This is bad though, because if I want to provide binary data in CSV as requested in #1034 I need some way of providing that data.

Which suggests to me that the base64 option is the only one that can make sense for arbitrary SQL queries represented as CSV. Download links won't work.

Providing a binary download link here is actually extremely difficult.

The problem is that the SQL query itself represents data that can change from one moment to the next. It's no good showing a "Binary: 55 bytes" message that links to that same SQL query but with a .blob extension and arguments to select the particular result, because the data may change in a way that causes that query to return a different row - at which point the download link will give you the wrong data, not the 55 bytes you asked for.

So providing a download link risks being misleading.

Idea: .blob output rendererer, where you tell it which column you want using ?_blob_column=x.

This is clearly a better solution than the one I implemented in #1040 - I don't have to add a new route, I don't have to implement permission checks, it reuses mechanism.

Codecov Report

Merging #1049 into main will not change coverage. The diff coverage is n/a.

```diff @@ Coverage Diff @@

main #1049 +/-

======================================= Coverage 84.72% 84.72%
======================================= Files 28 28
Lines 3942 3942
======================================= Hits 3340 3340
Misses 602 602
```

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 42f4851...50a743a. Read the comment docs.

I'm trying to run something behind a MyBinder proxy, but seem to have something set up incorrectly and not sure what the fix is?

I'm starting datasette with jupyter-server-proxy setup:

```

init.py

def setup_nbsearch():

return {
    "command": [
        "datasette",
        "serve",
        f"{_NBSEARCH_DB_PATH}",
        "-p",
        "{port}",
        "--config",
        "base_url:{base_url}nbsearch/"
    ],
    "absolute_url": True,
    # The following needs a the labextension installing.
    # eg in postBuild: jupyter labextension install jupyterlab-server-proxy
    "launcher_entry": {
        "enabled": True,
        "title": "nbsearch",
    },
}

```

where the base_url gets automatically populated by the server-proxy. I define the loaders as:

```

init.py

from datasette import hookimpl

@hookimpl def extra_css_urls(database, table, columns, view_name, datasette): return [ "/-/static-plugins/nbsearch/prism.css", "/-/static-plugins/nbsearch/nbsearch.css", ] ``` but these seem to also need a base_url prefix set somehow?

Currently, the generated HTML loads properly but internal links are incorrect; eg they take the form <link rel="stylesheet" href="/-/static-plugins/nbsearch/prism.css"> which resolves to eg https://notebooks.gesis.org/hub/-/static-plugins/nbsearch/prism.css rather than required URL of form https://notebooks.gesis.org/binder/jupyter/user/ouseful-testing-nbsearch-0fx1mx67/nbsearch/-/static-plugins/nbsearch/prism.css.

The main css is loaded correctly: <link rel="stylesheet" href="/binder/jupyter/user/ouseful-testing-nbsearch-0fx1mx67/nbsearch/-/static/app.css?404439">

SQLite actually has APIs that could help here: https://www.sqlite.org/c3ref/column_database_name.html - for any given SQL query they identify the origin/table/column that is the source of each resulting column.

Those aren't exposed in the Python sqlite3 module though, so using them could be extremely tricky.

For arbitrary CSV the only solution I can think of is to embed the base64 value.

What should happen for CSV export of arbitrary SQL queries, where there's no obvious BLOB to link to?

That documentation: https://docs.datasette.io/en/latest/internals.html#absolute-url-request-path

I should turn datasette.absolute_url(...) into a documented internal API on https://docs.datasette.io/en/stable/internals.html#datasette-class

Here's how those absolute next_url values are generated: https://github.com/simonw/datasette/blob/5db7ae3ce165ded57c7fb1cfbdb3258b1cf06c10/datasette/views/table.py#L774-L776

This is actually a bit tricky to implement, for a few reasons:

• Need to generate a full URL, including the https://host/ bit. I've done this for next_url in the JSON output before, thankfully.
• This only makes sense for CSV output for tables. If it's the CSV output of an arbitrary query there's no /db/table/-/blob/pk/column.blob page for me to link to.
• Need to generate those /.../-/blob/... URLs for the data that is being output as CSV.

Blocked awaiting #1036 (update: now unblocked)

I used https://primer.style/octicons/download-16 instead.