public void ExecutesQueriesViaHttpClient()
{
var httpClient = new Mock<IHttpClient>(MockBehavior.Strict);
var client = new WikimediaClient(httpClient.Object, "Foo", 5);
var task = client.Query("foo=bar");
httpClient.Verify(s => s.Get(client.BaseQueryUrl + "&foo=bar", "Foo"));
}
static async Task MainAsync(IMongoDatabase db)
{
var client = new WikimediaClient(USER_CLIENT, MAX_LAG);
client.LaggedResponse += (sender, response) =>
Console.WriteLine($"Response was lagged {response.Error.SecondsLagged} seconds. Waiting to issue next query.");
var articles = db.GetCollection<BsonDocument>("Articles");
await articles.Indexes.CreateOneAsync("{PageId: 1}");
var start = await articles.CountAsync("{}");
Console.WriteLine($"Starting download. {start.ToString("#,###,##0")} articles currently into the database. Press Ctrl+C to interrupt.");
// Keep the same number of download tasks running at the same time
int MAX_TASKS = 50;
var queue = new List<Task>(MAX_TASKS);
for(int i = 0; i < MAX_TASKS; i++) {
queue.Add(InsertRandomDocumentsIntoDatabase(client, articles));
}
int completedTasks = 0;
var sw = new Stopwatch();
sw.Start();
while (true) {
// keep replacing finished tasks with new ones until the process is interrupted by the user.
var finishedTask = await Task.WhenAny(queue);
queue.Remove(finishedTask);
queue.Add(InsertRandomDocumentsIntoDatabase(client, articles));
completedTasks++;
if(completedTasks % 100 == 0) {
var total = await articles.CountAsync("{}");
var soFar = total - start;
var totalSeconds = sw.Elapsed.TotalSeconds;
var speed = soFar / (double)totalSeconds;
Console.WriteLine($"Total of {total.ToString("#,###,##0")} articles inserted into the database ({speed.ToString("0.00")} articles/sec). Press Ctrl+C to interrupt.");
}
}
}
static async Task InsertRandomDocumentsIntoDatabase(WikimediaClient client, IMongoCollection<BsonDocument> collection)
{
var response = await client.GetMaxRandomArticles();
var updates = response.Pages.Values.Select(page => {
var doc = DocumentFromPage(page);
return collection.ReplaceOneAsync(
Builders<BsonDocument>.Filter.Eq(d => d["PageId"], page.PageId), doc,
new UpdateOptions { IsUpsert = true });
});
await Task.WhenAll(updates);
}
public async Task GetsRandomArticles()
{
var client = new WikimediaClient(USER_CLIENT, 5);
var response = await client.GetMaxRandomArticles();
Assert.IsNotEmpty(response.Pages);
}
public async Task ReturnsQueryResponsesWithPages()
{
var httpClient = new Mock<IHttpClient>(MockBehavior.Strict);
var client = new WikimediaClient(httpClient.Object, "Foo", 5);
httpClient.Setup(c => c.Get(client.BaseQueryUrl + "&foo=bar", "Foo")).ReturnsAsync("{'query':{'pages':{}}}");
var response = await client.Query("foo=bar");
Assert.NotNull(response.Pages);
}
public void HasBaseQueryUrl()
{
var httpClient = Mock.Of<IHttpClient>();
var client = new WikimediaClient(httpClient, "foo", 52);
Assert.AreEqual("https://en.wikipedia.org/w/api.php?format=json&maxlag=52&action=query&prop=extracts|info&inprop=url&exlimit=max&exintro=1", client.BaseQueryUrl);
}
public void UsesDefaultClientWhenNotSupplied()
{
var client = new WikimediaClient("foo", 2);
Assert.AreSame(HttpClient.Default, client.HttpClient);
}
public void HasMaxLag()
{
var httpClient = Mock.Of<IHttpClient>();
var client = new WikimediaClient(httpClient, "foo", 5);
Assert.AreEqual(5, client.MaxLag);
}
public void HasBaseUrl()
{
var httpClient = Mock.Of<IHttpClient>();
var client = new WikimediaClient(httpClient, "foo", 25);
Assert.AreEqual("https://en.wikipedia.org/w/api.php?format=json&maxlag=25", client.BaseUrl);
}
public void HasUserAgent()
{
var httpClient = Mock.Of<IHttpClient>();
var client = new WikimediaClient(httpClient, "foo", 5);
Assert.AreEqual("foo", client.UserAgent);
}
public void HasHttpClient()
{
var httpClient = Mock.Of<IHttpClient>();
var client = new WikimediaClient(httpClient, "foo", 5);
Assert.AreSame(httpClient, client.HttpClient);
}
public async Task RetriesHttpQueryWhileHasMaxLagError()
{
var httpClient = new Mock<IHttpClient>(MockBehavior.Strict);
var client = new WikimediaClient(httpClient.Object, "FooAgent", 5, minimumDelaySeconds: 0); // To avoid delays during tests
var errorResponse = "{'error':{'code':'maxlag','info':'0 seconds lagged'}}";
var successResponse = "{'query':{}}";
// Sets up mock responses in sequence to simulate a maxlag errors followed by a success response the third time around
httpClient
.SetupSequence(c => c.Get(client.BaseQueryUrl + "&foo=bar", "FooAgent"))
.Returns(Task.FromResult(errorResponse))
.Returns(Task.FromResult(errorResponse))
.Returns(Task.FromResult(successResponse));
var response = await client.Query("foo=bar");
Assert.NotNull(response);
}
public async Task QueriesMaxNumberOfRandomArticles()
{
var httpClient = new Mock<IHttpClient>(MockBehavior.Strict);
var client = new WikimediaClient(httpClient.Object, "Foo", 5);
var expectedUrl = client.BaseQueryUrl + "&generator=random&grnnamespace=0&grnlimit=max";
httpClient.Setup(c => c.Get(expectedUrl, "Foo")).ReturnsAsync("{'query':{}}");
var response = await client.GetMaxRandomArticles();
Assert.NotNull(response);
}
public async Task ReturnsQueryResponsesWithPageData()
{
var httpClient = new Mock<IHttpClient>(MockBehavior.Strict);
var client = new WikimediaClient(httpClient.Object, "Foo", 5);
httpClient.Setup(c => c.Get(client.BaseQueryUrl + "&foo=bar", "Foo")).ReturnsAsync("{'query':{'pages':{'a':{'title':'Hello world'}}}}");
var response = await client.Query("foo=bar");
Assert.AreEqual("Hello world", response.Pages["a"].Title);
}
