Users/oakeredolu/newpipdetector

src/Microsoft.ComponentDetection.Detectors/pip/SimplePipComponentDetector.cs

@@ -0,0 +1,133 @@
+namespace Microsoft.ComponentDetection.Detectors.Pip;
+
+using System;
+using System.Collections.Generic;
+using System.Linq;
+using System.Reactive.Linq;
+using System.Threading.Tasks;
+using Microsoft.ComponentDetection.Contracts;
+using Microsoft.ComponentDetection.Contracts.Internal;
+using Microsoft.ComponentDetection.Contracts.TypedComponent;
+using Microsoft.Extensions.Logging;
+
+public class SimplePipComponentDetector : FileComponentDetector, IDefaultOffComponentDetector
+{
+    private readonly IPythonCommandService pythonCommandService;
+    private readonly ISimplePythonResolver pythonResolver;
+
+    public SimplePipComponentDetector(
+        IComponentStreamEnumerableFactory componentStreamEnumerableFactory,
+        IObservableDirectoryWalkerFactory walkerFactory,
+        IPythonCommandService pythonCommandService,
+        ISimplePythonResolver pythonResolver,
+        ILogger<SimplePipComponentDetector> logger)
+    {
+        this.ComponentStreamEnumerableFactory = componentStreamEnumerableFactory;
+        this.Scanner = walkerFactory;
+        this.pythonCommandService = pythonCommandService;
+        this.pythonResolver = pythonResolver;
+        this.Logger = logger;
+    }
+
+    public override string Id => "SimplePip";
+
+    public override IList<string> SearchPatterns => new List<string> { "setup.py", "requirements.txt" };
+
+    public override IEnumerable<string> Categories => new List<string> { "Python" };
+
+    public override IEnumerable<ComponentType> SupportedComponentTypes { get; } = new[] { ComponentType.Pip };
+
+    public override int Version { get; } = 1;
+
+    protected override async Task<IObservable<ProcessRequest>> OnPrepareDetectionAsync(IObservable<ProcessRequest> processRequests, IDictionary<string, string> detectorArgs)
+    {
+        this.CurrentScanRequest.DetectorArgs.TryGetValue("Pip.PythonExePath", out var pythonExePath);
+        if (!await this.pythonCommandService.PythonExistsAsync(pythonExePath))
+        {
+            this.Logger.LogInformation($"No python found on system. Python detection will not run.");
+
+            return Enumerable.Empty<ProcessRequest>().ToObservable();
+        }
+
+        return processRequests;
+    }
+
+    protected override async Task OnFileFoundAsync(ProcessRequest processRequest, IDictionary<string, string> detectorArgs)
+    {
+        this.CurrentScanRequest.DetectorArgs.TryGetValue("Pip.PythonExePath", out var pythonExePath);
+        var singleFileComponentRecorder = processRequest.SingleFileComponentRecorder;
+        var file = processRequest.ComponentStream;
+
+        try
+        {
+            var initialPackages = await this.pythonCommandService.ParseFileAsync(file.Location, pythonExePath);
+            var listedPackage = initialPackages.Where(tuple => tuple.PackageString != null)
+                .Select(tuple => tuple.PackageString)
+                .Where(x => !string.IsNullOrWhiteSpace(x))
+                .Select(x => new PipDependencySpecification(x))
+                .Where(x => !x.PackageIsUnsafe())
+                .ToList();
+
+            var roots = await this.pythonResolver.ResolveRootsAsync(singleFileComponentRecorder, listedPackage);
+
+            RecordComponents(
+                singleFileComponentRecorder,
+                roots);
+
+            initialPackages.Where(tuple => tuple.Component != null)
+                .Select(tuple => new DetectedComponent(tuple.Component))
+                .ToList()
+                .ForEach(gitComponent => singleFileComponentRecorder.RegisterUsage(gitComponent, isExplicitReferencedDependency: true));
+        }
+        catch (Exception e)
+        {
+            this.Logger.LogError(e, "Error while parsing pip components in {File}", file.Location);
+        }
+    }
+
+    private static void RecordComponents(
+        ISingleFileComponentRecorder recorder,
+        IList<PipGraphNode> roots)
+    {
+        var nonRoots = new Queue<(DetectedComponent, PipGraphNode)>();
+
+        var explicitRoots = roots.Select(a => a.Value).ToHashSet();
+
+        foreach (var root in roots)
+        {
+            var rootDetectedComponent = new DetectedComponent(root.Value);
+
+            recorder.RegisterUsage(
+                rootDetectedComponent,
+                isExplicitReferencedDependency: true);
+
+            foreach (var child in root.Children)
+            {
+                nonRoots.Enqueue((rootDetectedComponent, child));
+            }
+        }
+
+        var registeredIds = new HashSet<string>();
+
+        while (nonRoots.Count > 0)
+        {
+            var (parent, item) = nonRoots.Dequeue();
+
+            var detectedComponent = new DetectedComponent(item.Value);
+
+            recorder.RegisterUsage(
+                detectedComponent,
+                parentComponentId: parent.Component.Id);
+
+            if (!registeredIds.Contains(detectedComponent.Component.Id))
+            {
+                foreach (var child in item.Children)
+                {
+                    nonRoots.Enqueue((detectedComponent, child));
+                }
+
+                registeredIds.Add(detectedComponent.Component.Id);
+            }
+        }
+    }
+}

