Support "create table if not exists"

When importing multiple CSV files into a SQL (which all share the same table schema), it would be great if csvsql used CREATE TABLE IF NOT EXISTS foo.

Current vs. Desired Usage

Consider this shell code, which dynamically generates CSV files (say, by making requests to a JSON API, munging the data, and spitting out a CSV file):

for thing in $THINGS; do
  # ...
  # process $thing, generate $thing.csv
  # ...

  # Load $thing.csv into SQLite database
  csvsql --db sqlite:///foo.db --insert $thing.csv
done

This will currently fail for every loop iteration after the first $thing, because it uses this SQL to create the table:

CREATE TABLE ... (
...
);

Ideally, I’d be able to have a flag like --create-if-not-exists (substitute the name of your choice here, like --create, --create-force, --create-silent, etc.), which generates this SQL:

CREATE TABLE IF NOT EXISTS ... (
...
);

Implementation Considerations

It’s possible this change would have to be made in accordance with an upstream change to agate-sql. If this is not the correct place to report an issue, let me know.

If this is a feature you’d consider, but you don’t want to do it yourself, I’d be glad to lend a hand.

[jpmckinney]

--create-if-not-exists is clear and good. I think we’d just need to add a keyword argument in agate-sql and use checkfirst= in the sql_table.create() line. Then add the command-line option to csvkit’s csvsql tool, add some tests, and update the changelog. If you can do that, it’d be much appreciated 😃

[jest]

Sure, #818.