public IEnumerable <WebSearchResult> Search(string text)
{
string sql =
"SELECT id, title, actors, year, genres, alternate_titles " +
"FROM movie_info " +
"WHERE title LIKE @search OR alternate_titles LIKE @search OR actors LIKE @search";
return(ReadList <WebSearchResult>(sql, delegate(SQLiteDataReader reader)
{
var result = new WebSearchResult()
{
Type = WebMediaType.Movie,
Id = reader.ReadIntAsString(0),
Title = reader.ReadString(1),
Details = new WebDictionary <string>()
{
{ "Year", reader.ReadIntAsString(3) },
{ "Genres", String.Join(", ", reader.ReadPipeList(4)) },
}
};
// simple score
int score = result.Title.Contains(text, false) ? (int)Math.Round(40 + (decimal)text.Length / result.Title.Length * 40) : 0;
var validAlternate = reader.ReadPipeList(5).Where(x => x.Contains(text, false));
score = Math.Max(score, validAlternate.Count() > 0 ? validAlternate.Max(x => 40 + (int)Math.Round((decimal)text.Length / x.Length * 30)) : 0);
// actors
var valid = reader.ReadPipeList(2).Where(x => x.Contains(text, false));
score = Math.Max(score, valid.Count() > 0 ? valid.Max(x => 40 + (int)Math.Round((decimal)text.Length / x.Length * 30)) : 0);
// set
result.Score = score;
return result;
}, new SQLiteParameter("@search", "%" + text + "%")));
}
public SearchResultsViewModel(WebSearchResult res, string url)
{
Score = res.Score;
URL = url;
Type = (SearchResultType)res.Type;
Title = GetTitle(res);
}
private string CreateLink(WebSearchResult result)
{
switch (result.Type)
{
case WebMediaType.Movie:
return(Url.Action("Details", "MovieLibrary", new { movie = result.Id }));
case WebMediaType.MusicAlbum:
return(Url.Action("Album", "MusicLibrary", new { album = result.Id }));
case WebMediaType.MusicArtist:
return(Url.Action("Albums", "MusicLibrary", new { artist = result.Id }));
case WebMediaType.MusicTrack:
return(Url.Action("Album", "MusicLibrary", new { album = result.Details["AlbumId"], track = result.Id }));
case WebMediaType.TVEpisode:
return(Url.Action("Details", "TVShowsLibrary", new { episode = result.Id }));
case WebMediaType.TVSeason:
return(Url.Action("Episodes", "TVShowsLibrary", new { season = result.Id }));
case WebMediaType.TVShow:
return(Url.Action("Seasons", "TVShowsLibrary", new { show = result.Id }));
default:
return(null);
}
}
public async Task <IActionResult> Search(string query)
{
var searchResults = new WebSearchResult
{
Query = query.Trim(),
};
await Task.Run(() =>
{
var results = EngineOptions.SearchEngine.Search(query);
foreach (var result in results)
{
var urlFile = EngineOptions.UrlFileStore.GetUrlFile(result.UrlFileId);
var searchResultItem = new SearchResultItem
{
Title = urlFile.Title,
Url = urlFile.Url,
PublishDate = DateTime.FromBinary((long)urlFile.PublishDate),
Score = result.Score.Value,
ProScore = result.ProScore.Value,
ScoreDebugInfo = result.Score.ToString(),
ProScoreDebugInfo = result.ProScore.ToString(),
};
if (result.WordPositions == null)
{
searchResultItem.Details = urlFile.TextContent
.Substring(0, Math.Min(50, urlFile.TextContent.Length)).Replace("\r", "").Replace("\n", " ");
}
else
{
var orderPos = result.WordPositions.OrderBy(o => o.Position);
var minWordPos = orderPos.First();
var minPos = Math.Max((int)minWordPos.Position - 50, 0);
var maxPos = orderPos.Last();
var content = urlFile.TextContent;
searchResultItem.Details = content.Substring((int)minPos,
