public override void chars(string text)
{
if (preStack.Count > 0)
{
text = ReplaceForPre(text);
}
else if (text.Trim() != "")
{
text = ReplaceMultipleWhitespaveWithOne(text);
var prevText = HtmlToMarkdownConverterHelper.peekTillNotEmpty(nodeStack.ToList());
if (prevText.EndsWith(" "))
{
text = text.TrimStart();
}
}
else
{
nodeStack.Push("");
return;
}
//if(blockquoteStack.length > 0 && peekTillNotEmpty(nodeList).endsWith("\n")) {
if (blockquoteStack.Count > 0)
{
var array = blockquoteStack.ToArray();
Array.Reverse(array);
nodeStack.Push(string.Join(string.Empty, array));
}
nodeStack.Push(ReplaceSpecialChars(text));
}
public bool removeIfEmptyTag(string start)
{
var cleaned = false;
if (start == HtmlToMarkdownConverterHelper.peekTillNotEmpty(nodeStack.ToList()))
{
while (nodeStack.Peek() != start)
{
nodeStack.Pop();
}
nodeStack.Pop();
cleaned = true;
}
return(cleaned);
}
private void block(bool isEndBlock)
{
if (nodeStack.Count == 0)
{
return;
}
var lastItem = nodeStack.Pop();
if (string.IsNullOrWhiteSpace(lastItem))
{
return;
}
if (!isEndBlock)
{
string block;
if (Regex.IsMatch(lastItem, @"\s*\n\n\s*$"))
{
lastItem = Regex.Replace(lastItem, @"\s*\n\n\s*$", "\n\n");
block = "";
}
else if (Regex.IsMatch(lastItem, @"\s*\n\s*$"))
{
lastItem = Regex.Replace(lastItem, @"\s*\n\s*$", "\n");
block = "\n";
}
else if (Regex.IsMatch(lastItem, @"\s+$"))
{
block = "\n\n";
}
else
{
block = "\n\n";
}
nodeStack.Push(lastItem);
nodeStack.Push(block);
}
else
{
nodeStack.Push(lastItem);
if (!HtmlToMarkdownConverterHelper.endsWith(lastItem, "\n"))
{
nodeStack.Push("\n\n");
}
}
}
public override void end(string tag)
{
tag = tag.ToLower();
switch (tag)
{
case "title":
case "h1":
case "h2":
case "h3":
case "h4":
case "h5":
case "h6":
if (!removeIfEmptyTag(Markdown.Tags[tag]))
{
block(true);
}
break;
case "p":
case "div":
case "table":
case "tbody":
case "tr":
case "td":
while (nodeStack.Count > 0 && nodeStack.Peek().Trim() == "")
{
nodeStack.Pop();
}
block(true);
break;
case "b":
case "strong":
case "i":
case "em":
case "dfn":
case "var":
case "cite":
if (!removeIfEmptyTag(Markdown.Tags[tag]))
{
nodeStack.Push(sliceText(Markdown.Tags[tag]).Trim());
nodeStack.Push(Markdown.Tags[tag]);
}
break;
case "a":
var text = sliceText("[");
text = Regex.Replace(text, @"\s+", " ");
text = text.Trim();
if (text == "")
{
nodeStack.Pop();
break;
}
var attrs = linkAttrStack.Pop();
var url = attrs.ContainsKey("href") && attrs["href"].Value != "" ? attrs["href"].Value : "";
if (url == "")
{
nodeStack.Pop();
nodeStack.Push(text);
break;
}
nodeStack.Push(text);
if (!inlineStyle && !HtmlToMarkdownConverterHelper.startsWith(nodeStack.Peek(), "!"))
{
var l = links.IndexOf(url);
if (l == -1)
{
links.Add(url);
l = links.Count - 1;
}
nodeStack.Push("][" + l + "]");
}
else
{
if (HtmlToMarkdownConverterHelper.startsWith(nodeStack.Peek(), "!"))
{
var localText = nodeStack.Pop();
localText = nodeStack.Pop() + localText;
block(false);
nodeStack.Push(localText);
}
var title = attrs.ContainsKey("title") && attrs["title"].Value != "" ? attrs["title"].Value : "";
var trimmedTitle = title.Trim();
nodeStack.Push("](" + url + (false == string.IsNullOrWhiteSpace(title) ? " \"" + Regex.Replace(trimmedTitle, @"\s+", " ") + "\"" : "") + ")");
if (HtmlToMarkdownConverterHelper.startsWith(nodeStack.Peek(), "!"))
