That passed on 3.8 but should have failed: https://github.com/simonw/datasette/actions/runs/6266341481/job/17017099801 - the "Test DATASETTE_LOAD_PLUGINS" test shows errors but did not fail the CI run.

Originally posted by @simonw in https://github.com/simonw/datasette/issues/2057#issuecomment-1730201226

There's a long justification for why this might be needed here: - https://github.com/simonw/datasette-auth-tokens/issues/10#issuecomment-1701820001

Short version: it would be neat if it was possible to stash some data on the request object such that a later plugin/middleware-type-thing could use that to influence the final returned response - similar to the kinds of things you can do with Django middleware.

The asgi_wrapper() mechanism doesn't have access to the request or response objects - it gets scope and can mess around with receive and send, but those are pretty low-level primitives.

Since Datasette has well-defined request and response objects now it might be nice to have a middleware layer that can manipulate those directly.

I just spotted a bug when using https://datasette.io/plugins/datasette-configure-fts and https://datasette.io/plugins/datasette-edit-schema at the same time.

Steps to reproduce:

• Configure FTS for a table, then run a test search
• Edit the schema for that table and change the order of columns
• Run the test search again

I got the wrong search results, which I think is because the _fts table pointed to the first table by rowid but those rowid values were entirely rewritten as a consequence of running table.transform() on the table.

Reconfiguring FTS on the table fixed the problem.

I think table.transform() should be able to preserve rowid values.

In building Datasette Cloud we realized that a bunch of the features we are building need a way of resolving an actor ID to the actual actor, in order to display something more interesting than just an integer ID.

Social plugins in particular need this - comments by X, CSV uploaded by X, that kind of thing.

I think the solution is a new plugin hook: actors_from_ids(datasette, ids) which can return a list of actor dictionaries.

The default implementation can return [{"id": "..."}] for the IDs passed to it.

Pluggy has a firstresult=True option which is relevant here, since this is the first plugin hook we will have implemented where only one plugin should provide an answer.

In thinking about the best way to implement https://github.com/simonw/datasette-auth-passwords/issues/6 (SQL-backed user accounts for datasette-auth-passwords) I realized that there are a few different use-cases where a plugin might want to store data that isn't visible to regular Datasette users:

• Storing password hashes
• Storing API tokens
• Storing secrets that are used for data import integrations (secrets for talking to the Twitter API for example)

Idea: allow one or more private database files to be attached to Datasette, something like this:

datasette github.db linkedin.db -s secrets.db -m metadata.yml

The secrets.db file would not be visible using any of the Datasette's usual interface or API routes - but plugins would be able to run queries against it.

So datasette-auth-passwords might then be configured like this:

yaml plugins: datasette-auth-passwords: database: secrets sql: "select password_hash from passwords where username = :username" The plugin could even refuse to operate against a database that hadn't been loaded as a secret database.

I use datasette in an iframe inside another HTML file that contains other ways to represent my data (mostly leaflets maps built with R on summarized data), and the datasette iframe is a tab in that page.

This particular use prevents users to access the full URLs of their datasette views and queries, which is a shame because the way datasette handles URLs to make every view or query easy to share is awesome. I know how to get the URL from the context menu of my browser, but I don't think many visitors would do it or even notice that datasette uses permalinks for pretty much every action they do. Would it be possible to add a "Share link" button to the interface, either in datasette itself or in a plugin?

I'm interested in making a plugin that saves every query that gets run to a table in the database. (I know about datasette-query-history but thought it would be good to have a server-side option.)

As far as I can tell reading the docs, there isn't really a hook setup to allow this.

Maybe I could hack it with some of the hooks that are passed requests, but that doesn't seem good.

I'm a little surprised this isn't possible, so I thought I would open an issue and see if that's a deeply considered decision or just "haven't needed it yet." I'm potentially interested in implementing the hook if the latter.

I find myself wanting two new features in sqlite-utils: - The ability to have the new transformed table set to a specific name, while keeping the old table around - The ability to rename a table (sqlite-utils doesn't have a table rename function at all right now)

Originally posted by @simonw in https://github.com/simonw/llm/issues/65#issuecomment-1618375042

e.g. in this section:

https://sqlite-utils.datasette.io/en/stable/cli.html#running-queries-directly-against-csv-or-json

The little copy button will also copy the $ which breaks the examples when copied.

https://github.com/Textualize/trogon

Notes can live in folders. These relationships should be exported too.

See this issue: - https://github.com/simonw/datasette.io/issues/141

Reported as a bug in #1098 because they caused 500 errors - but it would be even better if Datasette could hyperlink to related rows via compound foreign keys.

Hi there! I saw your presentation at Boston Python. I'm already a light user of Datasette (thank you!), but wasn't aware of this project.

I've been working on a "pull request dashboard" to get a comprehensive view of the state of open PR's, esp. related to reviews (i.e., pending, approved, changes requested). Currently it's a CLI command, but I thought a Datasette UI might be fun.

I see that PR's are available from the issues command, but I don't see reviews anywhere. From the API docs, it looks like there are separate endpoints for those (as well as pull requests in general). What do you think about adding that? Would you accept a PR? Any sense of the level of effort?

I need this for: - #1947

And I can refactor this to use it too: - #1855

By making this a documented internal API it can be used by other plugins too. It's also going to be really useful for writing tests.

I'm going to try to get this working on https://latest.datasette.io/ - it already has a way for people to sign in as root, but none of the databases there are writable.

So I'm going to build a plugin which adds a writable named in-memory database.

And some kind of mechanism for clearing out that database on a regular basis - maybe tables in that database get deleted automatically an hour after they are created?

(Would be neat to display their time-left-until-deleted too)

Ran inspect on a spatialite database and got these warnings:

ERROR: conn=<sqlite3.Connection object at 0x119e46110>, sql = 'select count(*) from [SpatialIndex]', params = None: no such module: VirtualSpatialIndex ERROR: conn=<sqlite3.Connection object at 0x119e46110>, sql = 'select count(*) from [ElementaryGeometries]', params = None: no such module: VirtualElementary ERROR: conn=<sqlite3.Connection object at 0x119e46110>, sql = 'select count(*) from [KNN]', params = None: no such module: VirtualKNN

It still worked, but probably want to catch this.

