public string _loc; // generated lazily.
public SourceLocation(string file, SectionRef section, MarkdownParagraph paragraph, MarkdownSpan span)
{
File = file;
Section = section;
Paragraph = paragraph;
Span = span;
}
private static int BugWorkaroundIndent(ref MarkdownParagraph mdp, int level)
{
if (!mdp.IsParagraph)
{
return(level);
}
var p = mdp as MarkdownParagraph.Paragraph;
var spans = p.Item;
if (spans.Count() == 0 || !spans[0].IsLiteral)
{
return(level);
}
var literal = spans[0] as MarkdownSpan.Literal;
if (!literal.Item.StartsWith("ceci-n'est-pas-une-indent"))
{
return(level);
}
//
var literal2 = MarkdownSpan.NewLiteral(literal.Item.Substring(25));
var spans2 = Microsoft.FSharp.Collections.FSharpList <MarkdownSpan> .Cons(literal2, spans.Tail);
var p2 = MarkdownParagraph.NewParagraph(spans2);
mdp = p2;
return(0);
}
public static Span FindSpan(string src, MarkdownSpan mds)
{
var mdwords = Span2Words(mds).ToList();
var srcwords = FindWords(src).ToList();
var srcwordsProjection = srcwords.Select(w => w.word);
var s = LevenshteinSearch(mdwords, srcwordsProjection);
if (s == null)
{
return(null);
}
var wordStart = s.start;
var wordEnd = s.start + s.length;
var spanStart = srcwords[wordStart].span.start;
var spanEnd = (wordEnd == srcwords.Count ? srcwords.Last().span.start + srcwords.Last().span.length : srcwords[wordEnd].span.start);
return(new Span(spanStart, spanEnd - spanStart));
}
private static IEnumerable <string> Span2Words(MarkdownSpan md)
{
if (md.IsLiteral)
{
var mdl = md as MarkdownSpan.Literal;
foreach (var s in String2Words(MarkdownSpec.mdunescape(mdl)))
{
yield return(s);
}
}
else if (md.IsStrong)
{
var mds = md as MarkdownSpan.Strong;
foreach (var s in Spans2Words(mds.body))
{
yield return(s);
}
}
else if (md.IsEmphasis)
{
var mde = md as MarkdownSpan.Emphasis;
foreach (var s in Spans2Words(mde.body))
{
yield return(s);
}
}
else if (md.IsInlineCode)
{
var mdi = md as MarkdownSpan.InlineCode;
foreach (var s in String2Words(mdi.code))
{
yield return(s);
}
}
else if (md.IsDirectLink)
{
var mdl = md as MarkdownSpan.DirectLink;
foreach (var s in Spans2Words(mdl.body))
{
yield return(s);
}
foreach (var s in String2Words(mdl.title.Option()))
{
yield return(s);
}
}
else if (md.IsIndirectLink)
{
var mdl = md as MarkdownSpan.DirectLink;
foreach (var s in Spans2Words(mdl.body))
{
yield return(s);
}
foreach (var s in String2Words(mdl.link))
{
yield return(s);
}
foreach (var s in String2Words(mdl.title.Option()))
{
yield return(s);
}
}
else if (md.IsAnchorLink)
{
var mdl = md as MarkdownSpan.AnchorLink;
foreach (var s in String2Words(mdl.link))
{
yield return(s);
}
}
else if (md.IsDirectImage)
{
var mdi = md as MarkdownSpan.DirectImage;
foreach (var s in String2Words(mdi.body))
{
yield return(s);
}
foreach (var s in String2Words(mdi.link))
{
yield return(s);
}
foreach (var s in String2Words(mdi.title.Option()))
{
yield return(s);
}
}
else if (md.IsIndirectImage)
{
var mdi = md as MarkdownSpan.IndirectImage;
foreach (var s in String2Words(mdi.body))
{
yield return(s);
}
foreach (var s in String2Words(mdi.link))
{
yield return(s);
}
foreach (var s in String2Words(mdi.key))
{
yield return(s);
}
}
else if (md.IsEmbedSpans)
{
}
else if (md.IsHardLineBreak)
{
}
else if (md.IsLatexDisplayMath)
{
var mdl = md as MarkdownSpan.LatexDisplayMath;
foreach (var s in String2Words(mdl.code))
{
yield return(s);
}
}
else if (md.IsLatexInlineMath)
{
var mdl = md as MarkdownSpan.LatexInlineMath;
foreach (var s in String2Words(mdl.code))
{
yield return(s);
}
}
}
IEnumerable <OpenXmlCompositeElement> Paragraph2Paragraphs(MarkdownParagraph md)
{
if (md.IsHeading)
{
var mdh = md as MarkdownParagraph.Heading;
var level = mdh.Item1;
var spans = mdh.Item2;
var sr = sections[new SectionRef(mdh, filename).Url];
var props = new ParagraphProperties(new ParagraphStyleId()
{
Val = $"Heading{level}"
});
var p = new Paragraph {
ParagraphProperties = props
};
maxBookmarkId.Value += 1;
p.AppendChild(new BookmarkStart {
Name = sr.BookmarkName, Id = maxBookmarkId.Value.ToString()
});
p.Append(Spans2Elements(spans));
p.AppendChild(new BookmarkEnd {
Id = maxBookmarkId.Value.ToString()
});
yield return(p);
Console.WriteLine(new string(' ', level * 4 - 4) + sr.Number + " " + sr.Title);
yield break;
}
else if (md.IsParagraph)
{
var mdp = md as MarkdownParagraph.Paragraph;
var spans = mdp.Item;
yield return(new Paragraph(Spans2Elements(spans)));
yield break;
}
else if (md.IsListBlock)
{
var mdl = md as MarkdownParagraph.ListBlock;
var flat = FlattenList(mdl);
// Let's figure out what kind of list it is - ordered or unordered? nested?
