private IEnumerable <Inline> ConvertInlines(HTML.Element e, Context ctx, bool collapseSpace, ParagraphSizeTracker pst)
{
return(from i in
(from n in e.Contents select ConvertInline(n, ctx, collapseSpace, pst))
where i != null
select i);
}
private Span RenderHyperlink(HTML.Element e)
{
if (e.Attributes.ContainsKey("href"))
{
Hyperlink h = new Hyperlink();
Uri TargetURI;
if (Uri.TryCreate(e.Attributes["href"], UriKind.RelativeOrAbsolute, out TargetURI))
{
if (!TargetURI.IsAbsoluteUri)
{
TargetURI = BaseUri != null ? new Uri(BaseUri, TargetURI) : null;
}
}
if (TargetURI != null)
{
h.NavigateUri = TargetURI;
h.RequestNavigate += RequestedNavigate;
h.ToolTip = TargetURI;
}
else
{
// tooltip contains URI even if it's bad
h.ToolTip = e.Attributes["href"];
}
return(h);
}
else
{
return(new Span());
}
}
private InlineUIContainer RenderImage(HTML.Element e, ParagraphSizeTracker pst)
{
Uri ImageURI;
if (e.Attributes.ContainsKey("src") &&
Uri.TryCreate(e.Attributes["src"], UriKind.RelativeOrAbsolute, out ImageURI))
{
Image image = new Image()
{
Stretch = Stretch.Fill
};
BitmapImage bitmapImage = new BitmapImage();
bitmapImage.BeginInit();
bitmapImage.UriSource = ImageURI;
bitmapImage.EndInit();
image.Source = bitmapImage;
if (ImageSizeCache.ContainsKey(ImageURI))
{
Tuple <int, int> size = ImageSizeCache[ImageURI];
image.Width = size.Item1;
image.Height = size.Item2;
}
// This hack forces image size to match pixel size,
// which undermines WPF's preference for honoring the
// embedded DPI value which leads to upscaled and
// mildly blurry images where that doesn't match
// the local display. Very high resolution displays
// might need another approach.
bitmapImage.DownloadCompleted += (s, ee) =>
{
ImageSizeCache[ImageURI] = new Tuple <int, int>(bitmapImage.PixelWidth, bitmapImage.PixelHeight);
image.Width = bitmapImage.PixelWidth;
image.Height = bitmapImage.PixelHeight;
};
// Communicate image size info back up to the container
if (pst != null)
{
pst.Add(image);
}
if (e.Attributes.ContainsKey("title"))
{
image.ToolTip = e.Attributes["title"];
}
else if (e.Attributes.ContainsKey("alt"))
{
image.ToolTip = e.Attributes["alt"];
}
// TODO arrange to display alt value if image can't be loaded
// (or before it has loaded)
return(new InlineUIContainer()
{
Child = image
});
}
else
{
return(null);
}
}
public void TestHTMLSimpleNoOpen()
{
HTML.Document d = HTML.Document.Parse("wibble");
HTML.Element p = d.HTML.Follow("body.p");
Assert.AreEqual(1, p.Contents.Count);
HTML.Cdata cdata = p.Contents[0] as HTML.Cdata;
Assert.AreEqual("wibble", cdata.Content);
}
private IEnumerable <Block> ConvertFlowsOrBlocks(HTML.Element root, Context ctx)
{
List <Block> blocks = new List <Block>();
foreach (HTML.Node n in root.Contents)
{
blocks.AddRange(ConvertFlowOrBlock(n as HTML.Element, ctx));
}
return(blocks);
}
public void TestHTMLEmpty()
{
HTML.Document d = HTML.Document.Parse("");
Assert.AreEqual("html", d.HTML.Name);
Assert.AreEqual(2, d.HTML.Contents.Count);
HTML.Element body = d.HTML.Contents[1] as HTML.Element;
Assert.IsNotNull(body);
Assert.AreEqual("body", body.Name);
Assert.AreEqual(0, body.Contents.Count);