Suggestion from Discord: https://discord.com/channels/823971286308356157/823971286941302908/1016725586351247430

Is it possible to limit the facet size per database or even per table?

This is a really good idea, it could be done in metadata.yml.

Imitate: - https://github.com/simonw/datasette/pull/1789 ```

would load default entrypoint like before

datasette data.db --load-extension ext

loads the extensions with the "sqlite3_foo_init" entrpoint

datasette data.db --load-extension ext:sqlite3_foo_init

loads the extensions with the "sqlite3_bar_init" entrpoint

datasette data.db --load-extension ext:sqlite3_bar_init ```

I noticed that my Pyodide Datasette prototype is downloading the same database file every single time rather than browser caching it:

I started sketching out a plugin to add a datasette query subcommand to export data from the command line. This is based on discussions in #1356 and #1605. Before I get too far down this rabbit hole, I figure it's worth getting some feedback here (unless this should happen in Discussions). Here's what I'm thinking:

At its most basic, it will write the results of a query to STDOUT.

sh datasette query -d data.db 'select * from data' > results.json

This isn't much improvement over using sqlite-utils. To make better use of datasette and its ecosystem, run datasette query using a canned query defined in a metadata.yml file.

For example, using the metadata file from alltheplaces-datasette:

sh cd alltheplaces-datasette datasette query -d alltheplaces.db -m metadata.yml count_by_spider

That query would be good to get as CSV, and we can auto-discover metadata and databases in the current directory:

sh cd alltheplaces-datasette datasette query count_by_spider -f csv

In this case, count_by_spider is a canned query defined on the alltheplaces database. If the same query is defined on multiple databases or its otherwise unclear which database query should use, pass the -d or --database option.

If a query takes parameters, I can pass them in at runtime, using the --param or -p option:

sh datasette query -d data.db -p value something 'select * from neighborhoods where some_column = :value'

I'm very interested in feedback on this, including whether it should be a plugin or in Datasette core. (I don't have a strong opinion about this, but I'm prototyping it as a plugin to start.)

Let the alpha bake for a while, since #1668 is a big last-minute change.

After shipping, release a new datasette-hashed-urls that depends on it, also this:

• https://github.com/simonw/datasette-hashed-urls/issues/11

% mkdir /tmp/data % cp ~/Dropbox/Development/datasette/fixtures.db /tmp/data % datasette /tmp/data/*.db /tmp/data/created.db --create -p 8852 ... INFO: Uvicorn running on http://127.0.0.1:8852 (Press CTRL+C to quit) ^CINFO: Shutting down % datasette /tmp/data/*.db /tmp/data/created.db --create -p 8852 ... INFO: 127.0.0.1:49533 - "GET / HTTP/1.1" 200 OK The first time I ran Datasette I got two databases - fixtures and created

BUT... when I ran Datasette the second time it looked like this:

This is the same result you get if you run:

datasette /tmp/data/fixtures.db /tmp/data/created.db /tmp/data/created.db

This is caused by this Datasette issue: - https://github.com/simonw/datasette/issues/509

So... either I teach Datasette to de-duplicate multiple identical file paths passed to the command, or I can't use /data/*.db in the Dockerfile here and I need to go back to other solutions for the challenge described in this comment: https://github.com/simonw/datasette-publish-fly/pull/12#issuecomment-1031971831

Originally posted by @simonw in https://github.com/simonw/datasette-publish-fly/pull/12#issuecomment-1032029874

Hi, love your work.

I am unable to lookup indexes in a database using sqlite-utils:

sqlite-utils indexes city_spec.db --table

or

sqlite-utils indexes city_spec.db MyTable

Software sqlite-utils, version 3.24 sqlite3 --version: 3.36.0

Output:

Traceback (most recent call last): File "/opt/app-root/bin/sqlite-utils", line 8, in <module> sys.exit(cli()) File "/opt/app-root/lib64/python3.8/site-packages/click/core.py", line 1128, in call return self.main(args, kwargs) File "/opt/app-root/lib64/python3.8/site-packages/click/core.py", line 1053, in main rv = self.invoke(ctx) File "/opt/app-root/lib64/python3.8/site-packages/click/core.py", line 1659, in invoke return _process_result(sub_ctx.command.invoke(sub_ctx)) File "/opt/app-root/lib64/python3.8/site-packages/click/core.py", line 1395, in invoke return ctx.invoke(self.callback, ctx.params) File "/opt/app-root/lib64/python3.8/site-packages/click/core.py", line 754, in invoke return __callback(args, kwargs) File "/opt/app-root/lib64/python3.8/site-packages/click/decorators.py", line 26, in new_func return f(get_current_context(), *args, kwargs) File "/opt/app-root/lib64/python3.8/site-packages/sqlite_utils/cli.py", line 2123, in indexes ctx.invoke( File "/opt/app-root/lib64/python3.8/site-packages/click/core.py", line 754, in invoke return __callback(args, kwargs) File "/opt/app-root/lib64/python3.8/site-packages/sqlite_utils/cli.py", line 1624, in query db.register_fts4_bm25() File "/opt/app-root/lib64/python3.8/site-packages/sqlite_utils/db.py", line 403, in register_fts4_bm25 self.register_function(rank_bm25, deterministic=True) File "/opt/app-root/lib64/python3.8/site-packages/sqlite_utils/db.py", line 399, in register_function register(fn) File "/opt/app-root/lib64/python3.8/site-packages/sqlite_utils/db.py", line 392, in register self.conn.create_function(name, arity, fn, *kwargs) sqlite3.NotSupportedError: deterministic=True requires SQLite 3.8.3 or higher

Datasette's documentation is aimed at people who install and configure it.