Math.Min((int)maxPos.Position - minPos + maxPos.Word.Length + 50, content.Length - minPos))
.Replace("\r", "").Replace("\n", " ");
}
searchResults.Items.Add(searchResultItem);
}
});
return(View("Index", searchResults));
}
private static IEnumerable <WebSearchResult> ParseResults(string data)
{
/*
* <div class="prew-film">
* <div style="padding: 0px 0px 0px 30px;">
* <div class="prew-film-content">
* <div class="box4"><a href="http://dcshara.ru/uzhasy/14831118-pod-kupolom-under-the-dome-1-sezon.html" class="tip-bottom item" ><img width="155" height="228" src="http://dcshara.ru/uploads/posts/2013-06/1372606952_1372224699_kupol.jpg" alt="Под куполом - Under the Dome (1 сезон)" />
* <span class="tt opacity"><h2>Под куполом - Under the Dome (1 сезон)</h2></span></a></div>
* </div>
* <div class="prew-film-title">
* <div style="clear:both; height:10px;"></div>
* <div align="center" class="date" style="height:30px; overflow:hidden;"><a href="http://dcshara.ru/uzhasy/">ужасы</a>, <a href="http://dcshara.ru/fantastika/">фантастика</a>, <a href="http://dcshara.ru/serial/">сериал</a></div>
* <div class="bt2"><a href="http://dcshara.ru/uzhasy/14831118-pod-kupolom-under-the-dome-1-sezon.html">СКАЧАТЬ</a></div>
* <div><a href="#" title=""><img width="70" height="18" src="/templates/Cinema/images/smt.jpg" alt="" /></a></div>
* </div>
* </div>
* </div>
*/
int startIndex = 0;
while ((startIndex = data.IndexOf("class=\"box4\"", startIndex)) != -1)
{
// class="box4"><a href="http://dcshara.ru/uzhasy/14831118-pod-kupolom-under-the-dome-1-sezon.html" class="tip-bottom item" ><img width="155" height="228" src="http://dcshara.ru/uploads/posts/2013-06/1372606952_1372224699_kupol.jpg" alt="Под куполом - Under the Dome (1 сезон)"
WebSearchResult res = new WebSearchResult();
var linkStartInd = data.IndexOf("http://", startIndex);
var linkEndInd = data.IndexOf("\" class", linkStartInd);
res.ReleaseUrl = data.Substring(linkStartInd, linkEndInd - linkStartInd);
var posterStartInd = data.IndexOf("http://", linkEndInd);
var posterEndInd = data.IndexOf("\" alt=", posterStartInd);
res.PosterUrl = data.Substring(posterStartInd, posterEndInd - posterStartInd);
var nameStartInd = data.IndexOf("alt=\"", posterEndInd) + 5;
var nameEndInd = data.IndexOf("\" />", nameStartInd);
res.Name = data.Substring(nameStartInd, nameEndInd - nameStartInd);
yield return(res);
startIndex++;
}
}
private string GetTitle(WebSearchResult result)
{
switch (result.Type)
{
case WebMediaType.MusicAlbum:
return(String.Format(FormStrings.SearchResultAlbum, result.Title, result.Details["Artist"]));
case WebMediaType.MusicTrack:
return(String.Format(FormStrings.SearchResultTrack, result.Title, result.Details["Artist"]));
case WebMediaType.TVEpisode:
return(String.Format("{0} ({1} {2}x{3})", result.Title, result.Details["ShowName"], result.Details["SeasonNumber"], result.Details["EpisodeNumber"]));
default:
return(result.Title);
}
}
public GridRow(WebSearchResult webSearchResult, int count) : this()
{
HeaderCell.Value = $"{count + 1}";
Cells.Add(new DataGridViewTextBoxCell());
Cells[0].Style.BackColor = webSearchResult.IsVIPad ? Color.DarkOrange : Color.White;
DataGridViewTextBoxCell dateCell = new DataGridViewTextBoxCell()
{
Value = webSearchResult.Date.ToString(CultureInfo.CurrentUICulture)
};
Cells.Add(dateCell);
dateCell.Style.BackColor = webSearchResult.IsHighlighted ? Color.Lavender : Color.White;
