public override IEnumerable <TokenBase> GetTokens()
{
var top = new TokenIndex();
string line;
while ((line = _reader.ReadLine()) != null)
{
if (string.IsNullOrEmpty(line))
{
continue;
}
var p = new XElement("p");
var property = new TextVisualProperties
{
TextIndent = 32.0,
Inline = false
};
yield return(new TagOpenToken(top.Index++, p, property, -1));
foreach (TokenBase baseToken in ParseText(line, top))
{
yield return(baseToken);
}
yield return(new TagCloseToken(top.Index++, -1));
}
}
public TagOpenToken(int id, HtmlNode node, TextVisualProperties properties, int parentID)
: base(id)
{
Name = node.Name;
TextProperties = properties;
ParentID = parentID;
}
public override IEnumerable <TokenBase> GetTokens()
{
_css = new CSS();
var html = new HtmlDocument
{
OptionOutputAsXml = true,
OptionReadEncoding = false
};
html.Load(_reader);
var root = html.DocumentNode.SelectSingleNode("//body");
var cssNode = html.DocumentNode.SelectSingleNode("//style");
if (cssNode != null)
{
_css.Analyze(cssNode.InnerText);
}
var stack = new Stack <TextVisualProperties>();
var item = new TextVisualProperties();
stack.Push(item);
return(ParseTokens(root, stack, new TokenIndex()));
}
public TagOpenToken(int id, XElement element, TextVisualProperties properties, int parentID)
: base(id)
{
Name = element.Name.LocalName;
TextProperties = properties;
ParentID = parentID;
}
private TagOpenToken(int id, string name, TextVisualProperties properties, int parentId)
: base(id)
{
Name = name;
TextProperties = properties;
ParentID = parentId;
}
private void AppendToLine(TextToken token, double textWidth)
{
string text = token.Text;
if (string.IsNullOrEmpty(text))
{
return;
}
_block = _block ?? new TextTokenBlock
{
TextAlign = _lastOpenTag.TextProperties.TextAlign,
MarginLeft = _marginLeft,
MarginRight = _marginRight,
FirstTokenID = _firstTokenID,
TextIndent = _textIndent
};
_block.LastTokenID = token.ID;
_block.UpdateHeight(GetTextHeight(text));
if (_separator)
{
TextVisualProperties properties = _lastOpenTag.TextProperties.Clone();
var inlineItem = _block.Inlines.OfType <TextElement>().LastOrDefault();
if (inlineItem != null && string.IsNullOrEmpty(inlineItem.LinkID))
{
properties.LinkID = string.Empty;
}
_block.AddText(" ", properties, _fontSize, GetTextSize(" ", properties));
}
_block.AddText(text, _lastOpenTag.TextProperties, _fontSize, GetTextSize(text, _lastOpenTag.TextProperties), token.Part, token.ID);
_textWidth += textWidth;
}
public override IEnumerable<TokenBase> GetTokens()
{
var propertiesStack = new Stack<TextVisualProperties>();
var item = new TextVisualProperties();
propertiesStack.Push(item);
var top = new TokenIndex();
return _root.Elements(_ns + "body").SelectMany(b => ParseNodes(b, propertiesStack, top, 0, "2"));
}
public static TokenBase Load(BinaryReader reader, int id)
{
var name = reader.ReadString();
int parentID = reader.ReadInt32();
TextVisualProperties properties = TextVisualProperties.Load(reader);
return(new TagOpenToken(id, name, properties, parentID));
}
public override IEnumerable <TokenBase> GetTokens()
{
var propertiesStack = new Stack <TextVisualProperties>();
var item = new TextVisualProperties();
propertiesStack.Push(item);
var top = new TokenIndex();
return(_root.Elements(_ns + "body").SelectMany(b => ParseNodes(b, propertiesStack, top, 0)));
}
private double GetTextWidth(string text, bool separator, TextVisualProperties properties)
{
if (separator)
{
text = " " + text;
}
double size = _fontSize * (properties.SubOption || properties.SupOption ? 0.5 : 1.0);
var width = GetTextWidth(text, properties, size);
return(width);
}
private static void ProcessLinks(TextVisualProperties properties, XElement xelement)
{
properties.LinkID = string.Empty;
XAttribute attribute = xelement.Attributes().FirstOrDefault(t => (t.Name.LocalName == "href"));
string str = ((attribute != null) ? attribute.Value : string.Empty).TrimStart('#');
if (!string.IsNullOrEmpty(str))
{
properties.LinkID = str;
}
}
public void AddText(string text, TextVisualProperties properties, double fontSize, Size size, string part = null, int tokenID = -1)