What about end users of preconfigured and deployed Datasette instances?

Something that can be linked to from the Datasette UI would be really useful.

It would be nice if the --cors option accepted an origin regex to more securely allow secure local development.

As an example, Observable notebooks namespace every user's notebooks by their username and user content is served from username.observableusercontent.com, so you would set --cors-origin username.observableusercontent.com to restrict access to a local development Datasette instance to only your own notebooks, rather than exposing the data to any website that makes a request.

Thank you for all of your work on Datasette!

First of all, thanks for creating this cool project, and also supporting publishing to various hosting services out of the box.

While testing out, I noticed legitimate queries such as select * from books where title like 'Pragmatic%' or select * from books where title = 'The Pragmatic Programmer' are blocked, due to the regular expression check here: https://github.com/simonw/datasette/blob/main/datasette/utils/init.py#L185

Example as seen from a Datasette instance: https://fivethirtyeight.datasettes.com/polls?sql=select+*+from+books+where+title+like+%27Pragmatic%25%27%0D%0A

I'd propose a regular expression like re.compile(f"pragma_(?!({'|'.join(allowed_pragmas)}))"), instead of re.compile(f"pragma(?!_({'|'.join(allowed_pragmas)}))"),

I can create a pull request with this change, unless the maintainers think it would allow unwanted queries to be executed.

Original title: Decide when to drop support for Python 3.6

context_vars can solve this but they were introduced in Python 3.7: https://www.python.org/dev/peps/pep-0567/

Python 3.6 support ends in a few days time, and it looks like Glitch has updated to 3.7 now - so maybe I can get away with Datasette needing 3.7 these days?

Tweeted about that here: https://twitter.com/simonw/status/1473761478155010048

Originally posted by @simonw in https://github.com/simonw/datasette/issues/1576#issuecomment-999878907

Using this as a tracking issue. I'm hoping to get the bulk of the JSON redesign work from the refactor in #1554 in for this release.

I've created a canned query with "write": true set. I've also created a custom template for it, but the template doesn't seem to be found. If I look in the HTML I see (stock_exchange is the db name):

<!-- Templates considered: query-stock_exchange.html, *query.html -->

My non-writeable canned queries pick up custom templates as expected, and if I look at their HTML I see the canned query name added to the templates considered (the canned query here is date_search):

<!-- Templates considered: query-stock_exchange-date_search.html, query-stock_exchange.html, *query.html -->

So it seems like the writeable canned query is behaving differently for some reason. Is it an authentication thing? I'm using the built in --root authentication.

Thanks!

I sometimes find the need to create an empty SQLite database file - for example if I want to enable WAL on it before using it with another script. I currently do that like this:

sqlite3 my.db vacuum
sqlite-utils enable-wal my.db

It would be nice if sqlite-utils had a convenience command for doing this.

Currently you have to do this: $ sqlite-utils insert dogs.db dogs dogs.json --convert ' row["is_good"] = 1 return row' Would be neat if this worked too: $ sqlite-utils insert dogs.db dogs dogs.json \ --convert 'row["is_good"] = 1'

Hovering over that could work a little bit like GitHub issue links:

I had to cancel the 0.59.2 release because ReadTheDocs was failing to build the documentation.

