[OSM-dev] Broken UTF-8 In Changesets

[Brett Henderson]

Florian Lohoff wrote:
> On Fri, Oct 24, 2008 at 12:49:47PM +1100, Brett Henderson wrote:
>   
>>    It looks like there's some broken UTF-8 in changesets again.  Presumably
>>    another tag has been truncated half way through a UTF-8 sequence when
>>    writing to the database.
>>
>>    I'm heading away for the weekend and won't have time to look into this
>>    before I leave.  I'll be back on Monday and will try to fix it then.  If
>>    anybody can identify the problematic node/way/relation in the meantime it
>>    will speed up the resolution.
>>
>>    Anybody needing minute/hourly/daily diffs in a hurry will have to manually
>>    edit the troublesome tag before use.
>>     
>
> 200810231353-200810231354.osc
>
> iconv -f utf-8 -t utf-8 <200810231353-200810231354.osc
> [...]
>     <way id="27932240" timestamp="2008-10-23T13:53:19Z" user="logictheo">
>       <nd ref="306591785"/>
>       <nd ref="306591800"/>
>       <nd ref="306591799"/>
>       <nd ref="306591786"/>
>       <nd ref="306591785"/>
>       <tag k="name" v="Εθνική Τράπεζα"/>
>       <tag k="source:opening_hours:en" v="Paper given on Oct 23 2008"/>
>       <tag k="atm" v="yes"/>
>       <tag k="amenity" v="bank"/>
>       <tag k="opening_hours" v="Mo-Th 08:00-14:30; Fr 08:00-14:00"/>
>       <tag k="note_opening_hours" v="Εάν υπάρχουν άτομα που έχουν πάρει νούμερα, και έχει λήξει το ωράριο της τράπεζας, τότε η τράπεζα επεκτείνει  το ωράριο μέχρι να έχουν εξυπη
>
> iconv: illegal input sequence at position 27213
>   
Thanks.

TomH has fixed the database and I've re-generated the broken hourly and 
daily changesets. The minute one had already been deleted.

If anybody sees any other problems please let me know.

Brett