/// <summary> /// TODO : Problem to fix, all lists are represented as Numbered... /// Process an array of markdown items of a list (with spaces to define the item level) /// The pattern can be detected as a CodeBlock pattern, so fix it with the try catch. /// I should probably not let defining a CodeBlock with spaces and tabulations, /// but only with tabulations... (WIP) /// </summary> /// <param name="core"></param> /// <param name="bulletedItems">Items of a list, split in an array by break lines</param> /// <param name="paragraphStyle"></param> public void MarkdownList(Md2MlEngine core, string[] bulletedItems, string paragraphStyle = "ParagraphList") { foreach (var item in bulletedItems) { // Detect if item is ordered or not var matchedPattern = PatternMatcher.GetMarkdownMatch(item); if (matchedPattern.Key != ParaPattern.OrderedList && matchedPattern.Key != ParaPattern.UnorderedList) { try { matchedPattern = PatternMatcher.GetMatchFromPattern(item, ParaPattern.OrderedList); } catch (Exception e) { matchedPattern = PatternMatcher.GetMatchFromPattern(item, ParaPattern.UnorderedList); } } // Then count spaces 3 by 3 to define the level of the item var nbSpaces = matchedPattern.Value.Groups[0].Value.TakeWhile(Char.IsWhiteSpace).Count();; var itemLvl = nbSpaces / 3; // Then Create paragraph, properties and format the text Paragraph paragraph1 = CreateParagraph(paragraphStyle); NumberingProperties numberingProperties1 = new NumberingProperties(); NumberingLevelReference numberingLevelReference1 = new NumberingLevelReference() { Val = itemLvl }; NumberingId numberingId1 = GetListType(matchedPattern.Key); numberingProperties1.Append(numberingLevelReference1); numberingProperties1.Append(numberingId1); paragraph1.ParagraphProperties.Append(numberingProperties1); MarkdownStringParser.FormatText(core, paragraph1, matchedPattern.Value.Groups[2].Value, new StyleProperties()); } }
/// <summary> /// Process a full string block of a markdown table. /// Split it by rows /// Create the Table with correct number of columns /// Fill the openXML table /// </summary> /// <param name="engine"></param> /// <param name="markdown">The complete markdown table, must be correctly formatted</param> private static void ProcessTable(Md2MlEngine engine, string markdown) { var rows = Regex.Split(markdown, "\r\n|\r|\n"); var firstLine = rows.First(); var secondLine = rows.Length >= 2 ? rows.Skip(1).First() : null; var table = engine.CreateTable(firstLine.Trim('|').Split('|').Count()); engine.AddTableRow(table, firstLine.Trim('|').Split('|').ToList()); var patternSecondLine = PatternMatcher.GetMarkdownMatch(secondLine).Key; if (string.IsNullOrEmpty(secondLine) || (patternSecondLine != ParaPattern.TableHeaderSeparation && patternSecondLine != ParaPattern.TableHeaderSeparation)) // TODO : Throw error : Table not well formatted { return; } // Define the table alignment properties List <JustificationValues> cellJustification = new List <JustificationValues>(); var nbCols = secondLine.Trim('|').Split('|').Count(); var secondLineCells = secondLine.Trim('|').Split('|').ToList(); for (int i = 0; i < nbCols; i++) { var justification = JustificationValues.Left; if (secondLineCells[i].StartsWith(":") && secondLineCells[i].EndsWith(":")) { justification = JustificationValues.Center; } else if (!secondLineCells[i].StartsWith(":") && secondLineCells[i].EndsWith(":")) { justification = JustificationValues.Right; } cellJustification.Add(justification); } // Process the rest of the table foreach (var row in rows.Skip(2).ToList()) { engine.AddTableRow(table, row.Trim('|').Split('|').ToList(), cellJustification); } }
/// <summary> /// This method allows you to rebuild a paragraph without interpreting breaking lines if there is not a double space before. /// Produces a counter in order to know how many lines are concatenated, in order to delete it after parsing it /// </summary> /// <param name="mdText">The text to parse line by line</param> /// <param name="pattern">The pattern to detect real new lines. Allows to create a block with same nature (lists, paragraphs...)</param> /// <returns></returns> private static (int counter, string textBlock) BuildWithoutBreakingLines(string mdText, ParaPattern pattern) { var lines = Regex.Split(mdText, "\r\n|\r|\n").ToArray(); var previousLine = lines.First().TrimStart('>'); var output = new StringBuilder(previousLine); int count = 1; if (string.IsNullOrEmpty(previousLine)) { return(counter : count, textBlock : output.ToString()); } foreach (var line in lines.Skip(1)) { // Break directly if processed line is empty if (string.IsNullOrEmpty(line)) { break; } var linePattern = PatternMatcher.GetMarkdownMatch(line).Key; // If first pattern is a table, do not break until pattern does not match to any table pattern bool isTableContinuing = pattern == ParaPattern.Table && ( linePattern == ParaPattern.Table || linePattern == ParaPattern.TableHeaderSeparation); // Quotes are continuing if first pattern is quote // And the current one equals to Quote or AnyChar bool isQuoteContinuing = pattern == ParaPattern.Quote && (linePattern == ParaPattern.Quote || linePattern == ParaPattern.AnyChar); // Paragraph types are continuing if previous line does not ends with double space // And first pattern matches with the current one bool isParagraphContinuing = (!previousLine.EndsWith(" ") && (pattern == ParaPattern.AnyChar && linePattern == pattern)); // If list is not continuing, then check if item is detected as a code block bool isListContinuing = (pattern == ParaPattern.OrderedList || pattern == ParaPattern.UnorderedList) && (linePattern == ParaPattern.OrderedList || linePattern == ParaPattern.UnorderedList); if ((pattern == ParaPattern.OrderedList || pattern == ParaPattern.UnorderedList) && !isListContinuing && linePattern == ParaPattern.CodeBlock) { try { linePattern = PatternMatcher.GetMatchFromPattern(line, ParaPattern.OrderedList).Key; } catch (Exception e) { linePattern = PatternMatcher.GetMatchFromPattern(line, ParaPattern.UnorderedList).Key; } isListContinuing = (linePattern == ParaPattern.OrderedList || linePattern == ParaPattern.UnorderedList); } if (!isTableContinuing && !isQuoteContinuing && !isParagraphContinuing && !isListContinuing) { break; } if (pattern == ParaPattern.TableHeaderSeparation || pattern == ParaPattern.Table || pattern == ParaPattern.OrderedList || pattern == ParaPattern.UnorderedList || pattern == ParaPattern.Quote) { output.AppendLine().Append(line.TrimStart('>')); } else { output.Append(line); } previousLine = line; count++; } return(counter : count, textBlock : output.ToString()); }
/// <summary> /// Parse the content, and detect all real lines breaks. /// In fact, within markdown it is possible to write a unique paragraph, with line breaks. /// If there is not a double space before the line break, it's the same paragraph, and the line break is not interpreted as is. /// /// So in this parsing method, cut the content by "block" with same nature. /// 1| Detect the paragraph type by the start of the line /// 2| Process the content by its type /// 3| Then remove the processed content, and continue to parse the next content /// /// A block is detected by a <see cref="ParaPattern"/> which have an associated regex. /// /// </summary> /// <param name="engine">Describe an openXML object by code</param> /// <param name="mdText">The markdown text to parse</param> internal static void Parse(Md2MlEngine engine, string mdText) { while (!string.IsNullOrEmpty(mdText)) { var firstLine = GetFirstLine(mdText); var matchedPattern = PatternMatcher.GetMarkdownMatch(firstLine); Paragraph para; (int counter, string textBlock)rebuildText = default; switch (matchedPattern.Key) { case ParaPattern.InfiniteHeading: string titleChars = matchedPattern.Value.Groups[1].Value; int titleLvl = titleChars.Count(c => c == '#'); titleLvl = titleLvl <= 9 ? titleLvl : 9; para = engine.CreateParagraph(new ParaProperties() { StyleName = string.Concat("Heading", titleLvl.ToString()) }); engine.WriteText(para, matchedPattern.Value.Groups[2].Value); mdText = DeleteLines(mdText); continue; case ParaPattern.Image: var link = matchedPattern.Value.Groups[2].Value; if (link.StartsWith("http://") || link.StartsWith("https://")) { engine.AddImage(new System.Net.WebClient().OpenRead(link)); } else { engine.AddImage(ConvertRelativeToAbsolutePath(link, engine.GetFileDirectory())); } mdText = DeleteLines(mdText); continue; case ParaPattern.Table: //case ParaPattern.TableHeaderSeparation: rebuildText = BuildWithoutBreakingLines(mdText, ParaPattern.Table); ProcessTable(engine, rebuildText.textBlock); mdText = DeleteLines(mdText, rebuildText.counter); continue; case ParaPattern.OrderedList: case ParaPattern.UnorderedList: // TODO : Ordered list can contain unordered items inside and vice versa rebuildText = BuildWithoutBreakingLines(mdText, ParaPattern.OrderedList); ProcessBullets(engine, rebuildText.textBlock); mdText = DeleteLines(mdText, rebuildText.counter); continue; case ParaPattern.CodeBlock: // TODO : Improve the rendering - not a priority for my needs rebuildText = BuildWithoutBreakingLines(mdText, ParaPattern.CodeBlock); para = engine.CreateParagraph(new ParaProperties() { StyleName = DocStyles.CodeBlock.ToDescriptionString() }); FormatText(engine, para, rebuildText.textBlock, new StyleProperties()); mdText = DeleteLines(mdText, rebuildText.counter); continue; case ParaPattern.Quote: // Markdown supports nested quotes, but word does not // So whatever, put the "nested" paragraph in the same quote rebuildText = BuildWithoutBreakingLines(mdText, ParaPattern.Quote); para = engine.CreateParagraph(new ParaProperties() { StyleName = DocStyles.Quote.ToDescriptionString() }); foreach (var text in Regex.Split(rebuildText.textBlock, "\r\n|\r|\n")) { para.AppendChild(new Break()); FormatText(engine, para, text, new StyleProperties()); } mdText = DeleteLines(mdText, rebuildText.counter); continue; case ParaPattern.AnyChar: default: rebuildText = BuildWithoutBreakingLines(mdText, matchedPattern.Key); para = engine.CreateParagraph(); FormatText(engine, para, rebuildText.textBlock, new StyleProperties()); // engine.WriteText(para, text.textBlock); mdText = DeleteLines(mdText, rebuildText.counter); continue; } } }