https://readthedocs.org/projects/datasette/builds/15268454/

``` /home/docs/checkouts/readthedocs.org/user_builds/datasette/envs/0.59.2/bin/python -m sphinx -T -b html -d _build/doctrees -D language=en . _build/html Running Sphinx v1.8.5 loading translations [en]... done making output directory... building [mo]: targets for 0 po files that are out of date building [html]: targets for 27 source files that are out of date updating environment: 27 added, 0 changed, 0 removed reading sources... [ 3%] authentication

Traceback (most recent call last): File "/home/docs/checkouts/readthedocs.org/user_builds/datasette/envs/0.59.2/lib/python2.7/site-packages/sphinx/cmd/build.py", line 304, in build_main app.build(args.force_all, filenames) File "/home/docs/checkouts/readthedocs.org/user_builds/datasette/envs/0.59.2/lib/python2.7/site-packages/sphinx/application.py", line 341, in build self.builder.build_update() File "/home/docs/checkouts/readthedocs.org/user_builds/datasette/envs/0.59.2/lib/python2.7/site-packages/sphinx/builders/init.py", line 347, in build_update len(to_build)) File "/home/docs/checkouts/readthedocs.org/user_builds/datasette/envs/0.59.2/lib/python2.7/site-packages/sphinx/builders/init.py", line 360, in build updated_docnames = set(self.read()) File "/home/docs/checkouts/readthedocs.org/user_builds/datasette/envs/0.59.2/lib/python2.7/site-packages/sphinx/builders/init.py", line 468, in read self._read_serial(docnames) File "/home/docs/checkouts/readthedocs.org/user_builds/datasette/envs/0.59.2/lib/python2.7/site-packages/sphinx/builders/init.py", line 490, in _read_serial self.read_doc(docname) File "/home/docs/checkouts/readthedocs.org/user_builds/datasette/envs/0.59.2/lib/python2.7/site-packages/sphinx/builders/init.py", line 534, in read_doc doctree = read_doc(self.app, self.env, self.env.doc2path(docname)) File "/home/docs/checkouts/readthedocs.org/user_builds/datasette/envs/0.59.2/lib/python2.7/site-packages/sphinx/io.py", line 318, in read_doc pub.publish() File "/home/docs/checkouts/readthedocs.org/user_builds/datasette/envs/0.59.2/lib/python2.7/site-packages/docutils/core.py", line 219, in publish self.apply_transforms() File "/home/docs/checkouts/readthedocs.org/user_builds/datasette/envs/0.59.2/lib/python2.7/site-packages/docutils/core.py", line 200, in apply_transforms self.document.transformer.apply_transforms() File "/home/docs/checkouts/readthedocs.org/user_builds/datasette/envs/0.59.2/lib/python2.7/site-packages/sphinx/transforms/init.py", line 90, in apply_transforms Transformer.apply_transforms(self) File "/home/docs/checkouts/readthedocs.org/user_builds/datasette/envs/0.59.2/lib/python2.7/site-packages/docutils/transforms/init.py", line 171, in apply_transforms transform.apply(**kwargs) File "/home/docs/checkouts/readthedocs.org/user_builds/datasette/envs/0.59.2/lib/python2.7/site-packages/sphinx/transforms/init.py", line 245, in apply apply_source_workaround(n) File "/home/docs/checkouts/readthedocs.org/user_builds/datasette/envs/0.59.2/lib/python2.7/site-packages/sphinx/util/nodes.py", line 94, in apply_source_workaround for classifier in reversed(node.parent.traverse(nodes.classifier)): TypeError: argument to reversed() must be a sequence

Exception occurred: File "/home/docs/checkouts/readthedocs.org/user_builds/datasette/envs/0.59.2/lib/python2.7/site-packages/sphinx/util/nodes.py", line 94, in apply_source_workaround for classifier in reversed(node.parent.traverse(nodes.classifier)): TypeError: argument to reversed() must be a sequence The full traceback has been saved in /tmp/sphinx-err-vkl0oE.log, if you want to report the issue to the developers. Please also report this if it was a user error, so that a better error message can be provided next time. A bug report can be filed in the tracker at https://github.com/sphinx-doc/sphinx/issues. Thanks! ```

It would be nice if sqlite-utils insert had a progress bar, for when it's churning through huge CSV files.

I followed the instructions here, and even after running python update-docs-help.py I get the following failed tests -- any thoughts?

FAILED tests/test_api.py::test_searchable[/fixtures/searchable.json?_search=te*+AND+do*&_searchmode=raw-expected_rows3] FAILED tests/test_api.py::test_searchmode[table_metadata1-_search=te*+AND+do*-expected_rows1] FAILED tests/test_api.py::test_searchmode[table_metadata2-_search=te*+AND+do*&_searchmode=raw-expected_rows2]

This is with python 3.9.7 and lots of other packages, as in attached environment listing from conda list. conda-installed.txt

I'm using the following command:

twitter-to-sqlite user-timeline -a twitter-auth.json twitter/tweets.db --since

Which gives the following error: Error: Use either --since or --since_id, not both

Running without --since.

Traceback (most recent call last): File "/usr/local/bin/twitter-to-sqlite", line 8, in <module> sys.exit(cli()) File "/usr/local/lib/python3.9/site-packages/click/core.py", line 1137, in __call__ return self.main(*args, **kwargs) File "/usr/local/lib/python3.9/site-packages/click/core.py", line 1062, in main rv = self.invoke(ctx) File "/usr/local/lib/python3.9/site-packages/click/core.py", line 1668, in invoke return _process_result(sub_ctx.command.invoke(sub_ctx)) File "/usr/local/lib/python3.9/site-packages/click/core.py", line 1404, in invoke return ctx.invoke(self.callback, **ctx.params) File "/usr/local/lib/python3.9/site-packages/click/core.py", line 763, in invoke return __callback(*args, **kwargs) File "/usr/local/lib/python3.9/site-packages/twitter_to_sqlite/cli.py", line 317, in user_timeline for tweet in bar: File "/usr/local/lib/python3.9/site-packages/click/_termui_impl.py", line 328, in generator for rv in self.iter: File "/usr/local/lib/python3.9/site-packages/twitter_to_sqlite/utils.py", line 234, in fetch_user_timeline yield from fetch_timeline( File "/usr/local/lib/python3.9/site-packages/twitter_to_sqlite/utils.py", line 202, in fetch_timeline raise Exception(str(tweets["errors"])) Exception: [{'code': 44, 'message': 'since_id parameter is invalid.'}]

Python 3.9.5 twitter-to-sqlite, version 0.21.3

A useful capability would be if it was super-easy to say "any pushes to branch X should be deployed to latest-X.datasette.io".

I'd like to use this for the column query information work in #1434

Since it's designed to be used in Jupyter or for rapid prototyping in an IDE (and it's still pretty small) sqlite-utils feels like a great candidate for me to finally try out Python type hints.

https://veekaybee.github.io/2019/07/08/python-type-hints/ is good.

It suggests the mypy docs for getting started: https://mypy.readthedocs.io/en/latest/existing_code.html plus this tutorial: https://pymbook.readthedocs.io/en/latest/typehinting.html

Hi, Recently got this error: Traceback (most recent call last): File "<stdin>", line 1, in <module> File "/home/ethan/git/music-metadata-indexer/src/mmindexer/__init__.py", line 38, in <module> start("/home/ethan/git/music-metadata-indexer/sample", "/home/ethan/git/music-metadata-indexer/test.db") File "/home/ethan/git/music-metadata-indexer/src/mmindexer/__init__.py", line 23, in start scanner.build_database() File "/home/ethan/git/music-metadata-indexer/src/mmindexer/scan.py", line 79, in build_database _import_song(self.db, Path(dirpath).joinpath(f), self.logger) File "/home/ethan/git/music-metadata-indexer/src/mmindexer/scan.py", line 23, in _import_song db.add_song(filepath) File "/home/ethan/git/music-metadata-indexer/src/mmindexer/index.py", line 166, in add_song for match in self.search("albums", album): File "/home/ethan/git/music-metadata-indexer/env/lib/python3.9/site-packages/sqlite_utils/db.py", line 1625, in search cursor = self.db.execute( File "/home/ethan/git/music-metadata-indexer/env/lib/python3.9/site-packages/sqlite_utils/db.py", line 243, in execute return self.conn.execute(sql, parameters) sqlite3.OperationalError: fts5: syntax error near "." So, the error seems to suggest there was a "." character somewhere in the SQL command that was causing the error. I did a little digging and found this in the docs: https://www.sqlite.org/fts5.html#fts5_strings. "." is one of the many prohibited characters.

My solution was to just strip these out of the query using this line query = query.translate({e: None for e in itertools.chain(range(0,26), range(27, 48), range(58,65), range(91,95), [96], range(123,128))})

Perhaps this could be included into the table.search() function?

I've tried several approaches, but this is the current one:

sh echo $json-line | sqlite-utils insert json.db jsontable --truncate --alter --detect-types - In all cases, I get this error:

sh OverflowError: Python int too large to convert to SQLite INTEGER Traceback (most recent call last): File "/home/sean/.local/bin/sqlite-utils", line 8, in <module> sys.exit(cli()) File "/usr/lib/python3/dist-packages/click/core.py", line 764, in __call__ return self.main(*args, **kwargs) File "/usr/lib/python3/dist-packages/click/core.py", line 717, in main rv = self.invoke(ctx) File "/usr/lib/python3/dist-packages/click/core.py", line 1137, in invoke return _process_result(sub_ctx.command.invoke(sub_ctx)) File "/usr/lib/python3/dist-packages/click/core.py", line 956, in invoke return ctx.invoke(self.callback, **ctx.params) File "/usr/lib/python3/dist-packages/click/core.py", line 555, in invoke return callback(*args, **kwargs) File "/home/sean/.local/lib/python3.8/site-packages/sqlite_utils/cli.py", line 841, in insert insert_upsert_implementation( File "/home/sean/.local/lib/python3.8/site-packages/sqlite_utils/cli.py", line 780, in insert_upsert_implementation db[table].insert_all( File "/home/sean/.local/lib/python3.8/site-packages/sqlite_utils/db.py", line 2145, in insert_all self.insert_chunk( File "/home/sean/.local/lib/python3.8/site-packages/sqlite_utils/db.py", line 1957, in insert_chunk result = self.db.execute(query, params) File "/home/sean/.local/lib/python3.8/site-packages/sqlite_utils/db.py", line 257, in execute return self.conn.execute(sql, parameters)

I googled the error and checked SO answers and advice, all good. I changed my JSON file to not use integers so I no longer get this error. Of course, that makes using the database a bit harder, so I also tried to solve the problem by modifying DB structure (while using integers in JSON).

If change all INTEGER Data Types to something else (STRING, TEXT) and try to import again using --truncate, I still get this error. I suppose I should tell sqlite-utils which columns should use non-INTEGER Data Type rather than rely on it to check my SQL table configuration.

If that is the case, can this error be a bit more specific for easier troubleshooting - maybe tell us which which record caused the problem when that error is thrown?

My table has 60+ columns, many of which use 64-bit integers (not all records are large or known in advance), so while I can modify JSON to use strings instead of integers, it decreases usability and finding out which records have values for which SQLite integers aren't sufficient requires some work (I'm thinking about parsing all integers with jq and sorting output by length to identify those columns, but I'd prefer if sqlite-utils could tell me that).

My environment:

• Python 3.8.10
• sqlite-utils 3.14
• pandas 1.3.1
• numpy 1.21.1
• sqlite-fts4 1.0.1
• sqlite 3.31.1-4ubuntu0.2

http://localhost:8001/index/summary?_search=language%3Aeng&_sort=title&_searchmode=raw

But I'm not able to manage it in the metadata file. Here is mine (note that the sort column is taken into account) Here it is:

