/// <summary>
/// Gets images from Brave Search.
/// </summary>
/// <remarks>This method returns at most 150 image results (unless Brave changes something in their API).</remarks>
/// <param name="query">The search query.</param>
/// <param name="safeSearch">The safe search level.</param>
/// <param name="country">The country. <see cref="BraveCountries"/> contains the countries that can be used here.</param>
/// <param name="size">The image size.</param>
/// <param name="type">The image type.</param>
/// <param name="layout">The image layout.</param>
/// <param name="color">The image color.</param>
/// <param name="license">The image license.</param>
/// <returns>A task representing the asynchronous operation. The result contains an <see cref="IEnumerable{T}"/> of <see cref="BraveImageResult"/>.</returns>
/// <exception cref="ArgumentNullException"><paramref name="query"/> is null or empty.</exception>
/// <exception cref="GScraperException">An error occurred during the scraping process.</exception>
public async Task <IEnumerable <BraveImageResult> > GetImagesAsync(string query, SafeSearchLevel safeSearch = SafeSearchLevel.Moderate,
string?country = null, BraveImageSize size = BraveImageSize.All, BraveImageType type = BraveImageType.All,
BraveImageLayout layout = BraveImageLayout.All, BraveImageColor color = BraveImageColor.All, BraveImageLicense license = BraveImageLicense.All)
{
GScraperGuards.NotNull(query, nameof(query));
byte[] bytes;
using (var request = new HttpRequestMessage())
{
string cookie = $"safesearch={safeSearch.ToString().ToLowerInvariant()}";
if (!string.IsNullOrEmpty(country))
{
cookie += $"; country={country}";
}
request.Method = HttpMethod.Get;
request.RequestUri = new Uri(BuildImageQuery(query, size, type, layout, color, license), UriKind.Relative);
request.Headers.Add("cookie", cookie);
var response = await _httpClient.SendAsync(request).ConfigureAwait(false);
response.EnsureSuccessStatusCode();
bytes = await response.Content.ReadAsByteArrayAsync().ConfigureAwait(false);
}
var document = JsonDocument.Parse(bytes);
var results = document.RootElement.GetPropertyOrDefault("results");
return(EnumerateResults(results));
}
/// <summary>
/// Initializes a new instance of the <see cref="BraveScraper"/> class using the provided <see cref="HttpClient"/> and API endpoint.
/// </summary>
public BraveScraper(HttpClient client, string apiEndpoint)
{
GScraperGuards.NotNull(client, nameof(client));
GScraperGuards.NotNullOrEmpty(apiEndpoint, nameof(apiEndpoint));
_httpClient = client;
_httpClient.BaseAddress = new Uri(apiEndpoint);
if (_httpClient.DefaultRequestHeaders.UserAgent.Count == 0)
{
_httpClient.DefaultRequestHeaders.UserAgent.ParseAdd(_defaultUserAgent);
}
}
/// <summary>
/// Gets images from DuckDuckGo.