}
public void TestHTMLExtraSlash()
{
HTML.Document d = HTML.Document.Parse("<img src=\"http://imgs.xkcd.com/comics/voyager_1.png\" title=\"what'ever'\" />");
HTML.Element img = d.HTML.Follow("body.p.img");
Assert.AreEqual(2, img.Attributes.Count);
Assert.IsTrue(img.Attributes.ContainsKey("src"));
Assert.AreEqual("http://imgs.xkcd.com/comics/voyager_1.png", img.Attributes["src"]);
Assert.IsTrue(img.Attributes.ContainsKey("title"));
Assert.AreEqual("what'ever'", img.Attributes["title"]);
Assert.AreEqual("<html><head></head><body><p><img src=\"http://imgs.xkcd.com/comics/voyager_1.png\" title=\"what'ever'\"></p></body></html>", d.ToString());
}
private Paragraph ConvertParagraph(HTML.Element e, Context ctx, bool collapseSpace)
{
LastCharacter = ' ';
Paragraph p = new Paragraph();
p.Inlines.AddRange(ConvertInlines(e, ctx, collapseSpace,
new ParagraphSizeTracker()
{
ParentScrollViewer = ParentScrollViewer
}));
return(p.Inlines.Count > 0 ? p : null);
}
private List ConvertList(HTML.Element e, Context ctx)
{
List l = new List();
foreach (HTML.Element ee in e.Contents)
{
ListItem li = new ListItem();
li.Blocks.AddRange(ConvertFlowOrBlock(ee, ctx));
l.ListItems.Add(li);
}
return(l);
}
public void TestHTMLTwoPara()
{
HTML.Document d = HTML.Document.Parse("<p>first</p> <p>second</p>");
Assert.AreEqual(2, d.HTML.Follow("body").Contents.Count);
HTML.Element p1 = d.HTML.Follow("body.p");
Assert.AreEqual(1, p1.Contents.Count);
HTML.Cdata cdata1 = p1.Contents[0] as HTML.Cdata;
Assert.AreEqual("first", cdata1.Content);
HTML.Element p2 = (HTML.Element)d.HTML.Follow("body").Contents[1];
HTML.Cdata cdata2 = p2.Contents[0] as HTML.Cdata;
Assert.AreEqual("second", cdata2.Content);
}
public void TestHTMLInline()
{
HTML.Document d = HTML.Document.Parse("<p>one <i>two</i> three</p>");
HTML.Element p = d.HTML.Follow("body.p");
Assert.AreEqual(3, p.Contents.Count);
HTML.Cdata cdata1 = p.Contents[0] as HTML.Cdata;
Assert.AreEqual("one ", cdata1.Content);
HTML.Element italic = p.Contents[1] as HTML.Element;
Assert.AreEqual("i", italic.Name);
Assert.AreEqual(1, italic.Contents.Count);
HTML.Cdata cdata2 = italic.Contents[0] as HTML.Cdata;
Assert.AreEqual("two", cdata2.Content);
HTML.Cdata cdata3 = p.Contents[2] as HTML.Cdata;
Assert.AreEqual(" three", cdata3.Content);
}
public void TestHTMLSimple()
{
HTML.Document d = HTML.Document.Parse("<p>wibble</p>");
Assert.AreEqual("html", d.HTML.Name);
Assert.AreEqual(2, d.HTML.Contents.Count);
HTML.Element body = d.HTML.Contents[1] as HTML.Element;
Assert.IsNotNull(body);
Assert.AreEqual("body", body.Name);
Assert.AreEqual(1, body.Contents.Count);
HTML.Element p = body.Contents[0] as HTML.Element;
Assert.AreEqual("p", p.Name);
Assert.AreEqual(1, p.Contents.Count);
HTML.Cdata cdata = p.Contents[0] as HTML.Cdata;
Assert.AreEqual("wibble", cdata.Content);
}
private IEnumerable <Block> ConvertDefinitionList(HTML.Element e, Context ctx)
{
List <Block> blocks = new List <Block>();
foreach (HTML.Element d in e.Contents)
{
IEnumerable <Block> dblocks = ConvertFlowOrBlock(d, ctx);
if (d.Name == "dd")
{
foreach (Block b in dblocks)
{
Indent(b, 12, 0);
