[dev-italia] Bad import in Emilia-Romagna

[Paolo F]

It's true !!!

One object with tags and without relation, the other without tags and with
relation.
With this schema it's possible to check automatically and resolve the
problem.

I've created a script in python for analyze all the object (only analyze,
for the moment!); for your example I found:

--------------------------------------------------------

id_1:  267932686
id_2:  268625084
*** way_1-tags:  {u'source': u'
http://dati.emilia-romagna.it/dato/item/11-11-uso-del-suolo-2008.html',
u'landuse': u'orchard', u'ref:corine_land_cover': u'2220'}
*** 1 NO RELATION !!
*** 2 NO TAGS !!
*** relation_2:  [{u'changeset': 21286678, u'uid': 480488, u'timestamp':
datetime.datetime(2014, 3, 24, 15, 7, 22), u'tag': {u'landuse':
u'farmland', u'type': u'multipolygon', u'ref:corine_land_cover': u'2121'},
u'member': [{u'role': u'outer', u'ref': 268625122, u'type': u'way'},
{u'role': u'inner', u'ref': 268625084, u'type': u'way'}, {u'role':
u'inner', u'ref': 268625098, u'type': u'way'}], u'visible': True,
u'version': 1, u'user': u'GiuseppeAmici_IT', u'id': 3600433}]

>>>>>>> update 2 - delete 1 !!

--------------------------------------------------------

It's possible to take the tags from the object id_1 and put them in the
object id_2 then update id_2 and delete id_1.

What do you think ??
It' the right procedure for correct the problems ??

Are about 5.2k the couple of objects that present this condition !!

Paolo


On Wed, Sep 14, 2016 at 3:59 PM, Martin Raifer <tyr.asd a gmail.com> wrote:

> > In many cases one record it's without "tags"
>
> Those could be cases where one of the way is an inner hole of a
> multipolygon and the other one is the landuse polygon that fills the
> hole. Like this example pair:
> http://www.openstreetmap.org/way/268625084 and
> http://www.openstreetmap.org/way/267932686
>
> Martin
>
> On Wed, Sep 14, 2016 at 9:59 AM, Paolo F <geofrizz a gmail.com> wrote:
> > Hi Martin,
> > thank you for your reporting.
> > I have done a spatial query for the region Emilia-Romagna for check all
> the
> > object (ways) with the same geometry.
> > The results are here:
> > http://osmit3.wmflabs.org/check/export_ways_multiple_RER.csv
> > Exist more than 11k records and probably half of these must be deleted.
> > In many cases one record it's without "tags" and I hope that this element
> > permit a massive/automatic deletion.
> > For the rest will be necessary a manual intervention.
> > Before deleting anything I provide abundant perform checks and
> inspections.
> >
> > Ciao,
> > GeoFrizz alias Paolo Frizzera
> >
> > On Tue, Sep 6, 2016 at 10:14 AM, Martin Raifer <tyr.asd a gmail.com>
> wrote:
> >>
> >> Hey,
> >>
> >> I've spotted a region in Emilia-Romagna where a landuse import
> >> obviously gone wrong:
> >> There  are many overlapping identical landuse-polygons, where for
> >> example one is tagged landuse=farmland and the other one
> >> landuse=forest. Another example that triggered my attention at first
> >> was overlapping leisure=sport_centre and natural=wood polygons.
> >>
> >> Here are some examples:
> >> * https://www.openstreetmap.org/way/259698006 and
> >> https://www.openstreetmap.org/way/259714904
> >> * https://www.openstreetmap.org/way/259718079 and
> >> https://www.openstreetmap.org/way/259701722
> >>
> >> Looking at the distribution of sport centres in the region makes it
> >> look like as if this is a widespread issue in the region:
> >> http://overpass-turbo.eu/s/ie9
> >>
> >> I couldn't find meta-information about this import in the respective
> >> changesets: e.g. https://www.openstreetmap.org/changeset/20349894 :-(
> >>
> >> Are those known problems of this import?
> >>
> >> Cheers,
> >> Martin
> >>
> >> _______________________________________________
> >> dev-italia mailing list
> >> dev-italia a openstreetmap.org
> >> https://lists.openstreetmap.org/listinfo/dev-italia
> >
> >
>
-------------- parte successiva --------------
Un allegato HTML è stato rimosso...
URL: <http://lists.openstreetmap.org/pipermail/dev-italia/attachments/20160916/56ee5347/attachment.html>