Also, without support for multiple tables & joins this is not very useful.
Also, without support for multiple tables & joins this is not very useful.
Not even slightly true. Can we stop crapping on v1 releases? I'd rather a developer release early and often than hold their code back until it has every possible feature.
- Multiple file/table support is coming. - I like the -tab setting. Will add. I also want to support -dlm=xFF format for arbitrary delimiters, which makes working with Hive text file formats easier.
Basically I recommend you to look at xmlstarlet sel (and find) as an example of converting complex languages into command line.
Did you think of loading every sqlite file on boot (-load?)
How about textql -header -source data.csv -uniq id -tstmp created,updated -index created 'SELECT ...'? - i.e. changes to table structure from command line switches?
Conventions also won't hurt. I.e. my-data.csv turns to my_data table that is saved to my-data.sqlite3 without me mentioning so. -save should do that.
The idea is to create a re-entrant workspace in the current directory which becomes richer with every invocation.
Also, does CREATE TABLE ... SELECT ... FROM tbl work? Plus -save.
Yes. You can even modify and append to existing sqlite datebases. I'd love for this to become, eventually, a general purpose tool for working on data from the CLI.
I will look into xmlstarlet sel and find for sure. I like the -load tag idea too. There's lots of possibilities here.