}
}
blocks.AddRange(dblocks);
}
return(blocks);
}
static private int GetSpan(HTML.Element e, string attribute)
{
string attributeValue;
if (e.Attributes.TryGetValue(attribute, out attributeValue))
{
int value;
if (int.TryParse(attributeValue, out value))
{
if (value < 1)
{
value = 1;
}
return(value);
}
}
return(1);
}
public void TestHTMLAttributes()
{
HTML.Document d = HTML.Document.Parse("<p><a href=\"http://www.example.com/\">target</a></p>");
HTML.Element a = d.HTML.Follow("body.p.a");
Assert.AreEqual(1, a.Attributes.Count);
Assert.IsTrue(a.Attributes.ContainsKey("href"));
Assert.AreEqual("http://www.example.com/", a.Attributes["href"]);
d = HTML.Document.Parse("<p><a href='http://www.example.com/&' >target</a></p>");
a = d.HTML.Follow("body.p.a");
Assert.AreEqual(1, a.Attributes.Count);
Assert.IsTrue(a.Attributes.ContainsKey("href"));
Assert.AreEqual("http://www.example.com/&", a.Attributes["href"]);
d = HTML.Document.Parse("<p><a href=\"http://www.example.com/\" Target=_blank>target</a></p>");
a = d.HTML.Follow("body.p.a");
Assert.AreEqual(2, a.Attributes.Count);
Assert.IsTrue(a.Attributes.ContainsKey("href"));
Assert.AreEqual("http://www.example.com/", a.Attributes["href"]);
Assert.IsTrue(a.Attributes.ContainsKey("target"));
Assert.AreEqual("_blank", a.Attributes["target"]);
}
public void TestHTMLLists()
{
HTML.Document d = HTML.Document.Parse("<ul><li>one</li><li>two</li></ul>");
HTML.Element ul = d.HTML.Follow("body.ul");
Assert.AreEqual(2, ul.Contents.Count);
HTML.Element li1 = ul.Contents[0] as HTML.Element;
Assert.AreEqual("li", li1.Name);
Assert.AreEqual(1, li1.Contents.Count);
Assert.AreEqual("one", (li1.Contents[0] as HTML.Cdata).Content);
HTML.Element li2 = ul.Contents[1] as HTML.Element;
Assert.AreEqual("li", li2.Name);
Assert.AreEqual(1, li2.Contents.Count);
Assert.AreEqual("two", (li2.Contents[0] as HTML.Cdata).Content);
d = HTML.Document.Parse("<ul><li><p>one<p>two</ul>");
ul = d.HTML.Follow("body.ul");
Assert.AreEqual(1, ul.Contents.Count);
li1 = ul.Contents[0] as HTML.Element;
Assert.AreEqual("li", li1.Name);
Assert.AreEqual(2, li1.Contents.Count);
Assert.AreEqual("one", ((li1.Contents[0] as HTML.Element).Contents[0] as HTML.Cdata).Content);
Assert.AreEqual("two", ((li1.Contents[1] as HTML.Element).Contents[0] as HTML.Cdata).Content);
}
private Table ConvertTable(HTML.Element e, Context ctx)
{
Table t = new Table()
{
CellSpacing = 0,
};
TableRowGroup trg = new TableRowGroup();
t.RowGroups.Add(trg);
int columns = 0;
HTML.Element caption = null;
foreach (HTML.Element r in e.Contents)
{
switch (r.Name)
{
case "tr":
TableRow tr = new TableRow();
trg.Rows.Add(tr);
int columnNumber = 0;
foreach (HTML.Element c in r.Contents)
{
TableCell tc = new TableCell()
{
ColumnSpan = GetSpan(c, "colspan"),
RowSpan = GetSpan(c, "rowspan"),
BorderBrush = Brushes.Black,
BorderThickness = new Thickness(0.5, 0.5, 0.5, 0.5),
Padding = new Thickness(1),
};
if (c.Name == "th")
{
tc.Blocks.AddRange(ConvertFlowOrBlock(c, new Context(ctx)
{
fontWeight = FontWeights.Bold
}));
}
else
{
tc.Blocks.AddRange(ConvertFlowOrBlock(c, ctx));
}
tr.Cells.Add(tc);
