[OSGeoJapan-discuss] Integration of Japanese GIS formats into mainline GDAL ?

Minoru Akagi akaginch @ gmail.com
2015年 8月 18日 (火) 01:05:37 PDT


Hi,

We had an online hackathon about this topic on the same day as QGIS
HackFests in Japan (2015-08-01). The purpose was to read FGD (GML)
vector data (v4) with the OGR GML driver.

I prepared .gfs schema files, then registered them to the GML registry
and made a few changes to the GML driver code. Finally, I managed to
read FGD (GML) vector data with the driver! See the work branch [1]
and the hackpad [2][3] for the details.

I think we should make efforts for the support of recent format rather
than old formats that are no longer distributed from GSI.

Any comments?

[1] https://github.com/minorua/gdal/commits/fgdgml_vector_1.11
    The branch is based on 1.11 (the current GDAL version of OSGeo4W).
[2] https://hackpad.com/JPGISGML-HackFest-2015-XgBUI6qfbGP  (In Japanese)
[3] https://hackpad.com/JPGIS-GML-Hackathon-2015-iI1ButdfjwX  (English summary)

Minoru

2015-07-20 23:23 GMT+09:00 Even Rouault <even.rouault @ spatialys.com>:
> Ko-san,
>
> Thanks a lot for bringing some light on this.
>
> I've created https://trac.osgeo.org/gdal/ticket/6046 to capture all findings up
> to now and trace tasks that I've in mind that should be completed for
> integretation into upstream GDAL.
>
> Best regards,
>
> Even
>
>> Hi Even,
>> CC: Hiroo-san, Yamate-san, list
>>
>> Excuse me for cutting in this topic.
>>
>> I am neither the developer nor the player, but I tried mext_gdal
>> test version (gdal110_test.zip) at last weekend and probably I understood
>> some data related part, so I mention some of your question about datas.
>> (But, if someone find my mistake, please point it out here...)
>>
>> 2015-05-01 18:59 GMT+09:00 Even Rouault <even.rouault @ spatialys.com>:
>> > I tried the FGD ogr driver on the file
>> > https://github.com/msherry/PyXB-1.1.4/blob/master/examples/unicode_jp/dat
>> > a/shift_jis/FG- GML-13-RailCL25000-20080331-0001.xml
>> > (found by googling FG-GML)
>> >
>> > It is not recognized by the driver. Looking more closely, the driver
>> > seems to expect a <GI> element that isn't found in that file. It appears
>> > that there are several versions of FGD GML, right ? The above file is
>> > apparently v3 (since there's a v3 schema next to it). Perhaps the driver
>> > is v4 (or a previous version) ? I tried to remove the test for the
>> > presence of <GI> but it crashes later during parsing of dates,
>> > apparently not liking the <gml:timePosition> element within <lfSpanFr>
>>
>> There are 2 format types (DLD / GML), 2 recent versions(v3 / v4)
>> and 2 (or more) data types (5/10m mesh DEM / General vector data .etc)
>> in FGD data,
>> and current mext_gdal test version seems to support the following pair.
>>
>> =====
>> 1. FGD driver: DLD v3 (5/10m mesh DEM)
>> 2. FG-GML driver: GML v3 (5/10m mesh DEM)
>> 3. FGD driver: DLD v3 (General vector data)
>> =====
>>
>> Above link which you found use GML v3, but it is not 5/10m mesh DEM,
>> so you couldn't convert it.
>>
>> > I found this project :
>> > http://mizutuu.github.io/jpgisgmlv4togml-converter/  /
>> > https://github.com/mizutuu/jpgisgmlv4togml-converter
>> > It includes the schema for FG-GML v4 and I can't see any reference to the
>> > <GI> element in it either.
>> > With some trickery related to encoding conversion, I managed to use that
>> > project to produce a standard GML file that the OGR GML driver handles
>> > well :
>> >
>> > cat FG-GML-13-RailCL25000-20080331-0001.xml | iconv --from-code=shift-jis
>> > -- to-code=utf-8 | sed "s/Shift_JIS/UTF-8/" | python jpgisgml2gml.py >
>> > out.gml
>>
>> And above GML v4 is not supported (both of 5/10m mesh DEM and
>> General vector data) by current mext_gdal for now.
>>
>> > I've also found this site http://nlftp.mlit.go.jp/ksj-e/ and downloaded a
>> > few datasets. Those I tried are not handled by the OGR FGD driver. Some
>> > of them look like GML with a particular application schema. They are
>> > unfortunately not standard enough for the OGR GML driver to understand
>> > them out of the box, although I believe it could be tweaked to parse
>> > them. Their header look like:
>> >
>> > <ksj:Dataset gml:id="A09Dataset"
>> > xmlns:ksj="http://nlftp.mlit.go.jp/ksj/schemas/ksj-app"
>> > xmlns:gml="http://www.opengis.net/gml/3.2"
>> > xmlns:xlink="http://www.w3.org/1999/xlink" xm
>> > lns:xsi="http://www.w3.org/2001/XMLSchema-instance"
>> > xsi:schemaLocation="http://nlftp.mlit.go.jp/ksj/schemas/ksj-app
>> > KsjAppSchema- A09-v3_1.xsd">
>> >
>> > There's a shapefile next to them, with apparently same content.
>>
>> And about above MLIT NLNI(National Land Numerical Information) data,
>> it is not FGD data...
>>
>> > So it appears there's a variety of GML-like formats ;-)
>>
>> Yes, sure...
>>
>> If you want to use mext_gdal test data, I just uploaded example dummy data