var format0 = new[] { "1", "1", "1", "1" };
foreach (var item in flat)
{
format0[item.Level] = (item.IsBulletOrdered ? "1" : "o");
}
var format = string.Join("", format0);
var numberingPart = wdoc.MainDocumentPart.NumberingDefinitionsPart ?? wdoc.MainDocumentPart.AddNewPart <NumberingDefinitionsPart>("NumberingDefinitionsPart001");
if (numberingPart.Numbering == null)
{
numberingPart.Numbering = new Numbering();
}
Func <int, bool, Level> createLevel;
createLevel = (level, isOrdered) =>
{
var numformat = NumberFormatValues.Bullet;
var levelText = new[] { "·", "o", "·", "o" }[level];
if (isOrdered && level == 0)
{
numformat = NumberFormatValues.Decimal; levelText = "%1.";
}
if (isOrdered && level == 1)
{
numformat = NumberFormatValues.LowerLetter; levelText = "%2.";
}
if (isOrdered && level == 2)
{
numformat = NumberFormatValues.LowerRoman; levelText = "%3.";
}
if (isOrdered && level == 3)
{
numformat = NumberFormatValues.LowerRoman; levelText = "%4.";
}
var r = new Level {
LevelIndex = level
};
r.Append(new StartNumberingValue {
Val = 1
});
r.Append(new NumberingFormat {
Val = numformat
});
r.Append(new LevelText {
Val = levelText
});
r.Append(new ParagraphProperties(new Indentation {
Left = (540 + 360 * level).ToString(), Hanging = "360"
}));
if (levelText == "·")
{
r.Append(new NumberingSymbolRunProperties(new RunFonts {
Hint = FontTypeHintValues.Default, Ascii = "Symbol", HighAnsi = "Symbol", EastAsia = "Times new Roman", ComplexScript = "Times new Roman"
}));
}
if (levelText == "o")
{
r.Append(new NumberingSymbolRunProperties(new RunFonts {
Hint = FontTypeHintValues.Default, Ascii = "Courier New", HighAnsi = "Courier New", ComplexScript = "Courier New"
}));
}
return(r);
};
var level0 = createLevel(0, format[0] == '1');
var level1 = createLevel(1, format[1] == '1');
var level2 = createLevel(2, format[2] == '1');
var level3 = createLevel(3, format[3] == '1');
var abstracts = numberingPart.Numbering.OfType <AbstractNum>().Select(an => an.AbstractNumberId.Value).ToList();
var aid = (abstracts.Count == 0 ? 1 : abstracts.Max() + 1);
var aabstract = new AbstractNum(new MultiLevelType()
{
Val = MultiLevelValues.Multilevel
}, level0, level1, level2, level3)
{
AbstractNumberId = aid
};
numberingPart.Numbering.InsertAt(aabstract, 0);
var instances = numberingPart.Numbering.OfType <NumberingInstance>().Select(ni => ni.NumberID.Value);
var nid = (instances.Count() == 0 ? 1 : instances.Max() + 1);
var numInstance = new NumberingInstance(new AbstractNumId {
Val = aid
})
{
NumberID = nid
};
numberingPart.Numbering.AppendChild(numInstance);
// We'll also figure out the indentation(for the benefit of those paragraphs that should be
// indendent with the list but aren't numbered). I'm not sure what the indent comes from.
// in the docx, each AbstractNum that I created has an indent for each of its levels,
// defaulted at 900, 1260, 1620, ... but I can't see where in the above code that's created?
Func <int, string> calcIndent = level => (540 + level * 360).ToString();
foreach (var item in flat)
{
var content = item.Paragraph;
if (content.IsParagraph || content.IsSpan)
{
var spans = (content.IsParagraph ? (content as MarkdownParagraph.Paragraph).Item : (content as MarkdownParagraph.Span).Item);
if (item.HasBullet)
{
yield return new Paragraph(Spans2Elements(spans))
{
ParagraphProperties = new ParagraphProperties(new NumberingProperties(new ParagraphStyleId {
Val = "ListParagraph"
}, new NumberingLevelReference {
Val = item.Level
}, new NumberingId {
Val = nid
}))
}
}
;
else
{
yield return new Paragraph(Spans2Elements(spans))
{
ParagraphProperties = new ParagraphProperties(new Indentation {
Left = calcIndent(item.Level)
})
}
};
}
else if (content.IsQuotedBlock || content.IsCodeBlock)
{
foreach (var p in Paragraph2Paragraphs(content))
{
var props = p.GetFirstChild <ParagraphProperties>();
if (props == null)
{
props = new ParagraphProperties(); p.InsertAt(props, 0);
}
var indent = props?.GetFirstChild <Indentation>();
if (indent == null)
{
indent = new Indentation(); props.Append(indent);
}
indent.Left = calcIndent(item.Level);
yield return(p);
}
}
else if (content.IsTableBlock)
{
foreach (var p in Paragraph2Paragraphs(content))
{
var table = p as Table;
if (table == null)
{
yield return(p); continue;
}
var tprops = table.GetFirstChild <TableProperties>();
var tindent = tprops?.GetFirstChild <TableIndentation>();
if (tindent == null)
{
throw new Exception("Ooops! Table is missing indentation");
}
tindent.Width = int.Parse(calcIndent(item.Level));
yield return(table);
}
}
else
{
throw new Exception("Unexpected item in list");
}
}
}
else if (md.IsCodeBlock)
{
var mdc = md as MarkdownParagraph.CodeBlock;
var code = mdc.Item1;
var lang = mdc.Item2;
code = BugWorkaroundDecode(code);
