78 lines
11 KiB
C#
78 lines
11 KiB
C#
using System.Text.RegularExpressions;
|
|
using DocumentAutomation.Shared;
|
|
using DocumentFormat.OpenXml;
|
|
using DocumentFormat.OpenXml.Packaging;
|
|
using DocumentFormat.OpenXml.Wordprocessing;
|
|
using DocxWorkbench.Core;
|
|
|
|
namespace DocxWorkbench.OpenXml;
|
|
|
|
public sealed class OpenXmlDocumentInspector : IDocumentInspector
|
|
{
|
|
private static readonly string[] KnownFields = ["PAGE", "NUMPAGES", "SECTIONPAGES", "FILENAME"];
|
|
|
|
public DocumentInspectionResult Inspect(string docxPath)
|
|
{
|
|
if (!File.Exists(docxPath)) throw new FileNotFoundException("DOCX was not found.", docxPath);
|
|
if (!docxPath.EndsWith(".docx", StringComparison.OrdinalIgnoreCase)) throw new UnsupportedDocumentFormatException($"Open XML inspector accepts DOCX only, received {Path.GetExtension(docxPath)}.");
|
|
var info = new FileInfo(docxPath); using var package = WordprocessingDocument.Open(docxPath, false); var main = package.MainDocumentPart ?? throw new InvalidDataException("The DOCX has no main document part."); var roots = GetRoots(main); var fields = roots.SelectMany(x => FindFields(x.Root, x.Part, x.Section)).ToArray(); var images = CountImages(main, roots);
|
|
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 = images.RasterImages, ImageSummary = images,
|
|
Headers = main.HeaderParts.Count(), Footers = main.FooterParts.Count(), HasPageFields = fields.Any(x => x.Type == "PAGE"), HasNumPagesFields = fields.Any(x => x.Type == "NUMPAGES"), HasSectionPagesFields = fields.Any(x => x.Type == "SECTIONPAGES"), Fields = fields,
|
|
Bookmarks = main.Document.Descendants<BookmarkStart>().Count(), ContentControls = main.Document.Descendants<SdtElement>().Count(), Comments = main.GetPartsOfType<WordprocessingCommentsPart>().Count(), HasTrackedChanges = main.Document.Descendants<Inserted>().Any() || main.Document.Descendants<Deleted>().Any(), HasProtectedParts = main.Document.Descendants<DocumentProtection>().Any(),
|
|
StyleDetails = GetStyles(main), SectionAssignments = GetSectionAssignments(main), InsertionPoint = FindInsertionPoint(main.Document), BodyImportPlan = CreateBodyImportPlan(main), Warnings = FindWarnings(roots)
|
|
};
|
|
}
|
|
|
|
private static IReadOnlyList<(OpenXmlPartRootElement Root, string Part, int? Section)> GetRoots(MainDocumentPart main)
|
|
{
|
|
var roots = new List<(OpenXmlPartRootElement, string, int?)> { (main.Document, "MainDocumentPart", null) }; roots.AddRange(main.HeaderParts.Select((p, i) => ((OpenXmlPartRootElement)p.Header, $"HeaderPart[{i + 1}]", (int?)null))); roots.AddRange(main.FooterParts.Select((p, i) => ((OpenXmlPartRootElement)p.Footer, $"FooterPart[{i + 1}]", (int?)null))); if (main.DocumentSettingsPart?.Settings is not null) roots.Add((main.DocumentSettingsPart.Settings, "SettingsPart", null)); if (main.FootnotesPart?.Footnotes is not null) roots.Add((main.FootnotesPart.Footnotes, "FootnotesPart", null)); if (main.EndnotesPart?.Endnotes is not null) roots.Add((main.EndnotesPart.Endnotes, "EndnotesPart", null)); foreach (var comments in main.GetPartsOfType<WordprocessingCommentsPart>()) if (comments.Comments is not null) roots.Add((comments.Comments, "CommentsPart", null)); return roots;
|
|
}
|
|
|
|
private static IEnumerable<DocumentFieldInfo> FindFields(OpenXmlElement root, string part, int? section)
|
|
{
|
|
foreach (var simple in root.Descendants<SimpleField>()) { var type = NormalizeField(simple.Instruction?.Value); if (type is not null) yield return new DocumentFieldInfo { Type = type, Part = part, Section = section, Instruction = simple.Instruction?.Value }; }
|
|
