using AIStudio.Chat; using AIStudio.Components; using AIStudio.Tools.Databases; using AIStudio.Tools.Services; using Microsoft.AspNetCore.Components; namespace AIStudio.Dialogs; /// /// Check how your file will be loaded. /// public partial class DocumentCheckDialog : MSGComponentBase { [CascadingParameter] private IMudDialogInstance MudDialog { get; set; } = null!; [Parameter] public FileAttachment? Document { get; set; } private void Close() => this.MudDialog.Cancel(); [Parameter] public string FileContent { get; set; } = string.Empty; /// /// Attaches the files the user drops onto this dialog, and answers which of them it attached. /// /// /// Null when our caller has no list of attachments to add to, which is the case for the prompt /// guide preview of the Prompt Optimizer. This dialog then shows its document and nothing else, /// exactly as it always did. /// [Parameter] public Func, Task>>? AttachPaths { get; set; } /// /// Decides, at the moment a drop arrives, whether attaching is possible right now. /// /// /// Asked rather than passed as a value, because the answer changes while this dialog is open: /// dropping a media file starts a transcription, and nothing else may be attached until that /// one is through. /// [Parameter] public Func? IsAttachingUnavailable { get; set; } /// /// Keeps what this dialog read of a document which was never sent, so that sending it does not /// read the file again, see FileAttachment.ProvenanceId. /// /// /// Gets the document, the text read from it, and the state of its file before it was read, so /// that a file which changed afterward is read again; never throws. Only the chat hands /// one over, for the attachments of the message being written: a sent message has its texts /// already, and elsewhere, nothing is sent later. /// [Parameter] public Func? KeepReadText { get; set; } /// /// The document we show right now. It starts out as the one we were opened with and changes /// whenever the user drops another file onto this dialog. /// /// /// Kept in a field rather than read from the parameter: the dialog fragment is rendered again /// with the parameters captured when it was opened, whenever something about the dialog stack /// changes. That happens in the middle of a drop, because attaching may open the Pandoc dialog /// or ask the user about a media file -- reading the parameter would undo the switch right /// after it was made. /// private FileAttachment? document; /// /// The content of the document we show, either handed to us by our caller or read by us. /// private string fileContent = string.Empty; /// /// How many characters we show at most. Rendering a huge document costs us a large Markdown /// syntax tree and an equally large render tree. This dialog answers the question of how we /// read the file, though — the beginning of the document is enough for that, and the AI still /// receives the entire content. /// private const int PREVIEW_CHARACTER_LIMIT = 200_000; /// /// Set when reading the file failed, so the dialog shows the reason instead of empty content. /// private string? loadFailureMessage; /// /// Ends the extraction when this dialog is gone, or when another document took the place of /// the one being read, before that file was read completely. /// private CancellationTokenSource extractionCancellation = new(); /// /// Numbers the loads, so that a load can tell whether it still owns this dialog. /// /// /// Cancelling ends the waiting, not the code behind it: what follows every await of an /// abandoned load runs regardless. Without this number, its final block would clear the loading /// state of the load which replaced it, and the new document would never leave its skeletons. /// private int loadGeneration; /// /// True once this dialog was disposed. The extraction runs across awaits, so it may return /// long after the user closed the dialog — it must not touch this component afterwards. /// private bool isDisposed; /// /// True while we extract the file content. Reading happens after the first render, so the /// dialog can tell the user that it is working instead of showing an empty document. /// private bool isLoadingContent; /// /// True when the content is the text the AI received from the document when it was sent, rather /// than what the file holds now, see FileAttachment.ProvenanceId. /// private bool showsStoredText; /// /// Whether the document we show is no longer at its place. /// private bool IsFileMissing => this.document is not null && !this.document.Exists; [Inject] private DatabaseClientProvider