``` { "databases": { "index": { "tables": { "summary": { "sort": "title", "searchmode": "raw" } } } } }

Originally posted by @Krazybug in https://github.com/simonw/datasette/issues/759#issuecomment-624860451

Most of this library deals with lists of Python dictionaries - .insert_all(), .rows, .rows_where(), .search().

The db.execute() method is the only thing that returns a sqlite3 cursor.

There is a clumsily named db.execute_returning_dicts(sql) method but it's not currently mentioned in the documentation.

It needs a better name, and needs to be properly documented.

It looks like there's an inconsistency when exporting to CSV via the the web interface. Say I'm looking at songs released in 1989 in the classic-rock/classic-rock-song-list table from the Five Thirty Eight data. The JSON and CSV export links at the top of the page both give me filtered data using Release+Year__exact=1989 in the URL. In the Advanced export tab, though, the CSV option gives me the whole data set, while the JSON options preserve the query.

It may be that this is intended behaviour related to the streaming CSV stuff discussed here, but if that's the case then I think it should be a little clearer.

While working on #260 I realized there's no command to show indexes in a database, even though there is one for showing tables and one for triggers.

I should implement #261 first.

Datasette currently only inflates foreign keys (into names hyperlinks) if it detects them as foreign key constraints in the underlying database.

It would be useful if you could specify additional "foreign keys" using both metadata.json and the querystring - similar time how you can pass ?_fts_table=x https://datasette.readthedocs.io/en/stable/full_text_search.html#configuring-full-text-search-for-a-table-or-view

I designed the config format before I had a good feel for CLI design using Click. --config max_page_size 2000 is better than --config max_page_size:2000.

For #192 and #197 I've decided I want to be able to order by relevance in FTS4 as well as FTS5.

This means I need to port over my work on bm25() from https://github.com/simonw/sqlite-fts4 (since I don't want to add a full dependency).

Added in 7a67bc7a569509d65b3a8661e0ad2c65f0b09166 refs #1026. Lots of tests are failing now.

Follow-on from #1066.

https://latest.datasette.io/fixtures?sql=select+rowid%2C+data+from+binary_data+order+by+rowid+limit+101

Problem: if these were larger fields that HTML page could have multiple megabytes of Python binary string representations on it.

It should behave more like the regular table view does:

Spun off from #782. These can go on all of the JSON endpoints that support pagination.

I stumbled into this while working on Dogsheep Beta: the requests it re-dispatched through TableView did not carry authentication cookies, and since this was against a private instance they were thus denied.

https://github.com/dogsheep/dogsheep-beta/blob/bed9df2b3ef68189e2e445427721a28f4e9b4887/dogsheep_beta/init.py#L223-L231

This made me think that datasette.client.get() (which Dogsheep Beta will start using shortly) could benefit from some kind of utility mechanism for passing through the cookies and general authenticated state from the current request.