{
part = part ?? string.Empty;
Inlines.Add(new TextElement
{
Text = text,
Width = size.Width,
Height = size.Height,
Bold = properties.Bold,
Italic = properties.Italic,
Size = fontSize,
SupOption = properties.SupOption,
SubOption = properties.SubOption,
Part = part,
LinkID = properties.LinkID,
TokenID = tokenID
});
}
public override IEnumerable <TokenBase> GetTokens()
{
var propertiesStack = new Stack <TextVisualProperties>();
var item = new TextVisualProperties();
propertiesStack.Push(item);
var top = new TokenIndex();
foreach (EpubSpineItem spineItem in GetSpineItems())
{
yield return(new NewPageToken(top.Index++));
AddAnchor(top, (_opfPath) + spineItem.Path);
foreach (TokenBase token in ParseSpineItem(spineItem, propertiesStack, top))
{
yield return(token);
}
}
}
private double GetTextWidth(string text, TextVisualProperties properties, double size)
{
return(text.Aggregate(0.0, (s, c) => s + _helper.GetSize(c, size, properties.Bold, properties.Italic).Width));
}
private IEnumerable <TokenBase> ParseNodes(HtmlNode container, Stack <TextVisualProperties> propertiesStack, TokenIndex top, EpubPath path, int parentID = -1)
{
foreach (HtmlNode child in container.ChildNodes)
{
var asText = child as HtmlTextNode;
if (asText != null && !string.IsNullOrEmpty(asText.Text))
{
foreach (TokenBase text in ParseText(asText.Text, top))
{
yield return(text);
}
}
else
{
TextVisualProperties properties = propertiesStack.Peek().Clone().Update(child, _css);
properties.LinkID = string.Empty;
if (child.Name == "a" || child.Name == "span")
{
ParseAnchors(top, path, child);
}
if (child.Name == "a")
{
string href = child.GetAttributeValue("href", string.Empty);
if (!string.IsNullOrEmpty(href))
{
if (href.StartsWith("#"))
{
properties.LinkID = path.CurrentFilePath + href;
}
else
{
properties.LinkID = path + href;
}
}
}
if (string.Equals(child.Name, "img"))
{
HtmlAttributeCollection attributes = child.Attributes;
string src = attributes.Contains("src") ? attributes["src"].Value : string.Empty;
var pictureToken = new PictureToken(top.Index++, (path) + src);
yield return(pictureToken);
}
else
{
if (child is HtmlCommentNode)
{
continue;
}
var tagOpenToken = new TagOpenToken(top.Index++, child, properties, parentID);
yield return(tagOpenToken);
propertiesStack.Push(properties);
foreach (TokenBase token in ParseNodes(child, propertiesStack, top, path, tagOpenToken.ID))
{
yield return(token);
}
propertiesStack.Pop();
yield return(new TagCloseToken(top.Index++, parentID));
}
}
}
}
private void LeaveMargin(TextVisualProperties properties)
{
_marginLeft -= properties.MarginLeft;
_marginRight -= properties.MarginRight;
}
private void EnterMargin(TextVisualProperties properties)
{
_marginLeft += properties.MarginLeft;
_marginRight += properties.MarginRight;
}
private static void ProcessLinks(TextVisualProperties properties, XElement xelement)
{
properties.LinkID = string.Empty;
XAttribute attribute = xelement.Attributes().FirstOrDefault(t => (t.Name.LocalName == "href"));
string str = ((attribute != null) ? attribute.Value : string.Empty).TrimStart('#');
if (!string.IsNullOrEmpty(str))
{
properties.LinkID = str;
}
}
private IEnumerable <TokenBase> ParseTokens(HtmlNode container, Stack <TextVisualProperties> propertiesStack, TokenIndex top, int parentID = -1)
{
foreach (HtmlNode child in container.ChildNodes)
{
var node = child as HtmlTextNode;
if (node != null)
{
if (!string.IsNullOrEmpty(node.Text))
{
foreach (TokenBase token in ParseText(node.Text, top))
{
yield return(token);
}
}
}
else
{
TextVisualProperties properties = propertiesStack.Peek().Clone().Update(child, _css);
properties.LinkID = string.Empty;
if (child.Name == "a" || child.Name == "span")
{
ParseAnchors(top, child);
}
if (child.Name == "a")
{
string attributeValue = child.GetAttributeValue("href", string.Empty);
if (!string.IsNullOrEmpty(attributeValue))
{
properties.LinkID = attributeValue;
}
}
//TODO: add images support
//if (string.Equals(child.Name, imageParser.ImageTag))
