feat: add DOCX workbench and Word automation

2026-07-30 13:18:54 +03:00
parent 1bda2c155d
commit 958bad6823
6 changed files with 240 additions and 0 deletions
@@ -0,0 +1,71 @@
+using System.Text.Json.Serialization;
+
+namespace DocumentAutomation.Shared;
+
+public sealed record DocumentInspectionResult
+{
+    public required string FilePath { get; init; }
+    public long FileSize { get; init; }
+    public DateTime LastWriteTimeUtc { get; init; }
+    public int Sections { get; init; }
+    public int Paragraphs { get; init; }
+    public int Tables { get; init; }
+    public int Images { get; init; }
+    public int Headers { get; init; }
+    public int Footers { get; init; }
+    public bool HasPageFields { get; init; }
+    public bool HasNumPagesFields { get; init; }
+    public bool HasSectionPagesFields { get; init; }
+    public int Bookmarks { get; init; }
+    public int ContentControls { get; init; }
+    public int Comments { get; init; }
+    public bool HasTrackedChanges { get; init; }
+    public bool HasProtectedParts { get; init; }
+    public IReadOnlyList<string> Styles { get; init; } = [];
+    public IReadOnlyList<string> Warnings { get; init; } = [];
+}
+
+public sealed record WordDocumentBuildRequest
+{
+    public required string TemplatePath { get; init; }
+    public required string BodyPath { get; init; }
+    public required string OutputDocxPath { get; init; }
+    public string? RulesProfilePath { get; init; }
+    public bool PreserveBodyFormatting { get; init; } = true;
+    public bool PreservePageBreaks { get; init; } = true;
+}
+
+public sealed record WordDocumentResult
+{
+    public required string OutputDocxPath { get; init; }
+    public string? OutputPdfPath { get; init; }
+    public required string LogPath { get; init; }
+    public required TimeSpan Duration { get; init; }
+    public required string InsertionPoint { get; init; }
+}
+
+public sealed record WordPreviewResult
+{
+    public required string PdfPath { get; init; }
+    public required TimeSpan Duration { get; init; }
+}
+
+public enum RuleEvaluationStatus { Pass, Fail, NotApplicable, NeedsReview, Error }
+
+public sealed record RuleEvaluationResult
+{
+    public required string RuleId { get; init; }
+    public required RuleEvaluationStatus Status { get; init; }
+    public required string Message { get; init; }
+    public string? ClauseId { get; init; }
+    public int? SourcePage { get; init; }
+    public string? DocumentLocation { get; init; }
+    public bool CanAutoFix { get; init; }
+}
+
+public sealed record OllamaSettings
+{
+    public string BaseUrl { get; init; } = "http://192.168.200.112:11434";
+    public string Model { get; init; } = "";
+    public int TimeoutSeconds { get; init; } = 120;
+}
@@ -0,0 +1,33 @@
+using System.Diagnostics;
+using System.Text;
+using System.Text.Json;
+using DocumentAutomation.Shared;
+using DocxWorkbench.Core;
+using DocxWorkbench.OpenXml;
+using DocxWorkbench.WordInterop;
+
+namespace DocxWorkbench.App;
+
+public sealed class MainForm : Form
+{
+    private readonly TextBox _template = new() { Width = 520 }; private readonly TextBox _body = new() { Width = 520 }; private readonly TextBox _output = new() { Width = 520 }; private readonly TextBox _log = new() { Multiline = true, ScrollBars = ScrollBars.Vertical, Dock = DockStyle.Fill }; private readonly Label _status = new() { AutoSize = true, Text = "Готово" }; private readonly IDocumentInspector _inspector = new DocumentInspector(); private readonly IWordDocumentService _word = new WordDocumentService();
+    public MainForm()
+    {
+        Text = "DOCX Workbench"; Width = 980; Height = 680; StartPosition = FormStartPosition.CenterScreen;
+        var grid = new TableLayoutPanel { Dock = DockStyle.Top, Height = 145, ColumnCount = 3, RowCount = 4, Padding = new Padding(8) }; grid.ColumnStyles.Add(new ColumnStyle(SizeType.Absolute, 150)); grid.ColumnStyles.Add(new ColumnStyle(SizeType.Percent, 100)); grid.ColumnStyles.Add(new ColumnStyle(SizeType.Absolute, 110));