var runs = new List <Run>();
var onFirstLine = true;
IEnumerable <ColorizedLine> lines;
if (lang == "csharp" || lang == "c#" || lang == "cs")
{
lines = Colorize.CSharp(code);
}
else if (lang == "vb" || lang == "vbnet" || lang == "vb.net")
{
lines = Colorize.VB(code);
}
else if (lang == "" || lang == "xml")
{
lines = Colorize.PlainText(code);
}
else if (lang == "antlr")
{
lines = Antlr.ColorizeAntlr(code);
}
else
{
throw new NotSupportedException($"unrecognized language {lang}");
}
foreach (var line in lines)
{
if (onFirstLine)
{
onFirstLine = false;
}
else
{
runs.Add(new Run(new Break()));
}
foreach (var word in line.Words)
{
var run = new Run();
var props = new RunProperties();
if (word.Red != 0 || word.Green != 0 || word.Blue != 0)
{
props.Append(new Color {
Val = $"{word.Red:X2}{word.Green:X2}{word.Blue:X2}"
});
}
if (word.IsItalic)
{
props.Append(new Italic());
}
if (props.HasChildren)
{
run.Append(props);
}
run.Append(new Text(word.Text)
{
Space = SpaceProcessingModeValues.Preserve
});
runs.Add(run);
}
}
var style = new ParagraphStyleId {
Val = (lang == "antlr" ? "Grammar" : "Code")
};
yield return(new Paragraph(runs)
{
ParagraphProperties = new ParagraphProperties(style)
});
}
else if (md.IsQuotedBlock)
{
var mdq = md as MarkdownParagraph.QuotedBlock;
var quoteds = mdq.Item;
var kind = "";
foreach (var quoted0 in quoteds)
{
var quoted = quoted0;
if (quoted.IsParagraph)
{
var p = quoted as MarkdownParagraph.Paragraph;
var spans = p.Item;
if (spans.Any() && spans.First().IsStrong)
{
var strong = (spans.First() as MarkdownSpan.Strong).Item;
if (strong.Any() && strong.First().IsLiteral)
{
var literal = mdunescape(strong.First() as MarkdownSpan.Literal);
if (literal == "Note")
{
kind = "AlertText";
}
else if (literal == "Annotation")
{
kind = "Annotation";
yield return(new Paragraph(Span2Elements(spans.Head))
{
ParagraphProperties = new ParagraphProperties(new ParagraphStyleId {
Val = kind
})
});
if (spans.Tail.Any() && spans.Tail.First().IsLiteral)
{
quoted = MarkdownParagraph.NewParagraph(new Microsoft.FSharp.Collections.FSharpList <MarkdownSpan>(MarkdownSpan.NewLiteral(mdunescape(spans.Tail.First() as MarkdownSpan.Literal).TrimStart()), spans.Tail.Tail));
}
else
{
quoted = MarkdownParagraph.NewParagraph(spans.Tail);
}
}
}
}
//
foreach (var qp in Paragraph2Paragraphs(quoted))
{
var qpp = qp as Paragraph;
if (qpp != null)
{
var props = new ParagraphProperties(new ParagraphStyleId()
{
Val = kind
}); qpp.ParagraphProperties = props;
}
yield return(qp);
}
}
else if (quoted.IsCodeBlock)
{
var mdc = quoted as MarkdownParagraph.CodeBlock;
var code = mdc.Item1;
var lang = mdc.Item2;
var ignoredAfterLang = mdc.Item3;
var lines = code.Split(new[] { "\r\n", "\r", "\n" }, StringSplitOptions.None).ToList();
if (string.IsNullOrWhiteSpace(lines.Last()))
{
lines.RemoveAt(lines.Count - 1);
}
var run = new Run()
{
RunProperties = new RunProperties(new RunStyle {
Val = "CodeEmbedded"
})
};
foreach (var line in lines)
{
if (run.ChildElements.Count() > 1)
{
run.Append(new Break());
}
run.Append(new Text(" " + line)
{
Space = SpaceProcessingModeValues.Preserve
});
}
yield return(new Paragraph(run)
{
ParagraphProperties = new ParagraphProperties(new ParagraphStyleId {
Val = kind
})
});
}
else if (quoted.IsListBlock)
{
if (!(quoted as MarkdownParagraph.ListBlock).Item1.IsOrdered)
{
throw new NotImplementedException("unordered list inside annotation");
}
var count = 1;
foreach (var qp in Paragraph2Paragraphs(quoted))
{
var qpp = qp as Paragraph;
if (qpp == null)
{
yield return(qp); continue;
}
qp.InsertAt(new Run(new Text($"{count}. ")
{
Space = SpaceProcessingModeValues.Preserve
}), 0);
count += 1;
var props = new ParagraphProperties(new ParagraphStyleId()
{
Val = kind
});
qpp.ParagraphProperties = props;
yield return(qp);
}
}
}
}
else if (md.IsTableBlock)
{
var mdt = md as MarkdownParagraph.TableBlock;
var header = mdt.Item1.Option();
var align = mdt.Item2;
var rows = mdt.Item3;
var table = new Table();
var tstyle = new TableStyle {
Val = "TableGrid"
};
var tindent = new TableIndentation {
Width = 360, Type = TableWidthUnitValues.Dxa
};
var tborders = new TableBorders();
tborders.TopBorder = new TopBorder {
Val = BorderValues.Single
};
tborders.BottomBorder = new BottomBorder {
Val = BorderValues.Single
};
tborders.LeftBorder = new LeftBorder {
Val = BorderValues.Single
};
tborders.RightBorder = new RightBorder {
Val = BorderValues.Single
};
tborders.InsideHorizontalBorder = new InsideHorizontalBorder {
Val = BorderValues.Single
};
tborders.InsideVerticalBorder = new InsideVerticalBorder {
Val = BorderValues.Single
};
var tcellmar = new TableCellMarginDefault();
tcellmar.Append();
table.Append(new TableProperties(tstyle, tindent, tborders));
var ncols = align.Length;
