/// <summary>
/// Retrieves the best match from the entity cache. If no matches exist this will attempt to download the file.
/// </summary>
/// <remarks>returns a cloned copy</remarks>
/// <returns>The entity cache result or the original token if it cannot be improved.</returns>
private async Task <JToken> GetEntityHelper(JToken originalToken, Uri entity, IEnumerable <Uri> properties)
{
bool?fetch = await _entityCache.FetchNeeded(entity, properties);
JToken token = originalToken;
if (fetch == true)
{
// we are missing properties and do not have the page
DataTraceSources.Verbose("[DataClient] GetFile required to Ensure {0}", entity.AbsoluteUri);
await EnsureFile(entity);
}
// null means either there is no work to do, or that we gave up, return the original token here
// if the original token is null, meaning this came from Ensure(uri, uri[]) then we need to get the token from the cache
if (fetch != null || originalToken == null)
{
JToken entityCacheResult = await _entityCache.GetEntity(entity);
// If the entity cache is unable to improve the result, return the original
if (entityCacheResult != null)
{
token = entityCacheResult.DeepClone();
}
}
return(token);
}
/// <summary>
/// Converts a json stream into a JObject.
/// </summary>
private async static Task <JObject> StreamToJson(Stream stream)
{
JObject jObj = null;
string json = string.Empty;
if (stream != null)
{
try
{
stream.Seek(0, SeekOrigin.Begin);
using (var reader = new StreamReader(stream))
{
json = await reader.ReadToEndAsync();
jObj = JObject.Parse(json);
}
}
catch (Exception ex)
{
DataTraceSources.Verbose("[StreamToJson] Failed {0}", ex.ToString());
jObj = new JObject();
jObj.Add("raw", json);
}
}
else
{
DataTraceSources.Verbose("[StreamToJson] Null stream!");
}
return(jObj);
}
/// <summary>
/// Returns a JToken for the given entity.
/// </summary>
/// <param name="entity">JToken @id</param>
/// <returns>The entity Json</returns>
public async Task <JToken> GetEntity(Uri entity)
{
if (entity == null)
{
throw new ArgumentNullException("entity");
}
JToken token = await _entityCache.GetEntity(entity);
if (token == null)
{
// we don't have any info on the given entity, try downloading it
await EnsureFile(entity);
// request the entity again
token = await _entityCache.GetEntity(entity);
if (token == null)
{
DataTraceSources.Verbose("[EntityCache] Unable to get entity {0}", entity.AbsoluteUri);
Debug.Fail("Unable to get entity");
}
}
if (token != null)
{
// clone our cache copy
token = token.DeepClone();
}
return(token);
}
/// <summary>
/// Return the entity from the graph
/// </summary>
/// <param name="entity"></param>
/// <returns></returns>
private JToken GetEntityFromGraph(Uri entity)
{
JToken token = null;
JsonLdTripleCollection triples = null;
DataTraceSources.Verbose("[EntityCache] GetEntity {0}", entity.AbsoluteUri);
// make sure everything added at this point has gone into the master graph
WaitForTasks();
lock (this)
{
// find the best JToken for this subject that we have
triples = _masterGraph.SelectSubject(entity);
// update the last access time for these pages
foreach (var page in triples.Select(t => t.Page).Distinct())
{
page.UpdateLastUsed();
}
}
// find the best jtoken for the subject
JsonLdTriple triple = triples.Where(n => n.JsonNode != null).OrderByDescending(t => t.HasIdMatchingUrl ? 1 : 0).FirstOrDefault();
if (triple != null)
{
token = triple.JsonNode;
}
return(token);
}
protected override async Task <HttpResponseMessage> SendAsync(HttpRequestMessage request, CancellationToken cancellationToken)
{
HttpResponseMessage response = null;
CacheEnabledRequestMessage cacheRequest = request as CacheEnabledRequestMessage;
if (cacheRequest != null && cacheRequest.CacheOptions.UseFileCache)
{
Uri uri = request.RequestUri;
using (var uriLock = new UriLock(uri, cancellationToken))
{
cancellationToken.ThrowIfCancellationRequested();
// check cache
Stream stream = null;
if (_fileCache.TryGet(uri, out stream))
{
DataTraceSources.Verbose("[HttpClient] Cached Length: {0}", "" + stream.Length);
response = new CacheResponse(stream);
}
else
{
// get the item and add it to the cache
DataTraceSources.Verbose("[HttpClient] GET {0}", uri.AbsoluteUri);
response = await base.SendAsync(request, cancellationToken);
if (response.IsSuccessStatusCode)
{
DataTraceSources.Verbose("[HttpClient] Caching {0}");
