/// <summary>
/// Get a completed post from the provided HTML div node.
/// </summary>
/// <param name="li">Div node that contains the post.</param>
/// <returns>Returns a post object with required information.</returns>
private PostComponents GetPost(HtmlNode li, IQuest quest)
{
if (li == null)
{
throw new ArgumentNullException(nameof(li));
}
string author;
string id;
string text;
int number;
id = li.GetAttributeValue("data-pid", "");
author = li.GetAttributeValue("data-username", "");
number = int.Parse(li.GetAttributeValue("data-index", "0"));
var content = li.GetChildWithClass("div", "content");
// Get the full post text.
text = PostText.ExtractPostText(content, n => false, Host);
PostComponents post;
try
{
post = new PostComponents(author, id, text, number);
}
catch
{
post = null;
}
return(post);
}
/// <summary>
/// Get a completed post from the provided HTML list item node.
/// </summary>
/// <param name="li">List item node that contains the post.</param>
/// <returns>Returns a post object with required information.</returns>
private PostComponents GetPost(HtmlNode li)
{
if (li == null)
{
throw new ArgumentNullException(nameof(li));
}
string author = "";
string id = "";
string text = "";
int number = 0;
// ID
id = li.Id.Substring("post_".Length);
// Number
var postCount = li.OwnerDocument.GetElementbyId($"postcount{id}");
if (postCount != null)
{
number = int.Parse(postCount.GetAttributeValue("name", "0"));
}
HtmlNode postDetails = li.Elements("div").FirstOrDefault(n => n.GetAttributeValue("class", "") == "postdetails");
if (postDetails != null)
{
// Author
HtmlNode userinfo = postDetails.GetChildWithClass("div", "userinfo");
HtmlNode username = userinfo?.GetChildWithClass("a", "username");
author = PostText.CleanupWebString(username?.InnerText);
// Text
string postMessageId = "post_message_" + id;
var message = li.OwnerDocument.GetElementbyId(postMessageId)?.Element("blockquote");
// Predicate filtering out elements that we don't want to include
var exclusion = PostText.GetClassExclusionPredicate("bbcode_quote");
// Get the full post text.
text = PostText.ExtractPostText(message, exclusion, Host);
}
PostComponents post;
try
{
post = new PostComponents(author, id, text, number);
}
catch
{
post = null;
}
return(post);
}
/// <summary>
/// Get a completed post from the provided HTML div node.
/// </summary>
/// <param name="div">Div node that contains the post.</param>
/// <returns>Returns a post object with required information.</returns>
private PostComponents GetPost(HtmlNode div, IQuest quest)
{
if (div == null)
{
throw new ArgumentNullException(nameof(div));
}
string author = "";
string id;
string text;
int number = 0;
// id="p12345"
id = div.Id.Substring(1);
var inner = div.GetChildWithClass("div", "inner");
var postbody = inner.GetChildWithClass("div", "postbody");
var authorNode = postbody.GetChildWithClass("p", "author");
var authorStrong = authorNode.Descendants("strong").FirstOrDefault();
var authorAnchor = authorStrong.Element("a");
author = PostText.CleanupWebString(authorAnchor.InnerText);
// No way to get the post number??
// Get the full post text. Two different layout variants.
var content = postbody.GetChildWithClass("div", "content");
if (content == null)
{
content = postbody.Elements("div").FirstOrDefault(n => n.Id.StartsWith("post_content", StringComparison.Ordinal));
}
text = PostText.ExtractPostText(content, n => false, Host);
PostComponents post;
try
{
post = new PostComponents(author, id, text, number);
}
catch
{
post = null;
}
return(post);
}
/// <summary>
/// Get a completed post from the provided HTML div node.
/// </summary>
/// <param name="postDiv">Div node that contains the post.</param>
/// <returns>Returns a post object with required information.</returns>
private PostComponents GetPost(HtmlNode postDiv)
{
if (postDiv == null)
{
throw new ArgumentNullException(nameof(postDiv));
}
string author = "";
string id;
string text;
int number = 0;
var postTable = postDiv.Descendants("table").FirstOrDefault(a => a.Id.StartsWith("post", StringComparison.Ordinal));
if (postTable == null)
{
return(null);
}
id = postTable.Id.Substring("post".Length);
string postAuthorDivID = "postmenu_" + id;
var authorAnchor = postTable.OwnerDocument.GetElementbyId(postAuthorDivID).Element("a");
if (authorAnchor != null)
{
author = authorAnchor.InnerText;
// ??
if (authorAnchor.Element("span") != null)
{
author = authorAnchor.Element("span").InnerText;
}
}
string postNumberAnchorID = "postcount" + id;
var anchor = postTable.OwnerDocument.GetElementbyId(postNumberAnchorID);
if (anchor != null)
{
string postNumText = anchor.GetAttributeValue("name", "");
number = int.Parse(postNumText);
}
string postMessageId = "post_message_" + id;
var postContents = postTable.OwnerDocument.GetElementbyId(postMessageId);
// Predicate filtering out elements that we don't want to include
var exclusion = PostText.GetClassExclusionPredicate("bbcode_quote");
// Get the full post text.
text = PostText.ExtractPostText(postContents, exclusion, Host);
PostComponents post;
try
{
post = new PostComponents(author, id, text, number);
}
catch
{
post = null;
}
return(post);
}
/// <summary>
/// Get a completed post from the provided HTML list item node.
