Ran into this bug again while writing tests for this: - #186

Demos: % sqlite-utils insert /tmp/all.db blah /tmp/log.log --convert '[1]' --text Error: Rows must all be dictionaries, got: 1 % sqlite-utils insert /tmp/all.db blah /tmp/log.log --convert '1' --text Error: --convert must return dict or iterator

This check should run inside the .insert_all() method. It should raise a custom exception which the CLI code can then catch and turn into a click error.

Actually no that doesn't work, because this line causes an error before we even get to .insert_all():

https://github.com/simonw/sqlite-utils/blob/7d928f83085fb285f294dbdaeb93fd94a44d5d44/sqlite_utils/cli.py#L1012-L1013

Much better: https://sqlite-utils.datasette.io/en/latest/python-api.html#insert-replacing-data

Closing this - it was something I was curious about, but evidently not curious enough to actually do the work!

This broke on Windows: https://github.com/simonw/sqlite-utils/runs/5056912641 ``` if recreate and os.path.exists(filename_or_conn):

          os.remove(filename_or_conn)

E PermissionError: [WinError 32] The process cannot access the file because it is being used by another process: 'C:\Users\runneradmin\AppData\Local\Temp\pytest-of-runneradmin\pytest-0\test_recreate_False_True_0\data.db' `` I'm going to revert it frommain` for the moment.

OK, I moved all the GIS helpers into db.py as methods on Database and Table, and I put find_spatialite back in utils.py. I deleted gis.py, since there's nothing left it. Docs and tests are updated and passing.

I think this is better.

Works for me. I was just looking at how the FTS extensions work and they're just methods, too. So this can be consistent with that.

I thought about adding these as methods on Database and Table, and I'm back and forth on it for the same reasons you are. It's certainly cleaner, and it's clearer what you're operating on. I could go either way.

I do sort of like having all the Spatialite stuff in its own module, just because it's built around an extension you might not have or want, but I don't know if that's a good reason to have a different API.

You could have init_spatialite add methods to Database and Table, so they're only there if you have Spatialite set up. Is that too clever? It feels too clever.

Yeah that's too clever. You know what? I'm pretty confident we are both massively over-thinking this. We should put the methods on Database and Table! API simplicity and consistency matters more than vague concerns about purity.

from sqlite_utils.utils import find_spatialite is part of the documented API already:

https://sqlite-utils.datasette.io/en/3.22.1/python-api.html#finding-spatialite

To avoid needing to bump the major version number to 4 to indicate a backwards incompatible change, we should keep a from .gis import find_spatialite line at the top of utils.py such that any existing code with that documented import continues to work.

This is fixed now. I had to take out the type annotations for Database and Table to avoid a circular import, but that's fine and may be moot if these become class methods.

Taking part of the conversation from #385 here.

Would sqlite-utils add-geometry-column ... be a good CLI enhancement. for example?

Yes. And also sqlite-utils create-spatial-index would be great to have. My plan would be to add those once the Python API is settled.

I thought about adding these as methods on Database and Table, and I'm back and forth on it for the same reasons you are. It's certainly cleaner, and it's clearer what you're operating on. I could go either way.

I do sort of like having all the Spatialite stuff in its own module, just because it's built around an extension you might not have or want, but I don't know if that's a good reason to have a different API.

You could have init_spatialite add methods to Database and Table, so they're only there if you have Spatialite set up. Is that too clever? It feels too clever.

I'm not sure I like name="geometry" as the default argument to add_geometry_column - mainly because of this example here: python add_geometry_column(db["locations"], "POINT") create_spatial_index(db["locations"], "geometry") I had to go and look at the code to figure out if "POINT" was the name of the column - and I don't like how inconsistent it looks next to the following create_spatial_index() call where you DO need to pass the column name.

