public void Analyze(DqDocument document)
{
var parts = new[]
{
document.Structure.Toc,
document.Structure.Introduction,
document.Structure.MainPart,
document.Structure.Conclusion,
document.Structure.Bibliography,
document.Structure.Appendixes,
}.Concat(document.Structure.Abstracts).Where(p => p != null);
foreach (var dqPart in parts)
{
if (dqPart is DqMainPart dqMainPart)
{
AnalyzeMainPart(dqMainPart, document);
}
else
{
AnalyzeNonMainPart(dqPart, document);
}
FindSpaces(dqPart);
}
}
private void AnalyzeNonMainPart(DqPart dqPart, DqDocument dqDocument)
{
var centeredParts = new[] { PartType.Bibliography, PartType.Abstract, PartType.Introduction, PartType.Conclusion, PartType.Toc };
if (!centeredParts.Contains(dqPart.Type))
{
return;
}
var dqParagraph = dqPart.Start;
if (!dqParagraph.Text.StartsWith("{PageBreak}") && !dqDocument.Paragraphs[dqParagraph.Index - 1].Text.EndsWith("{PageBreak}"))
{
dqParagraph.Meta.Errors.Add(new DqError("Перед разделом отсутсвует разрыв страницы."));
}
if (dqParagraph.Style.Aligment != DqAligment.Center)
{
dqParagraph.Meta.Errors.Add(new DqAlignmentError($"Заголовки разделов должны быть выровнены по центру."));
}
else if (dqParagraph.Style.Indent != 0)
{
dqParagraph.Meta.Errors.Add(new DqAlignmentError($"При выравнивании по центру должен отсутвовать абзацный отступ."));
}
if (string.Equals(dqParagraph.GetPureText(), "список использованной литературы", StringComparison.OrdinalIgnoreCase))
{
dqParagraph.Meta.Errors.Add(new DqError($"Нормативное название части — «Список использованных источников»."));
}
}
public void Analyze2(DqDocument dqDocument)
{
if (dqDocument.Structure.Bibliography == null)
{
return;
}
foreach (var content in dqDocument.Structure.Bibliography.Paragraphs)
{
var sources = content.Meta.SourceDeclarations.OfType <DqSource>().ToList();
if (sources.Count == 0)
{
continue;
}
Recognize(sources);
var source = sources.First();
if (source.SourceType == SourceType.Unknown)
{
content.Meta.Errors.Add(new DqError("Источник имеет неверный формат"));
}
else if (source.SourceType == SourceType.WebUnknown)
{
content.Meta.Errors.Add(new DqError("Web-источник имеет неверный формат"));
}
else if (source.Formatted != source.Paragraph.Text)
{
content.Meta.Errors.Add(new DqError($"Ожидаемый формат данных: {source.Formatted}"));
}
}
}
public IList <Token> GetHeaders(DqDocument document)
{
if (document.Structure.Introduction == null)
{
return(new List <Token>());
}
var dqParagraphs = document.Structure.Introduction.Paragraphs
.Skip(1)
.Where(IsNumberedParagraph)
.ToList();
var list = new List <Token>();
var mainPartHeaders = new List <Token>();
foreach (var dqParagraph in dqParagraphs)
{
if (!list.Any())
{
list.Add(CreateToken(dqParagraph));
continue;
}
if (dqParagraph.Index - list.Last().Paragraph.Index == 1 && !IsChapter(dqParagraph))
{
list.Add(new Token(dqParagraph, GetLevel(dqParagraph)));
}
else
{
var isGreater = true;
for (var i = 1; i < list.Count; ++i)
{
if (list[i - 1].Level >= list[i].Level)