if (webSearchResult.ImageBinary == null && webSearchResult.ImageBase64 != null)
{
webSearchResult.ImageBinary = ImageUtil.Base64StringToImage(webSearchResult.ImageBase64);
}
Cells.Add(new DataGridViewImageCell()
{
Value = (Image)webSearchResult.ImageBinary
});
DataGridViewLinkCell linkCell = new DataGridViewLinkCell
{
Value = webSearchResult.Title.Text,
Tag = webSearchResult.Title.Url
};
Cells.Add(linkCell);
Cells.Add(new DataGridViewTextBoxCell()
{
Value = $"{webSearchResult.Price:C}"
});
Cells.Add(new DataGridViewTextBoxCell()
{
Value = webSearchResult.Category
});
Cells.Add(new DataGridViewTextBoxCell()
{
Value = webSearchResult.Address
});
}
public IEnumerable <WebSearchResult> Search(string text)
{
using (DatabaseConnection connection = OpenConnection())
{
IEnumerable <WebSearchResult> results;
SQLiteParameter param = new SQLiteParameter("@search", "%" + text + "%");
// search for shows
string showSql = "SELECT ID, Pretty_Name FROM online_series WHERE Pretty_Name LIKE @search";
IEnumerable <WebSearchResult> shows = ReadList <WebSearchResult>(showSql, delegate(SQLiteDataReader reader)
{
string title = reader.ReadString(1);
return(new WebSearchResult()
{
Type = WebMediaType.TVShow,
Id = reader.ReadIntAsString(0),
Title = title,
Score = 40 + (int)Math.Round((decimal)text.Length / title.Length * 40),
});
}, param);
results = shows;
string actorShowSql = "SELECT ID, Pretty_Name, Actors FROM online_series WHERE Actors LIKE @search";
IEnumerable <WebSearchResult> actorShows = ReadList <WebSearchResult>(actorShowSql, delegate(SQLiteDataReader reader)
{
var valid = reader.ReadPipeList(2).Where(x => x.IndexOf(text, StringComparison.CurrentCultureIgnoreCase) >= 0); // contains is case sensitive
if (valid.Count() == 0)
{
return(null);
}
return(new WebSearchResult()
{
Type = WebMediaType.TVShow,
Id = reader.ReadIntAsString(0),
Title = reader.ReadString(1),
Score = valid.Max(x => 40 + (int)Math.Round((decimal)text.Length / x.Length * 30))
});
}, param);
results = results.Concat(actorShows);
#region Episodes
// initialize
List <SQLiteParameter> parameters = new List <SQLiteParameter>()
{
param
};
// search for episodes
Regex episodeRegex = new Regex(@"^(.*) ([0-9]{1,3})x([0-9]{1,3})$");
Match episodeResult = episodeRegex.Match(text);
string episodeRegexWhere = "0";
if (episodeResult.Success)
{
episodeRegexWhere = "((os.Pretty_Name LIKE @show OR ls.Parsed_Name LIKE @show) AND e.SeasonIndex = @season AND e.EpisodeIndex = @episode)";
parameters.Add(new SQLiteParameter("@show", "%" + episodeResult.Groups[1].Value.Trim() + "%"));
parameters.Add(new SQLiteParameter("@season", episodeResult.Groups[2].Value));
parameters.Add(new SQLiteParameter("@episode", episodeResult.Groups[3].Value));
}
// execute
string episodeSql =
"SELECT e.EpisodeID, e.EpisodeName, e.EpisodeIndex, e.SeasonIndex, e.SeriesID, e.GuestStars, os.Pretty_Name, MIN(ls.Parsed_Name) AS Parsed_Name " +
"FROM online_episodes e " +
"INNER JOIN online_series os ON os.ID = e.SeriesID " +
"INNER JOIN local_series ls ON ls.ID = e.SeriesID " +
"WHERE e.EpisodeName LIKE @search OR e.GuestStars LIKE @search OR " + episodeRegexWhere + " " +
"GROUP BY e.EpisodeID, e.EpisodeName, e.EpisodeIndex, e.SeasonIndex, e.SeriesID, e.GuestStars, os.Pretty_Name ";
IEnumerable <WebSearchResult> episodes = ReadList <WebSearchResult>(episodeSql, delegate(SQLiteDataReader reader)
{
// read the data
string title = reader.ReadString(1);
