Wikipedia:Bots/Requests for approval/DschwenBot

DschwenBot

Time filed: 22:38, Friday February 24, 2012 (UTC)

Automatic, Supervised, or Manual: automatic (with initial supervision)

Programming language(s): python (pywikipedia framework)

Source code available: will be made available

Function overview: parses TIGER Line US Census shape files and extracts individual US county outlines. The outlines are converted to KML and attached to the respective county articles using the {{AttachedKML}} template (note that the actual KML data does not appear in the article text). This is a new geocoding method that was developed during the last few weeks. The outlines will be displayed on the WikiMiniAtlas and can also be viewed using Google/Bing Maps

Links to relevant discussions (where appropriate): Wikipedia_talk:WikiProject_Geographical_coordinates#Autogeneration_of_U.S._highway_KML, Wikipedia_talk:WikiProject_Geographical_coordinates#Outlines

Edit period(s): one time (per US state)

Estimated number of pages affected: 3143 (the total number of US counties)

Exclusion compliant (Y/N): Y

Already has a bot flag (Y/N): N

Function details: Bot will create a new page Talk:Article/KML (unless it already exists), bot will generate the correct KML outline data and upload it on said page, bot will place {{AttachedKML}} in the article text at the bottom before the first category link.

Discussion

Example: Talk:Santa Fe County, New Mexico/KML; Santa Fe County, New Mexico; Santa Fe County, New Mexico (pre KML); diff to add KML

In the example, I don't like the positioning of the KML link; I think it should be in the External Links section.
The KML infobox is terrible, far too editor-centric (but this is not a bot issue in of itself).
Given the KML is associated with the article, why is it a subpage of the talk rather than the article?

Josh Parris 09:23, 25 February 2012 (UTC)[reply]

Thanks for the feedback, Josh. The examples were added fully manually, I mentioned that I intend to put the template before the cats, that would be a technically simple solution. Of course detecting the External links heading is not hard either, but is it really an external link (yes, I guess, if you think about it as links to google maps/bing maps)? The template design is not really my business and not a bot issue. I know there has been some discussion to revise it. There are no subpages in the article space (but yes, that would be the preferable location to me, too) so we had to resort to moving it into talk space. It won't be likely to interfere with article discussions and won't pollute the article namespace. The basic idea here is to keep momentum going for a useful new geocoding device that emerged from an initially heated but now very productive discussion process. All involved people are fully aware that the technical implementation could be better, but we also realize that if we wait for the necessary technical changes in the MediaWiki software (i.e. uploading of xml data on commons) we will have to wait for a looongg time and this thing will just die. This is about putting the idea out there, giving it exposure and demonstrating it's usefulness. --Dschwen 14:33, 25 February 2012 (UTC)[reply]

If it makes anybody more comfortable, I do have a record developing, maintaining, and running bots on commons: DschwenBot, VICbot, and QICbot. All those bots are custom developments and have a fairly high edit volume, for fairly complex edits. --Dschwen 20:23, 25 February 2012 (UTC)[reply]

Ok, I'm pretty much ready for a limited test run on a handful of pages, if that is ok. I programmed the bot to search for an external links section, and if it does not find one, it inserts it either after "Further reading" or "References", whatever is found first. I will skip articles that either already contain an {{AttachedKML}} template, or that already have an existing KML data page. Articles are just skipped and stored for manual processing if any of the conditions are not met. --Dschwen 20:51, 25 February 2012 (UTC)[reply]

A map of a county isn't really further reading, and it barely passes as a reference - is it referred to in the article? I suggest, in order:

External links
Further reading
References

But I'm one guy. We really need more people eyeballing this.

Have you considered adding it to {{Infobox U.S. County}}? That seems a more natural fit. If you were to go down that route, I'd suggest creating all the talk/kml pages and then modifying {{Infobox U.S. County}} to hard-reference the KML non-optionally. An entry like:

County boundaries on Google or Bing

might fit in nicely.

Please alert Wikipedia talk:WikiProject U.S. counties to this BRfA, but it looks like it's a pretty quiet wikiproject. I don't forsee any objections, but equally I'm not hopeful on others chiming in. Also raise it at Wikipedia talk:WikiProject United States in the hope that will attract more participation. Josh Parris 21:35, 25 February 2012 (UTC)[reply]

Ok, will do. And yes, we could add it to infobox county (but I'll leave that to the county guys). Right now the bot is creating a new external links section at the lostations I pointed out (it does not just put it into further reading!). --Dschwen 21:40, 25 February 2012 (UTC)[reply]

I think you should drop a note on USRoads as well. They were discussing doing something similar with Roads, Highways, INterstates, etc. --Kumioko (talk) 04:11, 26 February 2012 (UTC)[reply]