+        AddRow(grid, 0, "Шаблон DOCX", _template, BrowseTemplate); AddRow(grid, 1, "Тело документа", _body, BrowseBody); AddRow(grid, 2, "Выходной DOCX", _output, BrowseOutput);
+        var buttons = new FlowLayoutPanel { Dock = DockStyle.Fill }; buttons.Controls.Add(Button("Анализировать", async (_, _) => await InspectAsync())); buttons.Controls.Add(Button("Сформировать", async (_, _) => await BuildAsync())); buttons.Controls.Add(Button("Проверить", async (_, _) => await InspectOutputAsync())); buttons.Controls.Add(Button("Открыть DOCX", (_, _) => Open(_output.Text))); buttons.Controls.Add(Button("Открыть PDF", (_, _) => Open(Path.ChangeExtension(_output.Text, ".pdf")))); grid.Controls.Add(buttons, 1, 3); grid.SetColumnSpan(buttons, 2);
+        var tabs = new TabControl { Dock = DockStyle.Fill }; foreach (var title in new[] { "Предпросмотр шаблона", "Предпросмотр тела", "Структура", "Результат", "Нарушения", "Журнал" }) { var page = new TabPage(title); if (title == "Журнал") page.Controls.Add(_log); tabs.TabPages.Add(page); }
+        Controls.Add(tabs); Controls.Add(grid); Controls.Add(_status); _status.Dock = DockStyle.Bottom;
+    }
+    private static void AddRow(TableLayoutPanel p, int row, string title, TextBox box, EventHandler browse) { p.Controls.Add(new Label { Text = title, AutoSize = true, Anchor = AnchorStyles.Left }, 0, row); p.Controls.Add(box, 1, row); var b = new Button { Text = "Обзор...", Width = 95 }; b.Click += browse; p.Controls.Add(b, 2, row); }
+    private static Button Button(string text, EventHandler handler) { var b = new Button { Text = text, AutoSize = true }; b.Click += handler; return b; }
+    private void BrowseTemplate(object? s, EventArgs e) { using var d = new OpenFileDialog { Filter = "DOCX|*.docx" }; if (d.ShowDialog() == DialogResult.OK) _template.Text = d.FileName; }
+    private void BrowseBody(object? s, EventArgs e) { using var d = new OpenFileDialog { Filter = "DOCX/TXT|*.docx;*.txt" }; if (d.ShowDialog() == DialogResult.OK) _body.Text = d.FileName; }
+    private void BrowseOutput(object? s, EventArgs e) { using var d = new SaveFileDialog { Filter = "DOCX|*.docx", DefaultExt = "docx" }; if (d.ShowDialog() == DialogResult.OK) _output.Text = d.FileName; }
+    private async Task InspectAsync() => await RunAsync("Анализ", () => Task.Run(() => { var a = _inspector.Inspect(_template.Text); var b = _inspector.Inspect(_body.Text); return (object)new { Template = a, Body = b }; }));
+    private async Task InspectOutputAsync() => await RunAsync("Проверка результата", () => Task.Run(() => (object)_inspector.Inspect(_output.Text)));
+    private async Task BuildAsync() => await RunAsync("Формирование", async () => { var result = await _word.BuildDocumentAsync(new WordDocumentBuildRequest { TemplatePath = _template.Text, BodyPath = _body.Text, OutputDocxPath = _output.Text }, new Progress<string>(x => _status.Text = x), CancellationToken.None); var pdf = await _word.ExportPdfAsync(result.OutputDocxPath, Path.ChangeExtension(result.OutputDocxPath, ".pdf"), CancellationToken.None); return (object)new { result, pdf }; });
+    private async Task RunAsync(string name, Func<Task<object>> action) { try { _status.Text = name + "..."; var value = await action(); _log.AppendText(JsonSerializer.Serialize(value, new JsonSerializerOptions { WriteIndented = true }) + Environment.NewLine); _status.Text = name + ": готово"; } catch (Exception ex) { _log.AppendText($"{name}: {ex}\r\n"); _status.Text = name + ": ошибка"; MessageBox.Show(ex.Message, name, MessageBoxButtons.OK, MessageBoxIcon.Error); } }
+    private static void Open(string path) { if (File.Exists(path)) Process.Start(new ProcessStartInfo(path) { UseShellExecute = true }); }
+}
@@ -0,0 +1,16 @@
+namespace DocxWorkbench.App;
+
+static class Program
+{
+    /// <summary>
+    ///  The main entry point for the application.
