feat: `VVMDeployer` to deploy older vyper contract

[AlbertoCentonze]

What I did

Added capability to compile and deploy code from older vyper versions using Vyper Version Manager to fetch the required version of the compiler.

How I did it

• Added vvm as a dependency.
• Added a new type of contract deployer VVMDeployer that can deploy code compiled by vvm.
• Created load_vvm, loads_vvm, load_partial_vvm, loads_partial_vvm and exposed them in the public API.

How to verify it

Unit tests are available

Description for the changelog

feat: VVMDeployer to deploy legacy vyper contracts

Cute Animal Picture

[socket-security[bot]]

New and removed dependencies detected. Learn more about Socket for GitHub ↗︎

Package	New capabilities	Transitives	Size	Publisher
pypi/vvm@0.2.1	environment, eval, filesystem, network, shell	0	36.4 kB	charles-cooper, iamdefinitelyahuman

View full report↗︎

[charles-cooper]

with straight regex it could be much faster --

In [4]: r = re.compile(r"\s*#\s*pragma\s+version\s+\d+\.\d+\.\d+")
In [6]: r.match(s)
Out[6]: <re.Match object; span=(0, 23), match='# pragma version 0.3.10'>
In [7]: %timeit r.match(s)
341 ns ± 24.5 ns per loop (mean ± std. dev. of 7 runs, 1,000,000 loops each)

[DanielSchiavini]

with straight regex it could be much faster --

@charles-cooper Isn't the pragma guaranteed to be in the beginning of the file? As soon as we encounter anything but comments the rest of the file can be ignored.

[charles-cooper]

with straight regex it could be much faster --

@charles-cooper Isn't the pragma guaranteed to be in the beginning of the file? As soon as we encounter anything but comments the rest of the file can be ignored.

no it's not, the proper way is to use pre_parse. for example the regex method can produce a different result than the tokenization method. but i think it's ok, it's a fast way to get a heuristic of which compiler to use.

we could do something more failsafe like searching for all the regex matches, and in the rare case that the regex returns more than one match we fall back to the tokenization method. but i think it's overkill until somebody actually runs into the issue