fdup/FindDuplicates/ListCommand.cs

134 lines
4.6 KiB
C#
Raw Normal View History

2024-04-16 19:33:16 +00:00
using System.Collections.Concurrent;
using System.Diagnostics;
2024-04-16 19:15:07 +00:00
using System.Security.Cryptography;
using System.Text;
using Spectre.Console;
using Spectre.Console.Cli;
namespace FindDuplicates;
internal sealed class ListCommand : AsyncCommand<ListSettings>
{
private readonly ConcurrentDictionary<string, List<FileInfo>> _fileHashMap = new();
2024-04-16 19:15:07 +00:00
public override async Task<int> ExecuteAsync(CommandContext context, ListSettings settings)
{
var inputDirectory = new DirectoryInfo(settings.InputPath);
if (!inputDirectory.Exists)
{
AnsiConsole.MarkupLine($"[red]{inputDirectory} does not exist![/]");
return -1;
}
AnsiConsole.MarkupLineInterpolated($"Searching [cyan]{inputDirectory.FullName}[/]");
AnsiConsole.MarkupLine($"Recursive mode is {(settings.Recursive ? "[green]ON" : "[red]OFF")}[/]");
await AnsiConsole.Status()
.StartAsync("Waiting to hash files...", DoHashWaitAsync)
.ConfigureAwait(false);
2024-04-16 19:15:07 +00:00
AnsiConsole.WriteLine();
int duplicates = 0;
foreach ((string hash, List<FileInfo> files) in _fileHashMap)
{
int fileCount = files.Count;
if (fileCount > 1)
{
duplicates += fileCount;
AnsiConsole.MarkupLineInterpolated($"Found [cyan]{fileCount}[/] identical files");
AnsiConsole.MarkupLineInterpolated($"SHA512 [green]{hash}[/]:");
foreach (FileInfo file in files)
AnsiConsole.MarkupLineInterpolated($"- {file.FullName}");
AnsiConsole.WriteLine();
}
}
if (duplicates == 0)
AnsiConsole.MarkupLine("[green]No duplicates found![/]");
else
AnsiConsole.MarkupLineInterpolated($"[yellow]Found [cyan]{duplicates}[/] duplicates![/]");
return 0;
async Task DoHashWaitAsync(StatusContext ctx)
{
await WaitForHashCompletionAsync(settings, inputDirectory, ctx);
}
2024-04-16 19:15:07 +00:00
}
private async Task WaitForHashCompletionAsync(ListSettings settings,
DirectoryInfo inputDirectory,
StatusContext ctx)
2024-04-16 19:15:07 +00:00
{
var tasks = new List<Task>();
SearchDuplicates(inputDirectory, settings, tasks);
await Task.Run(() =>
{
int incompleteTasks;
do
{
incompleteTasks = tasks.Count(t => !t.IsCompleted);
ctx.Status($"Waiting to hash {incompleteTasks} {(incompleteTasks == 1 ? "file" : "files")}...");
ctx.Refresh();
} while (tasks.Count > 0 && incompleteTasks > 0);
ctx.Status("Hash complete");
}).ConfigureAwait(false);
}
private void SearchDuplicates(DirectoryInfo inputDirectory, ListSettings settings, ICollection<Task> tasks)
{
2024-04-16 19:15:07 +00:00
var directoryStack = new Stack<DirectoryInfo>([inputDirectory]);
while (directoryStack.Count > 0)
{
DirectoryInfo currentDirectory = directoryStack.Pop();
string relativePath = Path.GetRelativePath(inputDirectory.FullName, currentDirectory.FullName);
if (relativePath != ".")
AnsiConsole.MarkupLineInterpolated($"Searching [cyan]{relativePath}[/]");
if (settings.Recursive)
{
foreach (DirectoryInfo childDirectory in currentDirectory.EnumerateDirectories())
directoryStack.Push(childDirectory);
}
foreach (FileInfo file in currentDirectory.EnumerateFiles())
{
string relativeFilePath = Path.GetRelativePath(inputDirectory.FullName, file.FullName);
AnsiConsole.MarkupLineInterpolated($"Checking hash for [cyan]{relativeFilePath}[/]");
tasks.Add(Task.Run(() => ProcessFile(file)));
}
}
}
private void ProcessFile(FileInfo file)
{
Span<byte> buffer = stackalloc byte[64];
using FileStream stream = file.OpenRead();
using BufferedStream bufferedStream = new BufferedStream(stream, 1048576 /* 1MB */);
SHA512.HashData(bufferedStream, buffer);
string hash = ByteSpanToString(buffer);
Trace.WriteLine($"{file.FullName}: {hash}");
if (!_fileHashMap.TryGetValue(hash, out List<FileInfo>? cache))
_fileHashMap[hash] = cache = new List<FileInfo>();
lock (cache)
cache.Add(file);
}
private static string ByteSpanToString(ReadOnlySpan<byte> buffer)
{
var builder = new StringBuilder();
foreach (byte b in buffer)
builder.Append($"{b:X2}");
return builder.ToString();
}
}