+    /// </summary>
+    [STAThread]
+    static void Main()
+    {
+        // To customize application configuration such as set high DPI settings or default font,
+        // see https://aka.ms/applicationconfiguration.
+        ApplicationConfiguration.Initialize();
+    Application.Run(new MainForm());
+    }    
+}
@@ -0,0 +1,26 @@
+using DocumentAutomation.Shared;
+
+namespace DocxWorkbench.Core;
+
+public interface IDocumentInspector
+{
+    DocumentInspectionResult Inspect(string docxPath);
+}
+
+public interface IWordDocumentService
+{
+    Task<WordDocumentResult> BuildDocumentAsync(WordDocumentBuildRequest request, IProgress<string>? progress, CancellationToken cancellationToken);
+    Task<WordPreviewResult> ExportPdfAsync(string docxPath, string outputPdfPath, CancellationToken cancellationToken);
+}
+
+public static class DocumentInputValidation
+{
+    public static void Validate(WordDocumentBuildRequest request)
+    {
+        if (!File.Exists(request.TemplatePath)) throw new FileNotFoundException("Шаблон DOCX не найден.", request.TemplatePath);
+        if (!File.Exists(request.BodyPath)) throw new FileNotFoundException("Файл тела документа не найден.", request.BodyPath);
+        if (!request.TemplatePath.EndsWith(".docx", StringComparison.OrdinalIgnoreCase)) throw new ArgumentException("Шаблон должен быть DOCX.");
+        if (Path.GetFullPath(request.TemplatePath).Equals(Path.GetFullPath(request.OutputDocxPath), StringComparison.OrdinalIgnoreCase)) throw new InvalidOperationException("Нельзя перезаписывать исходный шаблон.");
+        Directory.CreateDirectory(Path.GetDirectoryName(Path.GetFullPath(request.OutputDocxPath))!);
+    }
+}
@@ -0,0 +1,39 @@
+using System.Text.RegularExpressions;
+using DocumentAutomation.Shared;
+using DocumentFormat.OpenXml.Packaging;
+using DocumentFormat.OpenXml.Wordprocessing;
+using DocxWorkbench.Core;
+
+namespace DocxWorkbench.OpenXml;
+
+public sealed class DocumentInspector : IDocumentInspector
+{
+    public DocumentInspectionResult Inspect(string docxPath)
+    {
+        if (!File.Exists(docxPath)) throw new FileNotFoundException("DOCX не найден.", docxPath);
+        var info = new FileInfo(docxPath);
+        using var package = WordprocessingDocument.Open(docxPath, false);
+        var main = package.MainDocumentPart ?? throw new InvalidDataException("В DOCX отсутствует основной документ.");
+        var body = main.Document.Body;
+        var text = main.Document.InnerText;
+        var headers = main.HeaderParts.Count();
+        var footers = main.FooterParts.Count();
+        var warnings = new List<string>();
+        if (Regex.IsMatch(text, "2019|старый|legacy", RegexOptions.IgnoreCase)) warnings.Add("Обнаружены возможные старые проектные реквизиты; проверьте вручную.");
+        if (main.Document.Descendants<SectionProperties>().Any(s => s.GetFirstChild<HeaderReference>() is null)) warnings.Add("Есть секция без явной ссылки на header.");
+        return new DocumentInspectionResult
+        {
+            FilePath = docxPath, FileSize = info.Length, LastWriteTimeUtc = info.LastWriteTimeUtc,
+            Sections = Math.Max(1, main.Document.Descendants<SectionProperties>().Count()),
+            Paragraphs = main.Document.Descendants<Paragraph>().Count(), Tables = main.Document.Descendants<Table>().Count(),
