Chrome browser extensions - hooking into C functions? - google-chrome

I'm writing a Chrome extension and some parts of it need to be super high performance. I'm trying to find documentation to see if it's possible to use C extensions within the Chrome extension. Is this currently possible?

Take a look at Native Client. You can write browser/web apps with NaCl via C or C++ (and Mono since 2.10). You can find a small tutorial here.

Related

Can I make calls to APIs such as youtube-dl and ffmpeg from a chrome-app?

First of all, I haven't started the implementation of the system I'm about to describe, as I didn't want to commit on implementing something I did not know if was possible.
So, what I'm trying to achieve is to build a chrome-app to download the audio from certain websites (e.g. youtube and soundcloud) using youtube-dl, post process it using ffmpeg and then upload it to a cloud service via some api. The reason I want to do it via a chrome-app is because I could do all the work on the client side (no need for servers) and I'd have the ability to insert javascript into the pages using content scripts, which would make the app pretty simple to use (I could create buttons such as 'download song' and stuff like that).
Although I have already read the documentation explaining the NaCl Technical Overview and some of the Application Structure, I still am not sure as to whether I would be able to make these calls via some C/C++ module or if I would get denied due to security reasons.
To summarize: considering that the user has the needed dependencies in his system (youtube-dl, python, ffmpeg and etc.), is it possible to make calls to third party APIs such as the ones described before via a chrome-app using NaCl ?
Thank you all in advance,
Chrome apps are normally sandboxed.
Less so than extensions - they can reach much more system resources via app APIs.
But still, what you mention is executing libraries / utilities out of browser, and it's not normally allowed.
(P)NaCl is tightly sandboxed in this regard. See this old question, it still applies: you can only use 3rd-party code that compiles into NaCl along with your app, not just link to a library. There are some library ports to NaCl, but it's not automatic.
Normally, a few years back you would use a mechanism like NPAPI to reach out and use a library out of browser. It's deprecated, and won't work anymore. In its place, Chrome offers a pipe-like (through stdio) connection to an external program called Native Messaging. You could use it to perform operations with system-level libraries and tools, but the downside is that you can't bundle the native host with your app, you'll need a separate installer.

How do you use existing C libraries with the Chrome Native Client?

I'm new to Chrome Application development and the Native Client/PNaCL pipeline. I'm a bit confused over the process of using existing C code/libraries in a chrome application. The FAQs and NaCL official docs suggest that using existing code is easy, and one of the advantages of developing Chrome applications.
However, there is also mention of nacl-ports, an official list of ported C libraries to be used in Native Client apps.
If I have some random C library I've used, what is the process for actually using it in my Chrome application? This question feels silly, but I'm quite confused over the process. Do I have to recompile the source with a NaCL compiler? Where in the SDK can I find this?
The FAQ also mentions that things like forks, file i/o is not allowed in the library, so I will have to rewrite any code that does these things, is that correct?
Bottomline: I have an existing C library. What is the process for using it correctly, and making calls to it, in a Chrome application?
The short answer is that you'll have to recompile your library with a NaCl C compiler.
You may want to take a look at naclports: this repository contains ports of many common C libraries to Native Client. You can browse the source more easily here.
As for File I/O, we often suggest using the nacl_io library when porting existing code. This provides a POSIX interface (e.g. fopen/fclose/fread, etc.) Many ported libraries require no modifications when using nacl_io.

Execute Shell Command, and read it's output in Chrome Packaged Apps

I need to develop a Chrome Packaged App that will work as a UI for a console application.
For this I need to find a way to be able to execute that console application inside my Packaged App and read it's outputs.
This is a simple task to do in another platforms such as .NET (which I use nowadays).
I've googled everywhere and searched here on Stackoverflow, but I can't seem to find anything related.
The only interoperability I found is NPAPI, (which is about to be discontinuead). So...
Does anyone have a knowledge about this and could give an example or documentarion on how to achieve it?
Thanks!
There's a feature as part of Chrome extensions that might suit you needs called native messaging.
http://developer.chrome.com/extensions/messaging.html#native-messaging-client

Google Chrome extensions interaction with Native Client Applications

Is it possible to write a Chrome Extension that will interact with a Native Client application?
I use Irssi through the NaCl Secure Shell application, and I would like to write an extension that simply looks for plain URLs and makes them clickable links.
I'm seeing some strange behavior with extensions and NaCl applications, so I wanted to know if this is something that is even possible.
Additionally, if anyone has a more elegant solution to this, please don't hesitate to let me know.
Thanks!
The fact that the NaCl Secure Shell app uses Native Client under the hood shouldn't matter; the page is rendered using standard HTML.
The documentation for Chrome Extensions here says you should be able to inject content scripts into pages with the chrome-extension scheme. I tried it, but it doesn't work. It seems the documentation is incorrect. See http://crbug.com/153245.
As far as I can tell, it is not possible to modify an extension this way. It is possible to communicate with other extensions/apps (see here), but it seems that the application must expose an API to communicate with.
Secure Shell supports clicking links now via Ctrl+Click. it will support OSC+8 in a future release.

Using Java/Python libraries in programming Firefox/Chrome Extensions

I have an idea of studying user behavior on the browser, for which I intend to make a Chrome/Firefox extension to study the behavior dynamically. I have some predefined libraries in Java and Python to analyze the results, which will be impossible to program in plain JavaScript.
Now for my question: is it possible to use third party libraries, especially those of Python or Java like plain function calls?
I have a vague idea about something like Java XPCOM or PyXPCOM for Firefox. However, for a beginner, it all looks so scary. I started making Add-On for Firefox, but got lost somewhere in the huge API.
I found Programming Chrome extensions easier than Firefox, but I couldn't come across something similar to XPCOM in Chrome.
How can I decide which one to go for?
Chrome - seems easy but I am not sure of its power.
Firefox - Seems powerful, but is it really possible to use any Java/Python Library?
Additionally, I came across this link that may be useful: How does someone use thirdparty libraries to be included in Firefox addons/extensions?
But seems like it mostly talks about C++ and XPCOM.
I have a vague idea about something like Java XPCOM or PyXPCOM for Firefox. But for a beginner, it all looks so scary.
I am not a beginner and JavaXPCOM/PyXPCOM are very scary (in addition to being barely maintained). As Firefox goes, it should be much easier to wrap your Java/Python library in an application and run it as an external process: https://developer.mozilla.org/en/XPCOM_Interface_Reference/nsIProcess. Note that you cannot get data back (other than an exit code) so the application should write it to a file that you can then read in your Firefox extension. Not very elegant but it has the advantage of being doable.
As to Chrome, its extensions run in a sandbox and using Java or Python isn't possible. Only option is adding an NPAPI plugin to your extension. It is binary code meaning that it could do anything.
When writing Chrome extensions, you're limited to JavaScript unless you choose to use an NPAPI plugin, which lets you do pretty much anything, but is not recommended.
The other approach you could take is to implement your Java or Python code on the server and make requests from the chrome extension's JavaScript.