{
isGreater = false;
}
}
if (isGreater)
{
mainPartHeaders.AddRange(list);
}
list.Clear();
list.Add(new Token(dqParagraph, GetLevel(dqParagraph)));
}
}
if (list.Count == 1)
{
mainPartHeaders.AddRange(list);
}
foreach (var token in mainPartHeaders.Concat(GetNonMainHeaders(document)))
{
token.Paragraph.Meta.IsHeader = true;
}
return(mainPartHeaders);
}
public void Parse(DqParagraph paragraph, DqDocument dqDocument)
{
var m = Regex.Matches(paragraph.Text, @"[^]]\[(\d+)(?:,\s*c(?:тр)\.\d+)?\]", RegexOptions.IgnoreCase);
if (m.Count != 0)
{
paragraph.Meta.Structure.AddRange(m.Cast <Match>().Select(mm => mm.Groups[1].Value).Select(t => new DqNumberedElement(paragraph, DqStructureElementType.SourceReference)
{
Number = t
}));
}
var bibliographyPart = dqDocument.Structure.Bibliography;
if (bibliographyPart == null || !bibliographyPart.Paragraphs.Contains(paragraph))
{
return;
}
var match = Regex.Match(paragraph.Text, @"^\s*(\d+)", RegexOptions.IgnoreCase);
if (!match.Success)
{
return;
}
paragraph.Meta.Structure.Add(new DqSource(paragraph, DqStructureElementType.SourceDeclaration)
{
Number = match.Groups[1].Value
});
}
private int?TryFindByTitle(DqDocument document, IReadOnlyCollection <string> titles, int startIndex)
{
for (var i = startIndex + 1; i < document.Paragraphs.Count; ++i)
{
var paragraph = document.Paragraphs[i];
var title = titles.FirstOrDefault(t => string.Equals(paragraph.GetPureText(), t, StringComparison.OrdinalIgnoreCase));
if (title != null)
{
return(i);
}
}
return(null);
}
private void AnalyzeMainPart(DqMainPart dqMainPart, DqDocument dqDocument)
{
foreach (var dqParagraph in dqMainPart.Paragraphs)
{
AnalyzeMainPartParagraph(dqParagraph, dqDocument);
}
foreach (var chapter in dqMainPart.Children)
{
var dqParagraph = chapter.Start;
if (!dqParagraph.Text.StartsWith("{PageBreak}") && !dqDocument.Paragraphs[dqParagraph.Index - 1].Text.EndsWith("{PageBreak}"))
{
dqParagraph.Meta.Errors.Add(new DqError("Перед разделом отсутсвует разрыв страницы."));
}
}
}
public void Analyze(DqDocument dqDocument)
{
var bibliography = dqDocument.Structure.Bibliography;
if (bibliography == null)
{
return;
}
var sources = bibliography.Paragraphs.Skip(1).Where(p => p.Meta.SourceDeclarations.Any());
foreach (var dqParagraph in sources)
{
Analyze(dqParagraph);
}
}
private IEnumerable <Token> GetNonMainHeaders(DqDocument dqDocument)
{
var report = dqDocument.Structure;
return(report.Abstracts.Concat(new[]
{
report.Toc,
report.Introduction,
report.Conclusion,
report.Bibliography,
report.Appendixes,
})
.Select(p => p?.Paragraphs.FirstOrDefault())
.Where(p => p != null)
.Select(CreateToken));
}
public DqDocument Parse(string path)
{
var tempPath = Path.GetTempFileName();
File.Copy(path, tempPath, overwrite: true);
try
{
using (var doc = WordprocessingDocument.Open(tempPath, isEditable: false))
{
var dqDocument = new DqDocument();
var fontScheme = _dqStyleParser.GetFontScheme(doc);