_fileCache.Add(uri, cacheRequest.CacheOptions.MaxCacheLife, await response.Content.ReadAsStreamAsync());
}
}
}
}
if (response == null)
{
// skip cache
DataTraceSources.Verbose("[HttpClient] GET {0}", request.RequestUri.AbsoluteUri);
response = await base.SendAsync(request, cancellationToken);
}
return(response);
}
private async Task <JObject> GetFileInternal(Uri uri, TimeSpan cacheTime, bool cacheInGraph = true, bool cloneJson = true)
{
if (uri == null)
{
throw new ArgumentNullException("uri");
}
if (cacheTime == null)
{
throw new ArgumentNullException("cacheTime");
}
bool cache = cacheTime.TotalSeconds > 0;
// request the root document
Uri fixedUri = Utility.GetUriWithoutHash(uri);
Stream stream = null;
JObject result = null;
JObject clonedResult = null; // the copy we give the caller
try
{
using (var uriLock = new UriLock(fixedUri))
{
if (!cache || !_fileCache.TryGet(fixedUri, out stream))
{
// the stream was not in the cache or we are skipping the cache
int tries = 0;
// try up to 5 times to be a little more robust
while (stream == null && tries < 5)
{
tries++;
try
{
HttpRequestMessage request = new HttpRequestMessage(HttpMethod.Get, fixedUri.AbsoluteUri);
DataTraceSources.Verbose("[HttpClient] GET {0}", fixedUri.AbsoluteUri);
var response = await _httpClient.SendAsync(request);
Debug.Assert(response.StatusCode == HttpStatusCode.OK, "Received non-OK status code response from " + request.RequestUri.ToString());
if (response.StatusCode == HttpStatusCode.OK)
{
stream = await response.Content.ReadAsStreamAsync();
if (stream != null)
{
if (cache)
{
DataTraceSources.Verbose("[HttpClient] Caching {0}");
_fileCache.Add(fixedUri, _lifeSpan, stream);
}
DataTraceSources.Verbose("[HttpClient] 200 OK Length: {0}", "" + stream.Length);
result = await StreamToJson(stream);
}
}
else
{
DataTraceSources.Verbose("[HttpClient] FAILED {0}", "" + (int)response.StatusCode);
result = new JObject();
result.Add("HttpStatusCode", (int)response.StatusCode);
}
}
catch (HttpRequestException ex)
{
Debug.Fail("WebRequest failed: " + ex.ToString());
DataTraceSources.Verbose("[HttpClient] FAILED {0}", ex.ToString());
// request error
result = new JObject();
result.Add("HttpRequestException", ex.ToString());
}
}
}
else
{
// the stream was in the cache
DataTraceSources.Verbose("[HttpClient] Cached Length: {0}", "" + stream.Length);
result = await StreamToJson(stream);
}
}
}
finally
{
if (stream != null)
{
stream.Dispose();
}
}
if (result != null)
{
// this must be called before the entity cache thread starts using it
if (cloneJson)
{
clonedResult = result.DeepClone() as JObject;
}
else
{
// in some scenarios we can skip cloning, such as when we are throwing away the result
clonedResult = result;
}
if (cacheInGraph)
{
// this call is only blocking if the cache is overloaded
_entityCache.Add(result, fixedUri);
}
}
return(clonedResult);
}
/// <summary>
/// Ensures that the given properties are on the JToken. If they are not inlined they will be fetched.
/// Other data may appear in the returned JToken, but the root level will stay the same.
/// </summary>
/// <param name="jToken">The JToken to expand. This should have an @id.</param>
/// <param name="properties">Expanded form properties that are needed on JToken.</param>
/// <returns>The same JToken if it already exists, otherwise the fetched JToken.</returns>
public async Task <JToken> Ensure(JToken token, IEnumerable <Uri> properties)
{
if (token == null)
{
throw new ArgumentNullException("token");
}
if (properties == null)
{
throw new ArgumentNullException("properties");
}
JObject jObject = token as JObject;
if (jObject != null)
{
CompactEntityReader compactEntity = new CompactEntityReader(jObject);
// if the entity is found on it's originating page we know it is already complete in this compact form
if (compactEntity.IsFromPage == false)
{
if (compactEntity.EntityUri != null)
{
// inspect the compact entity on a basic level to determine if it already has the properties it asked for
if (compactEntity.HasPredicates(properties) != true)
{
// at this point we know the compact token does not include the needed properties,
// we need to either download the file it lives on, or find it in the entity cache
// if the token is for an entity that just does not exist or is corrupted in some way
// the original token will be returned since the entity cache cannot improve it after
// trying all possible methods.
