Hi Mike, committed with r3435. I had to modify one of the unit tests, please remember to run ant test and check the results when you post a patch. I also hesitated because we are now trimming the values before we actually know if they are used at all, but the performance implication seems very small. I really don't like this repeated code around the keepTag() call in three readers. I think it should be possible to implement something like void addTagIfUsed(Element e, String tag, String val) which combines the code.
Gerd From: [email protected] To: [email protected] Date: Fri, 30 Jan 2015 11:25:25 +0000 Subject: Re: [mkgmap-dev] leading and trailing spaces Hi Gerd, good point. There isn’t any single place that can fix for all the input formats as far as I can see. The attached patch separately updates each of the readers. Regards,Mike From: Gerd Petermann [mailto:[email protected]] Sent: 29 January 2015 15:03 To: [email protected] Subject: Re: [mkgmap-dev] leading and trailing spaces Hi Mike, I'd prefer a patch that changes the behaviour also for xml and o5m input, not just for *.pbf. Maybe you find a place where this can be done once for all sources? GerdFrom: [email protected] To: [email protected] Date: Thu, 29 Jan 2015 12:10:40 +0000 Subject: Re: [mkgmap-dev] leading and trailing spacesHi Gerd, you are right; trim() removes all leading and trailing characters that are less than 0x20. However, the place I mention is reading the raw OSM data before any styles have been applied, so this should not be a problem. I agree with Paco that ideally the OSM data should be corrected, but that will always be playing catch up, so I think it is best to also remove them in the mkgmap processing. I have corrected a number of local ones in OSM, but there are too many to correct them all manually. I've attached a patch that seems to be OK to me. Please give it a try. Regards, Mike -----Original Message----- From: GerdP [mailto:[email protected]] Sent: 29 January 2015 06:18 To: [email protected] Subject: Re: [mkgmap-dev] leading and trailing spaces Hi Mike, I can't find it now, but I seem to remember a comment that trim() should be used with great care as it removes the highway shield codes. Gerd Mike Baggaley wrote > Good evening all, > > I've noticed that there are a number of unwanted leading and trailing > spaces > in OSM data. You can get a list of the ones on the name field using the > following: > > name~'.* ' {echotags "name ends with a space"} > name~' .*' {echotags "name starts with a space"} > > I can't see any facility in mkgmap to remove these, and was wondering > whether the best way would be simply to code them out by replacing the > following line in parseDense, parseNodes, ParseWays and parseRelations: > > String val = getStringById(...); > > With > > String val = getStringById(...).trim(); > > I can't think of any reason why a value would need to start or end with a > space, so it seems reasonable to just remove them. Are there any other > suggestions, or reasons why this should not be done? > > Regards, > Mike > > _______________________________________________ > mkgmap-dev mailing list > [email protected] > http://www.mkgmap.org.uk/mailman/listinfo/mkgmap-dev -- View this message in context: http://gis.19327.n5.nabble.com/leading-and-trailing-spaces-tp5831778p5831782 .html Sent from the Mkgmap Development mailing list archive at Nabble.com. _______________________________________________ mkgmap-dev mailing list [email protected] http://www.mkgmap.org.uk/mailman/listinfo/mkgmap-dev _______________________________________________ mkgmap-dev mailing list [email protected] http://www.mkgmap.org.uk/mailman/listinfo/mkgmap-dev
_______________________________________________ mkgmap-dev mailing list [email protected] http://www.mkgmap.org.uk/mailman/listinfo/mkgmap-dev