+            Images = main.ImageParts.Count(), Headers = headers, Footers = footers,
+            HasPageFields = text.Contains("PAGE", StringComparison.OrdinalIgnoreCase), HasNumPagesFields = text.Contains("NUMPAGES", StringComparison.OrdinalIgnoreCase),
+            HasSectionPagesFields = text.Contains("SECTIONPAGES", StringComparison.OrdinalIgnoreCase), Bookmarks = main.Document.Descendants<BookmarkStart>().Count(),
+            ContentControls = main.Document.Descendants<SdtElement>().Count(), Comments = package.MainDocumentPart.GetPartsOfType<WordprocessingCommentsPart>().Count(),
+            HasTrackedChanges = main.Document.Descendants<Inserted>().Any() || main.Document.Descendants<Deleted>().Any(),
+            HasProtectedParts = main.Document.Descendants<DocumentProtection>().Any(),
+            Styles = main.StyleDefinitionsPart?.Styles?.Elements<Style>().Select(s => s.StyleId?.Value ?? "").Where(s => s.Length > 0).Distinct().Order().ToArray() ?? [],
+            Warnings = warnings
+        };
+    }
+}
@@ -0,0 +1,55 @@
+using System.Diagnostics;
+using System.Runtime.InteropServices;
+using System.Text.Json;
+using DocumentAutomation.Shared;
+using DocxWorkbench.Core;
+using Word = Microsoft.Office.Interop.Word;
+
+namespace DocxWorkbench.WordInterop;
+
+public sealed class WordDocumentService : IWordDocumentService
+{
+    public Task<WordDocumentResult> BuildDocumentAsync(WordDocumentBuildRequest request, IProgress<string>? progress, CancellationToken cancellationToken)
+        => Task.Run(() => Build(request, progress, cancellationToken), cancellationToken);
+
+    public Task<WordPreviewResult> ExportPdfAsync(string docxPath, string outputPdfPath, CancellationToken cancellationToken)
+        => Task.Run(() => ExportPdf(docxPath, outputPdfPath, cancellationToken), cancellationToken);
+
+    private static WordDocumentResult Build(WordDocumentBuildRequest request, IProgress<string>? progress, CancellationToken token)
+    {
+        DocumentInputValidation.Validate(request); var started = Stopwatch.StartNew();
+        var temp = Path.Combine(Path.GetTempPath(), "DocumentAutomation", Guid.NewGuid().ToString("N")); Directory.CreateDirectory(temp);
+        var working = Path.Combine(temp, "working.docx"); File.Copy(request.TemplatePath, working, true);
+        Word.Application? app = null; Word.Document? document = null; string insertion = "none"; int? pid = null;
+        try
+        {
+            token.ThrowIfCancellationRequested(); progress?.Report("Запуск Word COM"); app = new Word.Application { Visible = false, DisplayAlerts = Word.WdAlertLevel.wdAlertsNone }; pid = GetProcessId(app);
+            document = app.Documents.Open(working, ReadOnly: false, AddToRecentFiles: false, Visible: false);
+            progress?.Report("Поиск точки вставки DOC_BODY"); var range = FindInsertionRange(document, out insertion);
+            if (range is null) throw new InvalidOperationException("В шаблоне не найдена точка вставки DOC_BODY.");
+            token.ThrowIfCancellationRequested(); range.InsertFile(request.BodyPath, ConfirmConversions: false, Link: false, Attachment: false);
+            progress?.Report("Обновление полей и пагинация"); document.Repaginate(); document.Fields.Update();
