/// <summary>
/// Builds a list tree.
/// </summary>
/// <param name="ordered"><c>true</c> for an ordered list.</param>
/// <param name="html">The input HTML.</param>
/// <returns>The list tree.</returns>
private static HtmlList BuildListTree(bool ordered, string html) {
string[] tags = new string[] { "<ol>", "<ul>", "<li>", "</li>", "</ul>", "</ol>" };
// IE seems to add new-lines after some elements
// \r\n are never added by the Formatter, so it is safe to remove all them
html = html.Replace("\r", "");
html = html.Replace("\n", "");
int index = 0;
int lastOpenListItemIndex = 0;
int stringFound;
HtmlList root = new HtmlList(ordered ? HtmlListType.Ordered : HtmlListType.Unordered);
HtmlList currentList = root;
do {
index = FirstIndexOfAny(html, index, out stringFound, tags);
if(index != -1) {
switch(stringFound) {
case 0: // <ol>
// Unless at the beginning, start a new sub-list
if(index != 0) {
// Set text of current element (sub-lists are added into the previous item)
if(lastOpenListItemIndex != -1) {
string text = html.Substring(lastOpenListItemIndex + 4, index - (lastOpenListItemIndex + 4));
currentList.Elements[currentList.Elements.Count - 1].Text = text;
}
currentList.Elements[currentList.Elements.Count - 1].SubList = new HtmlList(HtmlListType.Ordered);
currentList = currentList.Elements[currentList.Elements.Count - 1].SubList;
}
break;
case 1: // <ul>
// Unless at the beginning, start a new sub-list
if(index != 0) {
// Set text of current element (sub-lists are added into the previous item)
if(lastOpenListItemIndex != -1) {
string text = html.Substring(lastOpenListItemIndex + 4, index - (lastOpenListItemIndex + 4));
currentList.Elements[currentList.Elements.Count - 1].Text = text;
}
currentList.Elements[currentList.Elements.Count - 1].SubList = new HtmlList(HtmlListType.Unordered);
currentList = currentList.Elements[currentList.Elements.Count - 1].SubList;
}
break;
case 2: // <li>
lastOpenListItemIndex = index;
currentList.Elements.Add(new HtmlListElement());
break;
case 3: // </li>
// If lastOpenListItemIndex != -1 (i.e. there are no sub-lists) extract item text and set it to the last list element
// Otherwise, navigate upwards to parent list (if any)
if(lastOpenListItemIndex != -1) {
string text = html.Substring(lastOpenListItemIndex + 4, index - (lastOpenListItemIndex + 4));
currentList.Elements[currentList.Elements.Count - 1].Text = text;
}
else {
currentList = FindAnchestor(root, currentList);
}
break;
case 4: // </ul>
// Close last open list (nothing to do)
lastOpenListItemIndex = -1;
break;
case 5: // </ol>
// Close last open list (nothing to do)
lastOpenListItemIndex = -1;
break;
default:
throw new NotSupportedException();
}
index++;
}
} while(index != -1);
return root;
}
/// <summary>
/// Finds the anchestor of a list in a tree.
/// </summary>
/// <param name="root">The root of the tree.</param>
/// <param name="current">The current element.</param>
/// <returns>The anchestor of <b>current</b>.</returns>
private static HtmlList FindAnchestor(HtmlList root, HtmlList current) {
foreach(HtmlListElement elem in root.Elements) {
if(elem.SubList == current) return root;
else if(elem.SubList != null) {
HtmlList temp = FindAnchestor(elem.SubList, current);
if(temp != null) return temp;
}
}
//return root;
return null;
}
/// <summary>
/// Builds the WikiMarkup for a list.
/// </summary>
/// <param name="list">The root list.</param>
/// <param name="previousBullets">The previous bullets, used at upper levels.</param>
/// <returns>The WikiMarkup.</returns>
private static string BuildListWikiMarkup(HtmlList list, string previousBullets) {
previousBullets = previousBullets + (list.Type == HtmlListType.Ordered ? "#" : "*");
StringBuilder sb = new StringBuilder(500);
foreach(HtmlListElement elem in list.Elements) {
sb.Append(previousBullets);
sb.Append(" ");
sb.Append(elem.Text);
sb.Append("\r\n");
if(elem.SubList != null) {
sb.Append(BuildListWikiMarkup(elem.SubList, previousBullets));
}
}
// Remove empty lines in the middle of the list
string raw = sb.ToString().Replace("\r", "");
string[] lines = raw.Split(new char[] { '\n' }, StringSplitOptions.RemoveEmptyEntries);
return
string.Join("\r\n", lines) +
(raw.EndsWith("\r\n") || raw.EndsWith("\n") ? "\r\n" : "");
}