[sqlalchemy] SQLi concerns with using unsafe data to make tables

2019-04-06 Thread Zac Goldstein
Hello, I'd like to create tables based off user input, but I'm uncertain about the security implications. Most of the raw text data consists of column and table names, which will be restricted and validated to snake_case with no special characters. I assume I'm okay there, but confirmation

Re: [sqlalchemy] SQLi concerns with using unsafe data to make tables

2019-04-06 Thread Mike Bayer
On Sat, Apr 6, 2019 at 8:43 PM Zac Goldstein wrote: > > >they are severe. > I thought this would likely be the case. Exposure to DoS attacks is > interesting and something I hadn't thought of, so I'm glad you brought that > up. > > I do have a parser already set up that I could use for this,

Re: [sqlalchemy] SQLi concerns with using unsafe data to make tables

2019-04-06 Thread Mike Bayer
On Sat, Apr 6, 2019 at 6:56 PM Zac Goldstein wrote: > > Hello, > > I'd like to create tables based off user input, but I'm uncertain about the > security implications. they are severe.DDL is modification to the database schema structure and requires a lot of privileges too, depending on

Re: [sqlalchemy] SQLi concerns with using unsafe data to make tables

2019-04-06 Thread Zac Goldstein
>they are severe. I thought this would likely be the case. Exposure to DoS attacks is interesting and something I hadn't thought of, so I'm glad you brought that up. I do have a parser already set up that I could use for this, but I was hoping for something more foolproof. The other thought I

Re: [sqlalchemy] SQLi concerns with using unsafe data to make tables

2019-04-06 Thread Zac Goldstein
Ok, makes sense. Thanks for the help. On Saturday, April 6, 2019 at 7:04:42 PM UTC-7, Mike Bayer wrote: > > On Sat, Apr 6, 2019 at 8:43 PM Zac Goldstein > wrote: > > > > >they are severe. > > I thought this would likely be the case. Exposure to DoS attacks is > interesting and something I