Query any HTML table with PRQL

Once again I find myself searching through a giant table on Wikipedia. And again I think: it would be so much better if I could just query it with SQL or something. Wish someone made a browser extension for that!..

Aaand now I did. Here is the source code: https://codeberg.org/aka_dude/tableq. And here you can get it for your Firefox. Except this uses PRQL because it is much better that SQL. Here is an example:

*Source (Wikipedia)*
File system	Block journaling	Metadata-only journaling	Case-sensitive	Case-preserving	File Change Log	XIP
APFS	?	?	Optional	Yes	?	?
NTFS	No	Yes (2000)	Yes	Yes	Yes	?
ext4	Yes	Yes	Yes, optional	Yes	No	Yes

The button over upper-left corner activates the extension. Upon activation, column names are renamed to snake_case and a PRQL prompt appears. Try writing the following there and see what happens!

from t0
filter (file_change_log == 'Yes')

Note that the table still contains original HTML, e.g. the links are still there.

Some caveats:

~~The extension doesn’t do anything with column duplicates, so underlying DB (sqlite) will have a problem with that.~~ Fixed!
Since content is stored as HTML in the DB, operations over text will work with it, not rendered representation. This may lead to weird results when sorting or searching through rich text.

All caveats should (and maybe will) be taken care of, but current functionality is enough for me.

Additional features (since 2026-08-08)

For web pages that include TableQ as script, there are some options.

If table/thead includes <script type="tableq">, the contents of this script will be used instead of default from @table@. This can be used, for example, to make human-oriented textual data more convenient for querying. Here is the same table from above, but with Yes and No parsed as booleans:

*Source (Wikipedia)*
File system	Block journaling	Metadata-only journaling	Case-sensitive	Case-preserving	File Change Log	XIP
APFS	?	?	Optional	Yes	?	?
NTFS	No	Yes (2000)	Yes	Yes	Yes	?
ext4	Yes	Yes	Yes, optional	Yes	No	Yes

This allows for easier querying, like filter (file_change_log || xip).

If th has attribute tableq-id, its value will be used for the column name. Can be used to provide terser names than the auto-generated ones.
th and td may have non-zero colspan and rowspan. For td, the value will simply be copied into several sub-cells. For th, the column will be divided into sub-columns, and
1. if tableq-id exists, spaces will be used as separators for names of sub-columns;
2. otherwise, sub-columns will be named as ${column_generated_name}_${idx}.

For a real-world example of TableQ usage, check out this comparison of 3D printer filaments made by my friend: https://gospodin.codeberg.page/filaments.html.