diff --git a/NadekoBot/Classes/IMDB/ImdbMovie.cs b/NadekoBot/Classes/IMDB/ImdbMovie.cs index 18bb3478..92bdb6d0 100644 --- a/NadekoBot/Classes/IMDB/ImdbMovie.cs +++ b/NadekoBot/Classes/IMDB/ImdbMovie.cs @@ -1,78 +1,62 @@ -using System.Collections; +using NadekoBot.Extensions; using System.Collections.Generic; namespace NadekoBot.Classes.IMDB { public class ImdbMovie { - public bool status { get; set; } + public bool Status { get; set; } public string Id { get; set; } public string Title { get; set; } public string OriginalTitle { get; set; } public string Year { get; set; } public string Rating { get; set; } - public List Genres { get; set; } - public ArrayList Directors { get; set; } - public ArrayList Writers { get; set; } - public ArrayList Cast { get; set; } - public ArrayList Producers { get; set; } - public ArrayList Musicians { get; set; } - public ArrayList Cinematographers { get; set; } - public ArrayList Editors { get; set; } - public string MpaaRating { get; set; } - public string ReleaseDate { get; set; } public string Plot { get; set; } - public ArrayList PlotKeywords { get; set; } public string Poster { get; set; } - public string PosterLarge { get; set; } - public string PosterFull { get; set; } - public string Runtime { get; set; } - public string Top250 { get; set; } - public string Oscars { get; set; } - public string Awards { get; set; } - public string Nominations { get; set; } - public string Storyline { get; set; } - public string Tagline { get; set; } - public string Votes { get; set; } - public ArrayList Languages { get; set; } - public ArrayList Countries { get; set; } - public Dictionary ReleaseDates { get; set; } - public ArrayList MediaImages { get; set; } - public ArrayList RecommendedTitles { get; set; } + public List Genres { get; set; } + //public ArrayList Directors { get; set; } + //public ArrayList Writers { get; set; } + //public ArrayList Cast { get; set; } + //public ArrayList Producers { get; set; } + //public ArrayList Musicians { get; set; } + //public ArrayList Cinematographers { get; set; } + //public ArrayList Editors { get; set; } + //public string MpaaRating { get; set; } + //public string ReleaseDate { get; set; } + //public ArrayList PlotKeywords { get; set; } + //public string PosterLarge { get; set; } + //public string PosterFull { get; set; } + //public string Runtime { get; set; } + //public string Top250 { get; set; } + //public string Oscars { get; set; } + //public string Awards { get; set; } + //public string Nominations { get; set; } + //public string Storyline { get; set; } + //public string Tagline { get; set; } + //public string Votes { get; set; } + //public ArrayList Languages { get; set; } + //public ArrayList Countries { get; set; } + //public Dictionary ReleaseDates { get; set; } + //public ArrayList MediaImages { get; set; } + //public ArrayList RecommendedTitles { get; set; } public string ImdbURL { get; set; } public Dictionary Aka { get; set; } - public override string ToString() - { - return "`Title:` **" + EnglishTitle + " (" + OriginalTitle + ")" + - "**\n`Year:` " + Year + - "**\n`Rating:` " + Rating + - "**\n`Genre:` " + GenresAsString + - "\n`Link:` " + ImdbURL + - "\n`Plot:` " + Plot.Substring(0, Plot.Length > 500 ? 