internal HtmlPage(WebBytes wb, WebString prelimContent) : base(wb.Location)
{
HeadersCharset = wb.CharacterSet;
EncHeaders = prelimContent.UsedEncoding;
DetermineEncoding(wb.Data, prelimContent.Document);
}
/// <summary>
/// Get post associated with post number.
/// </summary>
/// <returns><see cref="ChanPost">ChanPost</see> detailing the post</returns>
///
/// <exception cref="ArgumentNullException">Thrown if board is null.</exception>
/// <exception cref="ArgumentException">Thrown if board is empty or whitespace.</exception>
/// <exception cref="ArgumentOutOfRangeException">Thrown if postNo <= 0.</exception>
///
/// <param name="board">Board where post is located.</param>
/// <param name="postNo">Post number.</param>
public static ChanPost GetPost(string board, int postNo)
{
board.ThrowIfNullOrWhiteSpace("board");
if (postNo < 1)
{
throw new ArgumentOutOfRangeException("post", "Cannot be 0 or negative.");
}
var jsonReq = string.Format("http://archive.moe/_/api/chan/post/?board={0}&num={1}", board, postNo);
var json = WebString.Download(jsonReq);
if (!json.Success)
{
return(new ChanPost(json));
}
dynamic postJson = JsonConvert.DeserializeObject(json.Document);
int threadNo = postJson.thread_num;
string subject = postJson.title;
string comment = postJson.comment_sanitized;
return(new ChanPost(json.Location,
board, ChanTools.GetBoardName(board),
threadNo, postNo,
subject, comment));
}
public WeatherConditions GetConditions(WeatherLocation location)
{
if (location == null)
{
throw new ArgumentNullException(nameof(location));
}
if (!location.IsValidQuery)
{
throw new ArgumentException("WeatherLocation must have a valid value.", nameof(location));
}
var queryResult = WebString.Download(MakeQuery(location));
if (!queryResult.Success)
{
return(new WeatherConditions(queryResult));
}
var json = JObject.Parse(queryResult.Document);
var message = (string)json["message"];
if (string.IsNullOrEmpty(message))
{
return(new WeatherConditions(queryResult.Location, json));
}
else
{
var ex = new JsonErrorException(message);
return(new WeatherConditions(queryResult.Location, ex));
}
}
FetcherResult Load(Uri uri, bool useCookies, int redirects, string previousPage)
{
var req = SetupRequest(uri, useCookies);
var wb = WebBytes.Create(req, MaxSizeHtml, MaxSizeNonHtml);
if (wb.ContentIsHtml)
{
var page = WebString.Create(wb, EncHelp.Windows1252);
Uri refreshUrl;
if (redirects < MaxRefreshes &&
TryGetMetaRefresh(page, out refreshUrl) &&
VerifyRefresh(page.Location, refreshUrl))
{
// If during the redirects we get a different page/HTML string, refrain from following more
// redirects. It probably means we've arrived, but only more real world testing will tell us
// if that's true.
// Otherwise keep trying to follow the redirects.
if (redirects == 0 || page.Document == previousPage)
{
return(Load(refreshUrl, useCookies, (redirects + 1), page.Document));
}
}
return(new FetcherResult(wb, page));
}
// Silently handle cookie exceptions, Mono/.NET can be very strict with which cookies it accepts.
if (!wb.Success && wb.Exception.InnerException is CookieException)
{
return(Load(uri, false, redirects, previousPage));
}
// If either not HTML or there was an error in getting the resource, return as-is.
return(new FetcherResult(wb));
}
/// <summary>
/// Get info of a Danbooru post.
/// </summary>
/// <returns><see cref="DanboPost">DanboPost</see> detailing a post.</returns>
/// <exception cref="ArgumentOutOfRangeException">Thrown if postNo is <= 0.</exception>
/// <param name="postNo">Post number.</param>
public static DanboPost GetPostInfo(int postNo)
{
if (postNo < 1)
{
throw new ArgumentOutOfRangeException(nameof(postNo), "Can't be 0 or negative.");
}
var jsonReq = string.Format("https://danbooru.donmai.us/posts/{0}.json", postNo);
var json = WebString.Download(jsonReq);
if (!json.Success)
{
return(new DanboPost(json));
}
dynamic postJson = JsonConvert.DeserializeObject(json.Document);
string copyrights = postJson.tag_string_copyright;
string characters = postJson.tag_string_character;
string artists = postJson.tag_string_artist;
string general = postJson.tag_string_general;
string meta = postJson.tag_string_meta;
string all = postJson.tag_string;
string rating = postJson.rating;
return(new DanboPost(json.Location, postNo,
copyrights, characters, artists, general, meta, all, rating));
}
/// <summary>
/// Get post associated with post number.