Relates to #1. Evernote XML looks like this:

```xml

<en-note>

</en-note> ``` That hash is the md5 we use to store resources. It should be possible to turn these into embedded image tags, especially if done in conjunction with the https://github.com/simonw/datasette-media plugin.

https://latest-with-plugins.datasette.io/fixtures?sql=select%0D%0A++dateutil_parse%28%2210+october+2020+3pm%22%29%2C%0D%0A++dateutil_easter%28%222020%22%29%2C%0D%0A++dateutil_parse_fuzzy%28%22This+is+due+10+september%22%29%2C%0D%0A++dateutil_parse%28%221%2F2%2F2020%22%29%2C%0D%0A++dateutil_parse%28%222020-03-04%22%29%2C%0D%0A++dateutil_parse_dayfirst%28%222020-03-04%22%29%2C%0D%0A++dateutil_easter%282020%29

This can hook in to the existing extra_css_urls and extra_js_urls mechanism:

https://github.com/simonw/datasette/blob/b2955d9065ea019500c7d072bcd9d49d1967f051/datasette/app.py#L304-L305

The plugins should be able to bundle their own assets though, so it will also have to integrate with the /static/ static mounts mechanism somehow:

https://github.com/simonw/datasette/blob/b2955d9065ea019500c7d072bcd9d49d1967f051/datasette/app.py#L1255-L1257

Refs #14

Some of the content I'm indexing is by me - photos I've taken, tweets I wrote, commits, comments I posted.

Some of it is stuff that I've "liked" or "bookmarked" in some way - favourited tweets, Pocket articles, starred GitHub repos.

It woud be useful to be able to differentiate between the two.

• [x] extra_css_urls(template, database, table, datasette)
• [x] extra_js_urls(template, database, table, datasette)
• [x] extra_body_script(template, database, table, view_name, datasette)
• [x] extra_template_vars(template, database, table, view_name, request, datasette)

Originally posted by @simonw in https://github.com/simonw/datasette/issues/938#issuecomment-674544691

Pretty nasty bug this: I'm getting 500 errors for all pages that try to render a template after installing the newly released Datasette 0.47 - both from pip install and via Homebrew.

Similar problem to #863 - calling datasette.add_message() in a view registered using the register_routes() plugin hook doesn't work, because the code that writes accumulated messages to the ds_messages signed cookie lives in the BaseView class here:

https://github.com/simonw/datasette/blob/28bb1c51897f3956861755e345e18b8e0b1423ac/datasette/views/base.py#L94-L97

It might be useful to have an startup hook which gets passed the datasette object as soon as Datasette has finished initializing.

My initial use-case for this is configuration verification - checking that the "plugins" configuration block for this plugin contains valid details.

I imagine there are plenty of other potential uses for this as well.

For https://github.com/simonw/russian-ira-facebook-ads-datasette/tree/af6d956995e14afd585c35a6a06bb01da32043ba I wrote a script to convert YAML to JSON because YAML is a better format for embedding multi-line HTML descriptions and canned SQL statements.

Example yaml metadata file: https://github.com/simonw/russian-ira-facebook-ads-datasette/blob/af6d956995e14afd585c35a6a06bb01da32043ba/russian-ads-metadata.yaml

It would be useful if Datasette could be fed a YAML file directly:

datasette -m metadata.yaml

Question is... should this be a native feature (hence adding a YAML dependency) or should it be handled by a datasette-metadata-yaml plugin, using a new plugin hook for loading metadata? If so, what would other use-cases for that plugin hook be?

See this conversation: https://twitter.com/psychemedia/status/1199707352540368896

I got this on my laptop: ```pytest ... /Users/simon/.local/share/virtualenvs/datasette-AWNrQs95/lib/python3.7/site-packages/jinja2/loaders.py:171: in get_source f = open_if_exists(filename)

filename = '/Users/simon/Dropbox/Development/datasette/datasette/templates/400.html', mode = 'rb'

def open_if_exists(filename, mode='rb'):
    """Returns a file descriptor for the filename if that file exists,
    otherwise `None`.
    """
    try:

      return open(filename, mode)

E OSError: [Errno 24] Too many open files: '/Users/simon/Dropbox/Development/datasette/datasette/templates/400.html'

/Users/simon/.local/share/virtualenvs/datasette-AWNrQs95/lib/python3.7/site-packages/jinja2/utils.py:154: OSError ``` Based on the conversation in https://github.com/pytest-dev/pytest/issues/2970 I'm worried that my tests are opening too many files without closing them.

In particular... I call sqlite3.connect(filepath) a LOT - and I don't ever call conn.close() on those opened connections:

https://github.com/simonw/datasette/blob/cf7a2bdb404734910ec07abc7571351a2d934828/datasette/database.py#L58-L60

Could this be resulting in my tests eventually opening too many unclosed file handles? How could I confirm this?

Idea: an "allow_sql" key with a SQL query that gets passed the actor JSON as :actor and can extract the relevant keys from it and return 1 or 0.

Originally posted by @simonw in https://github.com/simonw/datasette/issues/698#issuecomment-639787304

See also #800

I need this for datasette-auth-github: https://github.com/simonw/datasette-auth-github/issues/62#issuecomment-642152076

This command is meant to write out fixtures.db, metadata.json and a plugins directory: $ python tests/fixtures.py /tmp/fixtures.db /tmp/metadata.json /tmp/plugins/ Test tables written to /tmp/fixtures.db - metadata written to /tmp/metadata.json Traceback (most recent call last): File "tests/fixtures.py", line 833, in <module> ("my_plugin.py", PLUGIN1), NameError: name 'PLUGIN1' is not defined

I'd like plugins to be able to ship with a default that says "anonymous users cannot do this", but allow site administrators to over-ride that such that anonymous users can use the feature after all.

This is tricky because right now the anonymous user doesn't have an actor dictionary at all, so there's no key to match to an allow block.

Split from #811. It would be useful if finely-grained permissions configured in metadata.json could be used to specify if a user is allowed to execute arbitrary SQL queries.

We have a permission check call for this already: https://github.com/simonw/datasette/blob/9397d718345c4b35d2a5c55bfcbd1468876b5ab9/datasette/views/database.py#L159

But there's currently no way to implement this check without writing a plugin.

I think a "allow_sql": {...} block at the database level in metadata.json (sibling to the current "allow" block for that database implemented in #811) would be a good option for this.

Since that object is passed to the extra_template_vars hooks AND the classes registered by register_facet_classes it should be part of the documented interface on https://datasette.readthedocs.io/en/stable/internals.html

I could also start passing it to the register_output_renderer callback.

Hi Simon

Is there anywhere in the app-context where one can access the hashed urlpath of the database? Currently it's included in the template-context (databases[0]["path") when rendering urls of the database (eg. /db-44b06v9/cases...), but where can I find the hashed url when rendering the index-page? I'm trying to avoid redirects. Thanks!

Decimal types in Postgres cause a failure in db.py data type selection

I have a Django app using a MoneyField, which uses a numeric(14,0) data type in Postgres (https://www.postgresql.org/docs/9.3/datatype-numeric.html). When attempting to export that table I get the following error:

bash $ db-to-sqlite --table isaweb_proposal "postgres://connection" test.db .... column_type=COLUMN_TYPE_MAPPING[column_type], KeyError: <class 'decimal.Decimal'>

Looking at sql_utils.db.py at 292-ish it's clear that there is no matching type for what I assume SQLAlchemy interprets as Python decimal.Decimal.

From the SQLite docs it looks like DECIMAL in other DBs are considered numeric.

I'm not quite sure if it's as simple as adding a data type to that list or if there are repercussions beyond it.

Thanks for a great tool!

I really, really want to start grabbing this data: https://github.com/simonw/datasette/network/dependents

Noticed that the "Templates Considered" comment is missing in 0.37. Believe I traced it back to #664 as you can see it in https://v0-34.datasette.io/ but not https://v0-35.datasette.io/. Looking at the template context debug between the two you can see what is missing from 0.35 vs. 0.34:

```diff < "datasette_version": "0.34", < "app_css_hash": "ffa51a", < "select_templates": [ < "*index.html" < ], < "zip": "<class 'zip'>", < "body_scripts": [], < "extra_css_urls": "<generator object BaseView._asset_urls at 0x7f6529ac05f0>", < "extra_js_urls": "<generator object BaseView._asset_urls at 0x7f6529ac0660>", < "format_bytes": "<function format_bytes at 0x7f652a1588b0>", < "database_url": "<bound method BaseView.database_url of \<datasette.views.index.IndexView object at 0x7f6529b03e50>>", < "database_color": "<bound method BaseView.database_color of \<datasette.views.index.IndexView object at 0x7f6529b03e50>>"

"datasette_version": "0.35",
"database_url": "<bound method BaseView.database_url of <datasette.views.index.IndexView object at 0x7f6140dacd90>>",
"database_color": "<bound method BaseView.database_color of <datasette.views.index.IndexView object at 0x7f6140dacd90>>"

```

I was originally going to do this with a plugin - see #357 - but the more I work with metadata.json the more I want it to just accept YAML as an optional alternative to JSON.

The best example why is still this one: https://github.com/simonw/russian-ira-facebook-ads-datasette/blob/master/russian-ads-metadata.yaml

YAML is just SO much better than JSON for multi-line strings - in particular HTML and SQL, both of which are common in metadata.json files.

If I try to add a list of dicts to an empty table using upsert_all, I get an error:

```python import sqlite3 from sqlite_utils import Database import pandas as pd

conx = sqlite3.connect(':memory') cx = conx.cursor() cx.executescript('CREATE TABLE "test" ("Col1" TEXT);')

q="SELECT * FROM test;" pd.read_sql(q, conx) #shows empty table

db = Database(conx) db['test'].upsert_all([{'Col1':'a'},{'Col1':'b'}])

TypeError Traceback (most recent call last) <ipython-input-74-8c26d93d7587> in <module> 1 db = Database(conx) ----> 2 db['test'].upsert_all([{'Col1':'a'},{'Col1':'b'}])

/usr/local/lib/python3.7/site-packages/sqlite_utils/db.py in upsert_all(self, records, pk, foreign_keys, column_order, not_null, defaults, batch_size, hash_id, alter, extracts) 1157 alter=alter, 1158 extracts=extracts, -> 1159 upsert=True, 1160 ) 1161

/usr/local/lib/python3.7/site-packages/sqlite_utils/db.py in insert_all(self, records, pk, foreign_keys, column_order, not_null, defaults, batch_size, hash_id, alter, ignore, replace, extracts, upsert) 1040 sql = "INSERT OR IGNORE INTO {table} VALUES({pk_placeholders});".format( 1041 table=self.name, -> 1042 pks=", ".join(["[{}]".format(p) for p in pks]), 1043 pk_placeholders=", ".join(["?" for p in pks]), 1044 )

