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 +1,39 @@
+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
+        };
+    }
+}