bug in parenthesis

[aosokin]

following #35 and #36
SELECT name FROM stadium WHERE stadium_id NOT IN (SELECT stadium_id FROM concert)
to
SELECT name FROM stadium WHERE stadium_id NOT IN ((SELECT stadium_id FROM concert))

SELECT rid FROM routes WHERE dst_apid IN (SELECT apid FROM airports WHERE country = 'United States')
to
SELECT rid FROM routes WHERE dst_apid IN ((SELECT apid FROM airports WHERE country = 'United States'))

SELECT COUNT(*) FROM (SELECT cName FROM tryout INTERSECT SELECT cName FROM tryout)
to
SELECT COUNT(*) FROM SELECT cName FROM tryout INTERSECT SELECT cName FROM tryout

[klahnakoski]

Do you have a source for these? Maybe I should work on a whole suite of examples at once.

[aosokin]

Do you have a source for these? Maybe I should work on a whole suite of examples at once.

I'm trying to use your parser for several datasets for the task of translating English text to SQL queries:

• Spider, queries here and here
• ATIS dataset, queries here, here, here
• Geography dataset, queries here, here, here, here

I've already tried a number of parsers on these queries, and yours is actually very close to working on all these queries :-)
Thanks a lot for your great work!

I'm not sure whether it is feasible to just try the parser on all of these queries, because it might be hard to check whether the output of the formatted is equivalent to the input queries. I was going over these queries, running all the queries and their parsed-formatted versions on actual databases from the datasets, comparing the results, and building a more-or-less minimal example of where the error comes from.
Maybe my strategy of submitting individual issues was not convenient for you, sorry for that!
What would be a better way to proceed? Should I get together the list of all the problematic queries first?

[klahnakoski]

Maybe my strategy of submitting individual issues was not convenient for you, sorry for that!

No problem. Individual issues is best because it gives me discrete chunks to work on. I was only concerned about whatever work you were doing.

The biggest task in a program like mo-sql-parsing is writing the test suite. If you have managed to find a source of SQL, then I use it to make a more-comprehensive suite.

[klahnakoski]

Looking at the examples you posted, I wonder if this project should be using that simpler format

[aosokin]

Thanks! Then I'll continue creating issues and adding examples similar to already created issues.

[aosokin]

Looking at the examples you posted, I wonder if this project should be using that simpler format

Do you mean the format used in Spider? It suits the examples I posted because those were from Spider but it does not go well with other sources (ATIS and Geography). I like your format a lot more because it is more flexible.

[klahnakoski]

c816cbf#diff-77e357ea1ebd625dbc52532a43f82e67a856e499d3d0b1427b57a71c6f85056aR460

fixes all three