Name_Multilingual: Basque & Asturian languages

[Crashillo]

Is it possible to append a new multilingual style for the spanish basque regions: Navarre and Basque country? Since that comment, I've seen some languages new here (Belgium, Kosovo, Morocco & Djibouti, if I'm right)

The two names validation throws multiple syntax errors in these regions, in spite of the name tag is right according the community. The validation should match with the multilingual names.

Default separator: /
Default languages: eu / es

[frodrigo]

Yes it is possible.

A new multilingual setup should be added. It does not match any existing one.
https://github.com/osm-fr/osmose-backend/blob/master/plugins/Name_Multilingual.py

The language order is fixed eu then es ?

Is it for Euskadi as in https://www.openstreetmap.org/relation/349042 ?
https://github.com/osm-fr/osmose-backend/blob/master/osmose_config.py#L1510

[Crashillo]

The language order is fixed eu then es ?

Exact

Is it for Euskadi as in https://www.openstreetmap.org/relation/349042 ?
https://github.com/osm-fr/osmose-backend/blob/master/osmose_config.py#L1510

Euskadi: https://www.openstreetmap.org/relation/349042
Nafarroa: https://www.openstreetmap.org/relation/349027

And perhaps, Navarra should also include language=["es", "eu"]
https://github.com/osm-fr/osmose-backend/blob/master/osmose_config.py#L1509-L1510

[Crashillo]

In addition to basque, the asturian lang has a similar casuistry:

Asturias: https://www.openstreetmap.org/relation/6428094
Separator: /
Languages: es / ast
Config: es_comm("asturias", 349033, "ES-AS", proj=32629 language=["es", "ast"])

[frodrigo]

Space around / should or not be present ? or it doesn't matter ?

When there is only one name in name tag, should be Basque or Castilian ?

If there is name:eu and name:es tags, name can have only one of these ?

[frodrigo]

<tag k="name" v="CC San Andres GE" />
<tag k="name:es" v="Centro Cívico San Andrés" />
<tag k="name:eu" v="San Andres gizarte etxea" />
<tag action="modify" k="name:eu" v="CC San Andres GE" />
<tag action="modify" k="name:es" v="CC San Andres GE" />
<tag action="modify" k="name" v="San Andres gizarte etxea/Centro Cívico San Andrés" />
<tag k="name" v="Quintana" />
<tag k="name:es" v="Quintana" />
<tag k="name:eu" v="Kinta" />
<tag action="modify" k="name" v="Kinta/Quintana" />
<tag k="name" v="Barriobusto/Gorrebusto" />
<tag k="name:es" v="Barriobusto" />
<tag k="name:eu" v="Gorrebusto" />
<tag action="modify" k="name" v="Gorrebusto/Barriobusto" />
<tag k="name" v="Pipaón" />
<tag k="name:es" v="Pipaón" />
<tag k="name:eu" v="Pipaon" />
<tag action="modify" k="name" v="Pipaon/Pipaón" />
<tag k="name" v="Consierra de Árcena" />
<tag k="name:eu" v="Artzenako Mendizerra" />
<tag action="modify" k="name:eu" v="Consierra de Árcena" />
<tag action="modify" k="name:es" v="Consierra de Árcena" />
<tag action="modify" k="name" v="Artzenako Mendizerra" />
<tag action="modify" k="name" v="Artzenako Mendizerra/Consierra de Árcena" />
<tag k="name" v="Gorobel / Sálvada" />
<tag k="name:es" v="Sierra Sálvada" />
<tag k="name:eu" v="Gorobelgo mendilerroa" />
<tag action="modify" k="name:eu" v="Gorobel" />
<tag action="modify" k="name:es" v="Sálvada" />
<tag action="modify" k="name" v="Gorobelgo mendilerroa/Sierra Sálvada" />
<tag k="name" v="Vitoria-Gasteiz (Estación Bus/Bus Geltokia)" />
<tag action="modify" k="name:eu" v="Vitoria-Gasteiz (Estación Bus" />
<tag action="modify" k="name:es" v="Bus Geltokia)" />
<tag k="name" v="EG/TC-13 Zaballa → Vitoria-Gasteiz" />
<tag action="modify" k="name:eu" v="EG" />
<tag action="modify" k="name:es" v="TC-13 Zaballa → Vitoria-Gasteiz" />
<tag k="name" v="EG/TC-13 Zaballa" />
<tag action="modify" k="name:eu" v="EG" />
<tag action="modify" k="name:es" v="TC-13 Zaballa" />
<tag k="name" v="EG/TC-13 Vitoria-Gasteiz → Zaballa" />
<tag action="modify" k="name:eu" v="EG" />
<tag action="modify" k="name:es" v="TC-13 Vitoria-Gasteiz → Zaballa" />
<tag k="name" v="Jundiz" />
<tag k="name:es" v="Júndiz" />
<tag k="name:eu" v="Jundiz" />
<tag k="alt_name" v="Jundizko industrialdea; Polígono de Júndiz" />
<tag action="modify" k="name" v="Jundiz/Júndiz" />
<tag k="name" v="Zadorra Sistemako urtegiak KBE/ZEC Embalses del sistema del Zadorra" />
<tag k="name:es" v="ZEC Embalses del sistema del Zadorra" />
<tag k="name:eu" v="Zadorra sistemako urtegiak KBE" />
<tag k="official_name:es" v="Zona Especial de Conservación Embalses del sistema del Zadorra" />
<tag k="official_name:eu" v="Zadorra Sistemako urtegiak Kontserbazio Bereziko Eremua" />
<tag action="modify" k="name:eu" v="Zadorra Sistemako urtegiak KBE" />
<tag action="modify" k="name" v="Zadorra sistemako urtegiak KBE/ZEC Embalses del sistema del Zadorra" />