src/Microsoft.ComponentDetection.Detectors/pip/SimplePythonResolver.cs

@@ -14,7 +14,7 @@ namespace Microsoft.ComponentDetection.Detectors.Pip;
 
 public class SimplePythonResolver : ISimplePythonResolver
 {
-    private static readonly Regex VersionRegex = new(@"-(\d+\.\d+(\.\d)*)(.tar|-)", RegexOptions.Compiled);
+    private static readonly Regex VersionRegex = new(@"-(\d+(\.)\w+((\+|\.)\w*)*)(.tar|-)", RegexOptions.Compiled);
 
     private readonly ISimplePyPiClient simplePypiClient;
     private readonly ILogger<SimplePythonResolver> logger;
@@ -100,10 +100,18 @@ public async Task<IList<PipGraphNode>> ResolveRootsAsync(ISingleFileComponentRec
 
             var simplePythonProject = await this.simplePypiClient.GetSimplePypiProjectAsync(rootPackage);
 
-            if (simplePythonProject != null && simplePythonProject.Files.Any())
+            if (simplePythonProject == null || !simplePythonProject.Files.Any())
             {
-                var pythonProject = this.ConvertSimplePypiProjectToSortedDictionary(simplePythonProject, rootPackage);
+                this.logger.LogWarning(
+                    "Root dependency {RootPackageName} not found on pypi. Skipping package.",
+                    rootPackage.Name);
+                singleFileComponentRecorder.RegisterPackageParseFailure(rootPackage.Name);
+            }
+
+            var pythonProject = this.ConvertSimplePypiProjectToSortedDictionary(simplePythonProject, rootPackage);
 
+            if (pythonProject.Keys.Any())
+            {
                 state.ValidVersionMap[rootPackage.Name] = pythonProject;
 
                 // Grab the latest version as our candidate version
@@ -121,7 +129,7 @@ public async Task<IList<PipGraphNode>> ResolveRootsAsync(ISingleFileComponentRec
             else
             {
                 this.logger.LogWarning(
-                    "Root dependency {RootPackageName} not found on pypi. Skipping package.",
+                    "Unable to resolve package: {RootPackageName} gotten from pypi possibly due to invalid versions. Skipping package.",
                     rootPackage.Name);
                 singleFileComponentRecorder.RegisterPackageParseFailure(rootPackage.Name);
             }
@@ -169,9 +177,17 @@ private async Task<IList<PipGraphNode>> ProcessQueueAsync(ISingleFileComponentRe
                     // We haven't encountered this package before, so let's fetch it and find a candidate
                     var newProject = await this.simplePypiClient.GetSimplePypiProjectAsync(dependencyNode);
 