/// </summary>
/// <returns><see cref="ChanPost">ChanPost</see> detailing the post.</returns>
///
/// <exception cref="ArgumentNullException">Thrown if board is null.</exception>
/// <exception cref="ArgumentException">Thrown if board is empty or whitespace.</exception>
/// <exception cref="ArgumentOutOfRangeException">Thrown if threadNo and/or postNo <= 0.</exception>
///
/// <param name="board">Board where thread/post is located.</param>
/// <param name="threadNo">Thread number.</param>
/// <param name="postNo">Post number.</param>
public static ChanPost GetPost(string board, int threadNo, int postNo)
{
board.ThrowIfNullOrWhiteSpace("board");
if (threadNo < 1)
{
throw new ArgumentOutOfRangeException("thread", "Cannot be 0 or negative.");
}
else if (postNo < 1)
{
throw new ArgumentOutOfRangeException("post", "Cannot be 0 or negative.");
}
var jsonReq = string.Format("http://a.4cdn.org/{0}/res/{1}.json", board, threadNo);
var json = WebString.Download(jsonReq);
if (!json.Success)
{
return(new ChanPost(json));
}
dynamic threadJson = JsonConvert.DeserializeObject(json.Document);
string subject = null;
string comment = null;
// Simple linear search, seems fast enough for the 100s of posts it needs to look through.
foreach (var post in threadJson.posts)
{
if (post.no == postNo)
{
subject = post.sub;
if (!string.IsNullOrEmpty(subject))
{
subject = HttpUtility.HtmlDecode(subject);
}
comment = post.com;
if (!string.IsNullOrEmpty(comment))
{
comment = Fix4chanPost(comment);
}
break;
}
}
return(new ChanPost(json.Location,
board, ChanTools.GetBoardName(board),
threadNo, postNo,
subject, comment));
}
public static SearchResults Search(string query)
{
if (query == null)
{
throw new ArgumentNullException(nameof(query));
}
var results = WebString.Download(GoogleUrl(query));
if (!results.Success)
{
return(new SearchResults(results));
}
var matches = resultsRegexp.Matches(results.Document);
var parsedResults = ParseMatches(matches);
return(new SearchResults(results, Deduplication(parsedResults)));
}
/// <summary>
/// Get info of a Gelbooru post.
/// </summary>
/// <returns><see cref="BooruPost">BooruPost</see> detailing a post.</returns>
/// <exception cref="ArgumentOutOfRangeException">Thrown if postNo is <= 0.</exception>
/// <param name="postNo">Post number.</param>
public static BooruPost GetPostInfo(int postNo)
{
if (postNo < 1)
{
throw new ArgumentOutOfRangeException(nameof(postNo), "Can't be 0 or negative.");
}
var xmlReq = string.Format("https://gelbooru.com/index.php?page=dapi&s=post&q=index&id={0}", postNo);
var xml = WebString.Download(xmlReq);
if (!xml.Success)
{
return(new BooruPost(xml));
}
var postXml = XElement.Parse(xml.Document).Element("post");
string tags = postXml.Attribute("tags").Value;
string rated = postXml.Attribute("rating").Value;
return(new BooruPost(xml.Location, postNo, tags, rated));
}
static bool TryGetMetaRefresh(WebString page, out Uri refreshUrl)
{
string metaRefresh = HtmlTagExtract.GetMetaRefresh(page.Document);
if (metaRefresh != null)
{
// Relative refresh URL.
// Try this one first, because creating an absolute URL out of a relative one starting with a '/' can
// lead to it being interpreted as a file:/// URI.
if (Uri.TryCreate(page.Location, metaRefresh, out refreshUrl))
{
return(true);
}
// Absolute refresh URL.
if (Uri.TryCreate(metaRefresh, UriKind.Absolute, out refreshUrl))
{
return(true);
}
}
refreshUrl = null;
return(false);
}
internal FetcherResult(WebBytes bytes, WebString page)
{
Bytes = bytes;
Page = new HtmlPage(bytes, page);
}
