i5ik

:classical_building: 22120

:classical_building: 22120 - NodeJS product to self-host the Internet with an Offline Archive. Like binaries? https://github.com/dosyago/22120/releases Similar to ArchiveBox, SingleFile and WebMemex, but gooderer. Works well with WorldBrain/Memex to give you full-text search.
Under Other
By i5ik

nodejs archive web-archive archiver chrome-headless devtools-protocol

:classical_building: 22120

:classical_building: - An archivist browser controller that caches everything you browse, a library server with full text search to serve your archive.


If you use or like this, don't forget to show your appreciation by starring this repo, or following me 😹


News - 22120 plus interview featured in Console - the open source newsletter

License

This is released into the public domain.


Top

About

This project literally makes your web browsing available COMPLETELY OFFLINE. Your browser does not even know the difference. It's literally that amazing. Yes.


Save your browsing, then switch off the net and go to http://localhost:22120 and switch mode to serve then browse what you browsed before. It all still works.


warning: if you have Chrome open, it will close it automatically when you open 22120, and relaunch it. You may lose any unsaved work.


Top

Get 22120

3 ways to get it:



  1. Get binary from the releases page., or

  2. Run with npx: npx [email protected], or

  3. Clone this repo and run as a Node.JS app: npm i && npm start


Also, coming soon is a Chrome Extension.


Top

Using
Pick save mode or serve mode

Go to http://localhost:22120 in your browser,
and follow the instructions.


Top

Exploring your 22120 archive

Archive will be located in 22120-arc/public/library*


But it's not public, don't worry!


You can also check out the archive index, for a listing of every title in the archive. The index is accessible from the control page, which by default is at http://localhost:22120 (unless you changed the port).


*Note:22120-arc is the archive root of a single archive, and by defualt it is placed in your home directory. But you can change the parent directory for 22120-arc to have multiple archvies.


Top

Format

The archive format is:


22120-arc/public/library/<resource-origin>/<path-hash>.json


Inside the JSON file, is a JSON object with headers, response code, key and a base 64 encoded response body.


Top

Why not WARC (or another format like MHTML) ?

The case for the 22120 format.


Other formats (like MHTML and SingleFile) save translations of the resources you archive. They create modifications, such as altering the internal structure of the HTML, changing hyperlinks and URLs into "flat" embedded data URIs, or local references, and require other "hacks* in order to save a "perceptually similar" copy of the archived resource.


22120 throws all that out, and calls rubbish on it. 22120 saves a verbatim high-fidelity copy of the resources your archive. It does not alter their internal structure in any way. Instead it records each resource in its own metadata file. In that way it is more similar to HAR and WARC, but still radically different. Compared to WARC and HAR, our format is radically simplified, throwing out most of the metadata information and unnecessary fields these formats collect.


Why?


At 22120, we believe in the resources and in verbatim copies. We don't annoint ourselves as all knowing enough to modify the resource source of truth before we archive it, just so it can "fit the format we choose. We don't believe we need to decorate with obtuse and superfluous metadata. We don't believe we should be modifying or altering resources we archive. We belive we should save them exactly as they were presented. We believe in simplicity. We believe the format should fit (or at least accommodate, and be suited to) the resource, not the other way around. We don't believe in conflating metadata with content*; so we separate them. We believe separating metadata and content, and keeping the content pure and altered throughout the archiving process is not only the right thing to do, it simplifies every part of the audit trail, because we know that the modifications between archived copies of a resource of due to changes to the resources themselves, not artefacts of the format or archiving process.


Both SingleFile and MHTML require mutilatious modifications of the resources so that the resources can be "forced to fit" the format. At 22120, we believe this is not required (and in any case should never be performed). We see it as akin to lopping off the arms of a Roman statue in order to fit it into a presentation and security display box. How ridiculous! The web may be a more "pliable" medium but that does not mean we should treat it without respect for its inherent content.


Why is changing the internal structure of resources so bad?


