ipfs files add

[coryschwartz]

fixes: #8504

adds command ipfs files add using mostly the same parameters as in ipfs add

This is functionally equivilent to running ipfs add followed by ifps files cp with some slightly different defautls.

• IPFS pinning is not enabled by default, although it is an option.

• Keeping with the pattern established for other files command, -p is for parent creation. The progress bar is moved to -P (capital) instead. Seems unfortunate; if this doesnt make sense, let me know what would be better.

• There is no wrap option. Although, I could be convinced that it should be included here. When adding a directory recursively, files are "wrapped" implicitly by the MFS directory that contains them.

• Following the convention from cp command, if the destination ends in a forward slash, files are added under an MFS directory such that many files can be added like this:

ipfs files add /mfs/path/ file1 file2 file3...

The two encantations, although similar, result in a different dag for the containing directory. This is a subtle difference that might be confusing, maybe there is a better way to go about it.

Recursive adds

Ξ ~ → ipfs files add -rp /d1 website 
added QmeEte3ZFyKA4yhiQTnBP2x2XTxdDfZzivUiX86xqGT42X website/image.jpg
added QmdXozHCMxfsps9exhLPCzoJq2VeVRkZtuD6KLev7JhfSB website/index.html
added QmfGCw7KQkYGnVbz2h7d3T4TGCpDNza32NqUyoV8zqhWFr website

Adding several files with implicit directory creation

Ξ ~ → ipfs files add -p /d2/ website/*
added QmeEte3ZFyKA4yhiQTnBP2x2XTxdDfZzivUiX86xqGT42X image.jpg
added QmdXozHCMxfsps9exhLPCzoJq2VeVRkZtuD6KLev7JhfSB index.html

Results:

Ξ ~ → ipfs files ls -l /
d1/	QmfGCw7KQkYGnVbz2h7d3T4TGCpDNza32NqUyoV8zqhWFr	0
d2/	bafybeih3om227lpu74wyc7zl3f4sddq4atwyv4b3abq6oowgoynkdyygce	0
Ξ ~ → ipfs files ls -l /d1
image.jpg	QmeEte3ZFyKA4yhiQTnBP2x2XTxdDfZzivUiX86xqGT42X	5
index.html	QmdXozHCMxfsps9exhLPCzoJq2VeVRkZtuD6KLev7JhfSB	48
Ξ ~ → ipfs files ls -l /d2
image.jpg	QmeEte3ZFyKA4yhiQTnBP2x2XTxdDfZzivUiX86xqGT42X	5
index.html	QmdXozHCMxfsps9exhLPCzoJq2VeVRkZtuD6KLev7JhfSB	48

[lidel]

Thank you @coryschwartz, amazing to see this happening so soon! ❤️

I took it for a spin and works as expected. Subjectively, it feels like it was always part of the ipfs files commands, which is a good sign in my book.

Given that the feature request was written by me, asked for more eyes to review this, mainly to ensure ergonomics feel intuitive to the majority of users from different backgrounds.

For now, only two actionable asks:

• rebase on latest master – this should fix interop tests
• add a basic sharness test
  • for now just add a file via ipfs add and ipfs files add with default parameters and confirm they (1) produce the same CID (2) the low-level pin is not created when ipfs files add is used

[BigLep]

@lidel : it looks like your asks have been addressed. Can you retract your "request changes" please?

@schomatis : can you please give this a once over and then merge? @lidel wanted another opinion on the ergonomics.

[schomatis]

This looks like a giant code duplication. If it does what was requested I'm not standing in the way of landing this but I can't approve it, especially since we're duplicating one of the ugliest pieces of the CLI code: ipfs add.

[BigLep]

@coryschwartz : I'm assuming you won't have bandwidth to drive this forward. Totally understood. It helped exposed some issues for doing this write. I'm going to close this for now so it's clear in the issue that someone else can take a crack at this.