pep8/pycodestyle fixes for hanging indents in Summarization module

[SamriddhiJain]

Updated the code according to pycodestyle. Fixed occurrences of vertical indent to hanging indent.

PR in pycodestyle for detecting hanging indents: #632
References: #1081, #1017

[tmylk]

Thanks. When it is ready in due time, could you please link to a PR in pycodestyle repo that highlights error when indent is not a hanging indent?

[tmylk]

line break should be before binary operator

Please update your pycodestyle code change accordingly.

[SamriddhiJain]

will update that, should I also change line 51 as in this diff, or is this fine (no arguments on first line).

[tmylk]

yes, It's not a function call so no need for hanging indent.

[SamriddhiJain]

updating this part to something like this diff raises this error E128 continuation line under-indented for visual indent and using visual indent is not what we want. So should I keep this as it is?

[SamriddhiJain]

And just to confirm the hanging indents are applicable to multi-line return statements too?

[tmylk]

there should be no E128 error and hanging indent is applicable to all multi-line statements

[SamriddhiJain]

As per the implementation of check for non hanging indents, you can see the diff here.
For hanging indents there are two conditions, no argument in first line and continuation line check. The latter is already implemented in pycodestyle 1. So I have implemented the first check (which occurs in the case of visual indents), when arguments are present on the first line.
Although I have submitted the pull request, but the tests failed because most of the tests and code itself are in vertical indents format, that is why I think its difficult to merge these changes in their branch, unless the tests are also carefully changed.

[tmylk]

For pycodestyle PR it should be an optional new check which is off by default. That will cause no issues with existing tests.

The current implementation has caused a fase positive on bm25.py line 51 so more work is needed.

[tmylk]

Could you please link to the PR in the pycodestyle repo?

[SamriddhiJain]

@tmylk Link for PR in pycodestyle: 1.

I tried to keep this test in default ignore by editing in pycodestyle.py, setup.cfg, test_api.py, but it still runs. Is there any documentation which can be used to implement this?

[tmylk]

Thanks for the fixes and improving pycodestyle. We are now closer to achieving #965.

[piskvorky]

@tmylk that's cool, but in the meanwhile, please inspect & fix code style manually, before merging PRs.

[tmylk]

@piskvorky The hanging indent requirement has made it harder to automate the style check.

1. Gensim is the only Python open source project that enforces the hanging indent. It's not a trivial common practice. A specific extension is in the process of being added to pycodestyle as described in this PR.

2. Even in gensim codebase there is code from years ago that doesn't comply with hanging indent. (237 lines)

The next steps to automation are merging to pycodestyle and fixing the existing violations.

To address your point above - without automation we have to resort to the manual process that is bound to let mistakes slip through as shown in point 2 above.

[piskvorky]

Gensim didn't have any hanging indent requirement until recently. Its style was more casual, because there weren't as many contributors. We could rely more on common sense.

Fixing existing violations sounds great -- let's start with a consistent, clean slate.

And of course, don't merge in anything new that moves us farther away from that.