>> to the following repository. So, please try to use those files.
>> https://github.com/sanak/dev-notes/tree/master/datas/fgd
>> (The dummy data location is near by Japan Wakayama prefecture,
>> but on the pacific ocean.)
>>
>> Regards,
>>
>> >> Hi Even,
>> >>
>> >> I talk to Yamate-san who has been working on those three drivers. He
>> >> explained that each driver has little bit different copy right
>> >> background he need to resolve before he can merge them to the main.
>> >>
>> >> The first one, FGD driver, which he wrote for a project and he does not
>> >> have control on its copy right. So, he would like to leave it at this
>> >> moment.
>> >>
>> >> For the second one, FG-GML driver, he wrote it and he owns its copy
>> >> right at this moment. But he said he can release with the licence you
>> >> suggested.
>> >>
>> >> The last one, FGD ogr driver, which he also wrote and it is highly
>> >> possible he can release it under the MIT licence. But he need to make
>> >> sure his client can agree on this. So, he said he will contact to his
>> >> client.
>> >>
>> >> So, the last two drivers can be merge into the main GDAL driver. It is
>> >> little bit confusing background but I am sure we can sort these out.
>> >>
>> >> I will let you know when he is ready to work with you.
>> >>
>> >> Thanks!
>> >>
>> >> Hiroo
>> >>
>> >> On Thu, Apr 30, 2015 at 2:26 PM, Hiroo Imaki <hiroo @ angeli.org> wrote:
>> >> > Hi Even,
>> >> >
>> >> > I am not the developer but he and I working at a same company. If he
>> >> > can agree on the license and release his efforts, that would be
>> >> > wonderful for all Japanese data uses.
>> >> >
>> >> > He hesitates writing his message in English. So, I can be a translator
>> >> > between you guys. I hope no lost in translation!
>> >> >
>> >> > I will get back to you once I hear from him.
>> >> >
>> >> > Hiroo
>> >> >
>> >> > On Thu, Apr 30, 2015 at 10:45 AM, Even Rouault
>> >> > <even.rouault @ spatialys.com
>> >> >
>> >> > > wrote:
>> >> >> Hi,
>> >> >>
>> >> >> Sorry for writing in English, but I don't unfortunately speak
>> >> >> Japanese.
>> >> >>
>> >> >> I'm one of the GDAL developers.
>> >> >>
>> >> >> I've noticed that there are "un-official" versions of GDAL floating
>> >> >> around, with
>> >> >> a few drivers for new Japanese-specific formats. I was wondering if
>> >> >> there was a
>> >> >> reason to keep them outside of the official GDAL tree ? Integrating
>> >> >> them into
>> >> >> mainline GDAL would probably make life of Japanese users easier. GDAL
>> >> >> has already support for a lof of national specific formats, mainly in
>> >> >> Europe and
>> >> >> North America. Would be cool for Asia to join in ;-)
>> >> >>
>> >> >> In a recent discussion thread, the following link is mentionned:
>> >> >>  https://www.dropbox.com/s/4oyhncknavdk2if/gdal110_test.zip
>> >> >>
>> >> >> I can see that it contains 2 new GDAL drivers ( FGD: Japanese
>> >> >> Fundamental Geographic Data (.xml) & FG-GML :  Japanese Fundamental
>> >> >> Geographic Data GML(.xml) ) and one OGR driver (FGD).
>> >> >> Unfortunatelly there's no author or copyright information in the
>> >> >> header of the
>> >> >> corresponding source files. Would their author(s) be willing to
>> >> >> release them
>> >> >> under the X/MIT license and contribute them to the mainline GDAL
>> >> >> project ? Perhaps some test files to write minimalistic regressions
>> >> >> tests would be useful.
>> >> >>
>> >> >> Let me know if I can be useful for anything. I can also help for
>> >> >> writing new
>> >> >> drivers if needed.
>> >> >>
>> >> >> Best regards,
>> >> >>
>> >> >> Even
>> >> >>
>> >> >> --
>> >> >> Spatialys - Geospatial professional services
>> >> >> http://www.spatialys.com
>> >> >> _______________________________________________
>> >> >> OSGeoJapan-discuss mailing list
>> >> >> OSGeoJapan-discuss @ lists.osgeo.org
>> >> >> http://lists.osgeo.org/mailman/listinfo/osgeojapan-discuss
>> >> >
>> >> > --
>> >> > Hiroo Imaki
>> >> > Pacific Spatial Solutions, LLC
>> >> > 1523 Chatham Colony Ct.
>> >> > Reston, VA  20190
>> >> > hiroo.imaki @ pacificspatial.com
>> >> > http://www.pacificspatial.com
>> >> > http://www.geopacific.org (GIS info site)
>> >> > hiroo @ angeli.org (private)
>> >
>> > --
>> > Spatialys - Geospatial professional services
>> > http://www.spatialys.com
>> > _______________________________________________
>> > OSGeoJapan-discuss mailing list
>> > OSGeoJapan-discuss @ lists.osgeo.org
>> > http://lists.osgeo.org/mailman/listinfo/osgeojapan-discuss
>
> --
> Spatialys - Geospatial professional services
> http://www.spatialys.com
> _______________________________________________
> OSGeoJapan-discuss mailing list
> OSGeoJapan-discuss @ lists.osgeo.org
> http://lists.osgeo.org/mailman/listinfo/osgeojapan-discuss


OSGeoJapan-discuss メーリングリストの案内