[patch] Update railway gauge values

[gaben]

I updated the railway gauge values from Wikipedia ​https://en.wikipedia.org/wiki/Track_gauge.

Also, the narrow track has its own gauge values, as the values from other presets won't fit the narrow track requirement.

[mdk]

And there are much more (see ​https://de.wikipedia.org/wiki/Liste_der_Spurweiten), but I think the patch covers the mostly used values.

[skyper]

Most common values are listed here: ​https://de.wikipedia.org/wiki/Spurweite_(Bahn)
Interesting, everything below 1435 mm is considered a narrow gauge but we probably have to take a closer look at the actual use within the OSM data.

@Gabe
Did you plan any validation?

[gaben]

Yes, I'm planning some range restrictions, but only for narrow gauge railways. Also open to suggestions, I haven't thought much about it yet.

[gaben]

Something like this

/* #21866  */
*[/railway$/=~/^narrow_gauge$/][gauge][gauge<89],
*[/railway$/=~/^narrow_gauge$/][gauge][gauge>1434] {
  group: tr("suspicious tag combination");
  throwWarning:  tr("unusual gauge value on narrow gauge railway");
  assertNoMatch: "way railway=narrow_gauge gauge=88";
  assertMatch:   "way railway=narrow_gauge gauge=89";
  assertMatch:   "way railway=narrow_gauge gauge=1434";
  assertNoMatch: "way railway=narrow_gauge gauge=1435";
}

@skyper what dou you think?

[skyper]

Nice. Why do you use a regex for railway? Can we use placeholders? Do we need [gauge]?

/* #21866  */
*[railway=narrow_gauge][gauge<89],
*[railway=narrow_gauge][gauge>1434] {
  group: tr("suspicious tag combination");
  throwWarning:  tr("unusual {2} value on {1} {0}, should be between 89 and 1434 mm", "{0.key}", "{0.value}", "{1.key}");
  assertNoMatch: "way railway=narrow_gauge gauge=88";
  assertMatch:   "way railway=narrow_gauge gauge=89";
  assertMatch:   "way railway=narrow_gauge gauge=1434";
  assertNoMatch: "way railway=narrow_gauge gauge=1435";
}

I find some multiple values with taginfo (e.g. 127;184;260), so we probably have to split at the semicolon and check each value of the list.
Additionally some values are in feet and inches (e.g. 0'10.25";0'7.25" or 56' 1/2") which should be excluded or at least show a different warning.
Looks like it is not that simple but manageable.

[gaben]

Why do you use a regex for railway? Can we use placeholders?

To include lifecycle prefixed objects in the test because gauge applies to them as well. It seems placeholders are possible only if we drop regex.

I intentionally left out multi values as *it doesn't worth the hassle IMO. Same for non-metric units as OSM defaults to metric and the ​Key:gauge mentions millimetres only.

*But here you go, a new version with multi value support:

/* #21866  */
*[/railway$/=~/^narrow_gauge$/][gauge!~/^((14(?:3[0-4]|[4-9])|(?:14[0-2]|(?:1[0-3]|9)[0-9])[0-9]?|143|(?:[2-7][0-9]|1[5-9])[0-9]|8(?:[0-8][0-9]|9[0-9]?));?)+$/] {
  group: tr("suspicious tag combination");
  throwWarning:  tr("unusual gauge value on narrow gauge railway");
  assertMatch:   "way railway=narrow_gauge gauge=88";
  assertNoMatch: "way railway=narrow_gauge gauge=89";
  assertNoMatch: "way railway=narrow_gauge gauge=1434";
  assertMatch:   "way railway=narrow_gauge gauge=1435";
  assertMatch:   "way railway=narrow_gauge gauge=1435;1500";
  assertMatch:   "way railway=narrow_gauge gauge=60;600";
}

[skyper]

I think you little misunderstood my intention. I feared false positives.
Almost perfect.
I just noticed, that there is already a rule for gauge=* in numeric.mapcss​ which could/should be used as local class to get rid of the false positives, like non-metric values.

