public static void Parse(Md2MlEngine engine, string mdText) { var lineAndPattern = new List <KeyValuePair <ParaPattern, string[]> >(); var lines = mdText.Split('\n'); foreach (var line in lines) { lineAndPattern.Add(PatternMatcher.GetParagraphType(line)); } bool OrderedList = false; bool UnorderedList = false; bool Table = false; List <string> BulletItems = new List <string>(); List <string> NumberItems = new List <string>(); List <string> TableData = new List <string>(); foreach (var line in lineAndPattern) { Paragraph para; switch (line.Key) { case ParaPattern.OrderedList: if (Table) { ProcessTable(engine, TableData); Table = false; } if (UnorderedList) { ProcessBullets(engine, BulletItems, false); UnorderedList = false; } OrderedList = true; NumberItems.Add(line.Value[1]); break; case ParaPattern.UnorderedList: if (Table) { ProcessTable(engine, TableData); Table = false; } if (OrderedList) { ProcessBullets(engine, NumberItems, true); OrderedList = false; } UnorderedList = true; BulletItems.Add(line.Value[1]); break; case ParaPattern.Image: if (Table) { ProcessTable(engine, TableData); Table = false; } if (OrderedList || UnorderedList) { ProcessBullets(engine, OrderedList ? NumberItems : BulletItems, OrderedList); OrderedList = false; UnorderedList = false; } para = engine.CreateParagraph(); if (line.Value[2].StartsWith("http://") || line.Value[2].StartsWith("https://")) { engine.AddImage(new System.Net.WebClient().OpenRead(line.Value[2])); } else { engine.AddImage(System.IO.File.OpenRead(line.Value[2])); } break; case ParaPattern.Table: if (OrderedList || UnorderedList) { ProcessBullets(engine, OrderedList ? NumberItems : BulletItems, OrderedList); OrderedList = false; UnorderedList = false; } Table = true; TableData.Add(line.Value[1]); break; case ParaPattern.CodeBlock: if (Table) { ProcessTable(engine, TableData); Table = false; } if (OrderedList || UnorderedList) { ProcessBullets(engine, OrderedList ? NumberItems : BulletItems, OrderedList); OrderedList = false; UnorderedList = false; } para = engine.CreateParagraph(new ParaProperties() { StyleName = "CodeBlock" }); engine.WriteText(para, line.Value[1]); break; case ParaPattern.Heading1: if (Table) { ProcessTable(engine, TableData); Table = false; } if (OrderedList || UnorderedList) { ProcessBullets(engine, OrderedList ? NumberItems : BulletItems, OrderedList); OrderedList = false; UnorderedList = false; } para = engine.CreateParagraph(new ParaProperties() { StyleName = "Heading1" }); engine.WriteText(para, line.Value[1]); break; case ParaPattern.Heading2: if (Table) { ProcessTable(engine, TableData); Table = false; } if (OrderedList || UnorderedList) { ProcessBullets(engine, OrderedList ? NumberItems : BulletItems, OrderedList); OrderedList = false; UnorderedList = false; } para = engine.CreateParagraph(new ParaProperties() { StyleName = "Heading2" }); engine.WriteText(para, line.Value[1]); break; case ParaPattern.Heading3: if (Table) { ProcessTable(engine, TableData); Table = false; } if (OrderedList || UnorderedList) { ProcessBullets(engine, OrderedList ? NumberItems : BulletItems, OrderedList); OrderedList = false; UnorderedList = false; } para = engine.CreateParagraph(new ParaProperties() { StyleName = "Heading3" }); engine.WriteText(para, line.Value[1]); break; case ParaPattern.Quote: if (Table) { ProcessTable(engine, TableData); Table = false; } if (OrderedList || UnorderedList) { ProcessBullets(engine, OrderedList ? NumberItems : BulletItems, OrderedList); OrderedList = false; UnorderedList = false; } para = engine.CreateParagraph(new ParaProperties() { StyleName = "Quote" }); engine.WriteText(para, line.Value[1]); break; case ParaPattern.CommanBlock: default: if (Table) { ProcessTable(engine, TableData); Table = false; } if (OrderedList || UnorderedList) { ProcessBullets(engine, OrderedList ? NumberItems : BulletItems, OrderedList); OrderedList = false; UnorderedList = false; } para = engine.CreateParagraph(); FormatText(engine, para, line.Value[1], new FontProperties()); //core.WriteText(para, line.Value[1]); break; } } if (Table) { ProcessTable(engine, TableData); Table = false; } if (OrderedList || UnorderedList) { ProcessBullets(engine, OrderedList ? NumberItems : BulletItems, OrderedList); OrderedList = false; UnorderedList = false; } }
/// <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; } } }