var active = new List<string>(); foreach (var element in root.Descendants()) { if (element is FieldChar begin && begin.FieldCharType?.Value == FieldCharValues.Begin) active.Add(""); else if (element is FieldCode code && active.Count > 0) active[^1] += " " + code.Text; else if (element is FieldChar end && end.FieldCharType?.Value == FieldCharValues.End && active.Count > 0) { var raw = active[^1]; active.RemoveAt(active.Count - 1); var type = NormalizeField(raw); if (type is not null) yield return new DocumentFieldInfo { Type = type, Part = part, Section = section, Instruction = raw.Trim() }; } }
|
|
}
|
|
|
|
private static string? NormalizeField(string? value) { if (string.IsNullOrWhiteSpace(value)) return null; var normalized = Regex.Replace(value, @"\s+", " ").Trim(); var name = Regex.Match(normalized, @"(?i)^(PAGE|NUMPAGES|SECTIONPAGES|FILENAME)\b").Value.ToUpperInvariant(); return KnownFields.Contains(name) ? name : null; }
|
|
|
|
private static DocumentImageSummary CountImages(MainDocumentPart main, IReadOnlyList<(OpenXmlPartRootElement Root, string Part, int? Section)> roots)
|
|
{
|
|
var imageParts = new HashSet<OpenXmlPart>(); var drawings = roots.Sum(x => x.Root.Descendants<Drawing>().Count()); var shapes = roots.Sum(x => x.Root.Descendants<DocumentFormat.OpenXml.Vml.Shape>().Count()); var external = 0;
|
|
foreach (var image in main.GetPartsOfType<ImagePart>()) imageParts.Add(image); foreach (var header in main.HeaderParts) { foreach (var image in header.GetPartsOfType<ImagePart>()) imageParts.Add(image); external += header.ExternalRelationships.Count(x => x.Uri?.ToString().Contains("image", StringComparison.OrdinalIgnoreCase) == true); } foreach (var footer in main.FooterParts) { foreach (var image in footer.GetPartsOfType<ImagePart>()) imageParts.Add(image); external += footer.ExternalRelationships.Count(x => x.Uri?.ToString().Contains("image", StringComparison.OrdinalIgnoreCase) == true); }
|
|
external += main.ExternalRelationships.Count(x => x.Uri?.ToString().Contains("image", StringComparison.OrdinalIgnoreCase) == true); return new DocumentImageSummary { RasterImages = imageParts.Count, DrawingObjects = drawings, VmlShapes = shapes, ExternalImages = external };
|
|
}
|
|
|
|
private static IReadOnlyList<DocumentStyleInfo> GetStyles(MainDocumentPart main)
|
|
{
|
|
var usage = main.Document.Descendants<ParagraphStyleId>().GroupBy(x => x.Val?.Value ?? "").ToDictionary(x => x.Key, x => x.Count()); return main.StyleDefinitionsPart?.Styles?.Elements<Style>().Select(style => new DocumentStyleInfo { StyleId = style.StyleId?.Value ?? "", Name = style.StyleName?.Val?.Value, Type = style.Type is null ? null : style.Type.Value.ToString(), IsDefault = style.Default?.Value == true, BasedOn = style.BasedOn?.Val?.Value, UsageCount = usage.GetValueOrDefault(style.StyleId?.Value ?? "") }).Where(x => x.StyleId.Length > 0).ToArray() ?? [];
|
|
}
|
|
|
|
private static IReadOnlyList<SectionPartAssignment> GetSectionAssignments(MainDocumentPart main)
|
|
{
|
|
var result = new List<SectionPartAssignment>(); var sections = main.Document.Body?.Elements<SectionProperties>().ToArray() ?? []; for (var i = 0; i < sections.Length; i++) { var section = sections[i]; var headers = section.Elements<HeaderReference>().Select(x => $"{x.Type?.Value}:{PartName(main, x.Id?.Value)}").ToArray(); var footers = section.Elements<FooterReference>().Select(x => $"{x.Type?.Value}:{PartName(main, x.Id?.Value)}").ToArray(); result.Add(new SectionPartAssignment { Section = i + 1, Headers = headers, Footers = footers, LinkToPrevious = i > 0 && headers.Length == 0 && footers.Length == 0 }); } return result;
|
|
}
|
|
|
|
private static string PartName(MainDocumentPart main, string? id) { if (id is null) return "missing"; try { return main.GetPartById(id).Uri.ToString(); } catch { return "unresolved:" + id; } }