DatabaseClientProvider { get; init; } = null!; [Inject] private RustService RustService { get; init; } = null!; [Inject] private ILogger Logger { get; init; } = null!; [Inject] private PandocAvailabilityService PandocAvailability { get; init; } = null!; protected override async Task OnInitializedAsync() { this.document = this.Document; this.fileContent = this.FileContent; this.isLoadingContent = this.NeedsExtraction(); await base.OnInitializedAsync(); } protected override async Task OnAfterRenderAsync(bool firstRender) { if (!firstRender) return; if (this.document is null) { this.Logger.LogWarning("Document check dialog opened without a valid file path."); return; } await this.LoadDocumentContentAsync(); } /// /// Whether the document we show has to be read before we can show anything of it. Images are /// shown as they are, a missing file shows its own message unless the text the AI received from /// it is stored, and content a caller already handed us is reused instead of being extracted a /// second time. /// private bool NeedsExtraction() => this.document is not null && !this.document.IsImage && (this.document.ProvenanceId is not null || this.document.Exists) && string.IsNullOrWhiteSpace(this.fileContent); /// /// Reads the content of the document we show and puts it into the preview. /// /// /// Runs after a render, so the user sees that we are working instead of an empty document. It /// is called for the document this dialog was opened with, and again for every file the user /// drops onto it. /// private async Task LoadDocumentContentAsync() { if (this.document is null || !this.isLoadingContent) return; // // A drop may arrive while we are still reading the file before it. We number this load and // end the previous one, so that what is left of it recognizes that this dialog has moved on: // var generation = ++this.loadGeneration; var documentToLoad = this.document; var previousCancellation = this.extractionCancellation; this.extractionCancellation = new(); var cancellationToken = this.extractionCancellation.Token; await previousCancellation.CancelAsync(); previousCancellation.Dispose(); if (this.isDisposed || generation != this.loadGeneration) return; string? textToKeep = null; FileVersion? version = null; try { // // A document sent before shows the text the AI received from it, which the provenance // store keeps, also when the file changed or is gone by now. Without it, we read the file: // if (documentToLoad.ProvenanceId is { } provenanceId) { var storedText = await this.ReadStoredTextAsync(provenanceId); if (this.isDisposed || generation != this.loadGeneration) return; if (storedText is not null) { this.fileContent = storedText; this.showsStoredText = true; return; } // Nothing left to read, which the dialog says: if (!documentToLoad.Exists) return; } // Taken before reading, so that a change while reading counts as a change: if (this.KeepReadText is not null) version = await Task.Run(() => FileVersion.Of(documentToLoad.FilePath), cancellationToken); var extraction = await UserFile.LoadFileData(documentToLoad.FilePath, this.RustService, this.PandocAvailability, cancellationToken); if (this.isDisposed || generation != this.loadGeneration) return; this.fileContent = extraction.Content; // // This dialog exists so the user can check what we hand to the AI. Showing an // empty document when reading the file failed would answer that question wrong. // if (!extraction.HasUsableContent) this.loadFailureMessage = extraction.ToUserMessage(documentToLoad.FileName); else if (this.KeepReadText is not null && documentToLoad.ProvenanceId is null) textToKeep = extraction.Content; } catch (OperationCanceledException) { // Either the user closed this dialog, or another document took the place of this one // while we were reading it. Nothing left to do in both cases. } catch (Exception ex) { this.Logger.LogError(ex, "Failed to load file content from '{FilePath}'", documentToLoad.FilePath); if (this.isDisposed || generation != this.loadGeneration) return; this.fileContent = string.Empty; this.loadFailureMessage = FileExtractionErrorCode.INTERNAL.ToUserMessage(documentToLoad.FileName); } finally { if (!this.isDisposed && generation == this.loadGeneration) { this.isLoadingContent = false; this.StateHasChanged(); } } // // Kept once the user sees the text, since keeping may first have to create the chat. Reading // the file told the user already what could not be read, so sending the kept text tells // nothing of it again: // if (textToKeep is not null && version is not null) await this.KeepReadText!