[Crashillo]

Space around / should or not be present ? or it doesn't matter ?

It would be great to keep some format consistency, but I did some overpass queries and the results are pretty similar. I'd tend to keep the spaces around.

If there is name:eu and name:es tags, name can have only one of these ?

Let me explain, since this is a big matter of controversy:
Is osmose able to check the issues over an smaller area than the region itself? I say so because the order will vary depending on which subregion you are.

Euskadi:

• Bizkaia: https://www.openstreetmap.org/relation/349034 --> name:eu / name:es
• Gipuzkoa: https://www.openstreetmap.org/relation/349015 --> name:eu / name:es
• Araba: https://www.openstreetmap.org/relation/349011 --> name:es / name:eu

Navarre:

• Basque-speaker area: https://www.openstreetmap.org/relation/9089318 --> name:eu / name:es
• Spanish-speaker area: https://www.openstreetmap.org/relation/9089396 --> name:es / name:eu
• Mixed area: https://www.openstreetmap.org/relation/9089335 --> name:es / name:eu

When there is only one name in name tag, should be Basque or Castilian ?

It doesn't matter. Ignore the casuistry since the subregions I mentioned before have their own rules. For instance, the city of Bilbao should be Bilbo / Bilbao, however, and officially, the name is only Bilbao. Therefore, if there's only one name.... let it go.
Regarding this, it exists another exception: some names could be splitted by a dash sign with no spaces around, i.e. Vitoria-Gasteiz. This is official, so they shouldn't be checked.

To sum up: If the name tag has any slash, look for the language tags in order to display the issue, if it's possible, fulfilling the regions I exposed before.

Sorry all this mess, but as I said, it's quite controversial since not all municipalities speak basque.

[Crashillo]

For asturian language, it's pretty much easier: if there's any slash on the name, just look for name:es / name:ast (spaces around). Fallbacks to spanish.

[frodrigo]

Space around / should or not be present ? or it doesn't matter ?

It would be great to keep some format consistency, but I did some overpass queries and the results are pretty similar. I'd tend to keep the spaces around.

If there is name:eu and name:es tags, name can have only one of these ?

Let me explain, since this is a big matter of controversy:
Is osmose able to check the issues over an smaller area than the region itself? I say so because the order will vary depending on which subregion you are.

No. The area are defined in the configuration

https://github.com/osm-fr/osmose-backend/blob/master/osmose_config.py#L1494-L1515

Requered changes:

+es_comm("asturias", 349033, "ES-AS", proj=32629, language=["es", "ast"], **{'multilingual-style': 'sp_eu'})

language=["eu", "es"] is also required for Navarra ?

When there is only one name in name tag, should be Basque or Castilian ?

It doesn't matter. Ignore the casuistry since the subregions I mentioned before have their own rules. For instance, the city of Bilbao should be Bilbo / Bilbao, however, and officially, the name is only Bilbao. Therefore, if there's only one name.... let it go.
Regarding this, it exists another exception: some names could be splitted by a dash sign with no spaces around, i.e. Vitoria-Gasteiz. This is official, so they shouldn't be checked.

To sum up: If the name tag has any slash, look for the language tags in order to display the issue, if it's possible, fulfilling the regions I exposed before.

Sorry all this mess, but as I said, it's quite controversial since not all municipalities speak basque.

No worry, except Bruxelles, all other areas I see have complex rules.

What about : "Zeharkalea 26 (Kiroldegia/Polideportivo)", "Kurtzero (Geltokia/Estación)", how to split that ?

[Crashillo]

No. The area are defined in the configuration

Since you cannot divide this, I'd suggest to avoid checking the language order.

language=["eu", "es"] is also required for Navarra ?

Same behaviour as Basque country

**{'multilingual-style': 'sp_eu'}

Watch this out. Asturias speaks Asturian, it's different from basque. sp_eu should apply for Euskadi & Navarre. In Asturias, should be something like sp_ast

What about : "Zeharkalea 26 (Kiroldegia/Polideportivo)", "Kurtzero (Geltokia/Estación)", how to split that ?

