public async void TestForTwoHitsOnPageOf10ResultsForGoogle()
{
var scraper = new GoogleSearchScraper();
var occurences = await scraper.GetSiteOccurencesForKeyword("online title search", "www.infotrack.com.au", 10);
Assert.Equal(2, occurences.Count);
}
public void GoogleSearchScraper_CanParseMultiplePages()
{
var SUT = new GoogleSearchScraper(downloadService: null);
var parsedItems = SUT.GetAllSearchResultItems(new string[] { testHtml, testHtml }).ToList();
Assert.NotNull(parsedItems);
Assert.True(parsedItems.Count == 8);
}
public void GoogleSearchScraper_CanGenerateSearchUrls()
{
// ASSERT:
// Scraper can generate correct list of urls required to run a search
var SUT = new GoogleSearchScraper(downloadService: null);
var keyWords = new string[] { "KEYWORD1", "KEYWORD2" };
var maxSearchResults = 100;
var requestUrls = SUT.GetSearchRequestUrls(keyWords, maxSearchResults).ToList();
Assert.NotNull(requestUrls);
Assert.True(requestUrls.Count == 10);
Assert.True(requestUrls[0] == "https://www.google.com.au/search?q=KEYWORD1+KEYWORD2&start=0");
Assert.True(requestUrls[1] == "https://www.google.com.au/search?q=KEYWORD1+KEYWORD2&start=10");
Assert.True(requestUrls[9] == "https://www.google.com.au/search?q=KEYWORD1+KEYWORD2&start=90");
}
public async Task GoogleSearchScraper_BadSearch_ReturnsEmptyList()
{
// ASSERT:
// Scraper is able to handle responses with no search results.
_downloadService
.DownloadWebsitesParallelAsync(urls: Arg.Any <IEnumerable <string> >())
.Returns(new List <string> {
notFoundTestHtml
});
var SUT = new GoogleSearchScraper(downloadService: _downloadService);
var keyWords = new string[] { "KEYWORD1", "KEYWORD2" };
var maxSearchResults = 100;
var searchResults = await SUT.Search(keyWords, maxSearchResults) as List <string>;
Assert.NotNull(searchResults);
Assert.False(searchResults.Any());
}
public async Task GoogleSearchScraper_Search_ReturnsValidSearchResults()
{
// ASSERT:
// Scraper is able to process multiple search downloads.
// One of the results contains expected substring.
_downloadService
.DownloadWebsitesParallelAsync(urls: Arg.Any <IEnumerable <string> >())
.Returns(new List <string> {
testHtml, testHtml, testHtml
});
var SUT = new GoogleSearchScraper(downloadService: _downloadService);
var keyWords = new string[] { "KEYWORD1", "KEYWORD2" };
var maxSearchResults = 100;
var expectedSubstring = "www.sympli.com.au";
var searchResults = await SUT.Search(keyWords, maxSearchResults) as List <string>;
Assert.NotNull(searchResults);
Assert.True(searchResults.Any(r => r.Contains(expectedSubstring)));
}