dqDocument.NumberingTable = _dqNumberingParser.ParseNumberingTable(doc);
dqDocument.StyleTable = _dqStyleParser.ParseStyleTable(doc);
var body = doc.MainDocumentPart.Document.Body;
dqDocument.Sections.AddRange(GetSections(body));
List <DqParagraph> paragraphs = new List <DqParagraph>();
foreach (var element in body.Elements())
{
if (element is Paragraph p)
{
paragraphs.Add(Convert(p, dqDocument.StyleTable, fontScheme, dqDocument.NumberingTable));
}
else if (element is DocumentFormat.OpenXml.Wordprocessing.Table)
{
paragraphs.Add(new DqParagraph("{TBL}", dqDocument.StyleTable.Paragraph.Default));
}
}
dqDocument.Paragraphs.AddRange(paragraphs.Where(p => !string.IsNullOrWhiteSpace(p.Text)));
for (int i = 0; i < dqDocument.Paragraphs.Count; i++)
{
dqDocument.Paragraphs[i].Index = i;
}
return(dqDocument);
}
}
finally
{
File.Delete(tempPath);
}
}
public void RestoreNumbering(DqDocument document)
{
var numberedParagraphs = document.Paragraphs.Where(IsNumbered).ToList();
var countersById = numberedParagraphs
.Select(p => p.Style.Numbering?.Id)
.Distinct()
.ToDictionary(k => k, k => Enumerable.Repeat(element: 0, count: 9).ToList());
foreach (var paragraph in numberedParagraphs)
{
if (paragraph.Style.OutlineLevel < paragraph.Style.Numbering?.Levels.Count)
{
UpdateCounters(paragraph, countersById[paragraph.Style.Numbering.Id]);
paragraph.Number = CalculateNumbers(paragraph, countersById[paragraph.Style.Numbering.Id]);
paragraph.Text = paragraph.Number + " " + paragraph.Text;
}
}
}
private void AnalyzeMainPartParagraph(DqParagraph dqParagraph, DqDocument dqDocument)
{
if (string.IsNullOrWhiteSpace(dqParagraph.Text.Replace("{IMG}", "").Replace("{TBL}", "").Replace("{PageBreak}", "")))
{
return;
}
if (dqParagraph.Meta.FigureDeclarations.Any())
{
if (dqParagraph.Style.FontSize > Settings.Default.ExpectedFontSize)
{
dqParagraph.Meta.Errors.Add(new DqFontSizeError($"Неверный размер шрифта ({dqParagraph.Style.FontSize} пт). Подписи риcунков должны использовать шрифт не больше основного ({Settings.Default.ExpectedFontSize} пт)."));
}
}
else if (dqParagraph.Meta.TableDeclarations.Any())
{
if (dqParagraph.Style.FontSize > Settings.Default.ExpectedFontSize)
{
dqParagraph.Meta.Errors.Add(new DqFontSizeError($"Неверный размер шрифта ({dqParagraph.Style.FontSize} пт). Заголовки таблиц должны использовать шрифт не больше основного ({Settings.Default.ExpectedFontSize} пт)."));
}
}
else if (dqParagraph.Meta.IsHeader)
{
if (dqParagraph.Style.FontSize < Settings.Default.ExpectedFontSize)
{
dqParagraph.Meta.Errors.Add(new DqFontSizeError($"Неверный размер шрифта ({dqParagraph.Style.FontSize} пт). Заголовки должны использовать шрифт не меньше основного ({Settings.Default.ExpectedFontSize} пт)."));
}
if (dqParagraph.GetPureText().Trim().EndsWith("."))
{
dqParagraph.Meta.Errors.Add(new DqError($"В конце заголовков точка не ставится."));
}
var number = Regex.Match(dqParagraph.GetPureText().TrimStart(), @"^((?:\d+\.)*\d+?)");
if (number.Success && number.Value.EndsWith("."))