for (int irow = -1; irow < rows.Length; irow++)
{
if (irow == -1 && header == null)
{
continue;
}
var mdrow = (irow == -1 ? header : rows[irow]);
var row = new TableRow();
for (int icol = 0; icol < Math.Min(ncols, mdrow.Length); icol++)
{
var mdcell = mdrow[icol];
var cell = new TableCell();
var pars = Paragraphs2Paragraphs(mdcell).ToList();
for (int ip = 0; ip < pars.Count; ip++)
{
var p = pars[ip] as Paragraph;
if (p == null)
{
cell.Append(pars[ip]); continue;
}
var props = new ParagraphProperties(new ParagraphStyleId {
Val = "TableCellNormal"
});
if (align[icol].IsAlignCenter)
{
props.Append(new Justification {
Val = JustificationValues.Center
});
}
if (align[icol].IsAlignRight)
{
props.Append(new Justification {
Val = JustificationValues.Right
});
}
p.InsertAt(props, 0);
cell.Append(pars[ip]);
}
if (pars.Count == 0)
{
cell.Append(new Paragraph(new ParagraphProperties(new SpacingBetweenLines {
After = "0"
}), new Run(new Text(""))));
}
row.Append(cell);
}
table.Append(row);
}
yield return(new Paragraph(new Run(new Text("")))
{
ParagraphProperties = new ParagraphProperties(new ParagraphStyleId {
Val = "TableLineBefore"
})
});
yield return(table);
yield return(new Paragraph(new Run(new Text("")))
{
ParagraphProperties = new ParagraphProperties(new ParagraphStyleId {
Val = "TableLineAfter"
})
});
}
else
{
yield return(new Paragraph(new Run(new Text($"[{md.GetType().Name}]"))));
}
}
IEnumerable <OpenXmlElement> Span2Elements(MarkdownSpan md)
{
if (md.IsLiteral)
{
var mdl = md as MarkdownSpan.Literal;
var s = mdunescape(mdl);
yield return(new Run(new Text(s)
{
Space = SpaceProcessingModeValues.Preserve
}));
}
else if (md.IsStrong || md.IsEmphasis)
{
IEnumerable <MarkdownSpan> spans = (md.IsStrong ? (md as MarkdownSpan.Strong).Item : (md as MarkdownSpan.Emphasis).Item);
// Workaround for https://github.com/tpetricek/FSharp.formatting/issues/389 - the markdown parser
// turns *this_is_it* into a nested Emphasis["this", Emphasis["is"], "it"] instead of Emphasis["this_is_it"]
// What we'll do is preprocess it into Emphasis["this", "_", "is" "_", "it"]
if (md.IsEmphasis)
{
var spans2 = new List <MarkdownSpan>();
foreach (var s in spans)
{
if (!s.IsEmphasis)
{
spans2.Add(s); continue;
}
spans2.Add(MarkdownSpan.NewLiteral("_"));
foreach (var ss in (s as MarkdownSpan.Emphasis).Item)
{
spans2.Add(ss);
}
spans2.Add(MarkdownSpan.NewLiteral("_"));
}
spans = spans2;
}
foreach (var e in Spans2Elements(spans))
{
var style = (md.IsStrong ? new Bold() as OpenXmlElement : new Italic());
var run = e as Run;
if (run != null)
{
run.InsertAt(new RunProperties(style), 0);
}
yield return(e);
}
}
else if (md.IsInlineCode)
{
var mdi = md as MarkdownSpan.InlineCode;
var code = mdi.Item;
var txt = new Text(BugWorkaroundDecode(code))
{
Space = SpaceProcessingModeValues.Preserve
};
var props = new RunProperties(new RunStyle {
Val = "CodeEmbedded"
});
var run = new Run(txt)
{
RunProperties = props
};
yield return(run);
}
else if (md.IsDirectLink || md.IsIndirectLink)
{
IEnumerable <MarkdownSpan> spans;
string url = "", alt = "";
if (md.IsDirectLink)
{
var mddl = md as MarkdownSpan.DirectLink;
spans = mddl.Item1;
url = mddl.Item2.Item1;
alt = mddl.Item2.Item2.Option();
}
else
{
var mdil = md as MarkdownSpan.IndirectLink;
var original = mdil.Item2;
var id = mdil.Item3;
spans = mdil.Item1;
if (mddoc.DefinedLinks.ContainsKey(id))
{
url = mddoc.DefinedLinks[id].Item1;
alt = mddoc.DefinedLinks[id].Item2.Option();
}
}
var anchor = "";
if (spans.Count() == 1 && spans.First().IsLiteral)
{
anchor = mdunescape(spans.First() as MarkdownSpan.Literal);
}
else if (spans.Count() == 1 && spans.First().IsInlineCode)
{
anchor = (spans.First() as MarkdownSpan.InlineCode).Item;
}
else
{
throw new NotImplementedException("Link anchor must be Literal or InlineCode, not " + md.ToString());
}
if (sections.ContainsKey(url))
{
var section = sections[url];
if (anchor != section.Title)
{
throw new Exception($"Mismatch: link anchor is '{anchor}', should be '{section.Title}'");
}
var txt = new Text("§" + section.Number)
{
Space = SpaceProcessingModeValues.Preserve
};
var run = new Hyperlink(new Run(txt))
{
Anchor = section.BookmarkName
};
yield return(run);
}
else if (url.StartsWith("http:") || url.StartsWith("https:"))
{
var style = new RunStyle {
Val = "Hyperlink"
};
var hyperlink = new Hyperlink {
DocLocation = url, Tooltip = alt
};
foreach (var element in Spans2Elements(spans))
{
var run = element as Run;
if (run != null)
{
run.InsertAt(new RunProperties(style), 0);
}
hyperlink.AppendChild(run);
}
yield return(hyperlink);
}
else
{
throw new Exception("Absent hyperlink in " + md.ToString());
}
}
else if (md.IsHardLineBreak)
{
// I've only ever seen this arise from dodgy markdown parsing, so I'll ignore it...
