Infer column type two #135

mathemancer · 2021-05-19T13:58:34Z

Fixes #92

This adds a function to infer the type of the column by attempting to alter the column to various types in a logical order.

Technical details

This has the side effect of losing data in the case where the column is altered to a type via some non-injective mapping. For example, altering the strings '0123' and '123' to a numeric will result in the value 123 in both cases. It's not possible to recover the original strings from the resulting value.

We'll add mechanisms to recover the original version of an altered column in future PRs.

Checklist

My pull request has a descriptive title (not a vague title like Update index.md).
My pull request targets the master branch of the repository
My commit messages follow best practices.
My code follows the established code style of the repository.
I added tests for the changes I made (if applicable).
I added or updated documentation (if applicable).
I tried running the project locally and verified that there are no
visible errors.

Developer Certificate of Origin

Developer Certificate of Origin
Version 1.1

Copyright (C) 2004, 2006 The Linux Foundation and its contributors.
1 Letterman Drive
Suite D4700
San Francisco, CA, 94129

Everyone is permitted to copy and distribute verbatim copies of this
license document, but changing it is not allowed.


Developer's Certificate of Origin 1.1

By making a contribution to this project, I certify that:

(a) The contribution was created in whole or in part by me and I
    have the right to submit it under the open source license
    indicated in the file; or

(b) The contribution is based upon previous work that, to the best
    of my knowledge, is covered under an appropriate open source
    license and I have the right under that license to submit that
    work with modifications, whether created in whole or in part
    by me, under the same open source license (unless I am
    permitted to submit under a different license), as indicated
    in the file; or

(c) The contribution was provided directly to me by some other
    person who certified (a), (b) or (c) and I have not modified
    it.

(d) I understand and agree that this project and the contribution
    are public and that a record of the contribution (including all
    personal information I submit with it, including my sign-off) is
    maintained indefinitely and may be redistributed consistent with
    this project or the open source license(s) involved.

The merge with master was too hairy, started a new branch.

kgodey

Approving to unblock, please make requested changes before merging. Looks good overall!

kgodey · 2021-05-19T22:48:05Z

db/tests/types/test_inference.py

+
+
+@pytest.fixture
+def temporary_testing_schema(engine_with_types):


You're using this fixture in multiple files, I recommend putting it in conftest.py in the root of this folder so that you can reuse it.

Same with other reused fixtures.

My reasoning was that these fixtures use code from mathesar that is tested by test files with no access to these fixtures. (This makes sure we're not using fixtures in a way that invalidates the tests that test the code they use. Do you know of a way to separate those out (or make it difficult/obvious when you shouldn't use a given fixture for testing? In particular, I'd want to find some way to guarantee that these fixtures won't be used for testing the db/types/install.py file at some point in the future.

You could organize the files differently so that any tests that can use these fixtures are in a single directory with the appropriate conftest.py and files that shouldn't use these fixtures are in a different directory.

that seems like the obvious solution to me but I might be missing something.

We could also put these fixtures in a non-default file (i.e., not a conftest.py file), then import it. It's kind of non-standard, but I think it'd be less confusing than partially mirroring the directory structure of the tested files with the structure of the test files.

Actually, I think that's probably my favorite solution.

For that kind of reorganization, I think it'd merit its own PR to document what was done and why with the restructuring. @kgodey, what do you think?

@mathemancer I think importing a non-standard file makes sense, and doing it in a separate PR also makes sense. I was originally thinking you'd reorganize the real files as well, but I like your solution too.

I'll double-check that the organization of the real files makes sense and adjust accordingly.

kgodey · 2021-05-19T22:48:50Z

db/tests/types/test_inference.py

+    TEST_TABLE = "test_table"
+    TEST_COLUMN = "test_column"
+    metadata = MetaData(bind=engine)
+    print(value_list)


Is this a debug print?

oops, good eye.

kgodey · 2021-05-19T22:50:19Z

db/types/inference.py

+            )
+            break
+        except Exception:
+            logger.info(


Shouldn't this be logger.error?

This exception is an expected part of the control flow (it's how we know when a given type is inappropriate for a given column). I think info is reasonable for that. I do think I should remove the spot where I set the log level to INFO in this file, though (it should be set by the importer if they want a non-default level). What do you think?

Can you catch the specific exception you're expecting and add a comment explaining it's the expected flow?

I agree re: removing log level in the file.

Unfortunately, there's a bunch of different exceptions based on which type was casting to which other. I can try to figure out if they're sub-classes of something appropriate (I think python will catch the parent class in that case, I'll have to check)

Okay, we'll use sqlalchemy.exc.DatabaseError. It's not super-precise, but it at least restricts to "you asked to do something and the DB said no".

Looks good, thanks!

kgodey · 2021-05-20T10:00:42Z

db/types/inference.py

+            )
+            break
+        except Exception:
+            logger.info(


Looks good, thanks!

mathemancer added 4 commits May 19, 2021 18:07

bring over inference code from old branch

d8957ea

The merge with master was too hairy, started a new branch.

allow '0' and '1' strings to cast to boolean

1bcdeb8

let numerics be inferred to be boolean

a9cc190

add tests to make sure non-0,1 numerics stay numeric

82a9db8

mathemancer requested review from a team, kgodey and pavish May 19, 2021 13:58

github-actions bot requested a review from ghislaineguerin May 19, 2021 13:58

kgodey approved these changes May 19, 2021

View reviewed changes

mathemancer added 2 commits May 20, 2021 17:24

clean up logging setup, remove debug print

4f1a3ad

use more specific error for control flow

851009c

kgodey approved these changes May 20, 2021

View reviewed changes

db/types/inference.py

)

break

except Exception:

logger.info(

Copy link

Contributor

kgodey May 20, 2021

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Looks good, thanks!

mathemancer merged commit 5e8b206 into master May 20, 2021

mathemancer deleted the infer_column_type_two branch May 20, 2021 10:05

mathemancer mentioned this pull request May 20, 2021

Move repeated pytest fixtures to importable file #136

Closed

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Infer column type two #135

Infer column type two #135

mathemancer commented May 19, 2021

kgodey left a comment •

edited

Loading

kgodey May 19, 2021

kgodey May 19, 2021

mathemancer May 20, 2021 •

edited

Loading

kgodey May 20, 2021

kgodey May 20, 2021

mathemancer May 20, 2021

mathemancer May 20, 2021

mathemancer May 20, 2021

kgodey May 20, 2021

mathemancer May 20, 2021

kgodey May 19, 2021

mathemancer May 20, 2021

kgodey May 19, 2021

mathemancer May 20, 2021

kgodey May 20, 2021

mathemancer May 20, 2021

mathemancer May 20, 2021

kgodey May 20, 2021

kgodey May 20, 2021



		@pytest.fixture
		def temporary_testing_schema(engine_with_types):

Infer column type two #135

Infer column type two #135

Conversation

mathemancer commented May 19, 2021

Checklist

Developer Certificate of Origin

kgodey left a comment • edited Loading

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

mathemancer May 20, 2021 • edited Loading

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

kgodey left a comment •

edited

Loading

mathemancer May 20, 2021 •

edited

Loading