{
dqParagraph.Meta.Errors.Add(new DqError($"В конце номера (под)раздела точка не ставится."));
}
}
if (dqParagraph.Style.FontName != Settings.Default.ExpectedFontName)
{
dqParagraph.Meta.Errors.Add(new DqFontError($"Неверный шрифт ({dqParagraph.Style.FontName}). Иcпользуйте {Settings.Default.ExpectedFontName}."));
}
if (dqParagraph.Meta.IsHeader)
{
if (dqParagraph.Style.Aligment == DqAligment.Center || dqParagraph.Style.Aligment == DqAligment.Right)
{
dqParagraph.Meta.Errors.Add(new DqAlignmentError($"Заголовки главной части должны быть выровнены по левому краю."));
}
else if (dqParagraph.Style.Indent == 0)
{
dqParagraph.Meta.Errors.Add(new DqAlignmentError($"Отсутвует абзацный отступ."));
}
}
else if (dqParagraph.Meta.FigureDeclarations.Any())
{
if (dqParagraph.Style.Aligment != DqAligment.Center)
{
dqParagraph.Meta.Errors.Add(new DqAlignmentError($"Подпись рисунка должна быть выровнена по центру."));
}
else if (dqParagraph.Style.Indent != 0)
{
dqParagraph.Meta.Errors.Add(new DqAlignmentError($"При выравнивании по центру должен отсутвовать абзацный отступ."));
}
}
else if (dqParagraph.Meta.TableDeclarations.Any())
{
if (dqParagraph.Style.Aligment == DqAligment.Center || dqParagraph.Style.Aligment == DqAligment.Right)
{
dqParagraph.Meta.Errors.Add(new DqAlignmentError($"Заголовок таблицы должен быть выровнен по левому краю."));
}
else if (dqParagraph.Style.Indent != 0)
{
dqParagraph.Meta.Errors.Add(new DqAlignmentError($"Заголовок таблицы помещают без абзацного отступа."));
}
}
else if (!dqParagraph.Meta.IsHeader)
{
if (dqParagraph.Style.Indent == 0)
{
dqParagraph.Meta.Errors.Add(new DqAlignmentError($"Отсутствует абзацный отступ."));
}
}
if (dqParagraph.Style.SpacingBetweenLines != Settings.Default.ExpectedSpacingBetweenLines && !dqParagraph.Meta.IsHeader && !dqParagraph.Meta.FigureDeclarations.Any() && !dqParagraph.Meta.TableDeclarations.Any())
{
dqParagraph.Meta.Errors.Add(new DqError($"Неверный междустрочный интервал ({dqParagraph.Style.SpacingBetweenLines}). Ожидается {Settings.Default.ExpectedSpacingBetweenLines}-ый интервал."));
}
}
public DqStructure PrimaryParse(DqDocument document)
{
var partTypeByTitle = TitlesByPartType
.SelectMany(kv => kv.Value.Select(v => (kv.Key, v)))
.ToDictionary(kv => kv.v, kv => kv.Key);
var partStart = 0;
var dqParts = new List <DqPart>();
while (true)
{
var nextPartStart = TryFindByTitle(document, partTypeByTitle.Keys, partStart);
if (nextPartStart == null)
{
break;
}
var start = document.Paragraphs[nextPartStart.Value];
dqParts.Add(new DqPart
{
Type = partTypeByTitle[start.GetPureText().ToLower()],
Start = start,
});
partStart = nextPartStart.Value;
}
for (var i = 0; i < dqParts.Count - 1; ++i)
{
var dqPart = dqParts[i];
dqPart.Paragraphs.AddRange(document.Paragraphs.GetRange(dqPart.Start.Index, dqParts[i + 1].Start.Index - dqPart.Start.Index));
}
if (!dqParts.Any())
{
var mainPart = new DqMainPart();
mainPart.Paragraphs.AddRange(document.Paragraphs);
mainPart.Start = document.Paragraphs.First();
return(new DqStructure {
MainPart = mainPart
});
}
var lastPart = dqParts.Last();
lastPart.Paragraphs.AddRange(document.Paragraphs.GetRange(lastPart.Start.Index, document.Paragraphs.Count - lastPart.Start.Index));
var report = new DqStructure();
report.Title = CopyContent(new DqPart
{
Type = PartType.Title,
Start = document.Paragraphs.First()
}, document, 0, dqParts.First().Start.Index);
foreach (var dqPart in dqParts)
{
switch (dqPart.Type)
{
case PartType.Abstract:
report.Abstracts.Add(dqPart);
break;
case PartType.Toc:
report.Toc = dqPart;
break;
case PartType.Introduction:
report.Introduction = dqPart;
break;
case PartType.Conclusion:
report.Conclusion = dqPart;
break;
case PartType.Bibliography:
report.Bibliography = dqPart;
break;
case PartType.Annex:
report.Appendixes = dqPart;
break;
}
}
return(report);
}
private DqPart CopyContent(DqPart dqPart, DqDocument dqDocument, int partStart, int?partEnd)
{
dqPart.Paragraphs.AddRange(dqDocument.Paragraphs.GetRange(partStart, partEnd.Value - partStart));
return(dqPart);
}
public void ParseReferences(DqDocument document)
{
var figureReferenceParser = new DqFigureReferenceParser();
var tableReferenceParser = new DqTableReferenceParser();
var sourceReferenceParser = new DqSourceReferenceParser();
foreach (var paragraph in document.Paragraphs)
{
figureReferenceParser.Parse(paragraph);
tableReferenceParser.Parse(paragraph);
sourceReferenceParser.Parse(paragraph, document);
}
var structure = document.Paragraphs.SelectMany(p => p.Meta.Structure).OfType <DqNumberedElement>().ToList();
var allFigureDeclarations = structure.Where(s => s.Type == DqStructureElementType.FigureDeclaration).ToList();
var allFigureReferences = structure.Where(s => s.Type == DqStructureElementType.FigureReference).ToList();
foreach (var figureReference in allFigureReferences)
{
figureReference.IsMissing = allFigureDeclarations.All(fd => fd.Number != figureReference.Number);
}
for (var i = 1; i < allFigureDeclarations.Count; i++)
{
var n1 = DqNumber.TryParse(allFigureDeclarations[i - 1].Number);
var n2 = DqNumber.TryParse(allFigureDeclarations[i].Number);
if (n1 != null && n2 != null && n1.CompareTo(n2) >= 0)
{
allFigureDeclarations[i].Paragraph.Meta.Errors.Add(
new DqError($"Неправильный порядок нумерации ({allFigureDeclarations[i].Number} после {allFigureDeclarations[i - 1].Number})"));
}
}
foreach (var figureDeclaration in allFigureDeclarations)
{
var firstReference = allFigureReferences.FirstOrDefault(r => r.Number == figureDeclaration.Number);
if (firstReference == null)
{
figureDeclaration.IsMissing = true;
}
else if (firstReference.Paragraph.Index >= figureDeclaration.Paragraph.Index)
{
firstReference.Paragraph.Meta.Errors.Add(new DqError("Первая ссылка должна быть до рисунка."));
}
}
var allTableDeclarations = structure.Where(s => s.Type == DqStructureElementType.TableDeclaration).ToList();
var allTableReferences = structure.Where(s => s.Type == DqStructureElementType.TableReference).ToList();
foreach (var tableReference in allTableReferences)
{
tableReference.IsMissing = allTableDeclarations.All(fd => fd.Number != tableReference.Number);
}
for (var i = 1; i < allTableDeclarations.Count; i++)
{
var n1 = DqNumber.TryParse(allTableDeclarations[i - 1].Number);
var n2 = DqNumber.TryParse(allTableDeclarations[i].Number);
if (n1 != null && n2 != null && n1.CompareTo(n2) >= 0)
{
allTableDeclarations[i].Paragraph.Meta.Errors.Add(
new DqError($"Неправильный порядок нумерации ({allTableDeclarations[i].Number} после {allTableDeclarations[i - 1].Number})"));
}
}
foreach (var tableDeclaration in allTableDeclarations)
{
var firstReference = allTableReferences.FirstOrDefault(r => r.Number == tableDeclaration.Number);
if (firstReference == null)
{
tableDeclaration.IsMissing = true;
}
else if (firstReference.Paragraph.Index >= tableDeclaration.Paragraph.Index)
{
firstReference.Paragraph.Meta.Errors.Add(new DqError("Первая ссылка должна быть до таблицы."));
}
}
var allSourceDeclarations = structure.Where(s => s.Type == DqStructureElementType.SourceDeclaration).ToList();
var allSourceReferences = structure.Where(s => s.Type == DqStructureElementType.SourceReference).ToList();
foreach (var sourceReference in allSourceReferences)
{
sourceReference.IsMissing = allSourceDeclarations.All(fd => fd.Number != sourceReference.Number);
}
foreach (var sourceDeclaration in allSourceDeclarations)
{
var firstReference = allSourceReferences.FirstOrDefault(r => r.Number == sourceDeclaration.Number);
if (firstReference == null)
{
sourceDeclaration.IsMissing = true;
}
}
}