}
else
{
yield return(new Run(new Text($"[{md.GetType().Name}]")));
}
}
IEnumerable <OpenXmlElement> Span2Elements(MarkdownSpan md, bool nestedSpan = false)
{
reporter.CurrentSpan = md;
if (md.IsLiteral)
{
var mdl = md as MarkdownSpan.Literal;
var s = MarkdownUtilities.UnescapeLiteral(mdl);
foreach (var r in Literal2Elements(s, nestedSpan))
{
yield return(r);
}
}
else if (md.IsStrong || md.IsEmphasis)
{
IEnumerable <MarkdownSpan> spans = (md.IsStrong ? (md as MarkdownSpan.Strong).body : (md as MarkdownSpan.Emphasis).body);
// Workaround for https://github.com/tpetricek/FSharp.formatting/issues/389 - the markdown parser
// turns *this_is_it* into a nested Emphasis["this", Emphasis["is"], "it"] instead of Emphasis["this_is_it"]
// What we'll do is preprocess it into Emphasis["this_is_it"]
if (md.IsEmphasis)
{
var spans2 = spans.Select(s =>
{
var _ = "";
if (s.IsEmphasis)
{
s = (s as MarkdownSpan.Emphasis).body.Single();
_ = "_";
}
if (s.IsLiteral)
{
return(_ + (s as MarkdownSpan.Literal).text + _);
}
reporter.Error("MD15", $"something odd inside emphasis '{s.GetType().Name}' - only allowed emphasis and literal"); return("");
});
spans = new List <MarkdownSpan>()
{
MarkdownSpan.NewLiteral(string.Join("", spans2), FSharpOption <MarkdownRange> .None)
};
}
// Convention is that ***term*** is used to define a term.
// That's parsed as Strong, which contains Emphasis, which contains one Literal
string literal = null;
TermRef termdef = null;
if (!nestedSpan && md.IsStrong && spans.Count() == 1 && spans.First().IsEmphasis)
{
var spans2 = (spans.First() as MarkdownSpan.Emphasis).body;
if (spans2.Count() == 1 && spans2.First().IsLiteral)
{
literal = (spans2.First() as MarkdownSpan.Literal).text;
termdef = new TermRef(literal, reporter.Location);
if (context.Terms.ContainsKey(literal))
{
var def = context.Terms[literal];
reporter.Warning("MD16", $"Term '{literal}' defined a second time");
reporter.Warning("MD16b", $"Here was the previous definition of term '{literal}'", def.Loc);
}
else
{
context.Terms.Add(literal, termdef);
context.TermKeys.Clear();
}
}
}
// Convention inside our specs is that emphasis only ever contains literals,
// either to emphasis some human-text or to refer to an ANTLR-production
ProductionRef prodref = null;
if (!nestedSpan && md.IsEmphasis && (spans.Count() != 1 || !spans.First().IsLiteral))
{
reporter.Error("MD17", $"something odd inside emphasis");
}
if (!nestedSpan && md.IsEmphasis && spans.Count() == 1 && spans.First().IsLiteral)
{
literal = (spans.First() as MarkdownSpan.Literal).text;
prodref = productions.FirstOrDefault(pr => pr.Names.Contains(literal));
context.Italics.Add(new ItalicUse(literal, prodref != null ? ItalicUse.ItalicUseKind.Production : ItalicUse.ItalicUseKind.Italic, reporter.Location));
}
if (prodref != null)
{
var props = new RunProperties(new Color {
Val = "6A5ACD"
}, new Underline {
Val = UnderlineValues.Single
});
var run = new Run(new Text(literal)
{
Space = SpaceProcessingModeValues.Preserve
})
{
RunProperties = props
};
var link = new Hyperlink(run)
{
Anchor = prodref.BookmarkName
};
yield return(link);
}
else if (termdef != null)
{
context.MaxBookmarkId.Value += 1;
yield return(new BookmarkStart {
Name = termdef.BookmarkName, Id = context.MaxBookmarkId.Value.ToString()
});
var props = new RunProperties(new Italic(), new Bold());
yield return(new Run(new Text(literal)
{
Space = SpaceProcessingModeValues.Preserve
})
{
RunProperties = props
});
yield return(new BookmarkEnd {
Id = context.MaxBookmarkId.Value.ToString()
});
}
else
{
foreach (var e in Spans2Elements(spans, true))
{
var style = (md.IsStrong ? new Bold() as OpenXmlElement : new Italic());
var run = e as Run;
if (run != null)
{
run.InsertAt(new RunProperties(style), 0);
}
yield return(e);
}
}
}
else if (md.IsInlineCode)
{
var mdi = md as MarkdownSpan.InlineCode;
var code = mdi.code;
var txt = new Text(BugWorkaroundDecode(code))
{
Space = SpaceProcessingModeValues.Preserve
};
var props = new RunProperties(new RunStyle {
Val = "CodeEmbedded"
});
var run = new Run(txt)
{
RunProperties = props
};
yield return(run);
}
else if (md.IsLatexInlineMath)
{
var latex = md as MarkdownSpan.LatexInlineMath;
var code = latex.code;
// TODO: Make this look nice - if we actually need it. It's possible that it's only present
// before subscripts are replaced.
