/// <summary>
/// Returns Apple Touch Icons for a Uri.
/// </summary>
/// <param name="uri">Uri to explore.</param>
/// <param name="doc">Instance of <see cref="HtmlDocument"/> containing parsed Html for the Uri.</param>
/// <param name="settings">Extraction settings.</param>
/// <returns>List of Apple Touch Icons as instances of <see cref="WebImage"/>.</returns>
public static async Task <List <WebImage> > GetAppleTouchIcons(Uri uri, HtmlDocument doc, ExtractionSettings settings)
{
List <WebImage> arr = new List <WebImage>();
// Check default apple touch icon location
if (!settings.SvgOnly)
{
Uri atiUri = uri.AddHtmlLink(Constants.DefaultAppleTouchIconPath);
if (await ImageDownloader.IsOkUri(atiUri))
{
if (!Recurser.FoundUris.Contains(atiUri.ToString()))
{
Recurser.FoundUris.Add(atiUri.ToString());
arr.Add(new WebImage(atiUri.ToString(), false, true));
}
}
}
// Parse HTML Tree
if (doc == null)
{
return(arr);
}
// Get icons through link info
IEnumerable <HtmlLinkTag> htmlLinkTags = HtmlExtractor.ExtractLinks(doc);
if (htmlLinkTags != null)
{
foreach (HtmlLinkTag link in htmlLinkTags)
{
if (link.IsAppleTouchIcon())
{
Uri absUri = uri.AddHtmlLink(link.Link);
if (settings.SvgOnly && !absUri.HasSvgExtension())
{
continue;
}
if (!settings.SvgOnly)
{
if (!absUri.HasImageExtension())
{
continue;
}
}
if (Recurser.FoundUris.Contains(absUri.ToString()))
{
continue;
}
if (await ImageDownloader.IsOkUri(absUri))
{
Recurser.FoundUris.Add(absUri.ToString());
arr.Add(new WebImage(absUri.ToString(), false, true));
}
}
}
}
return(arr);
}
/// <summary>
/// Returns images for a Uri (excluding Favicons and Apple Touch Icons).
/// </summary>
/// <param name="uri">Uri to explore.</param>
/// <param name="doc">Instance of <see cref="HtmlDocument"/> containing parsed Html for the Uri.</param>
/// <param name="settings">Extraction settings.</param>
/// <returns>List of images as instances of <see cref="WebImage"/>.</returns>
public static async Task <List <WebImage> > GetPageImages(Uri uri, HtmlDocument doc, ExtractionSettings settings)
{
List <WebImage> arr = new List <WebImage>();
if (doc == null)
{
return(arr);
}
if (settings.GetLinkTagImages)
{
IEnumerable <HtmlLinkTag> htmlLinkTags = HtmlExtractor.ExtractLinks(doc);
if (htmlLinkTags != null)
{
foreach (HtmlLinkTag link in htmlLinkTags)
{
if (!link.IsFavicon() && !link.IsAppleTouchIcon())
{
Uri absUri = uri.AddHtmlLink(link.Link);
if (settings.SvgOnly && !absUri.HasSvgExtension())
{
continue;
}
if (!settings.SvgOnly)
{
if (!absUri.HasImageExtension())
{
continue;
}
}
if (Recurser.FoundUris.Contains(absUri.ToString()))
{
continue;
}
if (await ImageDownloader.IsOkUri(absUri))
{
Recurser.FoundUris.Add(absUri.ToString());
arr.Add(new WebImage(absUri.ToString(), false, false));
}
}
}
}
}
if (settings.GetMetaTagImages)
{
IEnumerable <HtmlMetaTag> htmlMetaTags = HtmlExtractor.ExtractMetadata(doc);
if (htmlMetaTags != null)
{
foreach (HtmlMetaTag meta in htmlMetaTags)
{
Uri absUri = uri.AddHtmlLink(meta.Content);
if (settings.SvgOnly && !absUri.HasSvgExtension())
{
continue;
}
if (!settings.SvgOnly)
{
if (!absUri.HasImageExtension())
{
continue;
}
}
if (Recurser.FoundUris.Contains(absUri.ToString()))
{
continue;
}
if (await ImageDownloader.IsOkUri(absUri))
{
Recurser.FoundUris.Add(absUri.ToString());
arr.Add(new WebImage(absUri.ToString(), false, false));
}
}
}
}
if (settings.GetInlineBackgroundImages)
{
// Inline background images
HtmlNode bodyNode = doc.DocumentNode.SelectSingleNode("//body");
if (bodyNode != null)
{
await TraverseNode(bodyNode, async (HtmlNode node) =>
{
string styleValue = node.GetAttributeValue("style", null);
if (!string.IsNullOrEmpty(styleValue))
{
MatchCollection matches = Regex.Matches(styleValue, @"(?<=url\()(.*)(?=\))");
foreach (Match match in matches)
