ENH : Allow to_sql to recognize single sql type #11886

[RahulHP]

Continued in #13252

---

This solves #11886
It checks whether the passed dtype variable is a dictionary.
If not, it creates a new dictionary with keys as the columns of the dataframe.
It then passes this dictionary to the pandasSQL_builder.

[jreback]

tests!

[TomAugspurger]

I think flake8 will complain about this line. Need to get rid of the spaces around the curly braces.
Oh, also space after the common above in the isinstance check

[max-sixty]

Also is_dictlike (which isn't a perfect function but at least improvements will be encapsulated)

[RahulHP]

Update: Writing the tests has been more difficult that I thought it would be. I am currently working on it though.

[jorisvandenbossche]

@RahulHP For the test, best have a look at the existing test_dtype test (https://github.com/RahulHP/pandas/blob/to_sql-dictionary/pandas/io/tests/test_sql.py#L1502). You can see there how it is checked that the dtype is actually used in the database.
And maybe you can just add your test to that function.

[RahulHP]

Update: I added the new test and it passes the Travis build.

[jorisvandenbossche]

missing space before 'If'

[RahulHP]

Edits done

[jreback]

can you rebase / squash

[RahulHP]

Done, I followed the Contributing guide as best as I could but I think I may have messed it up.

[jorisvandenbossche]

@RahulHP There indeed slipped in some changes from other PRs. Normally, by doing exactly this, it should be solved:

git checkout to_sql-dictionary
git fetch upstream
git rebase upstream/master
git push -f origin to_sql-dictionary

In the rebase step, you will possibly need to resolve some conflicts

[jreback]

can you rebase/update

[RahulHP]

Sorry, I'll work on this in a few days.

[jreback]

pls rebase / update

[RahulHP]

I have created a new pull request here #13252