Add flag to enable collisionless media downloads

DiscordChatExporter.Cli/Commands/Base/ExportCommandBase.cs

@@ -101,6 +101,13 @@ public abstract class ExportCommandBase : DiscordCommandBase
     )]
     public bool ShouldReuseAssets { get; init; } = false;
 
+    [CommandOption(
+        "new-media-paths",
+        Description = "Saves media with a new file naming convention to prevent collisions. "
+            + "Duplicate media may be downloaded if consecutive runs use different values for this flag."
+    )]
+    public bool ShouldUseNewMediaFilePaths { get; init; } = false;
+
     [CommandOption(
         "media-dir",
         Description = "Download assets to this directory. "
@@ -154,6 +161,12 @@ public abstract class ExportCommandBase : DiscordCommandBase
             throw new CommandException("Option --reuse-media cannot be used without --media.");
         }
 
+        // New media file path can only be enabled if the download assets option is set
+        if (ShouldUseNewMediaFilePaths && !ShouldDownloadAssets)
+        {
+            throw new CommandException("Option --new-media-paths cannot be used without --media.");
+        }
+
         // Assets directory can only be specified if the download assets option is set
         if (!string.IsNullOrWhiteSpace(AssetsDirPath) && !ShouldDownloadAssets)
         {
@@ -270,6 +283,7 @@ public abstract class ExportCommandBase : DiscordCommandBase
                                         ShouldFormatMarkdown,
                                         ShouldDownloadAssets,
                                         ShouldReuseAssets,
+                                        ShouldUseNewMediaFilePaths,
                                         Locale,
                                         IsUtcNormalizationEnabled
                                     );

DiscordChatExporter.Core/Exporting/ExportAssetDownloader.cs

@@ -13,7 +13,11 @@ using DiscordChatExporter.Core.Utils.Extensions;
 
 namespace DiscordChatExporter.Core.Exporting;
 
-internal partial class ExportAssetDownloader(string workingDirPath, bool reuse)
+internal partial class ExportAssetDownloader(
+    string workingDirPath,
+    bool reuse,
+    bool useNewMediaFilePaths
+)
 {
     private static readonly AsyncKeyedLocker<string> Locker = new();
 
@@ -25,8 +29,10 @@ internal partial class ExportAssetDownloader(string workingDirPath, bool reuse)
         CancellationToken cancellationToken = default
     )
     {
-        var fileName = GetFileNameFromUrl(url);
+        var localFilePath = useNewMediaFilePaths
-        var filePath = Path.Combine(workingDirPath, fileName);
+            ? GetFilePathFromUrl(url)
+            : GetFileNameFromUrlLegacy(url);
+        var filePath = Path.Combine(workingDirPath, localFilePath);
 
         using var _ = await Locker.LockAsync(filePath, cancellationToken);
 
@@ -44,6 +50,9 @@ internal partial class ExportAssetDownloader(string workingDirPath, bool reuse)
             {
                 // Download the file
                 using var response = await Http.Client.GetAsync(url, innerCancellationToken);
+                var directory = Path.GetDirectoryName(filePath);
+                if (directory != null)
+                    Directory.CreateDirectory(directory);
                 await using var output = File.Create(filePath);
                 await response.Content.CopyToAsync(output, innerCancellationToken);
             },
@@ -54,6 +63,37 @@ internal partial class ExportAssetDownloader(string workingDirPath, bool reuse)
     }
 }
 
+internal partial class ExportAssetDownloader
+{
+    private static string GetFilePathFromUrl(string url)
+    {
+        var uri = new Uri(url);
+
+        // Try to extract the file name from URL
+        var pathAndFileName = Regex.Match(uri.AbsolutePath, @"/(.+)/([^?]*)");
+        var path = pathAndFileName.Groups[1].Value;
+        var fileName = pathAndFileName.Groups[2].Value;
+
+        // If this isn't a Discord CDN URL, save the file to the `media/external` folder.
+        if (!string.Equals(uri.Host, "cdn.discordapp.com", StringComparison.OrdinalIgnoreCase))
+        {
+            File.AppendAllTextAsync("external_urls.txt", $"{url}\n");
+            return $"external/{uri.Host}{uri.AbsolutePath}";
+        }
+
+        // If it is a Discord URL, we're guaranteed to have matches for these groups. <see cref="ImageCdn"/>
+        if (string.IsNullOrWhiteSpace(path) || string.IsNullOrWhiteSpace(fileName))
+            throw new Exception("Invalid Discord URL shape");
+
+        // If there is a size parameter, add it as the final folder in the path.
+        // This prevents multiple sizes of an avatar, for example, from overwriting each other.
+        var sizeParam = HttpUtility.ParseQueryString(uri.Query)["size"];
+        var sizePathSegment = sizeParam != null ? $"{sizeParam}px/" : "";
+
+        return $"{path}/{sizePathSegment}{Path.EscapeFileName(fileName)}";
+    }
+}
+
 internal partial class ExportAssetDownloader
 {
     private static string GetUrlHash(string url)
@@ -80,7 +120,7 @@ internal partial class ExportAssetDownloader
             .Truncate(5);
     }
 
