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));
}
protected IEnumerable<TokenBase> ParseText(string text, TokenIndex top)
{
foreach (string word in text.BreakToWords())
{
if (string.IsNullOrEmpty(word))
yield return new WhitespaceToken(top.Index++);
else
yield return new TextToken(top.Index++, HttpUtility.HtmlDecode(word));
}
}
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;
}
}
}
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);
}
}
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 void AddAnchor(TokenIndex top, string path)
{
_anchors[path] = top.Index;
}
private void ParseAnchors(TokenIndex top, HtmlNode node)
{
foreach (string str in new[] { "id", "name" })
{
string attributeValue = node.GetAttributeValue(str, string.Empty);
if (!string.IsNullOrEmpty(attributeValue))
{
AddAnchor(top, attributeValue);
}
}
}
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);
}
}
}
private void ProcessTitleData(TokenIndex top, XElement xelement, int bookLevel)
{
var item = new BookChapter
{
Level = bookLevel,
Title = GetText(xelement),
TokenID = top.Index
};
_chapters.Add(item);
}
private void ProcessAnchors(TokenIndex top, XElement xelement)
{
XAttribute attribute = xelement.Attributes().FirstOrDefault(t => (t.Name.LocalName == "id"));
if (attribute != null)
{
_anchors[attribute.Value] = top.Index;
}
}
private IEnumerable<TokenBase> ParseSpineItem(EpubSpineItem item, Stack<TextVisualProperties> propertiesStack, TokenIndex top)
{
HtmlNode pageBody = GetPageBody(item);
return ParseNodes(pageBody, propertiesStack, top, _opfPath + item.Path);
}
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);
}
}
}
}