{
string value = match.Value;
if (!string.IsNullOrEmpty(value))
{
if ((value[0] == '\'' && value[value.Length - 1] == '\'') || (value[0] == '\"' && value[value.Length - 1] == '\"'))
{
value = value.Substring(1, value.Length - 2);
}
Uri link = uri.AddHtmlLink(value);
if (settings.SvgOnly && !link.HasSvgExtension())
{
continue;
}
if (!settings.SvgOnly)
{
if (!link.HasImageExtension())
{
continue;
}
}
if (Recurser.FoundUris.Contains(link.ToString()))
{
continue;
}
if (await ImageDownloader.IsOkUri(link))
{
Recurser.FoundUris.Add(link.ToString());
arr.Add(new WebImage(link.ToString(), false, false, true));
}
}
}
}
});
}
// TODO: Add Option to get background images from CSS files
}
IEnumerable <HtmlImgTag> imgTags = HtmlExtractor.ExtractImages(doc);
if (imgTags != null)
{
foreach (HtmlImgTag img in imgTags)
{
Uri absUri = uri.AddHtmlLink(img.Src);
if (settings.SvgOnly && !absUri.HasSvgExtension())
{
continue;
}
if (absUri.IsBadMagickType())
{
continue;
}
if (Recurser.FoundUris.Contains(absUri.ToString()))
{
continue;
}
if (await ImageDownloader.IsOkUri(absUri))
{
Recurser.FoundUris.Add(absUri.ToString());
arr.Add(new WebImage(absUri.ToString(), false, false));
}
}
}
return(arr);
}
/// <summary>
/// Extracts images for a page and linked pages.
/// </summary>
/// <param name="method">Method which extracts images from speicific Uri.</param>
/// <param name="uri">Uri to extract images and start hyperlink recursion from.</param>
/// <param name="settings">Extraction Settings.</param>
/// <param name="depth">Depth to recurse hyperlinks to.</param>
/// <param name="cancellationToken">Cancellation Token.</param>
/// <returns>Returns extracted images for given Uri and linked pages.</returns>
public static async Task <List <WebImage> > HyperlinkRecurse(ExtractionMethod method, Uri uri, ExtractionSettings settings, int depth, CancellationToken cancellationToken)
{
HtmlDocument doc = null;
bool gotDoc = false;
List <WebImage> images = new List <WebImage>();
if (!exploredUris.Contains(uri.ToString()))
{
exploredUris.Add(uri.ToString());
if (settings.OnStartNewPage != null)
{
if (cancellationToken.IsCancellationRequested)
{
return(null);
}
await settings.OnStartNewPage.Invoke(uri.ToString());
}
if (cancellationToken.IsCancellationRequested)
{
return(null);
}
gotDoc = true;
doc = await GetDocumnent(uri, cancellationToken);
images = await method.Invoke(uri, doc, settings);
if (cancellationToken.IsCancellationRequested)
{
return(null);
}
if (settings.OnEndNewPage != null)
{
await settings.OnEndNewPage.Invoke(uri.ToString(), images);
}
if (settings.ShouldStopOnFoundImage != null)
{
// Take all images up to the point where should stop
int index = images.TakeWhile(i => !settings.ShouldStopOnFoundImage.Invoke(i)).Count();
if (index != images.Count)
{
images.RemoveRange(index + 1, images.Count - index - 1);
stopAlg = true;
}
}
if (!settings.LazyDownload)
{
if (cancellationToken.IsCancellationRequested)
{
return(null);
}
await Task.WhenAll(images.Select(i => i.GetImageAsync(cancellationToken)).ToArray());
}
if (settings.OnFoundImage != null)
{
images.ForEach(i => settings.OnFoundImage.Invoke(i));
}
}
if (!stopAlg && settings.RecurseHyperlinks && depth < settings.HyperlinkRecursionDepth)
{
if (!gotDoc)
{
if (cancellationToken.IsCancellationRequested)
{
return(null);
}
doc = await GetDocumnent(uri, cancellationToken);
}
if (doc != null)
{
IEnumerable <HtmlATag> aTags = HtmlExtractor.ExtractATags(doc);
foreach (HtmlATag aTag in aTags)
{
if (cancellationToken.IsCancellationRequested)
{
return(null);
}
Uri newUri = uri.AddHtmlLink(aTag.Href);
List <WebImage> moreImages = await HyperlinkRecurse(method, newUri, settings, depth + 1, cancellationToken);
images.AddRange(moreImages);
}
}
}
return(images);
}