In our view, the internal structure of the resource as presented, is the cannon. Internal structure is not just substitutable "presentation" - no, in fact it encodes vital semantic information such as hyperlink relationships, source choices, and the "strokes" of the resource author as they create their content, even if it's mediated through a web server or web framework.


Why else is 22120 the obvious and natural choice?


22120 also archives resources exactly as they are sent to the browser. It runs connected to a browser, and so is able to access the full-scope of resources (with, currently, the exception of video, audio and websockets, for now) in their highest fidelity, without modification, that the browser receives and is able to archive them in the exact format presented to the user. Many resources undergo presentational and processing changes before they are presented to the user. This is the ubiquitous, "web app", where client-side scripting enabled by JavaScript, creates resources and resource views on the fly. These sorts of "hyper resources" or "realtime" or "client side" resources, prevalent in SPAs, are not able to be archived, at least not utilizing the normal archive flow, within traditional wget-based archiving tools.


In short, the web is an online medium, and it should be archived and presented in the same fashion. 22120 archives content exactly as it is received and presented by a browser, and it also replays that content exactly as if the resource were being taken from online. Yes, it requires a browser for this exercise, but that browser need not be connected to the internet. It is only natural that viewing a web resource requires the web browser. And because of 22120 the browser doesn't know the difference! Resources presented to the browser form a remote web site, and resources given to the browser by 22120, are seen by the browser as exactly the same. This ensures that the people viewing the archive are also not let down and are given the change to have the exact same experience as if they were viewing the resource online.


Top

How it works

Uses DevTools protocol to intercept all requests, and caches responses against a key made of (METHOD and URL) onto disk. It also maintains an in memory set of keys so it knows what it has on disk.


Top

FAQ
Do I need to download something?

Yes. But....If you like 22120, you might love the clientless hosted version coming in future. You'll be able to build your archives online from any device, without any download, then download the archive to run on any desktop. You'll need to sign up to use it, but you can jump the queue and sign up today.


Can I use this with a browser that's not Chrome-based?

No.


Top

How does this interact with Ad blockers?

Interacts just fine. The things ad blockers stop will not be archived.


Top

How secure is running chrome with remote debugging port open?

Seems pretty secure. It's not exposed to the public internet, and pages you load that tried to use it cannot use the protocol for anything (except to open a new tab, which they can do anyway). It seems there's a potential risk from malicious browser extensions, but we'd need to confirm that and if that's so, work out blocks. See this useful security related post for some info.


Top

Is this free?

Yes this is totally free to download and use. It's also open source (under AGPL-3.0) so do what you want with it. For more information about licensing, see the license section.


Top

What if it can't find my chrome?

See this useful issue.


Top

What's the roadmap?


Top

What about streaming content?

The following are probably hard (and I haven't thought much about):



Probably some way to do this tho.


Top

Can I black list domains to not archive them?

Yes! Put any domains into 22120-arc/no.json*, eg:


json
[
"*.horribleplantations.com",
"*.cactusfernfurniture.com",
"*.gustymeadows.com",
"*.nytimes.com",
"*.cnn.co?"
]


Will not cache any resource with a host matching those. Wildcards:



*Note: the no file is per-archive. 22120-arc is the archive root of a single archive, and by defualt it is placed in your home directory. But you can change the parent directory for 22120-arc to have multiple archvies, and each archive requires its own no file, if you want a blacklist in that archive.


Top

Is there a DEBUG mode for troubleshooting?

Yes, just make sure you set an environment variable called DEBUG_22120 to anything non empty.


So for example in posix systems:


bash
export DEBUG_22120=True


Top

Can I version the archive?

Yes! But you need to use git for versioning. Just initiate a git repo in your archive repository. And when you want to save a snapshot, make a new git commit.


Top

Can I change the archive path?

Yes, there's a control for changing the archive path in the control page: http://localhost:22120


Top

Can I change this other thing?

There's a few command line arguments. You'll see the format printed as the first printed line when you start the program.


For other things you can examine the source code.


Top