Ensure import lines with ';' are properly split and auto corrected

[superDross]

Resolves #176

Does so by import lines with semicolon separators are split into new lines before being processed.
e.g. import pdb; pdb.set_trace() would become import pdb\npdb.set_trace().

Checklist

• [ x] Add test cases to all the changes you introduce
• [ x] Update the documentation for the changes

[coveralls]

Pull Request Test Coverage Report for Build 1714771961

• 12 of 12 (100.0%) changed or added relevant lines in 1 file are covered.
• No unchanged relevant lines lost coverage.
• Overall coverage increased (+0.1%) to 96.104%

---

Totals
Change from base Build 1712054575:	0.1%
Covered Lines:	296
Relevant Lines:	308

---

💛 - Coveralls

[lyz-code]

Hi @superDross thank you again for opening the issue and the PR. Your implementation works perfectly, but I wonder if we could avoid another iteration over the code lines.

The import ... ; ... structure is not often used in python scripts despite the use of the breaking point you mentioned in #176. The way I define breakpoints is with __import__("pdb").set_trace() # XXX BREAKPOINT which doesn't get changed by autoimport.

My concerns are that with the current solution we need to iterate over all the lines of the code to discern if there are ; in import statements, which slows down the run on autoimport for all the source codes for a corner case. I know is what we do in every filter, but the less we have, the better.

I agree that your breaking point snippet is the most popular one, so it make sense that we handle it. Could you please take a look if you can implement the logic avoiding yet another iteration? Maybe tweaking _move_imports_to_top by sending the left operand of the ; to the self.imports.append and changing the line content to the left of the operand.

[superDross]

I agree this solution is supoptimal, in fact I am not terribly proud of the solution I proposed here. I simply wanted a quick and dirty solution as this bug was really irritating me when using it locally and thought I should share it.

I will work on creating a better solution at some point. I don't have an ETA for an update to the PR though, just depends when I get some free time.

Thanks for your input.

[superDross]

I have updated the PR with a more optimal solution.

It seems pushing this branch up did not trigger the pipeline.

[lyz-code]

Now the solution looks perfect.

I'm not sure why the CI didn't trigger, it will on the merge commit,

Thanks for your contribution :)