public async Task ChildElementLocatorTest()
{
// Arrange
const string address = "https://www.google.com.au/search?gl=au&hl=en&pws=0&num=100&q=";
const string keywords = "online title search";
var crawlerV2 = new CrawlerV2(address, keywords);
await crawlerV2.Search();
var response = crawlerV2.GetResponse();
var parentString = "<div class=\"ZINbbc xpd O9g5cc uUPGi\"><div class=\"kCrYT\">";
var childString = "<div class=\"kCrYT\"><a href=\"/url?q=";
var uri = "https://www.infotrack.com.au";
var childElementLocator = new ChildElementLocator(response, parentString, childString, uri);
// Act
childElementLocator.FindLocationsOfUri();
// Assert
CollectionAssert.AllItemsAreUnique(childElementLocator.GetUriLocations());
CollectionAssert.AllItemsAreNotNull(childElementLocator.GetUriLocations());
CollectionAssert.AllItemsAreInstancesOfType(childElementLocator.GetUriLocations(), typeof(int));
}
public async Task GetResponse_InvalidAddressAndKeywords_throw()
{
// Arrange
const string address = "not an address";
const string keywords = "some keyworkd";
var crawlerV2 = new CrawlerV2(address, keywords);
// Act and Assert
await Assert.ThrowsExceptionAsync <ArgumentException>(() => crawlerV2.Search());
}
/// <summary>
/// Form submission
/// </summary>
/// <param name="uri"></param>
/// <param name="keywords"></param>
/// <returns></returns>
public async Task <JsonResult> Crawl(string uri, string keywords)
{
// create a new crawler and do a crawl using the provided search terms
var crawlerV2 = new CrawlerV2(Address, keywords);
await crawlerV2.Search();
var data = crawlerV2.GetResponse();
if (data == string.Empty)
{
return new JsonResult()
{
Data = new { error = "Could not retrieve search engine results." }
}
}
;
// parent element
const string htmlOfDivContainingLink = "<div class=\"ZINbbc xpd O9g5cc uUPGi\"><div class=\"kCrYT\">";
// child element
const string htmlOfAnchorElement = "<div class=\"kCrYT\"><a href=\"/url?q=";
// create a new uri locator so we can figure out where all the links are
var uriLocator = new ChildElementLocator(data, htmlOfDivContainingLink, htmlOfAnchorElement, uri, new FindIndex());
uriLocator.FindLocationsOfUri();
uriLocator.CountUriOccurrences();
var uriLocations = uriLocator.GetUriLocations();
var uriCount = uriLocator.GetUriCount();
// display "no results" if there are no links to infotrack in the search results
if (uriCount == 0)
{
keywords = $"{keywords} (no results)";
}
return(new JsonResult
{
Data = new
{
success = new
{
uri,
keywords,
uriCount,
uriLocations
}
}
});
}
}
public async Task GetResponse_ValidInputs_NotNull()
{
// Arrange
const string address = "https://www.google.com.au/search?gl=au&hl=en&pws=0&num=100&q=";
const string keywords = "online title search";
CrawlerV2 crawlerV2 = new CrawlerV2(address, keywords);
// Act
await crawlerV2.Search();
// Assert
Assert.IsNotNull(crawlerV2.GetResponse());
}
public async Task CountUriOccurrencesTest_NotNull()
{
// Arrange
const string address = "https://www.google.com.au/search?gl=au&hl=en&pws=0&num=100&q=";
const string keywords = "online title search";
var crawlerV2 = new CrawlerV2(address, keywords);
await crawlerV2.Search();
var response = crawlerV2.GetResponse();
var parentString = "<div class=\"ZINbbc xpd O9g5cc uUPGi\"><div class=\"kCrYT\">";
var childString = "<div class=\"kCrYT\"><a href=\"/url?q=";
var uri = "https://www.infotrack.com.au";
var childElementLocator = new ChildElementLocator(response, parentString, childString, uri);
// Act
childElementLocator.CountUriOccurrences();
// Assert
Assert.IsNotNull(childElementLocator.GetUriCount());
}