These names are wrong. They're not about two names rule, they include the amenity type in the name... they should be a different osmose error 😄, when there's text (in a name) within parentheses... you bet something is wrong

[frodrigo]

Please could you review the configuration about language values
frodrigo@fe2e962

I think I have implement all the rules. Could you overview the result, please :
Analyser_Sax-spain_euskadi.xml.zip

[Crashillo]

This error contains a name and a name:es. Therefore the fix should suggest:

• Add a name:eu with the name
• Modify the name in order to include name:eu / name:es

Same way if it was name and name:eu

<error class="50604" subclass="0">
<location lat="43.294141" lon="-2.3539311" />
<node id="27552539" lat="43.294141" lon="-2.3539311" version="10" user="Juan_Carlos_G_F-imports">
<tag k="ele" v="4" />
<tag k="name" v="Deba" />
<tag k="place" v="village" />
<tag k="source" v="Instituto Geográfico Nacional" />
<tag k="capital" v="8" />
<tag k="name:es" v="Deva" />
<tag k="ref:ine" v="20029000100" />
<tag k="wikidata" v="Q1611504" />
<tag k="population" v="5185" />
<tag k="source:ele" v="MDT5" />
<tag k="admin_level" v="8" />
<tag k="postal_code" v="20820" />
<tag k="source:date" v="2011-06" />
<tag k="source:file" v="http://centrodedescargas.cnig.es/CentroDescargas/equipamiento/BD_Municipios-Entidades.zip" />
<tag k="source:name" v="Nomenclátor Geográfico de Municipios y Entidades de Población" />
</node>
<fixes>
<fix>
<node id="27552539">
<tag action="modify" k="name:eu" v="Deba" />
</node>
</fix>
<fix>
<node id="27552539">
<tag action="modify" k="name:es" v="Deba" />
</node>
</fix>
<fix>
<node id="27552539">
<tag action="modify" k="name" v="Deva" />
</node>
</fix>
</fixes>
</error>

[Crashillo]

In any case, I'd go along with this. I'll test it one in production, to see which tunings would be better

[frodrigo]

Deployed. Update in progress
http://osmose.openstreetmap.fr/en/issues/open?item=5060&country=spain%2a&class=50604

[frodrigo]

Now we hare result in
http://osmose.openstreetmap.fr/en/issues/open?item=5060&country=spain%2a&class=50604

I let you check.

[Crashillo]

Hey, the suggestions are working great! 👍
But I see the "two names" error don't deal with them: http://osmose.openstreetmap.fr/en/issues/open?item=5030&source=&class=705&country=spain%2a

For instance, look at this village. Multilingual settings are ok, but it keeps displaying "two names" error

[frodrigo]

Ok.
Disabling multiple name plugin on this area in next deploy.

[frodrigo]

Ok.
Disabling multiple name plugin on this area in next deploy.

Deployed.

Manuel clean up. Stats update tomorrow.

[Crashillo]

Since the regions were splitted these errors have come back again http://osmose.openstreetmap.fr/en/map/#item=5030&zoom=10&lat=43.019&lon=-2.442&level=1

[frodrigo]

Since the regions were splitted these errors have come back again http://osmose.openstreetmap.fr/en/map/#item=5030&zoom=10&lat=43.019&lon=-2.442&level=1

Yes, we have to update the list

not_for = ["ES-NC", "ES-PV", "ES-AS"]

Please can you give the new list to exclude ?

https://github.com/osm-fr/osmose-backend/blob/master/osmose_config.py#L1509-L1580

[Crashillo]

Mmmmmm, that not_for list was the previous one? If so, the new one will be: "ES-O", "ES-NA", "ES-BI", "ES-SS", "ES-VI"

[frodrigo]

Deployed. But manual clean up required.

[frodrigo]

Manualy removed.
Wating to check after next update.

asturias 13635
comunidad_foral_de_navarra 14042
euskadi/vizcaya 427247
euskadi/guipuzcoa 427302
euskadi/alava 427192

DELETE FROM markers where item = '5030' and class='705' and source_id in (13635, 14042, 427247, 427302, 427192);
DELETE FROM markers_counts where item = '5030' and class='705' and source_id in (13635, 14042, 427247, 427302, 427192);
-- DELETE FROM class where item = '5030' and class='705' and source_id in (13635, 14042, 427247, 427302, 427192);

[frodrigo]

Looks good.

[Crashillo]

I'm noticing an issue here that should't. The element has these tags:

name = Xichón / Gillón
name:ast = Xichón
name:es = Gillón 

I presume the issue comes up due to the order. Osmose expects name = Gillón / Xichón, but that's wrong. The official name is in Asturian, then Spanish. But the solution is not about reversing the order, since another places are named firstly in Spanish, then Asturian.

Either you duplicate the following condition to deal ["es", "ast"] and ["ast", "es"], or you can check just the presence of the tags, no matter the order.

https://github.com/frodrigo/osmose-backend/blob/16582a0245fb8a874044c55dcac47aab831515c4/plugins/Name_Multilingual.py#L72

Same case for Navarre, basque / spanish, or spanish / basque, both are right. Perhaps the same for Basque country