500 : Plot.Length) + "..." - //"\n`img:` " + Poster //imdb url do it for us I think its a discord auto thing - ; - } + public override string ToString() => +$@"`Title:` {Title} {(string.IsNullOrEmpty(OriginalTitle) ? "" : $"({OriginalTitle})")} +`Year:` {Year} +`Rating:` {Rating} +`Genre:` {GenresAsString} +`Link:` <{ImdbURL}> +`Plot:` {System.Net.WebUtility.HtmlDecode(Plot.TrimTo(500))} +`img:` " + Poster.ShortenUrl().Result; - public string EnglishTitle - { - get - { - return Aka.ContainsKey("USA") ? Aka["USA"] : - (Aka.ContainsKey("UK") ? Aka["UK"] : - (Aka.ContainsKey("(original title)") ? Aka["(original title)"] : - (Aka.ContainsKey("(original)") ? Aka["(original)"] : OriginalTitle))); - } - } - public string GenresAsString - { - get - { - string ret = ""; - Genres.ForEach(g => ret = ret + " " + g); - return ret; - } - } + //public string EnglishTitle => Aka.ContainsKey("USA") ? Aka["USA"] : + // (Aka.ContainsKey("UK") ? Aka["UK"] : + // (Aka.ContainsKey("(original title)") ? Aka["(original title)"] : + // (Aka.ContainsKey("(original)") ? Aka["(original)"] : OriginalTitle))); + public string GenresAsString => + string.Join(", ", Genres); } } \ No newline at end of file diff --git a/NadekoBot/Classes/IMDB/ImdbScraper.cs b/NadekoBot/Classes/IMDB/ImdbScraper.cs index 12ab9bf2..cd319e19 100644 --- a/NadekoBot/Classes/IMDB/ImdbScraper.cs +++ b/NadekoBot/Classes/IMDB/ImdbScraper.cs @@ -18,15 +18,10 @@ using System.Text.RegularExpressions; * Last Updated: Feb, 2016 *******************************************************************************/ - - namespace NadekoBot.Classes.IMDB { public static class ImdbScraper { - - - //Search Engine URLs private static string GoogleSearch = "https://www.google.com/search?q=imdb+"; private static string BingSearch = "http://www.bing.com/search?q=imdb+"; @@ -35,11 +30,11 @@ namespace NadekoBot.Classes.IMDB public static ImdbMovie ImdbScrape(string MovieName, bool GetExtraInfo = true) { ImdbMovie mov = new ImdbMovie(); - string imdbUrl = getIMDbUrl(System.Uri.EscapeUriString(MovieName)); - mov.status = false; + string imdbUrl = GetIMDbUrl(System.Uri.EscapeUriString(MovieName)); + mov.Status = false; if (!string.IsNullOrEmpty(imdbUrl)) { - parseIMDbPage(imdbUrl, GetExtraInfo, mov); + ParseIMDbPage(imdbUrl, GetExtraInfo, mov); } return mov; @@ -49,10 +44,10 @@ namespace NadekoBot.Classes.IMDB { ImdbMovie mov = new ImdbMovie(); string imdbUrl = "http://www.imdb.com/title/" + imdbId + "/"; - mov.status = false; + mov.Status = false; if (!string.IsNullOrEmpty(imdbUrl)) { - parseIMDbPage(imdbUrl, GetExtraInfo, mov); + ParseIMDbPage(imdbUrl, GetExtraInfo, mov); } return mov; @@ -60,98 +55,98 @@ namespace NadekoBot.Classes.IMDB public static string GetIMDBId(string MovieName) { - string imdbUrl = getIMDbUrl(System.Uri.EscapeUriString(MovieName)); + string imdbUrl = GetIMDbUrl(System.Uri.EscapeUriString(MovieName)); return match(@"http://www.imdb.com/title/(tt\d{7})", imdbUrl); } //Get IMDb URL from search results - private static string getIMDbUrl(string MovieName, string searchEngine = "google") + private static string GetIMDbUrl(string MovieName, string searchEngine = "google") { string url = GoogleSearch + MovieName; //default to Google search if (searchEngine.ToLower().Equals("bing")) url = BingSearch + MovieName; if (searchEngine.ToLower().Equals("ask")) url = AskSearch + MovieName; - string html = getUrlData(url); - ArrayList imdbUrls = matchAll(@".*?", html); + string html = GetUrlData(url); + ArrayList imdbUrls = MatchAll(@".