var txt = new Text(BugWorkaroundDecode(code))
{
Space = SpaceProcessingModeValues.Preserve
};
var props = new RunProperties(new RunStyle {
Val = "CodeEmbedded"
});
var run = new Run(txt)
{
RunProperties = props
};
yield return(run);
}
else if (md.IsDirectLink || md.IsIndirectLink)
{
IEnumerable <MarkdownSpan> spans;
string url = "", alt = "";
if (md.IsDirectLink)
{
var mddl = md as MarkdownSpan.DirectLink;
spans = mddl.body;
url = mddl.link;
alt = mddl.title.Option();
}
else
{
var mdil = md as MarkdownSpan.IndirectLink;
var original = mdil.original;
var id = mdil.key;
spans = mdil.body;
if (markdownDocument.DefinedLinks.ContainsKey(id))
{
url = markdownDocument.DefinedLinks[id].Item1;
alt = markdownDocument.DefinedLinks[id].Item2.Option();
}
}
var anchor = "";
if (spans.Count() == 1 && spans.First().IsLiteral)
{
anchor = MarkdownUtilities.UnescapeLiteral(spans.First() as MarkdownSpan.Literal);
}
else if (spans.Count() == 1 && spans.First().IsInlineCode)
{
anchor = (spans.First() as MarkdownSpan.InlineCode).code;
}
else
{
reporter.Error("MD18", $"Link anchor must be Literal or InlineCode, not '{md.GetType().Name}'");
yield break;
}
if (sections.ContainsKey(url))
{
var section = sections[url];
// If we're linking to something with a section number, we know what the link text should be.
// (There are a few links that aren't to numbered sections, e.g. to "Annex C".)
if (section.Number is object)
{
var expectedAnchor = "§" + section.Number;
if (anchor != expectedAnchor)
{
reporter.Warning("MD19", $"Mismatch: link anchor is '{anchor}', should be '{expectedAnchor}'");
}
}
var txt = new Text(anchor)
{
Space = SpaceProcessingModeValues.Preserve
};
var run = new Hyperlink(new Run(txt))
{
Anchor = section.BookmarkName
};
yield return(run);
}
else if (url.StartsWith("http:") || url.StartsWith("https:"))
{
var style = new RunStyle {
Val = "Hyperlink"
};
var hyperlink = new Hyperlink {
DocLocation = url, Tooltip = alt
};
foreach (var element in Spans2Elements(spans))
{
var run = element as Run;
if (run != null)
{
run.InsertAt(new RunProperties(style), 0);
}
hyperlink.AppendChild(run);
}
yield return(hyperlink);
}
else
{
// TODO: Make this report an error unconditionally once the subscript "latex-like" Markdown is removed.
if (url != "")
{
reporter.Error("MD28", $"Hyperlink url '{url}' unrecognized - not a recognized heading, and not http");
}
}
}
else if (md.IsHardLineBreak)
{
// I've only ever seen this arise from dodgy markdown parsing, so I'll ignore it...
}
else
{
reporter.Error("MD20", $"Unrecognized markdown element {md.GetType().Name}");
yield return(new Run(new Text($"[{md.GetType().Name}]")));
}
}
IEnumerable <OpenXmlCompositeElement> Paragraph2Paragraphs(MarkdownParagraph md)
{
reporter.CurrentParagraph = md;
if (md.IsHeading)
{
var mdh = md as MarkdownParagraph.Heading;
var level = mdh.size;
var spans = mdh.body;
var sr = sections[new SectionRef(mdh, filename).Url];
reporter.CurrentSection = sr;
var properties = new List <OpenXmlElement>
{
new ParagraphStyleId {
Val = $"Heading{level}"
}
};
if (sr.Number is null)
{
properties.Add(new NumberingProperties(new NumberingLevelReference {
Val = 0
}, new NumberingId {
Val = 0
}));
}
var props = new ParagraphProperties(properties);
var p = new Paragraph {
ParagraphProperties = props
};
context.MaxBookmarkId.Value += 1;
p.AppendChild(new BookmarkStart {
Name = sr.BookmarkName, Id = context.MaxBookmarkId.Value.ToString()
});
p.Append(Span2Elements(MarkdownSpan.NewLiteral(sr.TitleWithoutNumber, FSharpOption <MarkdownRange> .None)));
p.AppendChild(new BookmarkEnd {
Id = context.MaxBookmarkId.Value.ToString()
});
yield return(p);
var i = sr.Url.IndexOf("#");
string currentSection = $"{sr.Url.Substring(0, i)} {new string('#', level)} {sr.Title} [{sr.Number}]";
reporter.Log(currentSection);
yield break;
}
else if (md.IsParagraph)
{
var mdp = md as MarkdownParagraph.Paragraph;
var spans = mdp.body;
yield return(new Paragraph(Spans2Elements(spans)));
yield break;
}
else if (md.IsQuotedBlock)
{
var mdq = md as MarkdownParagraph.QuotedBlock;
// TODO: Actually make this a block quote.
