Better support for cuisine=*

[mdk]

I want to map the "Texas Steakhaouse", but the JOSM preset has no suitable value. So I looked into the Wiki page, which lists about 90 values and manually use cuisine=steak_house.

I wonder why these values are not in the preset (and validator complain about it), so I looked into taginfo. The cuisine tag is used 347272 times with 10628 (!) different values. And I saw several, cases where clearly are typos:

steak_house	2400
steakhouse	36
Steakhouse	23
Steak_House	12
steak-house	3
Steak_house	3
steak␣house	3
steack_house	2
Steak␣House	1
Steakhaus	1

And some similar values:

steak	359
Steak	27
steak_grill	12
steaks	12
Steaks	1
ステーキ␣(Steak)	1
Beefsteak_house	1
steack	1
Steakhouse_chain	1

Many off the different values are combinations like (as an example I just pick out the combination with seafood):

steak;seafood	15
seafood;steak	5
steak_house;seafood	3
steak_house;fish	3
Steak_and_Seafood	3
seafood;steak_house	2
seafood␣and␣steaks	2
seafood,␣steak	2
seafood;steaks	2
fisch,steaks	1
seafood,steak_house	1
seafood,steak	1
Seafood_and_Steaks	1
steak,␣seafood	1
steak_&_seafood	1
Steak,_Fisch_und_mehr	1
Steak,_sea_food	1
Steak_&_Fish	1
Steaks_and_Seafood	1
steak;fish	1
steak;␣sea_food;	1
seafood,_steak	1
Steak,_Seafood	1
steak;seafood;	1
fish;steaks	1
steak;␣seafood	1
steaks_and_seafood	1
seafood,␣steack_house	1
Steaks,_Seafood	1

What can we do to reduce this mess?

• Should we add more values to the preset? Nice if the preset and the wiki would be in sync.
• Should we mark deprecated values more clearly in the wiki (like fish or sub)?
• Should we warn about deprecated values and offer a fix (like replacing cuisine=fish with cuisine=seafood)
• Should we have some pretty type mechanism like for opening_hours?

If we have a value which is not in the preset and we do operations like:

• convert to lower case
• replace ' ' or '-' with '_'
• replace ',' with ';'
• remove trailing or leading special chars like ' ', ';', '_'

and we now find the result in the preset, warn and offer a fix to replace the value.

Additionally we could also have a test where we remove the '_' from the preset value. If we find a match in this case like "steakhouse" would match "steak_house" without '_', we can warn and offer a fix to replace cuisine=steakhouse with cuisine=steak_house.

We should split combined values (separated by ';') and check each part separatly. And perhas these parts should be sorted alphabetically (prefer cuisine=seafood;steak instead of cuisine=steak;seafood).

Perhaps this could be a generic validation rule for many tags with a given list of values plus user defined values.

[stoecker]

In 8353/josm:

fix #11433 - add steak_house as cuisine

[stoecker]

Added steak_house.

For the typos in database: You can load them in JOSM, verify them and fix entries. Please don't do automatic fixes, as usually bad values go together with other errors.

Jochen started a cleanup challenge some time ago: ​http://blog.jochentopf.com/2015-03-05-new-taginfo-features-and-a-challenge.html

[aceman]

Why is the "house" is the name needed? Is that the common name of the cuisine (not the restaurant facility) in the US?