string showTitle = reader.ReadString(6);
WebSearchResult result = new WebSearchResult()
{
Type = WebMediaType.TVEpisode,
Id = reader.ReadIntAsString(0),
Title = title,
Details = new WebDictionary <string>()
{
{ "EpisodeNumber", reader.ReadIntAsString(2) },
{ "SeasonNumber", reader.ReadIntAsString(3) },
{ "ShowId", reader.ReadIntAsString(4) },
{ "ShowName", !String.IsNullOrEmpty(showTitle) ? showTitle : reader.ReadString(7) }
}
};
// title
int score = title.Contains(text, false) ? 40 + (int)Math.Round((decimal)text.Length / title.Length * 40) : 0;
// guest stars
var valid = reader.ReadPipeList(5).Where(x => x.Contains(text, false)); // .Contains() is case sensitive
score = Math.Max(score, valid.Count() > 0 ? valid.Max(x => 20 + (int)Math.Round((decimal)text.Length / x.Length * 40)) : 0);
// regex match
if (episodeResult.Success &&
(result.Details["ShowName"].Contains(episodeResult.Groups[1].Value.Trim(), false) &&
result.Details["SeasonNumber"] == episodeResult.Groups[2].Value &&
result.Details["EpisodeNumber"] == episodeResult.Groups[3].Value))
{
score = 100;
}
// set score and return
result.Score = score;
return(result);
}, parameters.ToArray());
// when there are multiple matches with 100%, set them all to 95% as we apparantly aren't sure
if (episodes.Count(x => x.Score == 100) > 1)
{
episodes = episodes.Select(x =>
{
if (x.Score == 100)
{
x.Score = 95;
}
return(x);
});
}
results = results.Concat(episodes);
#endregion
// fancy season search: <showname> s<season>
Regex seasonRegex = new Regex(@"^(.*) s([0-9]{1,3})$");
Match seasonResult = seasonRegex.Match(text);
if (seasonResult.Success)
{
string sql =
"SELECT DISTINCT s.ID, s.SeasonIndex, s.SeriesID, o.Pretty_Name, MIN(l.Parsed_Name) AS Parsed_Name " +
"FROM season s " +
"INNER JOIN online_series o ON s.SeriesID = o.ID " +
"INNER JOIN local_series l ON s.SeriesID = l.ID " +
"WHERE s.HasLocalFiles = 1 AND (o.Pretty_Name = @show OR l.Parsed_Name = @show) AND s.SeasonIndex = @season " +
"GROUP BY s.ID, s.SeasonIndex, s.SeriesID, o.Pretty_Name ";
results = ReadList <WebSearchResult>(sql, delegate(SQLiteDataReader reader)
{
string showTitle = !String.IsNullOrEmpty(reader.ReadString(3)) ? reader.ReadString(3) : reader.ReadString(4);
return(new WebSearchResult()
{
Type = WebMediaType.TVSeason,
Id = reader.ReadString(0),
Title = showTitle + " (season " + reader.ReadInt32(1) + ")",
Score = 100,
Details = new WebDictionary <string>()
{
{ "SeasonNumber", reader.ReadIntAsString(1) },
{ "ShowId", reader.ReadIntAsString(2) },
{ "ShowName", showTitle }
}
});
}, new SQLiteParameter("@show", seasonResult.Groups[1].Value.Trim()),
new SQLiteParameter("@season", seasonResult.Groups[2].Value))
.Concat(results);
}
// return
return(results);
}
}
public IEnumerable <WebSearchResult> Search(string text)
{
using (DatabaseConnection connection = OpenConnection())
{
SQLiteParameter param = new SQLiteParameter("@search", "%" + text + "%");
string artistSql = "SELECT DISTINCT strArtist, strAlbumArtist, strAlbum FROM tracks WHERE strArtist LIKE @search";
IEnumerable <WebSearchResult> artists = ReadList <IEnumerable <WebSearchResult> >(artistSql, delegate(SQLiteDataReader reader)
{
IEnumerable <string> albumArtists = reader.ReadPipeList(1);
return(reader.ReadPipeList(0)
.Where(name => name.Contains(text))
.Select(name => new WebSearchResult()
{
Type = WebMediaType.MusicArtist,
Id = name,
