Feature/Add Tika server support

Testcontainers.sln

@@ -105,6 +105,8 @@ Project("{FAE04EC0-301F-11D3-BF4B-00C04F79EFBC}") = "Testcontainers.ServiceBus",
 EndProject
 Project("{FAE04EC0-301F-11D3-BF4B-00C04F79EFBC}") = "Testcontainers.Sftp", "src\Testcontainers.Sftp\Testcontainers.Sftp.csproj", "{7D5C6816-0DD2-4E13-A585-033B5D3C80D5}"
 EndProject
+Project("{FAE04EC0-301F-11D3-BF4B-00C04F79EFBC}") = "Testcontainers.Tika", "src\Testcontainers.Tika\Testcontainers.Tika.csproj", "{AC084DE2-1857-E200-EF47-8C4ADEB2173D}"
+EndProject
 Project("{FAE04EC0-301F-11D3-BF4B-00C04F79EFBC}") = "Testcontainers.Weaviate", "src\Testcontainers.Weaviate\Testcontainers.Weaviate.csproj", "{68F8600D-24E9-4E03-9E25-5F6EB338EAC1}"
 EndProject
 Project("{FAE04EC0-301F-11D3-BF4B-00C04F79EFBC}") = "Testcontainers.WebDriver", "src\Testcontainers.WebDriver\Testcontainers.WebDriver.csproj", "{64A87DE5-29B0-4A54-9E74-560484D8C7C0}"
@@ -233,14 +235,13 @@ Project("{FAE04EC0-301F-11D3-BF4B-00C04F79EFBC}") = "Testcontainers.WebDriver.Te
 EndProject
 Project("{FAE04EC0-301F-11D3-BF4B-00C04F79EFBC}") = "Testcontainers.Xunit.Tests", "tests\Testcontainers.Xunit.Tests\Testcontainers.Xunit.Tests.csproj", "{E901DF14-6F05-4FC2-825A-3055FAD33561}"
 EndProject
+Project("{FAE04EC0-301F-11D3-BF4B-00C04F79EFBC}") = "Testcontainers.Tika.Tests", "tests\Testcontainers.Tika.Tests\Testcontainers.Tika.Tests.csproj", "{FDD2E9F5-DAAC-2F8F-B8DD-6924CD1D9091}"
+EndProject
 Global
 	GlobalSection(SolutionConfigurationPlatforms) = preSolution
 		Debug|Any CPU = Debug|Any CPU
 		Release|Any CPU = Release|Any CPU
 	EndGlobalSection
-	GlobalSection(SolutionProperties) = preSolution
-		HideSolutionNode = FALSE
-	EndGlobalSection
 	GlobalSection(ProjectConfigurationPlatforms) = postSolution
 		{5365F780-0E6C-41F0-B1B9-7DC34368F80C}.Debug|Any CPU.ActiveCfg = Debug|Any CPU
 		{5365F780-0E6C-41F0-B1B9-7DC34368F80C}.Debug|Any CPU.Build.0 = Debug|Any CPU
@@ -682,6 +683,17 @@ Global
 		{E901DF14-6F05-4FC2-825A-3055FAD33561}.Debug|Any CPU.Build.0 = Debug|Any CPU
 		{E901DF14-6F05-4FC2-825A-3055FAD33561}.Release|Any CPU.ActiveCfg = Release|Any CPU
 		{E901DF14-6F05-4FC2-825A-3055FAD33561}.Release|Any CPU.Build.0 = Release|Any CPU
+		{AC084DE2-1857-E200-EF47-8C4ADEB2173D}.Debug|Any CPU.ActiveCfg = Debug|Any CPU
+		{AC084DE2-1857-E200-EF47-8C4ADEB2173D}.Debug|Any CPU.Build.0 = Debug|Any CPU
+		{AC084DE2-1857-E200-EF47-8C4ADEB2173D}.Release|Any CPU.ActiveCfg = Release|Any CPU
+		{AC084DE2-1857-E200-EF47-8C4ADEB2173D}.Release|Any CPU.Build.0 = Release|Any CPU
+		{FDD2E9F5-DAAC-2F8F-B8DD-6924CD1D9091}.Debug|Any CPU.ActiveCfg = Debug|Any CPU
+		{FDD2E9F5-DAAC-2F8F-B8DD-6924CD1D9091}.Debug|Any CPU.Build.0 = Debug|Any CPU
+		{FDD2E9F5-DAAC-2F8F-B8DD-6924CD1D9091}.Release|Any CPU.ActiveCfg = Release|Any CPU
+		{FDD2E9F5-DAAC-2F8F-B8DD-6924CD1D9091}.Release|Any CPU.Build.0 = Release|Any CPU
+	EndGlobalSection
+	GlobalSection(SolutionProperties) = preSolution
+		HideSolutionNode = FALSE
 	EndGlobalSection
 	GlobalSection(NestedProjects) = preSolution
 		{5365F780-0E6C-41F0-B1B9-7DC34368F80C} = {673F23AE-7694-4BB9-ABD4-136D6C13634E}
@@ -794,5 +806,7 @@ Global
 		{DDB41BC8-5826-4D97-9C5F-001151E3FFD6} = {7164F1FB-7F24-444A-ACD2-2C329C2B3CCF}
 		{EBA72C3B-57D5-43FF-A5B4-3D55B3B6D4C2} = {7164F1FB-7F24-444A-ACD2-2C329C2B3CCF}
 		{E901DF14-6F05-4FC2-825A-3055FAD33561} = {7164F1FB-7F24-444A-ACD2-2C329C2B3CCF}
+		{AC084DE2-1857-E200-EF47-8C4ADEB2173D} = {673F23AE-7694-4BB9-ABD4-136D6C13634E}
+		{FDD2E9F5-DAAC-2F8F-B8DD-6924CD1D9091} = {7164F1FB-7F24-444A-ACD2-2C329C2B3CCF}
 	EndGlobalSection
 EndGlobal