{
block(true);
}
}
break;
case "ul":
case "ol":
case "dl":
listBlock();
listTagStack.Pop();
break;
case "li":
case "dt":
var li = getListMarkdownTag();
if (!removeIfEmptyTag(li))
{
var liContent = sliceText(li).Trim();
if (liContent.StartsWith("[!["))
{
nodeStack.Pop();
block(false);
nodeStack.Push(liContent);
block(true);
}
else
{
nodeStack.Push(liContent);
listBlock();
}
}
break;
case "blockquote":
blockquoteStack.Pop();
break;
case "pre":
// //uncomment following experimental code to discard line numbers when syntax highlighters are used
// //notes this code thorough testing before production user
// /*
// var p=[];
// var flag = true;
// var count = 0, whiteSpace = 0, line = 0;
// console.log(">> " + peek(nodeList));
// while(peek(nodeList).startsWith(" ") || flag == true)
// {
// //console.log('inside');
// var text = nodeList.pop();
// p.push(text);
// if(flag == true && !text.startsWith(" ")) {
// continue;
// } else {
// flag = false;
// }
// //var result = parseInt(text.trim());
// if(!isNaN(text.trim())) {
// count++;
// } else if(text.trim() == ""){
// whiteSpace++;
// } else {
// line++;
// }
// flag = false;
// }
// console.log(line);
// if(line != 0)
// {
// while(p.length != 0) {
// nodeList.push(p.pop());
// }
// }
// */
// block(true);
// preStack.pop();
break;
case "code":
case "span":
if (preStack.Count > 0)
{
break;
}
else if (string.IsNullOrWhiteSpace(nodeStack.Peek()))
{
nodeStack.Pop();
nodeStack.Push(Markdown.Tags[tag]);
}
else
{
var item = nodeStack.Pop();
nodeStack.Push(item.Trim());
nodeStack.Push(Markdown.Tags[tag]);
}
break;
//case "table":
// nodeList.push("</table>");
// break;
//case "thead":
// nodeList.push("</thead>");
// break;
//case "tbody":
// nodeList.push("</tbody>");
// break;
//case "tr":
// nodeList.push("</tr>");
// break;
//case "td":
// nodeList.push("</td>");
// break;
case "br":
case "hr":
case "img":
break;
}
}
public override void start(string tag, Dictionary <string, HtmlAttribute> attrs, bool unary)
{
tag = tag.ToLower();
if (unary && (tag != "br" && tag != "hr" && tag != "img"))
{
return;
}
switch (tag)
{
case "p":
case "div":
case "table":
case "tbody":
case "tr":
case "td":
block(false);
break;
}
switch (tag)
{
case "br":
nodeStack.Push(Markdown.Tags[tag]);
break;
case "hr":
block(false);
nodeStack.Push(Markdown.Tags[tag]);
break;
case "title":
case "h1":
case "h2":
case "h3":
case "h4":
case "h5":
case "h6":
block(false);
nodeStack.Push(Markdown.Tags[tag]);
break;
case "b":
case "strong":
case "i":
case "em":
case "dfn":
case "var":
case "cite":
nodeStack.Push(Markdown.Tags[tag]);
break;
case "code":
case "span":
if (preStack.Count > 0)
{
break;
}
else if (false == nodeStack.SafePeek().EndsWith(" "))
{
nodeStack.Push(Markdown.Tags[tag]);
}
break;
case "ul":
case "ol":
case "dl":
listTagStack.Push(Markdown.Tags[tag]);
// lists are block elements
if (listTagStack.Count > 1)
{
listBlock();
}
else
{
block(false);
}
break;
case "li":
case "dt":
var li = getListMarkdownTag();
nodeStack.Push(li);
break;
case "a":
linkAttrStack.Push(HtmlToMarkdownConverterHelper.convertAttrs(attrs));
nodeStack.Push("[");
break;
case "img":
var attribs = HtmlToMarkdownConverterHelper.convertAttrs(attrs);
string alt, title, url;
url = attribs.ContainsKey("src") ? attribs["src"].Value : "";
if (string.IsNullOrWhiteSpace(url))
{
break;
}
alt = attribs.ContainsKey("alt") ? attribs["alt"].Value.Trim() : "";
title = attribs.ContainsKey("title") ? attribs["title"].Value.Trim() : "";
// if parent of image tag is nested in anchor tag use inline style
if (!inlineStyle && false == HtmlToMarkdownConverterHelper.startsWith(HtmlToMarkdownConverterHelper.peekTillNotEmpty(nodeStack.ToList()), "["))
{
var l = links.IndexOf(url);
if (l == -1)
{
links.Add(url);
l = links.Count - 1;
}
block(false);
nodeStack.Push("![");
if (alt != "")
{
nodeStack.Push(alt);
}
else if (title != null)
{
nodeStack.Push(title);
}
nodeStack.Push("][" + l + "]");
block(false);
}
else
{
//if image is not a link image then treat images as block elements
if (false == HtmlToMarkdownConverterHelper.startsWith(HtmlToMarkdownConverterHelper.peekTillNotEmpty(nodeStack.ToList()), "["))
{
block(false);
}
nodeStack.Push("![" + alt + "](" + url + (false == string.IsNullOrWhiteSpace(title) ? " \"" + title + "\"" : "") + ")");
if (false == HtmlToMarkdownConverterHelper.startsWith(HtmlToMarkdownConverterHelper.peekTillNotEmpty(nodeStack.ToList()), "["))
{
block(true);
}
}
break;
case "blockquote":
//listBlock();
block(false);
blockquoteStack.Push(Markdown.Tags[tag]);
break;
case "pre":
block(false);
preStack.Push(true);
nodeStack.Push(" ");
break;
//case "table":
// nodeStack.Push("<table>");
// break;
//case "thead":
// nodeStack.Push("<thead>");
// break;
//case "tbody":
// nodeStack.Push("<tbody>");
// break;
//case "tr":
// nodeStack.Push("<tr>");
// break;
//case "td":
// nodeStack.Push("<td>");
// break;
}
}