(documentToLoad, textToKeep, version); } /// /// Whether a dropped file can be both attached and shown here, which decides what this dialog /// says and shows -- and whether it takes drops at all. /// /// /// Without a document, this dialog offers a file to be loaded instead, and that field is the /// default target of this dialog. An area which reports a delegate claims that role for itself /// and would take every drop away from the field, so we stay a plain marker in that case. /// private bool CanAttach => this.AttachPaths is not null && this.document is not null; private EventCallback> DropCallback => this.CanAttach ? EventCallback.Factory.Create>(this, this.PathsDropped) : default; private bool IsZoneDisabled() => this.IsAttachingUnavailable?.Invoke() ?? false; /// /// Marks the part of this dialog which shows the document while a file hovers over it, so it is /// visible where that file would land. The frame keeps its width in both states; only its color /// changes, or the content would jump by a few pixels with every drag. /// /// Whether this dialog is the target of the drop being aimed right now. private string PreviewAreaClass(bool isDropTarget) { if (!this.CanAttach) return string.Empty; return isDropTarget && !this.IsZoneDisabled() ? "border-dashed border-2 rounded-lg pa-2 mud-border-primary" : "border-dashed border-2 rounded-lg pa-2 mud-border-lines-default"; } /// /// Attaches what the user dropped onto this dialog and shows the first file of it. /// /// The dropped paths, in the order the runtime delivered them. private async Task PathsDropped(List paths) { if (this.AttachPaths is null) return; var attached = await this.AttachPaths(paths); if (this.isDisposed) return; // // Nothing came of the drop: the file is of a kind we do not take, Pandoc is missing, the // validation refused it, or it is a media file whose transcript does not exist yet. The // reason is already on its way to the user, and the document they were looking at stays. // if (attached.Count is 0) return; this.ShowDocument(attached[0]); // // Render before reading: the skeletons of the loading state are what tells the user that // the preview switched at all, and reading a file may well take a moment. // this.StateHasChanged(); await this.LoadDocumentContentAsync(); } /// /// Shows another document, discarding everything that belonged to the previous one. /// /// The document to show from now on. private void ShowDocument(FileAttachment attachment) { this.document = attachment; this.fileContent = string.Empty; this.loadFailureMessage = null; this.showsStoredText = false; this.isLoadingContent = this.NeedsExtraction(); } /// /// Reads the text the AI received from a document when it was sent. /// /// The entry of the text. /// The text, or null when the provenance store no longer holds it, or cannot be read. private async Task ReadStoredTextAsync(Guid provenanceId) { try { var store = await this.DatabaseClientProvider.GetProvenanceStoreAsync(); var entries = await store.GetEntriesAsync([provenanceId], CancellationToken.None); return entries.GetValueOrDefault(provenanceId)?.Content; } catch (Exception exception) { this.Logger.LogWarning(exception, "Failed to read the stored text of the document '{ProvenanceId}'. Its file is read instead.", provenanceId); return null; } } /// /// An image is shown as it is; everything else is shown as the text we read from it. /// private ContentViewerKind ViewerKind => (this.document?.IsImage ?? false) ? ContentViewerKind.IMAGE : ContentViewerKind.TEXT; /// /// Explains why we show only the beginning of a large document. The preview only answers how we /// read the file, so it may leave out the rest; the AI does not. /// /// How many characters the preview leaves out. private string CutOffNotice(int cutOffCharacters) => string.Format(T("Your document is large, so we show you only its beginning. We hide the remaining {0} characters here. Rest assured: the AI still receives your entire document."), cutOffCharacters.CompactCount()); /// /// Ends a running extraction. Without this, reading a large document would continue after the /// user closed this dialog and would keep this component, the extracted content, and the /// response stream alive until the runtime is done. /// protected override void DisposeResources() { this.isDisposed = true; // // Only the running load is left to end here: every load we replaced was ended and disposed // the moment its successor started. // this.extractionCancellation.Cancel(); this.extractionCancellation.Dispose(); base.DisposeResources(); } }