src/Testcontainers.Tika/Testcontainers.Tika.csproj

@@ -0,0 +1,12 @@
+<Project Sdk="Microsoft.NET.Sdk">
+	<PropertyGroup>
+		<TargetFrameworks>net8.0;net9.0;netstandard2.0;netstandard2.1</TargetFrameworks>
+		<LangVersion>latest</LangVersion>
+	</PropertyGroup>
+	<ItemGroup>
+		<PackageReference Include="JetBrains.Annotations" VersionOverride="2023.3.0" PrivateAssets="All" />
+	</ItemGroup>
+	<ItemGroup>
+		<ProjectReference Include="../Testcontainers/Testcontainers.csproj" />
+	</ItemGroup>
+</Project>

src/Testcontainers.Tika/TikaBuilder.cs

@@ -0,0 +1,115 @@
+namespace Testcontainers.Tika;
+
+/// <inheritdoc cref="ContainerBuilder{TBuilderEntity, TContainerEntity, TConfigurationEntity}" />
+[PublicAPI]
+public sealed class TikaBuilder : ContainerBuilder<TikaBuilder, TikaContainer, TikaConfiguration>
+{
+    public const string TikaImage = "apache/tika:3.0.0.0-full";
+    public const ushort TikaHttpPort = 9998;
+
+    /// <summary>
+    /// Initializes a new instance of the <see cref="TikaBuilder" /> class.
+    /// </summary>
+    public TikaBuilder()
+        : this(new TikaConfiguration())
+    {
+        DockerResourceConfiguration = Init().DockerResourceConfiguration;
+    }
+
+    /// <summary>
+    /// Initializes a new instance of the <see cref="TikaBuilder" /> class.
+    /// </summary>
+    /// <param name="resourceConfiguration">The Docker resource configuration.</param>
+    private TikaBuilder(TikaConfiguration resourceConfiguration)
+        : base(resourceConfiguration)
+    {
+        DockerResourceConfiguration = resourceConfiguration;
+    }
+    /// <inheritdoc />
+
+    protected override TikaConfiguration DockerResourceConfiguration { get; }
+
+    /// <summary>
+    /// Sets the Tika server timeout.
+    /// </summary>
+    /// <param name="timeout">The timeout for the server in milliseconds.</param>
+    /// <returns>A configured instance of <see cref="TikaBuilder" />.</returns>
+    public TikaBuilder WithTimeout(int timeout)
+    {
+        return Merge(DockerResourceConfiguration, new TikaConfiguration(timeout: timeout))
+            .WithEnvironment("TIKA_TIMEOUT", timeout.ToString());
+    }
+
+    public override TikaContainer Build()
+    {
+        Validate();
+        return new TikaContainer(DockerResourceConfiguration);
+    }
+
+    protected override TikaBuilder Init()
+    {
+        return base.Init()
+            .WithImage(TikaImage)
+            .WithPortBinding(TikaHttpPort, true)
+            .WithWaitStrategy(Wait.ForUnixContainer().AddCustomWaitStrategy(new WaitUntil()));
+    }
+
+    protected override TikaBuilder Clone(IResourceConfiguration<CreateContainerParameters> resourceConfiguration)
+    {
+        return Merge(DockerResourceConfiguration, new TikaConfiguration(resourceConfiguration));
+    }
+
+    protected override TikaBuilder Clone(IContainerConfiguration resourceConfiguration)
+    {
+        return Merge(DockerResourceConfiguration, new TikaConfiguration(resourceConfiguration));
+    }
+
+    protected override TikaBuilder Merge(TikaConfiguration oldValue, TikaConfiguration newValue)
+    {
+        return new TikaBuilder(new TikaConfiguration(oldValue, newValue));
+    }
+
+    private sealed class WaitUntil : IWaitUntil
+    {
+        private const string HealthCheckPath = "tika";
+        private const int MaxRetryAttempts = 10;
+        private const int DelayInMilliseconds = 1000;
+
+        /// <summary>
+        /// Waits until the Tika server is available by checking the health check endpoint.
+        /// </summary>
+        /// <param name="container">The container instance to check.</param>
+        /// <returns>
+        /// A task that represents the asynchronous operation. The task result contains a boolean indicating whether the Tika server is available.
+        /// </returns>
+        /// <remarks>
+        /// This method sends HTTP GET requests to the Tika server's health check endpoint and retries up to a maximum number of times if the server is not available.
+        /// </remarks>
+        public async Task<bool> UntilAsync(DotNet.Testcontainers.Containers.IContainer container)
+        {
+            string endpoint = $"http://{container.Hostname}:{container.GetMappedPublicPort(TikaBuilder.TikaHttpPort)}/{HealthCheckPath}";
+
+            using var client = new HttpClient();
+
+            for (int i = 0; i < MaxRetryAttempts; i++)
+            {
+                try
+                {
+                    var response = await client.GetAsync(endpoint);
+
+                    response.EnsureSuccessStatusCode();
+                    string responseContent = await response.Content.ReadAsStringAsync(); // This is Tika Server (Apache Tika 3.0.0). Please PUT, volendo si può fare questo check
+                    return true;
+                }
+                catch
+                {
+                    // Ignore exceptions and retry
+                }
+
+                await Task.Delay(DelayInMilliseconds);
+            }
+
+            return false;
+        }
+    }
+}

