Make sure empty and null strings are not allowed

Author: Genbox

Src/FastData.Tests/FastDataGeneratorTests.cs

@@ -123,6 +123,46 @@ public void Generate_ThrowOnInvalidType()
         Assert.Throws<InvalidOperationException>(() => FastDataGenerator.Generate([DateTime.Now, DateTime.UtcNow], new FastDataConfig(StructureType.Array), new DummyGenerator()));
     }
 
+    [Fact]
+    public void Generate_StringKeys_EmptyString_ThrowsFriendlyException()
+    {
+        FastDataConfig config = new FastDataConfig();
+        Assert.Throws<InvalidOperationException>(() => FastDataGenerator.Generate([""], config, new DummyGenerator()));
+    }
+
+    [Fact]
+    public void Generate_StringKeys_Null_ThrowsFriendlyException()
+    {
+        string[] keys = [null!];
+        FastDataConfig config = new FastDataConfig();
+        Assert.Throws<InvalidOperationException>(() => FastDataGenerator.Generate(keys, config, new DummyGenerator()));
+    }
+
+    [Fact]
+    public void Generate_IgnoreCase_DeduplicatesCaseInsensitive()
+    {
+        FastDataConfig config = new FastDataConfig(StructureType.Array)
+        {
+            IgnoreCase = true,
+            DeduplicationMode = DeduplicationMode.SortThrowOnDup
+        };
+
+        Assert.Throws<InvalidOperationException>(() => FastDataGenerator.Generate(["abc", "ABC"], config, new DummyGenerator()));
+    }
+
+    [Fact(Skip = "Known issue: all-negative sparse inputs can select Elias-Fano and throw during construction.")]
+    public void Generate_Auto_AllNegativeSparse_DoesNotThrow()
+    {
+        int[] keys = Enumerable.Range(1, 1000).Select(static x => -x * 20).ToArray();
+        FastDataConfig config = new FastDataConfig();
+
+        ContextCaptureGenerator generator = new ContextCaptureGenerator();
+        Exception? error = Record.Exception(() => FastDataGenerator.Generate(keys, config, generator));
+
+        Assert.Null(error);
+        Assert.False(generator.Context is EliasFanoContext<int>);
+    }
+
     [Fact]
     public void Generate_Overloads_String_Work()
     {

Src/FastData/FastDataGenerator.cs

@@ -76,21 +76,35 @@ private static string GenerateStringInternal<TValue>(ReadOnlyMemory<string> keys
         if (!values.IsEmpty && keys.Length != values.Length)
             throw new InvalidOperationException("The number of values does not match the number of keys.");
 
+        ReadOnlySpan<string> keySpan = keys.Span;
+
+        for (int i = 0; i < keySpan.Length; i++)
+        {
+            string? key = keySpan[i];
+
+            if (key is null)
+                throw new InvalidOperationException("Keys cannot contain null values.");
+
+            if (key.Length == 0)
+                throw new InvalidOperationException("Keys cannot contain empty strings.");
+        }
+
         factory ??= NullLoggerFactory.Instance;
 
         ILogger logger = factory.CreateLogger(typeof(FastDataGenerator));
         LogUserStructureType(logger, fdCfg.StructureType);
 
         int oldCount = keys.Length;
 
-        DeduplicateKeys(fdCfg, keys, values, StringComparer.Ordinal, StringComparer.Ordinal, out keys, out values, out int newCount);
+        StringComparer comparer = fdCfg.IgnoreCase ? StringComparer.OrdinalIgnoreCase : StringComparer.Ordinal;
+        DeduplicateKeys(fdCfg, keys, values, comparer, comparer, out keys, out values, out int newCount);
 
         if (oldCount == newCount)
             LogNumberOfKeys(logger, newCount);
         else
             LogNumberOfUniqueKeys(logger, oldCount, newCount);
 
-        ReadOnlySpan<string> keySpan = keys.Span;
+        keySpan = keys.Span;
 
         const KeyType keyType = KeyType.String;
         LogKeyType(logger, keyType);