|
|
private static InsertionPointInfo FindInsertionPoint(Document root) { if (root.Descendants<SdtElement>().Any(x => string.Equals(x.SdtProperties?.GetFirstChild<Tag>()?.Val?.Value, "DOC_BODY", StringComparison.OrdinalIgnoreCase))) return new InsertionPointInfo { Type = "ContentControl", Location = "MainDocumentPart" }; if (root.Descendants<BookmarkStart>().Any(x => string.Equals(x.Name?.Value, "DOC_BODY", StringComparison.OrdinalIgnoreCase))) return new InsertionPointInfo { Type = "Bookmark", Location = "MainDocumentPart" }; if (root.InnerText.Contains("{{DOC_BODY}}", StringComparison.Ordinal)) return new InsertionPointInfo { Type = "TextMarker", Location = "MainDocumentPart" }; return new InsertionPointInfo { Type = "None" }; }
|
|
private static IReadOnlyList<InspectionWarning> FindWarnings(IReadOnlyList<(OpenXmlPartRootElement Root, string Part, int? Section)> roots) { var result = new List<InspectionWarning>(); foreach (var root in roots) foreach (Match match in Regex.Matches(root.Root.InnerText, @"(?i)\b(2019|legacy|старый)\b")) result.Add(new InspectionWarning { Code = "PossibleLegacyProjectData", Message = "A value matched the legacy-project heuristic.", Part = root.Part, Location = "text", MatchedValue = match.Value, DetectionRule = "word-boundary: 2019|legacy|старый" }); return result; }
|
|
private static BodyImportPlan CreateBodyImportPlan(MainDocumentPart main)
|
|
{
|
|
var body = main.Document.Body; if (body is null) return new BodyImportPlan { ParagraphCount = 0, TableCount = 0, InlineShapeCount = 0, FloatingShapeCount = 0, InternalSectionBreakCount = 0, PageBreakCount = 0, IncludedStories = ["wdMainTextStory"], ExcludedStories = ["wdPrimaryHeaderStory", "wdFirstPageHeaderStory", "wdEvenPagesHeaderStory", "wdPrimaryFooterStory", "wdFirstPageFooterStory", "wdEvenPagesFooterStory"], Warnings = [] };
|
|
var drawings = body.Descendants<Drawing>(); return new BodyImportPlan { ParagraphCount = body.Descendants<Paragraph>().Count(), TableCount = body.Descendants<Table>().Count(), InlineShapeCount = drawings.Count(x => x.Ancestors<DocumentFormat.OpenXml.Drawing.Wordprocessing.Inline>().Any()), FloatingShapeCount = drawings.Count(x => x.Ancestors<DocumentFormat.OpenXml.Drawing.Wordprocessing.Anchor>().Any()), InternalSectionBreakCount = Math.Max(0, body.Descendants<SectionProperties>().Count() - 1), PageBreakCount = body.Descendants<Break>().Count(x => x.Type?.Value == BreakValues.Page), IncludedStories = ["wdMainTextStory"], ExcludedStories = ["wdPrimaryHeaderStory", "wdFirstPageHeaderStory", "wdEvenPagesHeaderStory", "wdPrimaryFooterStory", "wdFirstPageFooterStory", "wdEvenPagesFooterStory"], Warnings = ["Source headers, footers, page setup and final section properties are excluded."] };
|
|
}
|
|
}
|
|
|
|
public sealed class TextDocumentInspector : IDocumentInspector
|
|
{
|
|
public DocumentInspectionResult Inspect(string path) { if (!File.Exists(path)) throw new FileNotFoundException("TXT was not found.", path); if (!path.EndsWith(".txt", StringComparison.OrdinalIgnoreCase)) throw new UnsupportedDocumentFormatException($"TXT inspector accepts TXT only, received {Path.GetExtension(path)}."); var info = new FileInfo(path); return new DocumentInspectionResult { FilePath = path, FileSize = info.Length, LastWriteTimeUtc = info.LastWriteTimeUtc, Sections = 1, Paragraphs = File.ReadAllLines(path).Length, InsertionPoint = new InsertionPointInfo { Type = "None" } }; }
|
|
}
|
|
|
|
public sealed class DocumentInspector : IDocumentInspector
|
|
{
|
|
private readonly OpenXmlDocumentInspector _inner = new();
|
|
public DocumentInspectionResult Inspect(string docxPath) => _inner.Inspect(docxPath);
|
|
}
|