• Website URL: https://www.ozbargain.com.au/
• Desired ZIM Title: OzBaragin
• Desired ZIM Description: OzBargain is Australia's bargain hunting community.
• Desired ZIM Icon – https://files.delvu.com/images/ozbargain/logo/Square%20Flat.png
• Language (ISO 639-3): eng
• Is this a MediaWiki?: no
• API information: https://www.ozbargain.com.au/wiki/help:api

Make a ZIM of http://api.plos.org/text-and-data-mining/

I propose to close this ticket as wontfix

Original website has been closed. It seems we have now two forks:

• https://en.uncyclopedia.co/
• https://uncyclopedia.ca/wiki/Main_Page

This is definitly the wrong place.

From @ScyllaHide on June 18, 2017 15:58

Proof Wiki seems to have a typical "wikipedia structure" so far i can see. I have used it a few times to learn proof methods and to look up things, but it would be nice to have an offline version as a zim file. Hope that is possible.

"ProofWiki is an online compendium of mathematical proofs! Our goal is the collection, collaboration and classification of mathematical proofs."

Copied from original issue: openzim/mwoffliner#92

https://farm.openzim.org/recipes/proofwiki

From @Popolechien on May 30, 2017 6:11

Following popular request...

Copied from original issue: openzim/mwoffliner#86

Recipe https://farm.openzim.org/recipes/zdoom

@rgaudin Great, thx

This kind of scenario happens pretty often. In that case Zimit fails, like designed.

This is not really nice for the user, and I believe we could/should make a special handling fo this use case.

I propose that if we have the error with foo.bar redirecting to www.foo.bar (or the other way around), then we handle it and scrape the web site properly.

• accepts redirects to same first-level domain
• accepts redirects matching scope

• accepts redirects to same first-level domain
• accepts redirects matching scope

Hello again,

I just realized that indeed, due to my config, only kiwix-serve was installed, the links on the home page were generated but nothing was installed nor working ... Sorry about that false hope ! ;-)

I tried to install other software stacks but I realized that it was not working and would be very hard since it rapidly became obvious that the hotspot functioning relies on a lot of specificities for those additional packages (specific node version for ARM downloaded manually, custom venv for the user 'user' which I previously removed since I believed it was used only for 'clock', ...). I came to the conclusion that (at least for now given the current state of the code), it is not reliable to make a fresh install of all components. Especially given the fact that all new components would involve a new adaptation of the code base. Your previous QEMU-based solution is probably more realistic on this matter, since it would run exactly the same software stack without any need to maintain side-by-side compatibility for hardware and not_hardware compatibility.

Regarding the security issue, I could imagine that it is involved by one of the multiple packages installed.

I don't know how / where we could move forward. Maybe looking at how ideascube is achieving to run its demo?

There seem to be a problem with video handling on the scraper at the moment. I've opened an upstream ticket at https://github.com/webrecorder/browsertrix-crawler/issues/4

@Popolechien we can't do much until this is resolved but your screenshot might indicate an additional problem on vimeo. We might have hit a quota or something.

No, qemu is not a requirement at all. Was just convenient as it allowed creating an almost identical setup.

What you've created sounds great. We can tweak it so it does all this from a kiwix-hotspot config and voila. As you suggested, filtering all those ansible tasks via an hardware tag would allow us to get that into the repo.

Only remaining concern would be the security issue.

@rgaudin Actually the VM is down, I will rebuild it soon and send you the link + the git branch I used to build the VM.

Basically I started a VM in the Cloud, removed everything from the ansible which is "hardware oriented", i.e. wifi, partitions, clock, captive-portal and let everything else install just like it does in QEMU image and let ansible run on the VM itself. To have some data, I hacked a python script which I'm running manually on the VM for now, to download content, place it where it is supposed to be and then let ansible finish the installation.

I don't see any value in using a real QEMU image in my case, just drawbacks (tons on complicated network stuff, additional security risks, ...). After all, I just want to show the interface to kiwix like it is on the hotspot. With what I've built, I even have access to EduPI, KAlite, and stuff like that, even if it is not my goal for now. So if you still want to use the QEMU method, then my attempt will indeed be useless for you.

@rgaudin These are ssh passwords, right?

yes