItemExtensions } var fieldName = field; - if (string.Equals(field, "id", StringComparison.CurrentCultureIgnoreCase) && requireStringId && !containsLowercaseIdField) + if (requireStringId && string.Equals(field, "id", StringComparison.CurrentCultureIgnoreCase)) { - value = value?.ToString(); - fieldName = "id"; + if (preserveMixedCaseIds) + { + if (string.Equals(field, "id", StringComparison.CurrentCulture)) + { + value = value?.ToString(); + } + } + else if (!containsLowercaseIdField) + { + value = value?.ToString(); + fieldName = "id"; + } + else + { + value = value?.ToString(); + } } else if (value is IDataItem child) {