//{
// int oldTopIndex;
// HtmlAttributeCollection attributes = child.Attributes;
// string imagePath = attributes.Contains("src") ? attributes["src"].Value : string.Empty;
// top.Index = (oldTopIndex = top.Index) + 1;
// var pictureToken = new PictureToken(oldTopIndex)
// {
// ImageID = imagePath
// };
// yield return pictureToken;
//}
//else
{
if (child is HtmlCommentNode)
{
continue;
}
var tagOpen = new TagOpenToken(top.Index++, child, properties, parentID);
yield return(tagOpen);
propertiesStack.Push(properties);
foreach (TokenBase token in ParseTokens(child, propertiesStack, top, tagOpen.ID))
{
yield return(token);
}
propertiesStack.Pop();
yield return(new TagCloseToken(top.Index++, parentID));
}
}
}
}
private IEnumerable <TokenBase> ParseNodes(XContainer container, Stack <TextVisualProperties> propertiesStack, TokenIndex top, int bookLevel, int parentID = -1)
{
foreach (XNode node in container.Nodes())
{
var text = node as XText;
if ((text != null) && !string.IsNullOrEmpty(text.Value))
{
foreach (TokenBase token in ParseText(text.Value, top))
{
yield return(token);
}
}
var element = node as XElement;
if (element == null)
{
continue;
}
TextVisualProperties properties = propertiesStack.Peek().Clone().Update(element, _styleSheet);
string localName = element.Name.LocalName;
int level = bookLevel;
if (localName == "a")
{
ProcessLinks(properties, element);
}
ProcessAnchors(top, element);
if (localName == "section")
{
yield return(new NewPageToken(top.Index++));
level++;
}
if (localName == "title")
{
ProcessTitleData(top, element, level);
}
if (localName == "image")
{
XAttribute hrefAttr = element.Attributes().FirstOrDefault(t => (t.Name.LocalName == "href"));
string href = ((hrefAttr != null) ? hrefAttr.Value : string.Empty).TrimStart('#');
var pictureToken = new PictureToken(top.Index++, href);
yield return(pictureToken);
}
else
{
var tagOpen = new TagOpenToken(top.Index++, element, properties, parentID);
yield return(tagOpen);
propertiesStack.Push(properties);
foreach (TokenBase token in ParseNodes(element, propertiesStack, top, level, tagOpen.ID))
{
yield return(token);
}
propertiesStack.Pop();
yield return(new TagCloseToken(top.Index++, parentID));
}
}
}
public void ApplyProperties(string[] selectors, TextVisualProperties properties)
{
foreach (CssRule rule in Rules.Where(r => r.Selectors.Any(selectors.Contains)))
{
string val;
if (rule.TryGetValueForRule("font-weight", out val))
{
properties.Bold = val == "bold";
}
if (rule.TryGetValueForRule("font-style", out val))
{
properties.Italic = val == "italic";
}
if (rule.TryGetValueForRule("display", out val))
{
properties.Inline = val == "inline";
}
if (rule.TryGetValueForRule("vertical-align", out val))
{
properties.SupOption = val == "super";
}
if (rule.TryGetValueForRule("vertical-align", out val))
{
properties.SubOption = val == "sub";
}
if (rule.TryGetValueForRule("font-size", out val))
{
double size;
FontSizeType type;
if (TryParseFontSize(val, out size, out type))
{
properties.FontSize = size;
properties.FontSizeType = type;
}
}
if (rule.TryGetValueForRule("text-align", out val) && (val != "inherit"))
{
properties.TextAlign = val;
}
if (rule.TryGetValueForRule("margin", out val))
{
double left;
double right;
double bottom;
ParseMargin(val, out left, out right, out bottom);
properties.MarginLeft = left;
properties.MarginRight = right;
properties.MarginBottom = bottom;
}
if (rule.TryGetValueForRule("margin-left", out val))
{
properties.MarginLeft = ParseMarginValue(val);
}
if (rule.TryGetValueForRule("margin-right", out val))
{
properties.MarginRight = ParseMarginValue(val);
}
if (rule.TryGetValueForRule("margin-bottom", out val))
{
properties.MarginBottom = ParseMarginValue(val);
}
if (rule.TryGetValueForRule("text-indent", out val))
{
properties.TextIndent = ParseMarginValue(val);
}
}
}
private Size GetTextSize(string text, TextVisualProperties properties)
{
return(new Size(
GetTextWidth(text, false, properties),
GetTextHeight(text, properties.Bold, properties.Italic)));
}