-    private static string GetFileNameFromUrl(string url)
+    private static string GetFileNameFromUrlLegacy(string url)
     {
         var urlHash = GetUrlHash(url);
 

DiscordChatExporter.Core/Exporting/ExportContext.cs

@@ -21,7 +21,8 @@ internal class ExportContext(DiscordClient discord, ExportRequest request)
 
     private readonly ExportAssetDownloader _assetDownloader = new(
         request.AssetsDirPath,
-        request.ShouldReuseAssets
+        request.ShouldReuseAssets,
+        request.ShouldUseNewMediaFilePaths
     );
 
     public DiscordClient Discord { get; } = discord;

DiscordChatExporter.Core/Exporting/ExportRequest.cs

@@ -39,6 +39,8 @@ public partial class ExportRequest
 
     public bool ShouldReuseAssets { get; }
 
+    public bool ShouldUseNewMediaFilePaths { get; }
+
     public string? Locale { get; }
 
     public CultureInfo? CultureInfo { get; }
@@ -58,6 +60,7 @@ public partial class ExportRequest
         bool shouldFormatMarkdown,
         bool shouldDownloadAssets,
         bool shouldReuseAssets,
+        bool shouldUseNewMediaFilePaths,
         string? locale,
         bool isUtcNormalizationEnabled
     )
@@ -72,6 +75,7 @@ public partial class ExportRequest
         ShouldFormatMarkdown = shouldFormatMarkdown;
         ShouldDownloadAssets = shouldDownloadAssets;
         ShouldReuseAssets = shouldReuseAssets;
+        ShouldUseNewMediaFilePaths = shouldUseNewMediaFilePaths;
         Locale = locale;
         IsUtcNormalizationEnabled = isUtcNormalizationEnabled;
 

DiscordChatExporter.Gui/Services/SettingsService.cs

@@ -66,6 +66,9 @@ public partial class SettingsService()
     [ObservableProperty]
     public partial bool LastShouldReuseAssets { get; set; }
 
+    [ObservableProperty]
+    public partial bool LastShouldUseNewMediaFilePaths { get; set; }
+
     [ObservableProperty]
     public partial string? LastAssetsDirPath { get; set; }
 

DiscordChatExporter.Gui/ViewModels/Components/DashboardViewModel.cs

@@ -276,6 +276,7 @@ public partial class DashboardViewModel : ViewModelBase
                             dialog.ShouldFormatMarkdown,
                             dialog.ShouldDownloadAssets,
                             dialog.ShouldReuseAssets,
+                            dialog.ShouldUseNewMediaFilePaths,
                             _settingsService.Locale,
                             _settingsService.IsUtcNormalizationEnabled
                         );

DiscordChatExporter.Gui/ViewModels/Dialogs/ExportSetupViewModel.cs

@@ -67,6 +67,9 @@ public partial class ExportSetupViewModel(
     [ObservableProperty]
     public partial bool ShouldReuseAssets { get; set; }
 
+    [ObservableProperty]
+    public partial bool ShouldUseNewMediaFilePaths { get; set; }
+
     [ObservableProperty]
     public partial string? AssetsDirPath { get; set; }
 
@@ -105,6 +108,7 @@ public partial class ExportSetupViewModel(
         ShouldFormatMarkdown = settingsService.LastShouldFormatMarkdown;
         ShouldDownloadAssets = settingsService.LastShouldDownloadAssets;
         ShouldReuseAssets = settingsService.LastShouldReuseAssets;
+        ShouldUseNewMediaFilePaths = settingsService.LastShouldUseNewMediaFilePaths;
         AssetsDirPath = settingsService.LastAssetsDirPath;
 
         // Show the "advanced options" section by default if any
@@ -183,6 +187,7 @@ public partial class ExportSetupViewModel(
         settingsService.LastShouldFormatMarkdown = ShouldFormatMarkdown;
         settingsService.LastShouldDownloadAssets = ShouldDownloadAssets;
         settingsService.LastShouldReuseAssets = ShouldReuseAssets;
+        settingsService.LastShouldUseNewMediaFilePaths = ShouldUseNewMediaFilePaths;
         settingsService.LastAssetsDirPath = AssetsDirPath;
 
         Close(true);

DiscordChatExporter.Gui/Views/Dialogs/ExportSetupView.axaml

@@ -274,6 +274,16 @@
                             <ToggleSwitch DockPanel.Dock="Right" IsChecked="{Binding ShouldReuseAssets}" />
                         </DockPanel>
 
+                        <!--  Use new media file paths  -->
+                        <DockPanel
+                            Margin="16,8"
+                            IsEnabled="{Binding ShouldDownloadAssets}"
+                            LastChildFill="False"
+                            ToolTip.Tip="Uses a new organization to save media files to avoid data loss. If you change this setting from run to run, duplicate media may be downloaded.">
+                            <TextBlock DockPanel.Dock="Left" Text="Use new media file paths" />
+                            <ToggleSwitch DockPanel.Dock="Right" IsChecked="{Binding ShouldUseNewMediaFilePaths}" />
+                        </DockPanel>
+
                         <!--  Assets path  -->
                         <TextBox
                             Margin="16,8"