-                    if (newProject != null && newProject.Files.Any())
+                    if (newProject == null || !newProject.Files.Any())
+                    {
+                        this.logger.LogWarning(
+                             "Dependency Package {DependencyName} not found in Pypi. Skipping package",
+                             dependencyNode.Name);
+                        singleFileComponentRecorder.RegisterPackageParseFailure(dependencyNode.Name);
+                    }
+
+                    var result = this.ConvertSimplePypiProjectToSortedDictionary(newProject, dependencyNode);
+                    if (result.Keys.Any())
                     {
-                        var result = this.ConvertSimplePypiProjectToSortedDictionary(newProject, dependencyNode);
                         state.ValidVersionMap[dependencyNode.Name] = result;
                         var candidateVersion = state.ValidVersionMap[dependencyNode.Name].Keys.Any()
                             ? state.ValidVersionMap[dependencyNode.Name].Keys.Last() : null;
@@ -183,7 +199,7 @@ private async Task<IList<PipGraphNode>> ProcessQueueAsync(ISingleFileComponentRe
                     else
                     {
                         this.logger.LogWarning(
-                            "Dependency Package {DependencyName} not found in Pypi. Skipping package",
+                            "Unable to resolve dependency package {DependencyName} gotten from pypi possibly due to invalid versions. Skipping package",
                             dependencyNode.Name);
                         singleFileComponentRecorder.RegisterPackageParseFailure(dependencyNode.Name);
                     }
@@ -202,7 +218,7 @@ private async Task<IList<PipGraphNode>> ProcessQueueAsync(ISingleFileComponentRe
     /// <returns> Returns a SortedDictionary of PythonProjectReleases. </returns>
     private SortedDictionary<string, IList<PythonProjectRelease>> ConvertSimplePypiProjectToSortedDictionary(SimplePypiProject simplePypiProject, PipDependencySpecification spec)
     {
-        var sortedProjectVersions = new SortedDictionary<string, IList<PythonProjectRelease>>();
+        var sortedProjectVersions = new SortedDictionary<string, IList<PythonProjectRelease>>(new PythonVersionComparer());
         foreach (var file in simplePypiProject.Files)
         {
             try

src/Microsoft.ComponentDetection.Orchestrator/Experiments/Configs/SimplePipExperiment.cs

@@ -0,0 +1,22 @@
+namespace Microsoft.ComponentDetection.Orchestrator.Experiments.Configs;
+
+using Microsoft.ComponentDetection.Contracts;
+using Microsoft.ComponentDetection.Detectors.Pip;
+
+/// <summary>
+/// Validating the <see cref="SimplePipComponentDetector"/>.
+/// </summary>
+public class SimplePipExperiment : IExperimentConfiguration
+{
+    /// <inheritdoc />
+    public string Name => "NewPipDetector";
+
+    /// <inheritdoc />
+    public bool IsInControlGroup(IComponentDetector componentDetector) => componentDetector is PipComponentDetector;
+
+    /// <inheritdoc />
+    public bool IsInExperimentGroup(IComponentDetector componentDetector) => componentDetector is SimplePipComponentDetector;
+
+    /// <inheritdoc />
+    public bool ShouldRecord(IComponentDetector componentDetector, int numComponents) => true;
+}

src/Microsoft.ComponentDetection.Orchestrator/Extensions/ServiceCollectionExtensions.cs

@@ -72,6 +72,7 @@ public static IServiceCollection AddComponentDetection(this IServiceCollection s
         services.AddSingleton<IExperimentProcessor, DefaultExperimentProcessor>();
         services.AddSingleton<IExperimentConfiguration, NewNugetExperiment>();
         services.AddSingleton<IExperimentConfiguration, NpmLockfile3Experiment>();
+        services.AddSingleton<IExperimentConfiguration, SimplePipExperiment>();
 
         // Detectors
         // CocoaPods
@@ -118,6 +119,7 @@ public static IServiceCollection AddComponentDetection(this IServiceCollection s
         services.AddSingleton<IPythonResolver, PythonResolver>();
         services.AddSingleton<ISimplePythonResolver, SimplePythonResolver>();
         services.AddSingleton<IComponentDetector, PipComponentDetector>();
+        services.AddSingleton<IComponentDetector, SimplePipComponentDetector>();
 
         // pnpm
         services.AddSingleton<IComponentDetector, PnpmComponentDetector>();