Title = name,
Score = (int)Math.Round(40 + (decimal)text.Length / name.Length * 40)
})
.Concat(new[] { new WebSearchResult()
{
Type = WebMediaType.MusicAlbum,
Id = (string)AlbumIdReader(reader, 2),
Title = reader.ReadString(2),
Score = (int)Math.Round(20 + (decimal)text.Length / reader.ReadString(0).Length * 40),
Details = new WebDictionary <string>()
{
{ "Artist", albumArtists.First() },
{ "ArtistId", albumArtists.First() }
}
} }));
}, param).SelectMany(x => x);
string songSql = "SELECT DISTINCT idTrack, strTitle, strAlbumArtist, strAlbum, iDuration, iYear FROM tracks WHERE strTitle LIKE @search";
IEnumerable <WebSearchResult> songs = ReadList <WebSearchResult>(songSql, delegate(SQLiteDataReader reader)
{
IEnumerable <string> allArtists = reader.ReadPipeList(2);
string title = reader.ReadString(1);
string artist = allArtists.Count() == 0 ? String.Empty : allArtists.First();
return(new WebSearchResult()
{
Type = WebMediaType.MusicTrack,
Id = reader.ReadIntAsString(0),
Title = title,
Score = (int)Math.Round(40 + (decimal)text.Length / title.Length * 40),
Details = new WebDictionary <string>()
{
{ "Artist", artist },
{ "ArtistId", artist },
{ "Album", reader.ReadString(3) },
{ "AlbumId", (string)AlbumIdReader(reader, 3) },
{ "Duration", reader.ReadIntAsString(4) },
{ "Year", reader.ReadIntAsString(5) }
}
});
}, param);
string albumsSql =
"SELECT DISTINCT t.strAlbumArtist, t.strAlbum, " +
"CASE WHEN i.iYear ISNULL THEN MIN(t.iYear) ELSE i.iYear END AS year " +
"FROM tracks t " +
"LEFT JOIN albuminfo i ON t.strAlbum = i.strAlbum AND t.strArtist LIKE '%' || i.strArtist || '%' " +
"WHERE t.strAlbum LIKE @search " +
"GROUP BY t.strAlbumArtist, t.strAlbum ";
IEnumerable <WebSearchResult> albums = ReadList <IEnumerable <WebSearchResult> >(albumsSql, delegate(SQLiteDataReader reader)
{
string title = reader.ReadString(1);
IEnumerable <string> artistList = reader.ReadPipeList(0);
var albumResult = new WebSearchResult()
{
Type = WebMediaType.MusicAlbum,
Id = (string)AlbumIdReader(reader, 1),
Title = title,
Score = (int)Math.Round(40 + (decimal)text.Length / title.Length * 40),
Details = new WebDictionary <string>()
{
{ "Artist", artistList.First().Trim() },
{ "ArtistId", artistList.First().Trim() },
{ "Year", reader.ReadIntAsString(2) }
}
};
string allArtists = String.Join("", artistList);
return(artistList
.Select(name => new WebSearchResult()
{
Type = WebMediaType.MusicArtist,
Id = name,
Title = name,
Score = (int)Math.Round((decimal)name.Length / allArtists.Length * 30)
}).Concat(new[] { albumResult }));
}, param).SelectMany(x => x);
return(artists.Concat(songs).Concat(albums));
}
}
public List <WebSearchResult> GetResults(Filter filter)
{
List <WebSearchResult> results = new List <WebSearchResult>();
HtmlNode node = new HtmlToHtmlNode(_content.Value);
if (node == null)
{
Log.Error("Can't parse content. Invalid HtmlNode.");
return(results);
}
// <div class="snippet-date-info" data-marker="item-date" data-shape="default" data-tooltip="27 августа 14:12" flow="down">
// 2 дня назад
// </div>
HtmlNodeCollection dates = node.SelectNodes("//div[@data-marker='item-date']");
if (dates == null)
{
Log.Error("Can't parse dates. Unexpected html structure.");
return(results);
}
DateTime prevDate = DateTime.Today;
foreach (HtmlNode dateNode in dates)
{
// <div class="description item_table-description">
// <div class="snippet-title-row">...