// See https://github.com/ECMA-TC49-TG2/conversion-to-markdown/issues/123
foreach (var paragraph in mdq.paragraphs.SelectMany(Paragraph2Paragraphs))
{
yield return(paragraph);
}
yield break;
}
else if (md.IsListBlock)
{
var mdl = md as MarkdownParagraph.ListBlock;
var flat = FlattenList(mdl);
// Let's figure out what kind of list it is - ordered or unordered? nested?
var format0 = new[] { "1", "1", "1", "1" };
foreach (var item in flat)
{
format0[item.Level] = (item.IsBulletOrdered ? "1" : "o");
}
var format = string.Join("", format0);
var numberingPart = wordDocument.MainDocumentPart.NumberingDefinitionsPart ?? wordDocument.MainDocumentPart.AddNewPart <NumberingDefinitionsPart>("NumberingDefinitionsPart001");
if (numberingPart.Numbering == null)
{
numberingPart.Numbering = new Numbering();
}
Func <int, bool, Level> createLevel;
createLevel = (level, isOrdered) =>
{
var numformat = NumberFormatValues.Bullet;
var levelText = new[] { "·", "o", "·", "o" }[level];
if (isOrdered && level == 0)
{
numformat = NumberFormatValues.Decimal; levelText = "%1.";
}
if (isOrdered && level == 1)
{
numformat = NumberFormatValues.LowerLetter; levelText = "%2.";
}
if (isOrdered && level == 2)
{
numformat = NumberFormatValues.LowerRoman; levelText = "%3.";
}
if (isOrdered && level == 3)
{
numformat = NumberFormatValues.LowerRoman; levelText = "%4.";
}
var r = new Level {
LevelIndex = level
};
r.Append(new StartNumberingValue {
Val = 1
});
r.Append(new NumberingFormat {
Val = numformat
});
r.Append(new LevelText {
Val = levelText
});
r.Append(new ParagraphProperties(new Indentation {
Left = (540 + 360 * level).ToString(), Hanging = "360"
}));
if (levelText == "·")
{
r.Append(new NumberingSymbolRunProperties(new RunFonts {
Hint = FontTypeHintValues.Default, Ascii = "Symbol", HighAnsi = "Symbol", EastAsia = "Times new Roman", ComplexScript = "Times new Roman"
}));
}
if (levelText == "o")
{
r.Append(new NumberingSymbolRunProperties(new RunFonts {
Hint = FontTypeHintValues.Default, Ascii = "Courier New", HighAnsi = "Courier New", ComplexScript = "Courier New"
}));
}
return(r);
};
var level0 = createLevel(0, format[0] == '1');
var level1 = createLevel(1, format[1] == '1');
var level2 = createLevel(2, format[2] == '1');
var level3 = createLevel(3, format[3] == '1');
var abstracts = numberingPart.Numbering.OfType <AbstractNum>().Select(an => an.AbstractNumberId.Value).ToList();
var aid = (abstracts.Count == 0 ? 1 : abstracts.Max() + 1);
var aabstract = new AbstractNum(new MultiLevelType()
{
Val = MultiLevelValues.Multilevel
}, level0, level1, level2, level3)
{
AbstractNumberId = aid
};
numberingPart.Numbering.InsertAt(aabstract, 0);
var instances = numberingPart.Numbering.OfType <NumberingInstance>().Select(ni => ni.NumberID.Value);
var nid = (instances.Count() == 0 ? 1 : instances.Max() + 1);
var numInstance = new NumberingInstance(new AbstractNumId {
Val = aid
})
{
NumberID = nid
};
numberingPart.Numbering.AppendChild(numInstance);
// We'll also figure out the indentation(for the benefit of those paragraphs that should be
// indendent with the list but aren't numbered). I'm not sure what the indent comes from.
// in the docx, each AbstractNum that I created has an indent for each of its levels,
// defaulted at 900, 1260, 1620, ... but I can't see where in the above code that's created?
Func <int, string> calcIndent = level => (540 + level * 360).ToString();
foreach (var item in flat)
{
var content = item.Paragraph;
if (content.IsParagraph || content.IsSpan)
{
var spans = (content.IsParagraph ? (content as MarkdownParagraph.Paragraph).body : (content as MarkdownParagraph.Span).body);
if (item.HasBullet)
{
yield return(new Paragraph(Spans2Elements(spans))
{
ParagraphProperties = new ParagraphProperties(new NumberingProperties(new ParagraphStyleId {
Val = "ListParagraph"
}, new NumberingLevelReference {
Val = item.Level
}, new NumberingId {
Val = nid
}))
});
}
else
{
yield return(new Paragraph(Spans2Elements(spans))
{
ParagraphProperties = new ParagraphProperties(new Indentation {
Left = calcIndent(item.Level)
})
});
}
}
else if (content.IsQuotedBlock || content.IsCodeBlock)
{
foreach (var p in Paragraph2Paragraphs(content))
{
var props = p.GetFirstChild <ParagraphProperties>();
if (props == null)
{
props = new ParagraphProperties();
p.InsertAt(props, 0);
}
var indent = props?.GetFirstChild <Indentation>();
if (indent == null)
{
indent = new Indentation();
props.Append(indent);
}
indent.Left = calcIndent(item.Level);
yield return(p);
}
}
else if (content.IsTableBlock)
{
foreach (var p in Paragraph2Paragraphs(content))
{
var table = p as Table;
if (table == null)
{
yield return(p);
continue;
}
var tprops = table.GetFirstChild <TableProperties>();
var tindent = tprops?.GetFirstChild <TableIndentation>();