*?", html); if (imdbUrls.Count > 0) return (string)imdbUrls[0]; //return first IMDb result else if (searchEngine.ToLower().Equals("google")) //if Google search fails - return getIMDbUrl(MovieName, "bing"); //search using Bing + return GetIMDbUrl(MovieName, "bing"); //search using Bing else if (searchEngine.ToLower().Equals("bing")) //if Bing search fails - return getIMDbUrl(MovieName, "ask"); //search using Ask + return GetIMDbUrl(MovieName, "ask"); //search using Ask else //search fails return string.Empty; } //Parse IMDb page data - private static void parseIMDbPage(string imdbUrl, bool GetExtraInfo, ImdbMovie mov) + private static void ParseIMDbPage(string imdbUrl, bool GetExtraInfo, ImdbMovie mov) { - string html = getUrlData(imdbUrl + "combined"); + string html = GetUrlData(imdbUrl + "combined"); mov.Id = match(@"", html); if (!string.IsNullOrEmpty(mov.Id)) { - mov.status = true; + mov.Status = true; mov.Title = match(@"(IMDb \- )*(.*?) \(.*?", html, 2); mov.OriginalTitle = match(@"title-extra"">(.*?)<", html); mov.Year = match(@".*?\(.*?(\d{4}).*?\).*?", match(@"(.*?)", html)); mov.Rating = match(@"(\d.\d)/10", html); - mov.Genres = matchAll(@"(.*?)", match(@"Genre.?:(.*?)(|See more)", html)).Cast().ToList(); ; - mov.Directors = matchAll(@"(.*?)", match(@"Directed by(.*?)", html)); - mov.Writers = matchAll(@"(.*?)", match(@"Writing credits(.*?)", html)); - mov.Producers = matchAll(@"(.*?)", match(@"Produced by(.*?)", html)); - mov.Musicians = matchAll(@"(.*?)", match(@"Original Music by(.*?)", html)); - mov.Cinematographers = matchAll(@"(.*?)", match(@"Cinematography by(.*?)", html)); - mov.Editors = matchAll(@"(.*?)", match(@"Film Editing by(.*?)", html)); - mov.Cast = matchAll(@"(.*?)", match(@"

Cast

(.*?)", html)); + mov.Genres = MatchAll(@"(.*?)", match(@"Genre.?:(.*?)(|See more)", html)).Cast().ToList(); mov.Plot = match(@"Plot:.*?
(.*?)((.*?)", match(@"Plot Keywords:.*?
(.*?).*?
.*?(\d{1,2} (January|February|March|April|May|June|July|August|September|October|November|December) (19|20)\d{2})", html); - mov.Runtime = match(@"Runtime:
(\d{1,4}) min[\s]*.*?
", html); - mov.Top250 = match(@"Top 250: #(\d{1,3})<", html); - mov.Oscars = match(@"Won (\d+) Oscars?\.", html); - if (string.IsNullOrEmpty(mov.Oscars) && "Won Oscar.".Equals(match(@"(Won Oscar\.)", html))) mov.Oscars = "1"; - mov.Awards = match(@"(\d{1,4}) wins", html); - mov.Nominations = match(@"(\d{1,4}) nominations", html); - mov.Tagline = match(@"Tagline:.*?
(.*?)(:
Rated (G|PG|PG-13|PG-14|R|NC-17|X) ", html); - mov.Votes = match(@">(\d+,?\d*) votes<", html); - mov.Languages = matchAll(@"(.*?)", match(@"Language.?:(.*?)(
|>.?and )", html)); - mov.Countries = matchAll(@"(.*?)", match(@"Country:(.*?)(
|>.?and )", html)); + //mov.Directors = matchAll(@"(.*?)", match(@"Directed by(.*?)", html)); + //mov.Writers = matchAll(@"(.*?)", match(@"Writing credits(.*?)", html)); + //mov.Producers = matchAll(@"(.*?)", match(@"Produced by(.*?)", html)); + //mov.Musicians = matchAll(@"(.*?)", match(@"Original Music by(.*?)", html)); + //mov.Cinematographers = matchAll(@"(.*?)", match(@"Cinematography by(.*?)", html)); + //mov.Editors = matchAll(@"(.*?)", match(@"Film Editing by(.*?)", html)); + //mov.Cast = matchAll(@"(.*?)", match(@"

Cast

(.*?)", html)); + //mov.PlotKeywords = matchAll(@"(.*?)", match(@"Plot Keywords:.*?