return(await GetEntityHelper(token, compactEntity.EntityUri, properties));
}
}
else
{
DataTraceSources.Verbose("[EntityCache] Unable to find entity @id!");
}
}
}
else if (token.Type == JTokenType.String)
{
// It's just a URL, so we definitely need to fetch it from the cache
string tokenString = token.ToString();
// make sure it is a url
if (!String.IsNullOrEmpty(tokenString) && tokenString.StartsWith("http", StringComparison.OrdinalIgnoreCase))
{
Uri entityUrl = new Uri(tokenString);
// the entity cache should either find the child entity or if this url is a root url the full page will be returned
return(await GetEntityHelper(token, entityUrl, properties));
}
}
else
{
DataTraceSources.Verbose("[EntityCache] Non-JObject, unable to use this!");
}
// give the original token back
return(token);
}
/// <summary>
/// Removes all pages not used within the given time span.
/// </summary>
private void CleanUp(TimeSpan keepPagesUsedWithin)
{
DateTime cutOff = DateTime.UtcNow.Subtract(keepPagesUsedWithin);
// lock to keep any new pages from being added during this
lock (this)
{
// just in case we show really late
if (_disposed)
{
return;
}
// create a working set of pages that can be considered locked
JsonLdPage[] pages = _pages.Values.ToArray();
// if pages are still loading we should skip the clean up
// TODO: post-preview this should force a clean up if the graph is huge
if (pages.All(p => p.IsLoaded))
{
// check if a clean up is needed
if (pages.Any(p => !p.UsedAfter(cutOff)))
{
List <JsonLdPage> keep = new List <JsonLdPage>(pages.Length);
List <JsonLdPage> remove = new List <JsonLdPage>(pages.Length);
// pages could potentially change last accessed times, so make the decisions in one shot
foreach (var page in pages)
{
if (page.UsedAfter(cutOff))
{
keep.Add(page);
}
else
{
remove.Add(page);
}
}
// second check to make sure we need to do this
if (remove.Count > 0)
{
DataTraceSources.Verbose("[EntityCache] EntityCache rebuild started.");
JsonLdGraph graph = new JsonLdGraph();
// graph merge
foreach (var page in keep)
{
graph.Merge(page.Graph);
}
_masterGraph = graph;
DataTraceSources.Verbose("[EntityCache] EntityCache rebuild complete.");
// remove and dispose of the old pages
foreach (var page in remove)
{
JsonLdPage removedPage = null;
if (_pages.TryRemove(page.Uri, out removedPage))
{
Debug.Assert(!removedPage.UsedAfter(cutOff), "Someone used a page that was scheduled to be removed. This should have been locked.");
removedPage.Dispose();
}
else
{
Debug.Fail(page.Uri.AbsoluteUri + " disappeared from the page cache.");
}
}
}
}
}
}
}
private void Load(Action <JsonLdPage> callback)
{
JObject workingCopy = _compacted;
try
{
if (!Utility.IsValidJsonLd(workingCopy))
{
DataTraceSources.Verbose("[EntityCache] Invalid JsonLd skipping {0}", Uri.AbsoluteUri);
// we can't parse this page, load a blank graph
_graph = new JsonLdGraph();
return;
}
// we have to modify the json to create the graph, since other people are free to use
// _compacted during this time we have to make a copy, after we finish we will throw
// away _compacted and provide the copy we used instead.
workingCopy = _compacted.DeepClone() as JObject;
Uri rootUri = Utility.GetEntityUri(workingCopy);
if (rootUri == null)
{
// remove the blank node
string blankUrl = "http://blanknode.nuget.org/" + Guid.NewGuid().ToString();
workingCopy["@id"] = blankUrl;
DataTraceSources.Verbose("[EntityCache] BlankNode Doc {0}", blankUrl);
}
DataTraceSources.Verbose("[EntityCache] Added {0}", Uri.AbsoluteUri);
// Load
_graph = JsonLdGraph.Load(workingCopy, this);
// make the callback which should merge us into the master graph
callback(this);
}
catch (Exception ex)
{
// Something horrible happened when parsing the json-ld.
// The original file may be corrupted. The best option here is to leave the page
// out of the entity cache. Requests for entities from the page should default to just returning
// the compacted JTokens back. Those jtokens have as much info as we can get in this bad state.
DataTraceSources.Verbose("[EntityCache] Unable to load!! {0} {1}", Uri.AbsoluteUri, ex.ToString());
}
finally
{
if (_graph == null)
{
_graph = new JsonLdGraph();
}
// replace the original with the copy we used for the graph
_compacted = workingCopy;
_loadWait.Set();
_isLoaded = true;
}
}
