duckdb is a single file with no dependencies and it's fast. Still blows my mind how quickly it can query GB sized gzipped CSVs.

A poor man's version of csvlens is something like:

    sqlite -column :memory: '.import --csv file.csv tmp' 'select * from tmp;' | bat

which imports the csv into sqlite and outputs it to bat, my favorite pager - use `less` or whatever else you desire.

In addition to using a pager with sqlite3's fantastic text-only output .modes, if the CSV contains hyperlinks I use a custom UNIX filter I wrote that outputs simple, minimalist HTML. Then I view with text-only browser.

For example, this is how I use YouTube. I never use the YouTube website, with its gigantic pages and its "Javascript player", not to mention all of the telemetry. All the search results and information about videos is stored in SQL or CSV, viewed with a text-only sqlite3 output .mode, and optionally converted to simple HTML.

For me, this is better than a "modern" web browser that's too large for me to compile.

Wow, this is awesome. Thanks a lot.