(.*?).*?
.*?(\d{1,2} (January|February|March|April|May|June|July|August|September|October|November|December) (19|20)\d{2})", html); + //mov.Runtime = match(@"Runtime:
(\d{1,4}) min[\s]*.*?
", html); + //mov.Top250 = match(@"Top 250: #(\d{1,3})<", html); + //mov.Oscars = match(@"Won (\d+) Oscars?\.", html); + //if (string.IsNullOrEmpty(mov.Oscars) && "Won Oscar.".Equals(match(@"(Won Oscar\.)", html))) mov.Oscars = "1"; + //mov.Awards = match(@"(\d{1,4}) wins", html); + //mov.Nominations = match(@"(\d{1,4}) nominations", html); + //mov.Tagline = match(@"Tagline:.*?
(.*?)(:
Rated (G|PG|PG-13|PG-14|R|NC-17|X) ", html); + //mov.Votes = match(@">(\d+,?\d*) votes<", html); + //mov.Languages = matchAll(@"(.*?)", match(@"Language.?:(.*?)(
|>.?and )", html)); + //mov.Countries = matchAll(@"(.*?)", match(@"Country:(.*?)(
|>.?and )", html)); mov.Poster = match(@"
.*?", html); if (!string.IsNullOrEmpty(mov.Poster) && mov.Poster.IndexOf("media-imdb.com") > 0) { mov.Poster = Regex.Replace(mov.Poster, @"_V1.*?.jpg", "_V1._SY200.jpg"); - mov.PosterLarge = Regex.Replace(mov.Poster, @"_V1.*?.jpg", "_V1._SY500.jpg"); - mov.PosterFull = Regex.Replace(mov.Poster, @"_V1.*?.jpg", "_V1._SY0.jpg"); + //mov.PosterLarge = Regex.Replace(mov.Poster, @"_V1.*?.jpg", "_V1._SY500.jpg"); + //mov.PosterFull = Regex.Replace(mov.Poster, @"_V1.*?.jpg", "_V1._SY0.jpg"); } else { mov.Poster = string.Empty; - mov.PosterLarge = string.Empty; - mov.PosterFull = string.Empty; + //mov.PosterLarge = string.Empty; + //mov.PosterFull = string.Empty; } mov.ImdbURL = "http://www.imdb.com/title/" + mov.Id + "/"; if (GetExtraInfo) { - string plotHtml = getUrlData(imdbUrl + "plotsummary"); - mov.Storyline = match(@"

(.*?)(|

)", plotHtml); - getReleaseDatesAndAka(mov); - mov.MediaImages = getMediaImages(mov); + string plotHtml = GetUrlData(imdbUrl + "plotsummary"); + //mov.Storyline = match(@"

(.*?)(|

)", plotHtml); + GetReleaseDatesAndAka(mov); + //mov.MediaImages = getMediaImages(mov); //mov.RecommendedTitles = getRecommendedTitles(mov); } } } //Get all release dates and aka-s - private static void getReleaseDatesAndAka(ImdbMovie mov) + private static void GetReleaseDatesAndAka(ImdbMovie mov) { Dictionary release = new Dictionary(); - string releasehtml = getUrlData("http://www.imdb.com/title/" + mov.Id + "/releaseinfo"); - foreach (string r in matchAll(@"(.*?)", match(@"\n*?(.*?)
", releasehtml))) + string releasehtml = GetUrlData("http://www.imdb.com/title/" + mov.Id + "/releaseinfo"); + foreach (string r in MatchAll(@"(.*?)", match(@"\n*?(.*?)