TypeError: 'NoneType' object is not iterable

```

A hacky workaround in use is:

python try: db['test'].upsert_all([{'Col1':'a'},{'Col1':'b'}]) except: db['test'].insert_all([{'Col1':'a'},{'Col1':'b'}])

I've sucessfully "uploaded" an SQLite database (with a metadata.json file) to heroku using:

$ datasette publish heroku so-sales.db -m metadata.json -n so-sales

The question is: how can I modify the (small) metadata.json file without having to upload the (large) SQLite database.

The directions on heroku indicate I should run:

heroku git:clone -a so-sales

But this just results in an empty directory with a warning: warning: You appear to have cloned an empty repository.

I've been able to "clone" the heroku "app" using the command:

$ heroku slugs:download -a so-sales

but this is not a git repository....

Ideally, it seems to me, there'd be an option of the datasette CLI to allow a file to be updated, or there'd be some way to create a local git "clone" of the app so that the heroku instructions for "Deploying with git" would apply.

(p.s. I ran datasette publish heroku -m metadata.json -n so-sales in the hope that that would not cause the .db file to be wiped, but of course it was.)

(p.p.s. Thanks for Datasette!)

The concept of a "service" in Cloud Run is crucial: if you deploy to the same service, you will over-write what you deployed there last!

As such, I'd like to make service a required positional argument for publish cloudrun:

datasette publish cloudrun my-service one.db two.db three.db

I have an sqlite table with 16 million rows in it. Having read @simonw article "Fast Autocomplete Search for Your Website" I was curious to try datasette to see what kind of query performance I could get out of it. In truth I don't need to do full text search since all I would like to do is give my users a way to search for the names of investors such as "Warren Buffet", or "Tim Cook" (who's names are in a single column).

On the first search, Datasette takes over 20 seconds to return all records associated with elon musk:

If I rerun the same search, it then takes almost 9 seconds:

That's far to slow to implement an autocomplete feature. I could reduce the latency by making a special table of only unique investor names, thereby reducing the search space to less than a million rows (then I'd need to implement a way to add only new investor names to the table as I received new data.. about 4,000 rows a day). If I did that, I'm still concerned the new table wouldn't be lean enough to lookup investor names quickly. Plus, even if I can implement the autocomplete feature, I would still finally have to lookup records for that investors which would take between 8 - 20 seconds.

Are there any tricks for speeding this up?

Here's my hardware:

$ twitter-to-sqlite search dogsheep

I was messing around with a custom _rows_and_columns.html template and ended up with this: ```html {% for row in display_rows %}

{% endfor %} This is nasty. I'd like to be able to do something like this instead: {% for row in display_rows %}

{{ row["First_Name"] }} {{ row["Last_Name"] }}

... ```

I'm looking to import data from XML with a pre-defined schema that maps fairly closely to a relational database. In particular, it has explicit annotations for when fields are required, optional, or when a default value should be inferred.

Would there be value in adding the ability to define NOT NULL and DEFAULT column constraints to sqlite-utils?

If a table has a ["foo", "bar", "baz"] JSON column allow that to be faceted against.

• [x] Support ?column__arraycontains=x filter queries
• [x] Support ?_facet_array=column faceting

Relates to #255 - in particular supporting facets across M2M relationships.

It should be possible for .inspect() to notice when a table has two foreign keys to two different tables, and assume that this means there is a M2M relationship between those tables.

When rendering a table with a m2m relationship we could display the first X associated records as a comma separated list of hyperlinks in a new column on the table view, with a column name derived from the table on the other side.

Since SQLite doesn't have RANK or an equivalent of https://www.xaprb.com/blog/2006/12/02/how-to-number-rows-in-mysql/ this would be implemented as N+1 queries (one query per cell that we want to display an m2m summary). This should be OK in SQLite: https://sqlite.org/np1queryprob.html

The version of SQLite that ships with Python 3 is a bit limited - it doesn't support row values for example https://www.sqlite.org/rowvalue.html

Figure out how to bundle a more recent SQLite engine with datasette. We need to figure out two cases:

• Bundling a recent version in a Dockerfile build. I expect this to be quite easy.
• Making a more recent version available to people hacking around in Mac OS X. I have no idea how to start on this.

I want it working on Mac OS X too because I don't want to force Docker as a dependency for anyone who just want to hack around with Datasette a little and run the test suite.

We can use Travis CI and Zeit Now to ensure there is always a live demo of current master. We can ship archived demos for releases as well.

In #14 and 33c7c53ff87c2 I've added working support for setuptools entry_points plugins. These can be installed from PyPI using pip install ....

I imagine some projects will benefit from being able to add plugins without first publishing them to PyPI. Datasette already supports loading custom templates like so:

datasette serve --template-dir=mytemplates/ mydb.db

I propose an additional option, --plugins-dir= which specifies a directory full of blah.py files which will be loaded into Datasette when the application server starts.

datasette serve --plugins-dir=myplugins/ mydb.db

This will also need to be supported by datasette publish as those Python files should be copied up as part of the deployment.

e.g. https://datasette-bwnojrhmmg.now.sh/dk3-bde9a9a/dk?source__contains=http

The JSON output shows that the column is there, but is being displayed incorrectly:

https://datasette-bwnojrhmmg.now.sh/dk3-bde9a9a/dk.jsono?source__contains=http

https://stackoverflow.com/a/44430157/6083 documents the PRAGMA needed to extract foreign key references for a table.

At a minimum we can link column values known to be foreign keys to the corresponding row page. We could try to summarize the linked row in some way too - somehow extracting a sensible link title, maybe based on additional configuration in the metadata.json file.

Still todo:

• [x] Fix it to csvs-to-sqlite refactoring command correctly creates primary key on generated tables
• [x] Ship new csvs-to-sqlite with refactoring command
• [x] Refactor column logic to be more predictable in our templates (the rowid special case)
• [x] Mechanism by which table metadata can specify the "label" column for a table
• [x] Automatically set the label column as the first column that isn't a primary key (falling back on primary key)
• [x] Code which runs a "select id, label from table where id in (...)" query as part of the tableview and populates a lookup dictionary
• [x] Modify templates to use values from that lookup dictionary

Version one should have the following characteristics:

• Looks OK
• Works great on mobile
• Loads extremely fast
• No JavaScript! At least not in v1.

e.g

/database/table?name__contains=Simon&age__gte=4

Same format as Django: double underscore as the split.

If you need to match against a column that happens to contain a double underscore in its official name, do this:

/database/table?weird__column__exact=Simon

__exact is the default operation if none is supplied.

/database-name-7sha256/table-name/compound-pk
/database-name-7sha256/table-name/compound-pk.json

Tricky part will be figuring out what the private key is - especially since it could be a compound primary key and it might involve different data types.