Code is now organized by api version. So formatting has moved to
the api.v1 module. Instead of holding a separate ResultFormatter
object per result format, simply move the functions to the
formater collector and hand in the requested format as a parameter.
Thus reorganized, the api.v1 module can export three simple functions
for result formatting which in turn makes the code that uses
the formatters much simpler.
Use a directory for the submodule where the __init__ file contains
the public API. This makes it easier to separate public interface
from the internal implementation.
When a associatedStreet relation has multiple street members
always take the closest one. Avoid geometry operations for
the frequent case that there is only one street.
This makes Nominatim compatible with osm2pgsql's default update
modus operandi of deleting and reinserting data. Deletes are diverted
into a TODO table instead of executing them. When data is reinserted,
the corresponding entry in the TODO table is deleted. After updates are
finished, the remaining entries in the TODO table are executed, doing
the same work as the delete trigger did before.
The new behaviour also works against the gazetteer output with its
insert-only mechanism.
The search 'landstr' produces many duplicates so that with
some bad luck 4 or less results may appear. Disable deduplication
to make it more predictable.
The index over geometry_sectors are mainly used for ordering
the places which need indexing. That means they function effectively
as a TODO list. Consolodate them so that they always only contain
the places which are still to do. Also add the appropriate index
for the boundary indexing phase.
When deciding if an address interpolation has address information, only
look for addr:street and addr:place. If they are not there go looking
for the address on the address nodes. Ignores irrelevant tags like
addr:inclusion.
Fixes#2797.
When a boundary or place changes its address rank, all places where
it participates as address need to be potentially reindexed.
Also use the computed rank when testing place nodes against
boundaries. Boundaries are computed earlier.
Fixes#2794.
The functional split betweenthe two functions is now that the
first one creates the ID that is used in the word table and
the second one creates the variants. There no longer is a
requirement that the ID is the normalized version. We might
later reintroduce the requirement that a normalized version be available
but it doesn't necessarily need to be through the ID.
The function that creates the ID now gets the full PlaceName. That way
it might take into account attributes that were set by the sanitizers.
Finally rename both functions to something more sane.