if (tindent == null)
{
throw new Exception("Ooops! Table is missing indentation");
}
tindent.Width = int.Parse(calcIndent(item.Level));
yield return(table);
}
}
else
{
reporter.Error("MD08", $"Unexpected item in list '{content.GetType().Name}'");
}
}
}
else if (md.IsCodeBlock)
{
var mdc = md as MarkdownParagraph.CodeBlock;
var code = mdc.code;
var lang = mdc.language;
code = BugWorkaroundDecode(code);
var runs = new List <Run>();
var onFirstLine = true;
IEnumerable <ColorizedLine> lines;
switch (lang)
{
case "csharp":
case "c#":
case "cs":
lines = Colorize.CSharp(code);
break;
case "vb":
case "vbnet":
case "vb.net":
lines = Colorize.VB(code);
break;
case "":
case "console":
case "xml":
lines = Colorize.PlainText(code);
break;
case "ANTLR":
case "antlr":
lines = Antlr.ColorizeAntlr(code);
break;
default:
reporter.Error("MD09", $"unrecognized language {lang}");
lines = Colorize.PlainText(code);
break;
}
foreach (var line in lines)
{
if (onFirstLine)
{
onFirstLine = false;
}
else
{
runs.Add(new Run(new Break()));
}
foreach (var word in line.Words)
{
var run = new Run();
var props = new RunProperties();
if (word.Red != 0 || word.Green != 0 || word.Blue != 0)
{
props.Append(new Color {
Val = $"{word.Red:X2}{word.Green:X2}{word.Blue:X2}"
});
}
if (word.IsItalic)
{
props.Append(new Italic());
}
if (props.HasChildren)
{
run.Append(props);
}
run.Append(new Text(word.Text)
{
Space = SpaceProcessingModeValues.Preserve
});
runs.Add(run);
}
}
if (lang == "antlr")
{
var p = new Paragraph()
{
ParagraphProperties = new ParagraphProperties(new ParagraphStyleId {
Val = "Grammar"
})
};
var prodref = productions.Single(prod => prod.Code == code);
context.MaxBookmarkId.Value += 1;
p.AppendChild(new BookmarkStart {
Name = prodref.BookmarkName, Id = context.MaxBookmarkId.Value.ToString()
});
p.Append(runs);
p.AppendChild(new BookmarkEnd {
Id = context.MaxBookmarkId.Value.ToString()
});
yield return(p);
}
else
{
var p = new Paragraph()
{
ParagraphProperties = new ParagraphProperties(new ParagraphStyleId {
Val = "Code"
})
};
p.Append(runs);
yield return(p);
}
}
else if (md.IsTableBlock)
{
var mdt = md as MarkdownParagraph.TableBlock;
var header = mdt.headers.Option();
var align = mdt.alignments;
var rows = mdt.rows;
var table = new Table();
if (header == null)
{
reporter.Error("MD10", "Github requires all tables to have header rows");
}
if (!header.Any(cell => cell.Length > 0))
{
header = null; // even if Github requires an empty header, we can at least cull it from Docx
}
var tstyle = new TableStyle {
Val = "TableGrid"
};
var tindent = new TableIndentation {
Width = 360, Type = TableWidthUnitValues.Dxa
};
var tborders = new TableBorders();
tborders.TopBorder = new TopBorder {
Val = BorderValues.Single
};
tborders.BottomBorder = new BottomBorder {
Val = BorderValues.Single
};
tborders.LeftBorder = new LeftBorder {
Val = BorderValues.Single
};
tborders.RightBorder = new RightBorder {
Val = BorderValues.Single
};
tborders.InsideHorizontalBorder = new InsideHorizontalBorder {
Val = BorderValues.Single
};
tborders.InsideVerticalBorder = new InsideVerticalBorder {
Val = BorderValues.Single
};
var tcellmar = new TableCellMarginDefault();
tcellmar.Append();
table.Append(new TableProperties(tstyle, tindent, tborders));
var ncols = align.Length;
for (int irow = -1; irow < rows.Length; irow++)
{
if (irow == -1 && header == null)
{
continue;
}
var mdrow = (irow == -1 ? header : rows[irow]);
var row = new TableRow();
for (int icol = 0; icol < Math.Min(ncols, mdrow.Length); icol++)
{
var mdcell = mdrow[icol];
var cell = new TableCell();
var pars = Paragraphs2Paragraphs(mdcell).ToList();
for (int ip = 0; ip < pars.Count; ip++)
{
var p = pars[ip] as Paragraph;
if (p == null)
{
cell.Append(pars[ip]);
continue;
}
var props = new ParagraphProperties(new ParagraphStyleId {
Val = "TableCellNormal"
});
if (align[icol].IsAlignCenter)
{
props.Append(new Justification {
Val = JustificationValues.Center
});
}
if (align[icol].IsAlignRight)
{
props.Append(new Justification {
Val = JustificationValues.Right
});
}
p.InsertAt(props, 0);
cell.Append(pars[ip]);
}
if (pars.Count == 0)
{
cell.Append(new Paragraph(new ParagraphProperties(new SpacingBetweenLines {
After = "0"
}), new Run(new Text(""))));
}
row.Append(cell);
}
table.Append(row);
}
yield return(new Paragraph(new Run(new Text("")))
{
ParagraphProperties = new ParagraphProperties(new ParagraphStyleId {
Val = "TableLineBefore"
})
});
yield return(table);
yield return(new Paragraph(new Run(new Text("")))
{
ParagraphProperties = new ParagraphProperties(new ParagraphStyleId {
Val = "TableLineAfter"
})
});
}
else
{
reporter.Error("MD11", $"Unrecognized markdown element {md.GetType().Name}");
yield return(new Paragraph(new Run(new Text($"[{md.GetType().Name}]"))));
}
}