src/Testcontainers.Tika/TikaConfiguration.cs

@@ -0,0 +1,54 @@
+namespace Testcontainers.Tika;
+
+/// <inheritdoc cref="ContainerConfiguration" />
+[PublicAPI]
+public sealed class TikaConfiguration : ContainerConfiguration
+{
+  /// <summary>
+  /// Initializes a new instance of the <see cref="TikaConfiguration" /> class.
+  /// </summary>
+  /// <param name="timeout">The timeout for the Tika server.</param>
+  public TikaConfiguration(int timeout = 30000)
+  {
+    Timeout = timeout;
+  }
+
+  /// <summary>
+  /// Initializes a new instance of the <see cref="TikaConfiguration" /> class.
+  /// </summary>
+  /// <param name="resourceConfiguration">The Docker resource configuration.</param>
+  public TikaConfiguration(IResourceConfiguration<CreateContainerParameters> resourceConfiguration)
+      : base(resourceConfiguration)
+  {
+    // Passes the configuration upwards to the base implementations to create an updated immutable copy.
+  }
+
+  /// <summary>
+  /// Initializes a new instance of the <see cref="TikaConfiguration" /> class.
+  /// </summary>
+  /// <param name="resourceConfiguration">The Docker resource configuration.</param>
+  public TikaConfiguration(IContainerConfiguration resourceConfiguration)
+      : base(resourceConfiguration)
+  {
+    // Passes the configuration upwards to the base implementations to create an updated immutable copy.
+  }
+
+  /// <summary>
+  /// Initializes a new instance of the <see cref="TikaConfiguration" /> class,
+  /// combining properties from two existing configurations.
+  /// </summary>
+  /// <param name="oldValue">The previous configuration values.</param>
+  /// <param name="newValue">The new configuration values to merge with the old ones.</param>
+  public TikaConfiguration(TikaConfiguration oldValue, TikaConfiguration newValue)
+      : base(oldValue, newValue)
+  {
+    // Combine values manually
+    Timeout = BuildConfiguration.Combine(oldValue.Timeout, newValue.Timeout);
+  }
+
+
+  /// <summary>
+  /// Gets the Tika server timeout.
+  /// </summary>
+  public int Timeout { get; }
+}

