PEP8-format generated revisions

[sqlalchemy-bot]

Migrated issue, originally created by Adrian (@thiefmaster)

Even though they usually need to be modified a bit, it would be nice if the generated statements were PEP8-valid by default (besides maybe line-length..)

[sqlalchemy-bot]

Michael Bayer (@zzzeek) wrote:

see also https://bitbucket.org/zzzeek/alembic/pull-request/11/changed-autogenerated-code-so-it-fits-80/diff. line length is very tough. E2 rules should be working as well as E11*, but for E12, E13, E3, E5, I'm not optimistic this is ever going to be "fixed". I know there are code formatters but those are going to just create more headaches.

[sqlalchemy-bot]

Changes by Michael Bayer (@zzzeek):

• removed labels: low priority

[sqlalchemy-bot]

Changes by Michael Bayer (@zzzeek):

• added labels: autogenerate - rendering

[sqlalchemy-bot]

jmagnusson (@jmagnusson) wrote:

What about having one argument per line of a function call in the generated code? Should fix most problems of non PEP-8 compliant code. Example with what I mean:

op.alter_column(
        'blocki18n', 'links_to_text',
        existing_type=sa.TEXT(),
        server_default='',
        nullable=False
    )

[sqlalchemy-bot]

Adrian (@thiefmaster) wrote:

Exactly. The most common indentation in that case is this though:

op.alter_column(
    'blocki18n',
    'links_to_text',
    existing_type=sa.TEXT(),
    server_default='',
    nullable=False
)

[sqlalchemy-bot]

jmagnusson (@jmagnusson) wrote:

Agreed. Just missed that ;-)

[sqlalchemy-bot]

Michael Bayer (@zzzeek) wrote:

Issue #393 was marked as a duplicate of this issue.

[sqlalchemy-bot]

ukrutt (@ukrutt) wrote:

UPDATE: oops, sorry, saw now that this code has changed in script.py.mako -- however when I tested this I had failed to update my own old mako file. Please disregard...

** please disregard comments below... **

It looks like a fair amount of work was done on this leading up to https://bitbucket.org/zzzeek/alembic/commits/tag/rel_0_8_9. I still get a few pep8 warnings though,

project/alembic/versions/175a74063cca_test.py:15:1: E402 module level import not at top of file
project/alembic/versions/175a74063cca_test.py:16:1: E402 module level import not at top of file

These are due to this code in the top of the generated file:

# revision identifiers, used by Alembic.
revision = '175a74063cca'
down_revision = '99c454c52997'
branch_labels = None
depends_on = None

from alembic import op
import sqlalchemy as sa

(specifically, the two imports.) It seems like this is a fairly straightforward thing to fix; however I'm new to this code in particular (and autogenerated code in general...)

[sqlalchemy-bot]

Wes Downey wrote:

Here is another example:

def upgrade():
    # ### commands auto generated by Alembic - please adjust! ###
    op.create_table('table_name',
    sa.Column('id', sa.Integer(), nullable=False),

This fails linting for these two rules:

• E128 continuation line under-indented for visual indent
• E122 continuation line missing indentation or outdented

The passing format would be like this:

def upgrade():
    # ### commands auto generated by Alembic - please adjust! ###
    op.create_table(
        'table_name',
        sa.Column('id', sa.Integer(), nullable=False),

[sqlalchemy-bot]

Mike Waites (@mikeywaites) wrote:

Hey

I've recently started helping Mike out with sqlalchemy and alembic. On first inspection I agree with @ThiefMaster

op.alter_column(
    'blocki18n',
    'links_to_text',
    existing_type=sa.TEXT(),
    server_default='',
    nullable=False
)

Would solve the vast majority of peps -- That said there might be things I'm over looking. I'll get started on this on Tuesday and try and get a couple of "worst case scenario" tests cases together that highlight the issues. From there, the rendering of op instructions is fairly centralised so it should be fairly straight forward to prove out.

The work here will certainly be trying to produce test cases that produce all of the formatting issues. Perhaps it would make sense to deal with the really obvious ones first rather than going to deep.

It would be a great help if anyone interested in this fix could list any pep errors they have encountered not already listed above.

I'll report back here with my findings. 🤘

[sqlalchemy-bot]

Neil Basu (@nbasu02) wrote:

@mikeywaites As a user what would also be a huge help to me is if we add a trailing comma. i.e.

op.alter_column(
    'blocki18n',
    'links_to_text',
    existing_type=sa.TEXT(),
    server_default='',
    nullable=False,
)