2017-05-31 16:34 GMT+12:00 David G. Johnston <david.g.johns...@gmail.com>:
> On Tue, May 30, 2017 at 9:17 PM, Patrick B <patrickbake...@gmail.com> > wrote: > >> >> regexp_matches(name, '((main|medium).name/\d+.\d+)') as filename, >> >> >> Example here: http://sqlfiddle.com/#!15/5f4f0/4 >> >> 1. can only get the jpg file name >> >> Because those are the only two file names consisting of one or more > numbers, something else, and ending with one or more numbers > > I'm pretty sure you mean for the "." in that regex to be "\." so it is > treated as the period before the extension and not the regex "any" meta > character. > > \d means number, 1-9. You probably want something like "[\d\w\s]+" to > match digits, letters (including underscore), and white-space. Or maybe > "[^\r\n/\.]+" - specify what it is that won't be in the file name. > >> >> 1. I don't get only the file name but the rest as well, which is not >> what I need >> >> You get more than just "the rest"...you get an array with two entries > corresponding to the two parenthetical captures (the fiddle apparently > doesn't display the "{}" that would make this much more obvious). > > (regexp_matches(...))[#] > > The above will let you extract only the array position that you specify. > You will need to add more parens to delimit exactly what you need. > > You can use "(?:regex)" instead of normal parens to group without > capturing. > > David J. > > Thanks David! That helped. See example: http://dbfiddle.uk/?rdbms=postgres_9.6&fiddle=a74cd219191444f0228b9df581548a37 Will do some more tests but i think that's all i need. Cheers Patrick