The assert(No)Match seem to be mixed up. JOSM seems to have problem with them though as I do not get warnings on the console

Personally I thought about something like below but it does not work. I have no clue what I do wrong.

way[/railway$/=~/^narrow_gauge$/][get(sort_list(uniq_list(trim_list(split(";", tag("gauge"))))), 0)<89],
way[/railway$/=~/^narrow_gauge$/][get(sort_list(uniq_list(trim_list(split(";", tag("gauge"))))), count(uniq_list(trim_list(split(";", tag("gauge"))))-1))>1434] {
  throwWarning:  tr("wrong {0} value on {1}", "{1.key}", "narrow gauge railway");
  group: tr("suspicious tag combination");
  assertNoMatch: "way railway=narrow_gauge gauge=88'";
  assertMatch: "way railway=narrow_gauge gauge=88";
  assertNoMatch:   "way railway=narrow_gauge gauge=89";
  assertNoMatch:   "way railway=narrow_gauge gauge=1434";
  assertMatch: "way railway=narrow_gauge gauge=1435";
  assertMatch: "way railway=narrow_gauge gauge=1435;1500";
  assertMatch:   "way railway=narrow_gauge gauge=60;600";
}

[gaben]

Thanks! Updated assertions in my last comment.

Yeah, I forgot the non-metric units handled in the numeric tests, that's another reason I left them out.

Please give examples on false positives.

[skyper]

Oh, I missed the object type selector. Is way enough or do you want to catch the route=railway and route=tracks relations?

The wiki was changed 1½ years ago, demanding on millimeters only. The translated pages still mention other values.
With false positives I mean values in feet and inches or narrow which is regarded as ok by the existing rule.

[gaben]

JOSM seems to have problem with them though as I do not get warnings on the console

Just enable the validator.check_assert_local_rules setting in the advanced preferences.

I want to catch *, it doesn't harm.

If an object is tagged with railway=narrow_gauge + gauge=narrow and the validator complains with unusual gauge value on narrow gauge railway message, that's OK for me. The problem is the grouping because this time it isn't a suspicious tag combination, just an unusual value.

[skyper]

validator rules using local classes

[skyper]

Here is my proposal for validator: josm_21866_numeric.patch​

• using classes to get rid of duplicates and incorrect warnings
• add a new warning for imprecise values
• only check ways and relations

I prefer to only run tests on certain objects instead of all. Nodes with railway=narrow_gauge are suspicious but better placed in a separate warning. The relations could even be restricted to type=route, I guess.

Replying to gaben:

JOSM seems to have problem with them though as I do not get warnings on the console

Just enable the validator.check_assert_local_rules setting in the advanced preferences.

Thanks, I was using the wrong profile. Now, it is working.

[skyper]

version 2: fix typo in article and add type=route for relations

[skyper]

version 2​ fixes a typo and adds type=route for relations.

[gaben]

Replying to skyper:

Nodes with railway=narrow_gauge are suspicious

Oopsie, you are right. I was thinking the whole time about switches with gauge, don't know why.

[gaben]

Original patch with gauge validator with the help of @skyper: josm_21866_track_gauge_with_validator.patch​

Changes in validator rule:

• add comment for warned range
• exclude gauge from narrow {gauge} as it's an expression, cannot be translated separately
• reordered rules following node, way, relation order
• reworded the millimetre suggestion message

[gaben]

Hm, I'm getting the "unusual gauge value on narrow gauge railway" warning twice, but not with the original version.

[skyper]

Replying to gaben:

Hm, I'm getting the "unusual gauge value on narrow gauge railway" warning twice, but not with the original version.

What kind of values produces two warnings? I have tested with some values and everything seems to be alright.

gauge=1000;1400
gauge=1600
gauge=2''11'
gauge=70;1600
gauge=narrow

[gaben]

Now three times, what's going on.

Please tag a new way and see if it works as intended. It may be just my environments...

railway=narrow_gauge
gauge=60

[skyper]

I only get one warning like all my examples above. By chance, do you have several local files with identical/similar rules active? Quick info, by hovering over the warning, should give you a hint about the origin of each warning.

[gaben]

Thanks and sorry for the noise. Yes, there was a copy in the combinations validator file. The popup hint is very useful.

I have 10 different changelist in my IDE, and it's getting harder to maintain.

[skyper]

Nice, ready for commit!

[stoecker]

In 18385/josm:

fix #21866 - railway gauge improvements - patch by gaben