+            for (var i = 1; i <= document.Sections.Count; i++) { var section = document.Sections[i]; foreach (Word.HeaderFooter hf in section.Headers) hf.Range.Fields.Update(); foreach (Word.HeaderFooter hf in section.Footers) hf.Range.Fields.Update(); }
+            document.SaveAs2(request.OutputDocxPath, Word.WdSaveFormat.wdFormatXMLDocument);
+            var logPath = Path.ChangeExtension(request.OutputDocxPath, ".generation.json"); File.WriteAllText(logPath, JsonSerializer.Serialize(new { operationId = Guid.NewGuid(), template = request.TemplatePath, body = request.BodyPath, output = request.OutputDocxPath, insertionPoint = insertion, wordPid = pid, durationMs = started.ElapsedMilliseconds, completedAtUtc = DateTime.UtcNow }, new JsonSerializerOptions { WriteIndented = true }));
+            return new WordDocumentResult { OutputDocxPath = request.OutputDocxPath, LogPath = logPath, Duration = started.Elapsed, InsertionPoint = insertion };
+        }
+        finally { if (document is not null) { try { document.Close(Word.WdSaveOptions.wdDoNotSaveChanges); } catch { } Marshal.FinalReleaseComObject(document); } if (app is not null) { try { app.Quit(Word.WdSaveOptions.wdDoNotSaveChanges); } catch { } Marshal.FinalReleaseComObject(app); } try { Directory.Delete(temp, true); } catch { } }
+    }
+
+    private static WordPreviewResult ExportPdf(string docxPath, string outputPdfPath, CancellationToken token)
+    {
+        if (!File.Exists(docxPath)) throw new FileNotFoundException("DOCX не найден.", docxPath); Directory.CreateDirectory(Path.GetDirectoryName(Path.GetFullPath(outputPdfPath))!); var sw = Stopwatch.StartNew(); Word.Application? app = null; Word.Document? doc = null;
+        try { token.ThrowIfCancellationRequested(); app = new Word.Application { Visible = false, DisplayAlerts = Word.WdAlertLevel.wdAlertsNone }; doc = app.Documents.Open(docxPath, ReadOnly: true, AddToRecentFiles: false, Visible: false); doc.ExportAsFixedFormat(outputPdfPath, Word.WdExportFormat.wdExportFormatPDF); return new WordPreviewResult { PdfPath = outputPdfPath, Duration = sw.Elapsed }; }
+        finally { if (doc is not null) { try { doc.Close(Word.WdSaveOptions.wdDoNotSaveChanges); } catch { } Marshal.FinalReleaseComObject(doc); } if (app is not null) { try { app.Quit(Word.WdSaveOptions.wdDoNotSaveChanges); } catch { } Marshal.FinalReleaseComObject(app); } }
+    }
+
+    private static Word.Range? FindInsertionRange(Word.Document document, out string point)
+    {
+        foreach (Word.ContentControl control in document.ContentControls) if (string.Equals(control.Tag, "DOC_BODY", StringComparison.OrdinalIgnoreCase) || string.Equals(control.Title, "DOC_BODY", StringComparison.OrdinalIgnoreCase)) { point = "ContentControl:DOC_BODY"; return control.Range; }
+        if (document.Bookmarks.Exists("DOC_BODY")) { point = "Bookmark:DOC_BODY"; return document.Bookmarks["DOC_BODY"].Range; }
+        var search = document.Content; search.Find.ClearFormatting(); search.Find.Text = "{{DOC_BODY}}"; if (search.Find.Execute()) { point = "Marker:{{DOC_BODY}}"; search.Text = ""; return search; } point = "none"; return null;
+    }
+
+    private static int? GetProcessId(Word.Application app) { try { return Process.GetProcessesByName("WINWORD").OrderByDescending(p => p.StartTime).FirstOrDefault()?.Id; } catch { return null; } }
+}