Handle file/directory collisions

[bensteinberg]

See #1 for another error; @luminous8 reports

Error: ENOTDIR: not a directory, mkdir 'stripe/archive/data/https:/stripe.com/en-fr/contact'

and reports that

/en-fr is not a folder but a file

I wrote

OK, I'm going to guess that the WARC file contains an exchange like stripe.com/en-fr and another one like stripe.com/en-fr/contact -- so when the extractor gets to the second one and wants to make a directory, it can't because it's already made the file contact. I think we probably need to do something to watch for and handle collisions like this.

[bensteinberg]

@luminous8 you or whoever produced that WACZ should log out of Stripe; the capture contains session cookies. Sorry; I should have been more clear about asking for you to send it privately.

[luminous8]

Thanks for noticing! What's the best way to send you a Wacz file privately?

[bensteinberg]

I typically use Keybase for something like this (where I'm bens), or in some cases, the command-line program magic wormhole (when there's another, relatively private way to convey the command to receive a file, like a Slack DM). Some password managers (Bitwarden, others?) have a way to generate a one-time or time-limited link, but you still have to have a way to convey the link and sometimes a password. Some email systems have a way to send encrypted messages. (My email address is on my profile page.)

I think the service you used generates one-time links? If so, then just sending me a link via email is probably fine; you can be pretty well assured that if I downloaded it, no one else did. I occasionally use a DIY arrangement like this, flahspaper, which is my version of go-flashpaper. If the material is too sensitive to risk someone else downloading it, I'd find a more secure channel than email for conveying the one-time link, maybe Signal (after providing my Signal username via email).

Sorry if that's more than you bargained for! I bet there are other reasonable options.

[luminous8]

Thanks @bensteinberg
I think this one is cookie free (the link expires at the end of the week) https://www.dropbox.com/scl/fi/pvtwpjgt69ml0myyrwy2c/sstripe-1.wacz?rlkey=z0ahhdqrh4hu7eqfhk8gs3nw0&st=zkpb5t3k&dl=0

[bensteinberg]

Thanks, got it, and nice timing -- I just sat down to look at this issue again.