", releasehtml))) { Match rd = new Regex(@"(.*?)\n*?.*?(.*?)", RegexOptions.Multiline).Match(r); release[StripHTML(rd.Groups[1].Value.Trim())] = StripHTML(rd.Groups[2].Value.Trim()); } - mov.ReleaseDates = release; + //mov.ReleaseDates = release; Dictionary aka = new Dictionary(); - ArrayList list = matchAll(@".*?(.*?)", match(@"\n*?(.*?)
", releasehtml)); + ArrayList list = MatchAll(@".*?(.*?)", match(@"\n*?(.*?)
", releasehtml)); foreach (string r in list) { Match rd = new Regex(@"\n*?.*?(.*?)\n*?.*?(.*?)", RegexOptions.Multiline).Match(r); @@ -163,15 +158,15 @@ namespace NadekoBot.Classes.IMDB } //Get all media images - private static ArrayList getMediaImages(ImdbMovie mov) + private static ArrayList GetMediaImages(ImdbMovie mov) { ArrayList list = new ArrayList(); string mediaurl = "http://www.imdb.com/title/" + mov.Id + "/mediaindex"; - string mediahtml = getUrlData(mediaurl); - int pagecount = matchAll(@"", match(@"(.*?)", mediahtml)).Count; + string mediahtml = GetUrlData(mediaurl); + int pagecount = MatchAll(@"", match(@"(.*?)", mediahtml)).Count; for (int p = 1; p <= pagecount + 1; p++) { - mediahtml = getUrlData(mediaurl + "?page=" + p); + mediahtml = GetUrlData(mediaurl + "?page=" + p); foreach (Match m in new Regex(@"src=""(.*?)""", RegexOptions.Multiline).Matches(match(@"
(.*?)
", mediahtml))) { String image = m.Groups[1].Value; @@ -181,12 +176,12 @@ namespace NadekoBot.Classes.IMDB return list; } //Get Recommended Titles - private static ArrayList getRecommendedTitles(ImdbMovie mov) + private static ArrayList GetRecommendedTitles(ImdbMovie mov) { ArrayList list = new ArrayList(); string recUrl = "http://www.imdb.com/widget/recommendations/_ajax/get_more_recs?specs=p13nsims%3A" + mov.Id; - string json = getUrlData(recUrl); - list = matchAll(@"title=\\""(.*?)\\""", json); + string json = GetUrlData(recUrl); + list = MatchAll(@"title=\\""(.*?)\\""", json); HashSet set = new HashSet(); foreach (String rec in list) set.Add(rec); return new ArrayList(set.ToList()); @@ -198,7 +193,7 @@ namespace NadekoBot.Classes.IMDB return new Regex(regex, RegexOptions.Multiline).Match(html).Groups[i].Value.Trim(); } //Match all instances and return as ArrayList - private static ArrayList matchAll(string regex, string html, int i = 1) + private static ArrayList MatchAll(string regex, string html, int i = 1) { ArrayList list = new ArrayList(); foreach (Match m in new Regex(regex, RegexOptions.Multiline).Matches(html)) @@ -211,7 +206,7 @@ namespace NadekoBot.Classes.IMDB return Regex.Replace(inputString, @"<.*?>", string.Empty); } //Get URL Data - private static string getUrlData(string url) + private static string GetUrlData(string url) { WebClient client = new WebClient(); Random r = new Random(); diff --git a/NadekoBot/Modules/Searches.cs b/NadekoBot/Modules/Searches.cs index fc4c8115..fe5b158d 100644 --- a/NadekoBot/Modules/Searches.cs +++ b/NadekoBot/Modules/Searches.cs @@ -93,11 +93,12 @@ $@"🌍 **Weather for** 【{obj["target"]}】 .Do(async e => { if (!(await SearchHelper.ValidateQuery(e.Channel, e.GetArg("query")))) return; + await e.Channel.SendIsTyping(); string result; try { var movie = ImdbScraper.ImdbScrape(e.GetArg("query"), true); - if (movie.status) result = movie.ToString(); + if (movie.Status) result = movie.ToString(); else result = "Failed to find that movie."; } catch