V8 Support about anglesharp

[FlorianRappl]

The main library will only expose the DOM, however, there could be a V8 integration library (which is why I decorate every DOM element (type, property and method) with a DOMAttribute).
If you want to write a V8 integration library it would be great, otherwise it will probably take some time until I start doing it.

[philbritton]

V8 integration would be great, good thinking for decorating DOM elements with a DOMAttribute.

[FlorianRappl]

Alright so I think I will probably include a V8 integration sample in the samples application soon.

[FlorianRappl]

I will update the samples within the next week - there I will showcase a sample implementation of Jurassic, which is something like V8 (but all managed).

[pro100sanya]

It will be just an example or full featured implementation?

[FlorianRappl]

It will be an example AND it will use another library called (name still to be found) AngleSharp.Jurassic, which will also be distributed over NuGet. This library will depend on both, Jurassic and AngleSharp, and do all the necessary binding. Is that sufficient?

Initially I planned V8, but back then I did not know about Jurassic and it seems very good. I also started writing my own JS engine, but in fact it now useless, since it would do the same as Jurassic, which is already established. If you require / need V8, then I hope that the implementation of using Jurassic will give you enough to get started.

[pro100sanya]

Sounds really good!

Please clarify - DOM will be exposed to Jurassic or COULD be exposed?
I mean - if I add AngleShart.Jurassic can i suppose dom in js environment?

[devmondo]

will this make anglesharp have the ability to behave like a headless browser so we can use it to improve SEO ?

[FlorianRappl]

Exactly. But please keep in mind that this ability comes with 2 things (and I don't know if those 2 things will be implemented right away, however, they are on the roadmap and will be implemented soon otherwise):

1.) the ability to extend AngleSharp by registering elements e.g. for HTTP requests
2.) the scripting sections of AngleSharp

Some things in 2.) are already implemented (you can check yourself that it makes a difference calling e.g. the DocumentBuilder.Html() method (has scripting disabled at the moment) and e.g. the HtmlParser with a fresh HtmlDocument() that has scripting enabled - the noscript section for instance is treated differently). The rest is performing scripting actions and more on the DOM, i.e. running the code given in <script> tags.

tl;dr: Yes in the long run, maybe no in the short run (but quite close).

[devmondo]

thanks a lot for answering, do you mean that in the current time that HtmlParser with a fresh HtmlDocument() that has scripting enabled will execute the scripts in page like in Browser ?

regarding the implementation above that you mentioned, this would be awesome, a lot of us are facing many problems, like

• ability to extract styles and embed them in html document to send it as an email
• ability to execute scripts for AJAX application like mentioned above to improve SEO

currently we tend to use html agility pack and premailer to solve this, but if AngleSharp would enable us to do this nativly or through extensible points then this would be one of the best things happened in this domain for .NET

another thing, i dont see any documentation section :)

[FlorianRappl]

Hi, documentation is currently given in form of the samples, however, I would welcome if somebody would start something like a documentation section (that would be very helpful!). So if you are willing to contribute I would really find that amazing!

Regarding the list of problems:
1.) Yes that will be easily possible, but I have to integrate "getComputedStyle" style. This will happen in the next couple of weeks so stay tuned!
2.) That will be possible as well (as discussed in this thread).

Right now nothing executes scripts since no scripting engine is integrated. However, there is a flag if scripting is enabled or not. This flag determines the behavior for some tags e.g. noscript. If scripting is disabled everything is parsed as usual, otherwise everything inside is parsed in rawtext mode, which is a big difference.

In the future the DocumentBuilder will automatically switch on scripting if a scripting engine has been registered. Right now this has to be done manually by creating a fresh document, setting the flag, instantiating the parser etc.

[Rajbandi]

I can help with the documentation. Please let me know.

Registering a script engine will be helpful instead of sticking to particular engine.

[devmondo]

thanks a lot for reply, this sound so promising, i have downloaded the source and i am overwhelmed by the amount of work you have done, this library is so great and it is obviously way beyond Html Agility back and it needs more attention from the community. thanks a lot for putting out this gem :)

that aside, i cant access Htmldocument, Visual studio says it is internal so something like this cant work
var htmlDoc = new HtmlDocument();

[FlorianRappl]

The scripting flag is also internal - so this was just to illustrate (AngleSharp aims to follow the W3C spec, hence I usually do not expose internal stuff, which also does not follow the spec -- helper methods to do certain things are exceptions). As I wrote at the moment this is more or less useless - in the future when you can register a scripting engine AngleSharp will auto-detect this and handle requests accordingly (with scripting enabled / disabled depending on the scripting engine state).
Like in a browser once a scripting engine is registered one can also de-activate it just in a certain context. Hence one could load a webpage with scripting and without.

By the way: The DOM gives you the possibility of creating a new HTML document - but it is tricky (I think it is not known by many since it is quite usless in a real world context). You need the implementation property of a Document (or derived) instance and call the createDocument method (if you want an HTMLDocument and not a Document then you have also to enter the HTML namespace URI).

[devmondo]

this is really great to hear all the best to you on your great work with this library :)

[FlorianRappl]

Thanks!