Dennis Lee Bieber wrote on January 02, 2017 8:30 AM > > On Sun, 1 Jan 2017 18:30:03 -0800, "Deborah Swanson" > <pyt...@deborahswanson.net> declaimed the following: >
<snip> > Out of curiosity -- don't those listings have street > addresses? There must be lots of "2 br" units in the city. > <snip> > Part of the hassle in your schema is that you are > treating the type of unit (2 br) AND the city (as shown in > your sample; I'd hope a full address is available) as a > single data field. Similarly, you have some obscure code > attached to the state. haha - what schema? But seriously, I've loosely had one in mind all along, and there are currently 14 columns for each listing. I've just only mentioned the fields involved in the conditional problem I was having. One of those fields (br) is for bedrooms, and another couple I'll add later will be for address, phone, and local features, which may be pulled from my locations dictionary. The Descriptions I made up for my example are very simple short ones. Real webpage titles from Craigslist are frequently very long, complex and full of typos. I go fishing in them for most of the data, but I don't plan on making any effort to clean them up. I think it was you who asked another question that I didn't answer at the time, of why I used those weird 2-letter codes for the field names. It's rather stupidly simple. Since the "database" will be in flux for some time, with columns being moved around, added and deleted, I came up with this little scheme that executes right after the csv has been opened and read into ls[], to avoid having to change any of the subscripts in my code whenever the columns changed: flds = len(ls[0]) cl, ur, d1, de, lo, st, mi, ki, re, da, br, no, yn, ma, ar = range(0,flds) ls[0] is the title row, so the range is automatically sized to the number of columns in the current csv. If I open a csv with more columns than last time, I hit an IndexError right off, which is a handy reminder if I've forgotten to update the field codes. I'd welcome anyone's improvement on this scheme, if it accomplishes the same result. Also, 'kind' is for the kind of property. I want a house, but I'd consider a cottage, trailer or manufactured home if it's nice and has a garage or a carport. And I want 2 bedrooms (one for my office), but I'm keeping track of 1 brs & even nice studio cabins, just in case I really need to get out of here and into something I can afford quickly. I've also tried to screen the plexes (duplex, triplex, six-plex and what all), apartments, condos and other undesirable "kinds" out of my searches, but they still slip into the results anyway. Craigslist's search function leaves much to be desired. So 'kind' also serves as a place to snag them for later deletion. And it's a red flag when one listing for a property comes up with 'kind' = 'house' and another comes up 'duplex'. For some reason people like to pretend their duplexes are houses, and sometimes they change their titles from duplex to house after they've had to relist it a couple of times. Or they don't say what kind of property it is, and I have to pull up the ad and look at it. Details, details, details. ;) D. -- https://mail.python.org/mailman/listinfo/python-list