src/Testcontainers.Tika/TikaContainer.cs

@@ -0,0 +1,28 @@
+namespace Testcontainers.Tika;
+
+/// <inheritdoc cref="DockerContainer" />
+[PublicAPI]
+public sealed class TikaContainer : DockerContainer
+{
+  private readonly TikaConfiguration _configuration;
+
+  /// <summary>
+  /// Initializes a new instance of the <see cref="TikaContainer" /> class.
+  /// </summary>
+  /// <param name="configuration">The container configuration.</param>
+  public TikaContainer(TikaConfiguration configuration)
+      : base(configuration)
+  {
+    _configuration = configuration;
+  }
+
+  /// <summary>
+  /// Gets the Tika connection string.
+  /// </summary>
+  /// <returns>The Tika connection string.</returns>
+  public string GetConnectionString()
+  {
+    var endpoint = new UriBuilder(Uri.UriSchemeHttp, Hostname, GetMappedPublicPort(TikaBuilder.TikaHttpPort));
+    return endpoint.ToString();
+  }
+}

src/Testcontainers.Tika/Usings.cs

@@ -0,0 +1,8 @@
+global using System;
+global using Docker.DotNet.Models;
+global using DotNet.Testcontainers.Builders;
+global using DotNet.Testcontainers.Configurations;
+global using DotNet.Testcontainers.Containers;
+global using JetBrains.Annotations;
+global using System.Threading.Tasks;
+global using System.Net.Http;

tests/Testcontainers.Tika.Tests/Testcontainers.Tika.Tests.csproj

@@ -0,0 +1,19 @@
+<Project Sdk="Microsoft.NET.Sdk">
+  <PropertyGroup>
+    <TargetFrameworks>net9.0</TargetFrameworks>
+    <IsPackable>false</IsPackable>
+    <IsPublishable>false</IsPublishable>
+    <Configurations>Debug;Release</Configurations>
+  </PropertyGroup>
+  <ItemGroup>
+    <PackageReference Include="Microsoft.NET.Test.Sdk" />
+    <PackageReference Include="coverlet.collector" />
+    <PackageReference Include="xunit.runner.visualstudio" />
+    <PackageReference Include="xunit" />
+  </ItemGroup>
+  <ItemGroup>
+    <ProjectReference Include="../Testcontainers.Commons/Testcontainers.Commons.csproj" />
+    <ProjectReference Include="../../src/Testcontainers.Tika/Testcontainers.Tika.csproj" />
+  </ItemGroup>
+</Project>
+

tests/Testcontainers.Tika.Tests/TikaContainerTest.cs

@@ -0,0 +1,49 @@
+namespace Testcontainers.Tika.Tests;
+
+public sealed class TikaContainerTests : IAsyncLifetime
+{
+  private readonly TikaContainer _tikaContainer = new TikaBuilder().Build();
+
+  public Task InitializeAsync()
+  {
+    return _tikaContainer.StartAsync();
+  }
+
+  public Task DisposeAsync()
+  {
+    return _tikaContainer.DisposeAsync().AsTask();
+  }
+
+  [Fact]
+  [Trait(nameof(DockerCli.DockerPlatform), nameof(DockerCli.DockerPlatform.Linux))]
+  public async Task GetConnectionStringReturnsValidUrl()
+  {
+    // When
+    var connectionString = await Task.Run(() => _tikaContainer.GetConnectionString());
+
+    // Then
+    Assert.False(string.IsNullOrEmpty(connectionString));
+    Assert.StartsWith("http://", connectionString, StringComparison.OrdinalIgnoreCase);
+  }
+
+  [Fact]
+  [Trait(nameof(DockerCli.DockerPlatform), nameof(DockerCli.DockerPlatform.Linux))]
+  public async Task TikaHealthCheckShouldBeSuccessful()
+  {
+    {
+      // Given
+      var httpClient = new HttpClient();
+      var connectionString = await Task.Run(() => _tikaContainer.GetConnectionString());
+      var requestUrl = $"{connectionString}tika";
+
+      // When
+      var response = await httpClient.GetAsync(requestUrl);
+
+      // Then
+      response.EnsureSuccessStatusCode();
+      var content = await response.Content.ReadAsStringAsync();
+      Assert.False(string.IsNullOrEmpty(content));
+      Assert.StartsWith("This is Tika Server", content);
+    }
+  }
+}

tests/Testcontainers.Tika.Tests/Usings.cs

@@ -0,0 +1,5 @@
+global using DotNet.Testcontainers.Commons;
+global using System;
+global using System.Net.Http;
+global using System.Threading.Tasks;
+global using Xunit;