static async Task SaveVids(ChannelStored2 c, IReadOnlyCollection <VideoItem> vids, JsonlStore <VideoStored2> vidStore, DateTime?uploadedFrom,
ILogger log)
{
var updated = DateTime.UtcNow;
var vidsStored = vids.Select(v => new VideoStored2 {
VideoId = v.Id,
Title = v.Title,
Description = v.Description,
Duration = v.Duration,
Keywords = v.Keywords.ToList(),
Statistics = v.Statistics,
ChannelId = c.ChannelId,
ChannelTitle = c.ChannelTitle,
UploadDate = v.UploadDate,
AddedDate = v.AddedDate,
Updated = updated
}).ToList();
if (vidsStored.Count > 0)
{
await vidStore.Append(vidsStored, log);
}
var newVideos = vidsStored.Count(v => uploadedFrom == null || v.UploadDate > uploadedFrom);
log.Information("Collect - {Channel} - Recorded {VideoCount} videos. {NewCount} new, {UpdatedCount} updated",
c.ChannelTitle, vids.Count, newVideos, vids.Count - newVideos);
}
public static async Task Process(ISimpleFileStore store, YtWeb ytWeb, ILogger log)
{
var blobs = await store.List("rec_exports").SelectManyList();
//blobs = blobs.Where(b => b.Path == "rec_exports/Traffic source 2019-07-01_2019-08-01 David Pakman Show.zip").ToList();
var fileInfoRegex = new Regex("^Traffic source (?'from'\\d+-\\d+-\\d+)_(?'to'\\d+-\\d+-\\d+) (?'channel'[^.]+)", RegexOptions.Compiled);
var appendStore = new JsonlStore <TrafficSourceRow>(store, "rec_exports_processed", r => r.FileUpdated.FileSafeTimestamp(), log);
var md = await appendStore.LatestFile();
var latestModified = md?.Ts.ParseFileSafeTimestamp();
var newBlobs = latestModified != null
? blobs.Where(b => b.Modified > latestModified).ToList()
: blobs;
log.Information("Processing {NewExports}/{AllExports} exports", newBlobs.Count, blobs.Count);
foreach (var b in newBlobs)
{
log.Information("Processing {Path}", b.Path);
var m = fileInfoRegex.Match(b.Path.Name);
if (m.Groups.Count < 3)
{
throw new InvalidOperationException($"unable to parse export info from file name '{b.Path.Name}'");
}
var exportInfo = new {
Channel = m.Groups["channel"].Value,
From = m.Groups["from"].Value.ParseDate(),
To = m.Groups["to"].Value.ParseDate()
};
var stream = await store.Load(b.Path);
var zip = new ZipArchive(stream);
using var csvStream = new StreamReader(
zip.GetEntry("Table data.csv")?.Open() ?? throw new InvalidOperationException("expected export to have 'Table data.csv'"),
Encoding.UTF8);
using var csvReader = new CsvReader(csvStream, CsvExtensions.DefaultConfig);
var records = csvReader.GetRecords <TrafficSourceExportRow>().ToList();
var rows = (await records.BlockFunc(ToTrafficSourceRow, 4,
progressUpdate: p => log.Debug("Processing traffic sources for {Path}: {Rows}/{TotalRows}", b.Path, p.Completed, records.Count)))
.NotNull().ToList();
await appendStore.Append(rows, log);
log.Information("Completed processing traffic source exports for {Path}", b.Path);
async Task <TrafficSourceRow> ToTrafficSourceRow(TrafficSourceExportRow row)
{
var source = row.Source.Split(".");
if (source.Length != 2 || source[0] != "YT_RELATED")
{
return(null); // total at the top or otherwise. not interested
}
var videoId = source[1];
var fromVideo = await ytWeb.GetVideo(videoId, log);
return(new TrafficSourceRow {
ToChannelTitle = exportInfo.Channel,
From = exportInfo.From,
To = exportInfo.To,
Impressions = row.Impressions,
Source = row.Source,
AvgViewDuration = row.AvgViewDuration,
Views = row.Views,
SourceType = row.SourceType,
FromChannelId = fromVideo?.ChannelId,
FromChannelTitle = fromVideo?.ChannelTitle,
FromVideoId = fromVideo?.Id,
FromVideoTitle = fromVideo?.Title,
ImpressionClickThrough = row.ImpressionClickThrough,
WatchTimeHrsTotal = row.WatchTimeHrsTotal,
FileUpdated = b.Modified?.UtcDateTime ?? DateTime.MinValue
});
}
}
log.Information("Completed processing traffic source exports");
}