/// </summary>
/// <param name="li">List item node that contains the post.</param>
/// <returns>Returns a post object with required information.</returns>
private PostComponents GetPost(HtmlNode li)
{
if (li == null)
{
throw new ArgumentNullException(nameof(li));
}
string author;
string id;
string text;
int number;
// Author and ID are in the basic list item attributes
author = PostText.CleanupWebString(li.GetAttributeValue("data-author", ""));
id = li.Id.Substring("post-".Length);
if (AdvancedOptions.Instance.DebugMode)
{
author = $"{author}_{id}";
}
// Get the primary content of the list item
HtmlNode primaryContent = li.GetChildWithClass("primaryContent");
// On one branch, we can get the post text
HtmlNode messageContent = primaryContent.GetChildWithClass("messageContent");
HtmlNode postBlock = messageContent.Element("article").Element("blockquote");
// Predicate filtering out elements that we don't want to include
List <string> excludedClasses = new List <string> {
"bbCodeQuote", "messageTextEndMarker", "advbbcodebar_encadre",
"advbbcodebar_article", "adv_tabs_wrapper", "adv_slider_wrapper"
};
if (AdvancedOptions.Instance.IgnoreSpoilers)
{
excludedClasses.Add("bbCodeSpoilerContainer");
}
var exclusions = PostText.GetClassesExclusionPredicate(excludedClasses);
// Get the full post text.
text = PostText.ExtractPostText(postBlock, exclusions, Host);
// On another branch of the primary content, we can get the post number.
HtmlNode messageMeta = primaryContent.GetChildWithClass("messageMeta");
// HTML parsing of the post was corrupted somehow.
if (messageMeta == null)
{
return(null);
}
HtmlNode publicControls = messageMeta.GetChildWithClass("publicControls");
HtmlNode postNumber = publicControls.GetChildWithClass("postNumber");
string postNumberText = postNumber.InnerText;
// Skip the leading # character.
if (postNumberText.StartsWith("#", StringComparison.Ordinal))
{
postNumberText = postNumberText.Substring(1);
}
number = int.Parse(postNumberText);
PostComponents post;
try
{
post = new PostComponents(author, id, text, number);
}
catch (Exception e)
{
ErrorLog.Log(e);
post = null;
}
return(post);
}
/// <summary>
/// Get a completed post from the provided HTML list item.
/// </summary>
/// <param name="li">List item that contains the post.</param>
/// <returns>Returns a post object with required information.</returns>
private PostComponents GetPost(HtmlNode li)
{
if (li == null)
{
throw new ArgumentNullException(nameof(li));
}
string author = "";
string id = "";
string text = "";
int number = 0;
// ID
id = li.GetAttributeValue("data-node-id", "");
if (string.IsNullOrEmpty(id))
{
return(null);
}
// Author
var postAuthorNode = li.Descendants("div").FirstOrDefault(a => a.GetAttributeValue("itemprop", "") == "author");
var authorNode = postAuthorNode?.GetDescendantWithClass("div", "author");
if (authorNode != null)
{
author = PostText.CleanupWebString(authorNode.InnerText);
}
var contentArea = li.GetDescendantWithClass("div", "b-post__content");
// Number
var postCountAnchor = contentArea.GetDescendantWithClass("a", "b-post__count");
if (postCountAnchor != null)
{
string postNumText = postCountAnchor.InnerText;
if (postNumText.StartsWith("#", StringComparison.Ordinal))
{
postNumText = postNumText.Substring(1);
}
number = int.Parse(postNumText);
}
// Text
var postTextNode = contentArea.Descendants("div").FirstOrDefault(a => a.GetAttributeValue("itemprop", "") == "text");
// Predicate filtering out elements that we don't want to include
var exclusion = PostText.GetClassExclusionPredicate("bbcode_quote");
// Get the full post text.
text = PostText.ExtractPostText(postTextNode, exclusion, Host);
PostComponents post;
try
{
post = new PostComponents(author, id, text, number);
}
catch
{
post = null;
}
return(post);
}