HtmlNode entryNode = dateNode.ParentNode.ParentNode.ParentNode;
HtmlNode titleANode = entryNode.SelectSingleNode(".//div[contains(@class,'snippet-title-row')]/h3/a");
WebLink titleValue = GetTitleValue(titleANode);
string title = new LowerText(titleValue.Text);
bool @break = false;
foreach (string contains in filter.TitleContains)
{
if (!title.Contains(contains))
{
@break = true;
break;
}
}
if (@break)
{
continue;
}
foreach (string contains in filter.TitleDoesNotContain)
{
if (title.Contains(contains))
{
@break = true;
break;
}
}
if (@break)
{
continue;
}
DateTime?dateValue = AvitoParser.GetDateValue(dateNode.InnerText, prevDate, DateTime.Now);
if (dateValue == null || dateValue < filter.DatesAfter)
{
continue;
}
// <div class="snippet-price-row"><span itemprop="offers" itemtype="http://schema.org/Offer" itemscope="" class="snippet-price " data-shape="default" data-marker="item-price"><meta itemprop="priceCurrency" content="RUB"><meta itemprop="price" content="6490"><meta itemprop="availability" content="https://schema.org/LimitedAvailability">
// 6 490 ₽
HtmlNode priceNode = entryNode.SelectSingleNode(".//span[@data-marker='item-price']");
int? priceValue = AvitoParser.GetPriceValue(priceNode?.InnerText);
if (priceValue != null)
{
if (priceValue < filter.MinPrice || priceValue > filter.MaxPrice)
{
continue;
}
}
bool isVIPad = priceNode.ParentNode.ParentNode.GetAttributeValue("class", "") == "options" && priceNode.ParentNode.ParentNode.GetAttributeValue("itemprop", "") == "offers";
// <div class="data">
// <p>Часы и украшения</p>
HtmlNode categoryNode = entryNode.SelectSingleNode(".//div[@class='data']/p");
// <span class="item-address-georeferences-item__content">Комендантский проспект</span>
HtmlNode addressNode = entryNode.SelectSingleNode(".//span[@class='item-address-georeferences-item__content']");
// <div class="item-photo" data-marker="item-photo">
HtmlNode imgNode = entryNode.ParentNode.ParentNode.SelectSingleNode(".//div[@data-marker='item-photo']");
imgNode = imgNode?.SelectSingleNode(".//img");
if (isVIPad)
{
if (imgNode == null)
{
imgNode = entryNode.ParentNode.ParentNode.SelectSingleNode(".//div[@class='img-container']");
imgNode = imgNode?.SelectSingleNode(".//img");
}
}
WebSearchResult result = new WebSearchResult();
result.Title = titleValue;
result.IsVIPad = isVIPad;
result.Date = dateValue.Value;
prevDate = result.Date.Date;
result.Category = AvitoParser.GetCategoryValue(categoryNode?.InnerText);
result.Address = AvitoParser.GetAddressValue(addressNode?.InnerText);
result.ImageUrl = GetImageUrl(imgNode);
result.Price = priceValue;
results.Add(result);
}
return(results);
}