What do you think about adding these as methods on the Database class instead? Then you could do:

```python

This is with an optional argument, which if omitted runs find_spatialite() for you:

db.init_spatialite()

Instead of:

init_spatialite(db, find_spatialite()) Likewise, the `add_geometry_column` and `create_spatial_index` methods could live on `Table`:python

Instead of this:

add_geometry_column(db["locations"], "POINT", "geometry") create_spatial_index(db["locations"], "geometry")

Could have this:

db["locations"].add_geometry_column("POINT") db["locations"].create_spatial_index("geometry") `` On the one hand, this is much more consistent with the existingsqlite-utils` Python API.

But on the other hand... this is mixing SpatiaLite functionality directly into the core classes. Is that a good idea, seeing as SpatiaLite is both an optional extension (which can be tricky to install) AND something that has a very different release cadence and quality-of-documentation from SQLite itself?

There's a third option: the SpatiaLite could exist on subclasses of Database and Table - so the above examples would look something like this:

```python from sqlite_utils.gis import SpatiaLiteDatabase

db = SpatiaLiteDatabase("geo.db") db.init_spatialite() db["locations"].add_geometry_column("POINT") db["locations"].create_spatial_index("geometry") ```

On the one hand, this would keep the SpatiaLite-specific stuff out of the core Database/Table classes. But it feels a bit untidy to me, especially since it raises the spectre of someone who was already subclassing Database for some reason now needing to instead subclass SpatiaLiteDatabase (not too keen on that capitalization) - or even (horror) trying to dabble with multiple inheritance, which can only lead to pain.

So I don't have a strong opinion formed on this question yet!

from sqlite_utils.utils import find_spatialite is part of the documented API already:

https://sqlite-utils.datasette.io/en/3.22.1/python-api.html#finding-spatialite

To avoid needing to bump the major version number to 4 to indicate a backwards incompatible change, we should keep a from .gis import find_spatialite line at the top of utils.py such that any existing code with that documented import continues to work.

OK, this change makes a bunch of sense to me - and also raises some interesting questions about future additions to sqlite-utils with regards to SpatiaLite. Would sqlite-utils add-geometry-column ... be a good CLI enhancement. for example?

I see you've already talked about that in #79 - moving this conversation there!

Fixed my spelling. That's a useful thing.

Sorry had missed this - tests should run now.

@simonw Not sure if you've seen this, but any chance you can run the tests?

Filed an issue with Jinja suggesting a documentation update: - https://github.com/pallets/jinja/issues/1578

Codecov Report

Merging #1617 (af293c9) into main (2aa686c) will increase coverage by 0.06%. The diff coverage is 100.00%.

```diff @@ Coverage Diff @@

main #1617 +/-

========================================== + Coverage 92.09% 92.16% +0.06%
========================================== Files 34 34
Lines 4518 4531 +13
========================================== + Hits 4161 4176 +15
+ Misses 357 355 -2
```

| Impacted Files | Coverage Δ | | |---|---|---| | datasette/app.py | 95.37% <100.00%> (ø) | | | datasette/views/table.py | 96.19% <0.00%> (ø) | | | datasette/utils/__init__.py | 94.79% <0.00%> (+<0.01%) | :arrow_up: | | datasette/views/base.py | 95.49% <0.00%> (+0.07%) | :arrow_up: | | datasette/views/special.py | 95.09% <0.00%> (+2.38%) | :arrow_up: |

Continue to review full report at Codecov.

Legend - Click here to learn more Δ = absolute <relative> (impact), ø = not affected, ? = missing data Powered by Codecov. Last update 2aa686c...af293c9. Read the comment docs.

I was hoping to get the test suite running on Windows before merging this PR but that looks like it will be a BIG job, see: - #1627

So I'm going to merge this one as-is for the moment.

I understand the problem now! https://github.com/pallets/jinja/issues/1378#issuecomment-812410922

Jinja template names/paths are not always filesystem paths. So regardless of the OS Jinja always uses forward slashes.

Aha I understand the problem now! https://github.com/pallets/jinja/issues/1378#issuecomment-812410922

Jinja template names/paths are not always filesystem paths. So regardless of the OS Jinja always uses forward slashes.

OK, the tests do NOT pass against Windows! https://github.com/simonw/datasette/runs/5044105941