Break out CLI support from the main library

[0xDEC0DE]

Move the CLI tool into a cli.py, so that consumers can import the
library without needing to pull in Click, etc.

Also move the CLI-specific requirements into an extras section, and
regenerate/bump requirements

Fixes: Issue #20

---

This change is 

[bigjools]

Reviewed 13 of 13 files at r1, all commit messages.
Reviewable status: all files reviewed, 4 unresolved discussions (waiting on @0xDEC0DE and @juledwar)

---

requirements/requirements.txt line 321 at r1 (raw file):

    --hash=sha256:4a7317d5e3b17a3dccb6a8cfe67dab65b20551404c52c8ed41279fa4f0cb4cda \
    --hash=sha256:d1377122a5a00e2f940ee482999518efe16d745d423a670c27773dfbc3c9a7d9
    # via stevedore

Grrr. dogpile-cache pulling in stuff that uses PBR.

---

soufi/cli.py line 328 at r1 (raw file):

if __name__ == "__main__":
    main()

Did you do a git move on this file? I can't remember if it tracks those properly or not, and we lose the blame history.

---

soufi/finders/alpine.py line 249 at r1 (raw file):

        tmp_file_name = pathlib.Path(temp_dir) / name
        with closing(
            # B310 restricts permitted schemes, but we only call with ftp here.

S310 now

---

tools/compile-requirements line 43 at r1 (raw file):

compile ${REQ_DIR}/requirements-test.txt --extra test
compile ${REQ_DIR}/requirements-bootstrap.txt --extra bootstrap
compile ${REQ_DIR}/requirements-cli.txt --extra cli

You need to update the README since it references the CLI and the installation instruction is now incorrect.

[0xDEC0DE]

soufi/cli.py line 328 at r1 (raw file):

Previously, bigjools (Julian Edwards) wrote…

Did you do a git move on this file? I can't remember if it tracks those properly or not, and we lose the blame history.

I did, but I also dropped in an empty __init__.py, which apparently confuses and frightens git.

[0xDEC0DE]

soufi/cli.py line 328 at r1 (raw file):

Previously, 0xDEC0DE (Nicolas Simonds) wrote…

I did, but I also dropped in an empty __init__.py, which apparently confuses and frightens git.

I suppose the obvious tacky workaround is to move in one commit, and add in another, and accept that there will be a commit in the stream that "doesn't work"

[0xDEC0DE]

soufi/finders/alpine.py line 249 at r1 (raw file):

Previously, bigjools (Julian Edwards) wrote…

S310 now

Whoops

[0xDEC0DE]

tools/compile-requirements line 43 at r1 (raw file):

Previously, bigjools (Julian Edwards) wrote…

You need to update the README since it references the CLI and the installation instruction is now incorrect.

Whoopsie!

[0xDEC0DE]

Reviewable status: 12 of 14 files reviewed, 4 unresolved discussions (waiting on @bigjools and @juledwar)

---

requirements/requirements.txt line 321 at r1 (raw file):

Previously, bigjools (Julian Edwards) wrote…

Grrr. dogpile-cache pulling in stuff that uses PBR.

Done.

[bigjools]

Reviewed 2 of 2 files at r2, all commit messages.
Reviewable status: complete! all files reviewed, all discussions resolved (waiting on @juledwar)