Web Image Extractor

.NET library to extract images and icons from websites. Provides options to select Favicons, Apple Touch Icons, or all images on a page.

Blazor Demo

Available on NuGet

Features (all are optional and can be speicified in ExtractionSettings):

• Extracts Favicons and Apple touch icons.
• Extracts images from <img> tags.
• Extracts images from <link> and <meta> tags within HTML head.
• Extracts images from background images in CSS background-image: uri('https://...'). Currently this only works for inline styles and not referenced .css files.
• Supports Uri Recursion. E.g. Extracting from https://example.com/dir/page.html also extracts from https://example.com/dir and https://example.com.
• Supports following Hyperlinks (specified in <a> tags) up to a certain depth and extracting images from those pages too.
• Supports triggering events when new images are found or new pages are explored.
• Supports ending the extraction algorithm when a condition is met (e.g. enough images found).
• Supports Blazor (including Client Side Blazor / Blazor WebAssembly).
• Includes example console application.

To see an online example use of this tool, checkout this Blazor demo. You can find the source code for the demo here.

Note: If you are after something specific, e.g. a way of getting 48px square .ico icons from websites ... please note that some websites might only have one icon available and others might not have any. This library isn't magic, so you might not get exactly what you want. It's always best to give a user choice over the best icon/image, or even let them provide their own, don't rely entirely on the algorithms.

How does it work?

Favicons are found from the standard places. First the path /favicon.ico is checked, and then it checks for <link> tags in the HTML head. Any link with a valid Favicon rel property will be returned as a Favicon.

It works in a similar way for Apple Touch Icons.

When looking for all images on a page, Favicons and Apple touch icons are retrieved, then all other images in <link> and <meta> tags are retrieved, followed by all images in <img> tags, and finally all images specified using background-image: url('https://...') within inline styles (if enabled in settings). Note that images are not pulled from background-image styles set in separate .css files. It only works for inline styles.

Recursion works by exploring Uris with segments removed, and by following hyperlinks (<a> tags) on the page, providing these are enabled in the settings.

Installation

Install using the NuGet Package Manager:

Install-Package WebImageExtractor -Version 1.0.2

Or

To use in your project you can download in the library here or clone the repository and add the WebImageExtractor C# Project to your solution (then use a project reference in your own project).

The library is a .NET Standard 2.0 library so works with both .NET Framework and .NET Core.

How to Use

Add the using reference to the top of your file:

using WebImageExtractor;

Then use one of the following methods:

Method	Description
Extractor.GetFavicons	Extracts all Favicons.
Extractor.GetAppleTouchIcons	Extracts all Apple Touch Icons.
Extractor.GetAllIcons	Extracts all Favicons and Apple Touch Icons from a page.
Extractor.GetPageImages	Extracts all images from a page - excluding Favicons and Apple Touch Icons.
Extractor.GetAllImages	Extracts all images from a page - including Favicons and Apple Touch Icons.

Each method is awaitable and accepts the Uri to a page (starting with https://) as a string. They return a list of images using the type IEnumerable<WebImage>.

IEnumerable<WebImage> images = await Extractor.GetFavicons("https://github.com");

To download an image using the WebImage class you can use the WebImage.GetImageAsync method:

using ImageMagick;
...
MagickImage image = await images[0].GetImageAsync();

Although this is awaitable, by default the result will be immediate as all images will have been downloaded in advance. But you can change this to download the image when GetImageAsync is called by enabling lazy loading in the settings.

If you have not changed this setting and cannot use await, instead use WebImage.GetImageIfDownloaded.

The WebImage class also contains information about whether the image is a Favicon, Apple touch icon, or background image.

Once you have a MagickImage you can convert it to your prefered format using Magick.NET. When using Blazor Client Side you may experience issues using Magick.NET this way. Instead you can use the Uri from WebImage.Uri to download the images manually. See the source code of the Blazor demo.

The above extraction methods accept settings through an instance of ExtractionSettings. For example:

ExtractionSettings settings = new ExtractionSettings();
IEnumerable<WebImage> images = await Extractor.GetFavicons("https://github.com", settings);

Settings

The ExtractionSettings class allows you to customize how images are extracted.

ExtractionSettings Properties:

Method	Type	Description
SvgOnly	bool	Only extract Svgs?
RecurseUri	bool	Recurse Uri segments and extract images from all?
RecurseHyperlinks	bool	Extract images from any pages linked to by the given Uri?
HyperlinkRecursionDepth	int	Number of layers of hyperlinks to explore for image extraction.
LazyDownload	bool	Download images immediatley after extraction? Or download them when required?
GetMetaTagImages	bool	When extracting all images on a page, should the extractor get images from <meta> tags in the html <head>
GetLinkTagImages	bool	When extracting all images on a page, should the extractor get images from <link> tags in addition to favicons and apple touch icons.
GetInlineBackgroundImages	bool	When extracting all images on a page, should the extractor get images from 'background-image' styles? Note this only works for inline styles, and not for images specified in separate css files.
UseCorsAnywhere	bool	Should Cors Anywhere (https://cors-anywhere.herokuapp.com/) be used? Only required for Web Applications (e.g. Blazor).
DisableValidityCheck	bool	Disables an additional check (that the image url returns OK) before images are returned. Setting to true improves performance, but returns more false positives.
HttpClient	HttpClient	An instance of HttpClient to use. If not set, a new HttpClient is created.

ExtractionSettings Events:

Event	Description
OnFoundImage	Callback event when an image is found by the extractor.
OnStartNewPage	Callback event when the extractor begins to explore a new Uri.
OnEndNewPage	Callback event when the extractor has finished exploring a Uri.
ShouldStopOnFoundImage	Callback event when a new image is found. Checks if the extractor should continue or if enough images have been found.

Example Use

The below example shows how to retrieve all images from a Uri, and search linked pages up to depth 3, along with Uri recursion. A line is written to the console each time an image is found. Images are saved to files as jpg images.

using WebImageExtractor;
using ImageMagick;

...

ExtractionSettings settings = new ExtractionSettings()
{
    RecurseUri = true,
    RecurseHyperlinks = true,
    HyperlinkRecursionDepth = 3,
};
settings.OnFoundImage += async (WebImage image) =>
{
    Console.WriteLine($"Found Image at Uri: {image.Uri}");
};
IEnumerable<WebImage> images = await Extractor.GetAllImages("https://github.com", settings);
int imageNum = 0;
foreach (WebImage image in images) {
    MagickImage downloadedImage = await image.GetImageAsync();
    downloadedImage.Write($"image-{imageNum}.jpg");
    imageNum++;
}

License

This code is released under MIT license. This means you can use this for whatever you want. Modify, distribute, sell, fork, and use this as much as you like. Both for personal and commercial use. I hold no responsibility if anything goes wrong.

If you use this, you don't need to refer to this repo, or give me any kind of credit but it would be appreciated.

Contributing

Pull Requests are welcome. But, note that by creating a pull request you are giving me permission to merge your code and release it under the MIT license mentioned above. At no point will you be able to withdraw merged code from the repository, or change the license under which it has been made available.