/// </summary>
/// <remarks>This method returns at most 100 image results.</remarks>
/// <param name="query">The search query.</param>
/// <param name="safeSearch">The safe search level.</param>
/// <param name="time">The image time.</param>
/// <param name="size">The image size.</param>
/// <param name="color">The image color.</param>
/// <param name="type">The image type.</param>
/// <param name="layout">The image layout.</param>
/// <param name="license">The image license.</param>
/// <param name="region">The region. <see cref="DuckDuckGoRegions"/> contains the regions that can be used here.</param>
/// <returns>A task representing the asynchronous operation. The result contains an <see cref="IEnumerable{T}"/> of <see cref="DuckDuckGoImageResult"/>.</returns>
/// <exception cref="ArgumentNullException"><paramref name="query"/> is null or empty.</exception>
/// <exception cref="ArgumentOutOfRangeException"><paramref name="query"/> is larger than <see cref="MaxQueryLength"/>.</exception>
/// <exception cref="GScraperException">An error occurred during the scraping process.</exception>
public async Task <IEnumerable <DuckDuckGoImageResult> > GetImagesAsync(string query, SafeSearchLevel safeSearch = SafeSearchLevel.Moderate,
DuckDuckGoImageTime time = DuckDuckGoImageTime.Any, DuckDuckGoImageSize size = DuckDuckGoImageSize.All, DuckDuckGoImageColor color = DuckDuckGoImageColor.All,
DuckDuckGoImageType type = DuckDuckGoImageType.All, DuckDuckGoImageLayout layout = DuckDuckGoImageLayout.All, DuckDuckGoImageLicense license = DuckDuckGoImageLicense.All,
string region = DuckDuckGoRegions.UsEnglish)
{
GScraperGuards.NotNull(query, nameof(query));
GScraperGuards.NotNullOrEmpty(region, nameof(region));
GScraperGuards.ArgumentInRange(query.Length, MaxQueryLength, nameof(query), $"The query cannot be larger than {MaxQueryLength}.");
string token = await GetTokenAsync(query);
var uri = new Uri($"i.js{BuildImageQuery(token, query, safeSearch, time, size, color, type, layout, license, region)}", UriKind.Relative);
byte[] bytes;
using (var request = new HttpRequestMessage())
{
request.Method = HttpMethod.Get;
request.RequestUri = uri;
if (safeSearch == SafeSearchLevel.Strict)
{
request.Headers.Add("cookie", "p=1");
}
else if (safeSearch == SafeSearchLevel.Off)
{
request.Headers.Add("cookie", "p=-2");
}
var response = await _httpClient.SendAsync(request).ConfigureAwait(false);
response.EnsureSuccessStatusCode();
bytes = await response.Content.ReadAsByteArrayAsync().ConfigureAwait(false);
}
var document = JsonDocument.Parse(bytes);
var results = document.RootElement.GetPropertyOrDefault("results");
return(EnumerateResults(results));
}
/// <summary>
/// Gets images from Google Images.
/// </summary>
/// <remarks>This method returns at most 100 image results.</remarks>
/// <param name="query">The search query.</param>
/// <param name="safeSearch">The safe search level.</param>
/// <param name="size">The image size.</param>
/// <param name="color">The image color. <see cref="GoogleImageColors"/> contains the colors that can be used here.</param>
/// <param name="type">The image type.</param>
/// <param name="time">The image time.</param>
/// <param name="license">The image license. <see cref="GoogleImageLicenses"/> contains the licenses that can be used here.</param>
/// <param name="language">The language code to use. <see cref="GoogleLanguages"/> contains the language codes that can be used here.</param>
/// <returns>A task representing the asynchronous operation. The result contains an <see cref="IEnumerable{T}"/> of <see cref="GoogleImageResult"/>.</returns>
/// <exception cref="ArgumentNullException"><paramref name="query"/> is null or empty.</exception>
/// <exception cref="GScraperException">An error occurred during the scraping process.</exception>
public async Task <IEnumerable <GoogleImageResult> > GetImagesAsync(string query, SafeSearchLevel safeSearch = SafeSearchLevel.Off, GoogleImageSize size = GoogleImageSize.Any,
string?color = null, GoogleImageType type = GoogleImageType.Any, GoogleImageTime time = GoogleImageTime.Any,
string?license = null, string?language = null)
{
GScraperGuards.NotNull(query, nameof(query));
var uri = new Uri(BuildImageQuery(query, safeSearch, size, color, type, time, license, language), UriKind.Relative);
string page = await _httpClient.GetStringAsync(uri).ConfigureAwait(false);
JsonElement rawImages;
try
{
rawImages = ExtractDataPack(page);
}
catch (Exception e) when(e is ArgumentOutOfRangeException or JsonException)
{
throw new GScraperException("Failed to unpack the image object data.", "Google", e);
}
return(EnumerateResults(rawImages));
}
