/// <summary>
/// From a JSON results string (must be from GetSearchResults()), grab all the urls
/// </summary>
/// <param name="rawJsonResults">The JSON string</param>
/// <returns>A list of dynamic objects</returns>
/// <remarks>
/// Url = url["unescapedUrl"].ToString(),
/// Index = index,
/// Domain = WebProcessor.GetDomainOfUrl(url["unescapedUrl"].ToString())
/// </remarks>
public IList <dynamic> GetResultUrls(string rawJsonResults)
{
IEnumerable temp = null;
IList <dynamic> result = new List <dynamic>();
if (!Functions.IsEmptyString(rawJsonResults))
{
JObject jsonResults = JObject.Parse(rawJsonResults);
temp = jsonResults["responseData"]["results"].Children().Select(
(url, index) => new
{
Url = url["unescapedUrl"].ToString(),
Index = index,
Domain = WebProcessor.GetDomainOfUrl(url["unescapedUrl"].ToString())
});
//
// "temp" is some crazy type of variable. IEnumerable doesn't have a "Count" property, so it's pretty much useless.
// Todo: figure more of this Linq stuff out. It seems cool, but so hard to use.
//
foreach (object item in temp)
{
result.Add(item);
}
}
return(result);
}
/// <summary>
/// Get all the sounds in the list of urls from the passed IDataSource and add the sounds to the passed websearch
/// </summary>
/// <param name="urls">The list of urls</param>
/// <param name="dataSource">The datasource to use to look for sounds</param>
/// <param name="currentSearch">The current search</param>
private static void GetSoundsOnPages(IList <dynamic> urls, IDataSource dataSource, websearch currentSearch, IList <websearchsound> searchResultList, Functions.LogMessageDelegate LogMessage, int maxDepthToFollow)
{
//const int MAX_URLS_TO_SEARCH = 20;
const int MAX_SOUNDS_PER_URL = 150;
//int urlsProcessed = 0;
HashSet <string> urlsOfObjectsSearched = new HashSet <string>();
//
// Multithreading here for requesting the pages works pretty well speed-wise. Unfortunately, the regexes bog down the
// server so badly that it becomes unresponsive for other users. So, don't do parallel on this outside loop.
//
// However, once the first page is processed, the sounds are webrequested asynchronously. So, the next page will
// start being processed while the first page's sounds are still being downloaded. This works quite well, and
// the performance is just about the same. So, let's stick with that.
//
foreach (dynamic url in urls)
{
string theUrl = url.Url;
string domain = WebProcessor.GetDomainOfUrl(theUrl);
if (unprocessableDomains.Contains(domain))
{
LogMessage(string.Format("Skipping crappy domain: {0}", domain));
}
else
{
LogMessage(string.Format("About to search for sounds on page: \"{0}\"", theUrl));
// string pageContent = WebProcessor.GetUrlContents(theUrl, null, null, LogMessage);
//
// todo: test this, make sure it works
//
string pageContent = dataSource.GetUrlContents(theUrl, null, GetUserAgent(), LogMessage);
bool wasAborted = false;
//
// todo: combine sound links func with above function
//
IList <string> linksOnPage = GetSoundLinksOnPage(pageContent, ref wasAborted);
//
// For generating test case files, set breakpoint on if (wasAborted) below with condition:
//
// maxDepthToFollow == 1
//
if (wasAborted)
{
LogMessage(string.Format("Had to abort link search on domain: {0}", domain));
lock (unprocessableDomains)
{
unprocessableDomains.Add(domain);
}
}
LogMessage(string.Format("Found {0} links on \"{1}\"", linksOnPage.Count, theUrl));
#if MULTITHREADED
Parallel.ForEach <string>(linksOnPage.Take(MAX_SOUNDS_PER_URL), partialLink => // <=-- normal operation - multithreaded
#else
foreach (string partialLink in linksOnPage.Take(MAX_SOUNDS_PER_URL)) // <=-- for debugging stuff, it's easier when not multithreaded
#endif
{
string soundLink = WebProcessor.GetUrlForObject(theUrl, partialLink);
LogMessage(string.Format("About to grab a potential sound here: \"{0}\"", soundLink));
if (!unprocessableDomains.Contains(domain) && IsNewSoundToGrab(urlsOfObjectsSearched, soundLink))
{
websearchsound receivedObject = GetWebObjectAtUrl(soundLink, null, null);
//
// enhanced search: if not a sound and is text/html and response code is 200, search for sounds on THAT page
//
if (receivedObject.issound)
{
receivedObject.sourceurl = theUrl;
receivedObject.sourceDomain = domain;
receivedObject.searchResultOrder = url.Index;
//
// Check for dups
//
string md5Hash = Functions.GetMd5Hash(receivedObject.soundbytes);
if (!HaveMd5ForSound(dataSource.CurrentSoundMd5s, md5Hash))
{
dataSource.SetSoundInSearch(currentSearch, receivedObject);
//
// Performance optimization: we're not going to return the sound data itself with the search
// so let's free up the mem here
//
receivedObject.soundbytes = null;
searchResultList.Add(receivedObject);
}
else
{
LogMessage("Not adding sound - already in collection");
}
}
else if (receivedObject.contenttype.ToLower().StartsWith("text/html"))
{
//
// We have another HTML page. Check that too?
//
if (maxDepthToFollow > 0)
{
LogMessage(string.Format("Going to drill down in this page - we're at max level: {0}", maxDepthToFollow));
GetSoundsOnPages(new List <dynamic>()
{
new { Url = soundLink, Index = url.Index }
}, dataSource, currentSearch, searchResultList, LogMessage, maxDepthToFollow - 1);
}
else
{
LogMessage(string.Format("No more drilling down, we're as low as we can go"));
}
}
}
else
{
LogMessage("Won't process: already had sound from that url, or the domain is unprocessable!");
}
#if MULTITHREADED
});