/// <summary>
/// This method basically does what it can to possibly sanitise the input.
/// Tries to match a program, extract a series and episode number, extract attributes like S for subtitles etc.
/// It should be specific to each feed probably. The regexes are quite specific thouogh.
/// </summary>
/// <param name="tag"></param>
/// <param name="handler"></param>
/// <param name="channelUriLookup"></param>
/// <param name="results"></param>
private void AddPrograms(TvProgramBroadCastListingImport tag, AddTvBroadCastListingInfoHandler handler, Dictionary<string, TvChannelListingSource> channelUriLookup, tvProgramme[] results)
{
foreach (var program in results) {
// the xmltv sometimes have missing end times but these are recognised by start == stop
var start = BlebXmltvProvider.ParseDateStringToUtc(program.start);
DateTime? end = BlebXmltvProvider.ParseDateStringToUtc(program.stop);
// we use a null in the DB if we don't know the end time
if (start == end) {
end = null;
}
// try and clean up the program titles etc.
var programTitle = program.title.Value;
var episodeTitle = program.subtitle?.Value;
var episodeDescription = program.desc?.Value;
// and see if we can extract any of this metadata
int? seriesNumber = null;
int? episodeNumber = null;
BroadCastListingAttributes attribs = 0;
// try and clean up the program titles etc.
if (episodeDescription != null) {
// the bleb import feed titles sometimes run into the descriptions using ...s
if (programTitle.EndsWith("...") && episodeDescription.StartsWith("...")) {
// find when the runon ends, ignoring the start ...
var endOfRunOn = episodeDescription.IndexOf('.', 3);
if (endOfRunOn != -1) {
// trim out the ...s and append the episode description
programTitle = programTitle.Substring(0, programTitle.Length - 3) + ' ' + episodeDescription.Substring(3, endOfRunOn - 3);
// then trim off title from the start of the description, and the full stop and any spaces
episodeDescription = episodeDescription.Substring(endOfRunOn + 1).Trim(' ');
}
}
// do a bit of basic sanitisation (necessary for the runons above! but shouldn't be harmful here)
programTitle = programTitle.Trim(' ', '.');
// AlsoInHdRegex
try {
var match = AlsoInHdRegex.Match(episodeDescription);
if (match.Success) {
attribs = attribs | BroadCastListingAttributes.AvailableInHD;
episodeDescription = match.Groups["pre"].Value + match.Groups["post"].Value;
}
}
catch { }
// BBCAttribsRegex
try {
// then try and pull out BBC style attribs like HD, subtitles, audio descriptions like [S, AD]
var match = BBCAttribsRegex.Match(episodeDescription);
if (match.Success) {
if (match.Groups["attribs"].Success) {
var strings = match.Groups["attribs"].Value.Split(new[] { ", " }, StringSplitOptions.RemoveEmptyEntries);
if (strings.Contains("S")) attribs = attribs | BroadCastListingAttributes.Subtitles;
if (strings.Contains("AD")) attribs = attribs | BroadCastListingAttributes.AudioDescription;
}
episodeDescription = match.Groups["pre"].Value + match.Groups["post"].Value;
}
}
catch { }
try {
// then try and pull out some episode information with the N/N format
var match = BBCEpisodeRegex.Match(episodeDescription);
if (match.Success) {
episodeNumber = int.Parse(match.Groups["episode"].Value);
episodeDescription = match.Groups["post"].Value;
}
else {
// and then try SN, EN format
match = OtherEpisodeRegex.Match(episodeDescription);
if (match.Success) {
seriesNumber = int.Parse(match.Groups["series"].Value);
episodeNumber = int.Parse(match.Groups["episode"].Value);
episodeDescription = match.Groups["pre"].Value + match.Groups["post"].Value;
}
}
}
// ignore any metadata that fails extraction but still save the rest
catch (ArgumentNullException) { }
catch (FormatException) { }
}
// and remove any double spaces we might have introduced
episodeDescription = episodeDescription?.Replace(" ", " ").Trim();
try {
handler(
tag,
channelUriLookup[program.channel],
new AddBroadcastListingInfo {
StartAt = start,
EndAt = end,
ProgramTitle = programTitle,
EpisodeTitle = episodeTitle,
EpisodeDescription = episodeDescription,
SeriesNumber = seriesNumber,
EpisodeNumber = episodeNumber,
Attributes = (int)attribs
}
);
}
catch (TvBroadCastListingAlreadyExistsException) {
// the feed just violated the PK for this entry, carry on with the next one
continue;
}
}
}
/// <summary>
/// Performs an update for the specified channels for the specified days!
/// </summary>
/// <param name="tag"></param>
/// <param name="channels"></param>
/// <param name="days"></param>
/// <param name="handler"></param>
/// <returns></returns>
public void ImportPrograms(TvProgramBroadCastListingImport tag, IEnumerable<TvChannelListingSource> channels, IEnumerable<int> days, AddTvBroadCastListingInfoHandler handler)
{
var importer = new BlebXmltvProvider();
// get a list of the source names to pass to the importer
var blebChannelNames = channels.Select(x => x.SourceChannelName);
var results = importer.GetFeedResults(blebChannelNames, days).Result;
// create a dictionary of channel URI -> TvChannel for the importer to find the TvChannelId from the URI
var channelUriLookup = channels.ToDictionary(x => x.SourceChannelUri, x => x);
// add each program
if (results.programme != null) {
AddPrograms(tag, handler, channelUriLookup, results.programme);
}
}