columnNumber += tc.ColumnSpan;
}
if (columnNumber > columns)
{
columns = columnNumber;
}
break;
case "caption":
caption = r;
break;
}
}
if (caption != null && columns == 0)
{
columns = 1;
}
for (int i = 0; i < columns; ++i)
{
t.Columns.Add(new TableColumn());
}
if (caption != null)
{
TableRow tr = new TableRow();
TableCell tc = new TableCell()
{
ColumnSpan = columns,
TextAlignment = TextAlignment.Center,
};
Paragraph p = ConvertParagraph(caption,
new Context(ctx)
{
fontStyle = FontStyles.Italic
},
true);
tc.Blocks.Add(p);
tr.Cells.Add(tc);
trg.Rows.Add(tr);
}
return(t);
}
private IEnumerable <Block> ConvertFlowOrBlock(HTML.Element e, Context ctx)
{
if (e.IsBlockElement() ||
e.IsListElement() ||
e.IsTableElement())
{
Block block = null;
switch (e.Name)
{
case "ul":
block = ConvertList(e, ctx);
break;
case "ol":
List l = ConvertList(e, ctx);
l.StartIndex = 1;
block = l;
break;
case "dl":
return(ConvertDefinitionList(e, ctx));
case "h1":
case "h2":
case "h3":
case "h4":
case "h5":
case "h6":
block = ConvertParagraph(e,
new Context(ctx)
{
fontFamily = HeadingFont, fontWeight = FontWeights.Bold, fontSize = TextSize * Math.Pow(1.125, '7' - e.Name[1])
},
true);
break;
case "pre":
block = ConvertParagraph(e,
new Context(ctx)
{
fontFamily = MonospaceFont
},
false);
break;
case "table":
block = ConvertTable(e, ctx);
break;
case "hr":
block = HorizontalRule();
break;
default:
block = ConvertParagraph(e, ctx, true);
break;
}
if (block != null)
{
return(new List <Block>()
{
block
});
}
else
{
return(new List <Block>());
}
}
else
{
IEnumerable <Block> content;
if (e.Contents.Count > 0 &&
(e.Contents[0] is HTML.Cdata ||
(e.Contents[0] as HTML.Element).IsInlineElement()))
{
content = new List <Block>()
{
ConvertParagraph(e, ctx, true)
};
}
else
{
content = ConvertFlowsOrBlocks(e, ctx);
}
switch (e.Name)
{
case "blockquote":
foreach (Block b in content)
{
Indent(b, 12, 12);
break;
}
break;
// TODO other kinds of flow container
}
return(content);
}
}
private Inline ConvertInlineElement(HTML.Element e, Context ctx, bool collapseSpace, ParagraphSizeTracker pst)
{
Span s;
switch (e.Name)
{
case "b":
case "strong":
s = new Span();
ctx = new Context(ctx)
{
fontWeight = FontWeights.Bold
};
break;
case "i":
case "em":
s = new Span();
ctx = new Context(ctx)
{
fontStyle = FontStyles.Italic
};
break;
case "u":
s = new Underline();
break;
case "big": {
s = new Span();
double newSize = ctx.fontSize * 1.25;
ctx = new Context(ctx)
{
fontSize = newSize
};
break;
}
case "small": {
s = new Span();
double newSize = ctx.fontSize / 1.25;
ctx = new Context(ctx)
{
fontSize = newSize
};
break;
}
case "code":
case "tt":
s = new Span();
ctx = new Context(ctx)
{
fontFamily = MonospaceFont
};
break;
case "sub":
s = new Span();
s.Typography.Variants = FontVariants.Subscript;
break;
case "sup":
s = new Span();
s.Typography.Variants = FontVariants.Superscript;
break;
case "a":
s = RenderHyperlink(e);
break;
case "img":
return(RenderImage(e, pst));
case "br":
return(new LineBreak());
// TODO map, anything else?
default: s = new Span(); break;
}
s.Inlines.AddRange(ConvertInlines(e, ctx, collapseSpace, pst));
return(s);
}