/// <summary>
/// localise every x.html file that has been marked WebPages.Localise=2
/// </summary>
/// <param name="repo">
/// IRepository to perform database work
/// </param>
/// <param name="batchSize">
/// number of files in request from db
/// </param>
/// <param name="localise">
/// object that actually performs the localise (find+alter each link)
/// </param>
/// <returns>
/// Task although activity is heavily CPU-bound and HAP methods all sync, there is some database I/O conducted async
/// </returns>
/// <remarks>
/// 1. batchSize is set by caller [from App.config
/// </remarks>
async Task HtmlLocalise(IRepository repo, int batchSize, Localiser localise, bool getMissing = false)
{
List <WebPage> batch;
batch = await repo.GetWebPagesToLocaliseAsync(batchSize); // get first batch (as List<WebPage>)
//batch = new List<WebPage>(dbctx.WebPages.Where(w => w.Url == "https://www.ligonier.org/learn/scripture/").ToList());
while (batch.Count > 0)
{
foreach (var webpage in batch) // iterate through [re-]obtained List
{
if (webpage.Download != WebPage.DownloadEnum.Downloaded || // this page already fully downloaded ?
webpage.Localise != WebPage.LocaliseEnum.ToLocalise) // and needs localisation ?
{
continue; // no. [sproc should not have included it]/ TODO: make Debug.Assert instead
}
var htmlFile = webpage.Filespec;
var backupFile = backupdir + Path.DirectorySeparatorChar + Path.GetFileName(htmlFile);
Console.WriteLine($"<<<{webpage.Url}\t~~>\t{htmlFile }>>>");
try
{
var changedLinks = await localise.Translate(webpage, MaxLinks, getMissing); // [async because of Downloader] but complete current page before starting the next
webpage.Localise = (changedLinks)
? WebPage.LocaliseEnum.Localised // show Localise success
: WebPage.LocaliseEnum.Ignore; // pretend it wasn't wanted anyway
}
catch (Exception excp) // either explicit from FetchFileAsync or HTTP timeout [TODO: Polly retries]
{
Console.WriteLine($"HtmlLocalise EXCEPTION\t{excp.Message}"); // see Filespec like '~%'
webpage.Localise = WebPage.LocaliseEnum.Ignore; // pretend it wasn't wanted anyway
}
}
var finalcnt = repo.SaveChanges(); // flush to update any pending "webpage.Localise = Ignore/Localised" rows
batch = await repo.GetWebPagesToLocaliseAsync(batchSize); // get next batch
}
}
static int MaxFileSize; // don't download files bigger than 10 MB
static async Task Main(string[] _)
{
//string fs1 = @"C:\Ligonier\webcache\state - theology - does - sin - deserve - damnation.html",
// fs2 = @"C:\Ligonier\webcache\assets\bible - plan.pdf";
//var rel = Utils.GetRelativePath(fs1, fs2);
//Console.WriteLine(rel);
dbctx = new WebModel(); // EF context defaults to config: "name=DefaultConnection"
IAsyncPolicy AdoRetryPolicy = // TODO: probably should configure based on App.config
Policy.Handle <Exception>(ex => true) // retry every exception! TODO: improve
.WaitAndRetryAsync(5, retryAttempt => TimeSpan.FromSeconds(Math.Pow(2, retryAttempt) / 4)); // i.e. 0.5, 1, 2, 4, 8 second retries
//IRepository repo = new Repository(dbctx);
IRepository repo = new BulkRepository(dbctx, AdoRetryPolicy);
MimeCollection.Load(await repo.GetContentTypeToExtnsAsync());
//var ct = new CancellationToken();
htmldir = ConfigurationManager.AppSettings["htmldir"] ?? @"C:\Ligonier\webcache";
if (!Directory.Exists(htmldir))
{
Directory.CreateDirectory(htmldir);
}
var otherdir = ConfigurationManager.AppSettings["otherdir"] ?? (htmldir + Path.DirectorySeparatorChar + OTHFOLDER);
if (!Directory.Exists(otherdir))
{
Directory.CreateDirectory(otherdir);
}
backupdir = ConfigurationManager.AppSettings["backupdir"] ?? (htmldir + Path.DirectorySeparatorChar + BACKUPFOLDER);
if (!Directory.Exists(backupdir))
{
Directory.CreateDirectory(backupdir);
}
if (!int.TryParse(ConfigurationManager.AppSettings["batchsize"], out var batchSize))
{
batchSize = 4;
}
if (!int.TryParse(ConfigurationManager.AppSettings["maxlinks"], out MaxLinks))
{
MaxLinks = 1500;
}
if (!int.TryParse(ConfigurationManager.AppSettings["maxfilesize"], out MaxFileSize))
{
MaxFileSize = 10_000_000; // 10 MB
}
var ValidRetry = new HttpStatusCode[] {
HttpStatusCode.Ambiguous, // 300
HttpStatusCode.Conflict, // 409
HttpStatusCode.InternalServerError, // 500
HttpStatusCode.NotImplemented, // 501
HttpStatusCode.BadGateway, // 502
HttpStatusCode.ServiceUnavailable, // 503
HttpStatusCode.GatewayTimeout
}; // 504
IAsyncPolicy <HttpResponseMessage> HttpRetryPolicy = // TODO: probably should configure based on App.config
Policy.HandleResult <HttpResponseMessage>(rsp => ValidRetry.Contains(rsp.StatusCode))
.WaitAndRetryAsync(0, retryAttempt => TimeSpan.FromSeconds(Math.Pow(2, retryAttempt) / 2)); // i.e. 1, 2, 4 seconds
#pragma warning disable GCop302 // Since '{0}' implements IDisposable, wrap it in a using() statement
//TODO: plug-in Polly as MessageProcessingHandler / whatever !
var Client = new HttpClient(
new HttpClientHandler {
AutomaticDecompression = DecompressionMethods.GZip | DecompressionMethods.Deflate, AllowAutoRedirect = true
})
{
Timeout = new TimeSpan(0, 0, 20)
};
#pragma warning restore GCop302 // Since '{0}' implements IDisposable, wrap it in a using() statement
var p = new Program();
var retrycount = 2;
Downloader download;
do
{
HParser = new HapParser(MaxLinks);
download = new Downloader(repo, Client, HttpRetryPolicy, HParser, htmldir, otherdir, backupdir, MaxFileSize);
var dlresult = await p.DownloadAndParse(repo, batchSize, download);
if (!dlresult) // failure may be due to tainted EF context so have to reset all these
{
dbctx = new WebModel(); // EF context defaults to config: "name=DefaultConnection"
repo = new BulkRepository(dbctx, AdoRetryPolicy);
retrycount--;
}
else
{
break;
}
} while (retrycount >= 0);
Console.WriteLine("*** DownloadAndParse FINISHED ***");
var localise = new Localiser(HParser, htmldir, backupdir, download);
await p.HtmlLocalise(repo, batchSize, localise, getMissing : true);
Console.WriteLine("*** HtmlLocalise FINISHED ***");
#if DEBUG
foreach (var extn in MimeCollection.MissingExtns.OrderBy(e => e))
{
Console.WriteLine($"missing extn\t{extn}");
}
#endif
Console.ReadLine();
}
