]> git.openstreetmap.org Git - nominatim.git/log
nominatim.git
4 years agorevert to && comparison for geometries
Sarah Hoffmann [Fri, 16 Oct 2020 07:49:48 +0000 (09:49 +0200)]
revert to && comparison for geometries

Postgis 3 picks the wrong index when using ~ or @.

4 years agouse computed centroid for location_area_large
Sarah Hoffmann [Thu, 15 Oct 2020 15:30:52 +0000 (17:30 +0200)]
use computed centroid for location_area_large

The new address computation assumes that the centroid is inside
the area. Therefore we cannot use the centroid function. Use the
pre-computed centroid instead which has already been corrected to
be inside the area.

4 years agocorrectly set from area column in place_addressline
Sarah Hoffmann [Thu, 15 Oct 2020 10:06:53 +0000 (12:06 +0200)]
correctly set from area column in place_addressline

This was always set to true which brings us to the question
if it is even still needed.

4 years agofix use of geometry operators
Sarah Hoffmann [Thu, 15 Oct 2020 10:06:18 +0000 (12:06 +0200)]
fix use of geometry operators

@ is contained by while ~ is contains.

4 years agodemote admin boundaries for place areas
Sarah Hoffmann [Wed, 14 Oct 2020 09:33:47 +0000 (11:33 +0200)]
demote admin boundaries for place areas

Also demote the address rank of an admin boundary when there
is a place area of higher rank that completely contains the
area. This catches the case where city boundaries do not exactly
align with administrative units (see for example Moscow).

4 years agoexclude rank 25 when computing addresses of streets
Sarah Hoffmann [Tue, 13 Oct 2020 20:36:17 +0000 (22:36 +0200)]
exclude rank 25 when computing addresses of streets

Address rank 25 is used for squares which are address-wise on the
same level as streets.

4 years agooverhaul address computation
Sarah Hoffmann [Tue, 13 Oct 2020 20:10:07 +0000 (22:10 +0200)]
overhaul address computation

This is a complete rewrite of the selection of address parts to
be inserted into the place_addressline table.

The new algorithm selects for each rank:
* the boundary overlapping with the addressee and contained
  in the already selected boundaries of lower rank, or failing that
* the place node closest to the addressee that is contained in
  the already selected boundaries and in the influence radius
  of already selected place nodes of lower rank

Place nodes that are not contained in already selected boundaries
of lower rank are completely thrown away. All other candidates are
added as non-address parts.

4 years agomove ordering out of getNearFeatures
Sarah Hoffmann [Tue, 13 Oct 2020 12:57:11 +0000 (14:57 +0200)]
move ordering out of getNearFeatures

The two places where the function is called have different ordering
requirement.

4 years agoMerge pull request #2012 from lonvia/format-reverse-debug-output
Sarah Hoffmann [Tue, 13 Oct 2020 07:23:43 +0000 (09:23 +0200)]
Merge pull request #2012 from lonvia/format-reverse-debug-output

use Debug class for formatting reverse debug output

4 years agouse Debug class for formatting reverse debug output
Sarah Hoffmann [Mon, 12 Oct 2020 15:12:03 +0000 (17:12 +0200)]
use Debug class for formatting reverse debug output

4 years agoMerge pull request #2011 from lonvia/increase-city-radius
Sarah Hoffmann [Mon, 12 Oct 2020 14:17:39 +0000 (16:17 +0200)]
Merge pull request #2011 from lonvia/increase-city-radius

Increase radius of influence around city nodes

4 years agoincrease radius of influence around city nodes
Sarah Hoffmann [Mon, 12 Oct 2020 12:17:37 +0000 (14:17 +0200)]
increase radius of influence around city nodes

The current radius does not cover cities with more than a
million inhabitants well.

4 years agoMerge pull request #2004 from lonvia/demote-place-nodes-in-admin-areas
Sarah Hoffmann [Mon, 12 Oct 2020 09:54:52 +0000 (11:54 +0200)]
Merge pull request #2004 from lonvia/demote-place-nodes-in-admin-areas

demote place nodes in admin areas

4 years agowhen linking always check against original address rank
Sarah Hoffmann [Sun, 11 Oct 2020 10:29:49 +0000 (12:29 +0200)]
when linking always check against original address rank

4 years agomake housenumber interpolation tests more lenient
Sarah Hoffmann [Sun, 11 Oct 2020 08:32:35 +0000 (10:32 +0200)]
make housenumber interpolation tests more lenient

4 years agodemote place nodes in admin areas
Sarah Hoffmann [Wed, 7 Oct 2020 15:33:52 +0000 (17:33 +0200)]
demote place nodes in admin areas

If a place node of city rank and above finds itself in an
administrative boundary of the same address rank, then
increase the address rank by 2. This catches the rather
frequent case where city suburbs are tagged for historical
reasons as towns or villages.

4 years agoMerge pull request #2008 from lonvia/docs-missing-index-migration
Sarah Hoffmann [Sun, 11 Oct 2020 09:05:59 +0000 (11:05 +0200)]
Merge pull request #2008 from lonvia/docs-missing-index-migration

docs: migration to new wikipedia needs new index

4 years agodocs: migration to new wikipedia needs new index
Sarah Hoffmann [Sun, 11 Oct 2020 08:40:23 +0000 (10:40 +0200)]
docs: migration to new wikipedia needs new index

Fixes #1998.

4 years agoMerge pull request #2006 from mtmail/ubuntu-20-postgresql-contrib
Sarah Hoffmann [Sun, 11 Oct 2020 07:45:37 +0000 (09:45 +0200)]
Merge pull request #2006 from mtmail/ubuntu-20-postgresql-contrib

Ubuntu 20: use postgresql-contrib-12 so no version higher gets installed

4 years agoMerge pull request #2003 from lonvia/admin-levels-indonesia
Sarah Hoffmann [Sun, 11 Oct 2020 07:43:33 +0000 (09:43 +0200)]
Merge pull request #2003 from lonvia/admin-levels-indonesia

Adapt address levels for admin boundaries in Indonesia

4 years agoUbuntu 20: use postgresql-contrib-12 so no version higher gets installed
marc tobias [Sat, 10 Oct 2020 00:42:46 +0000 (02:42 +0200)]
Ubuntu 20: use postgresql-contrib-12 so no version higher gets installed

4 years agoadapt address levels for admin boundaries in Indonesia
Sarah Hoffmann [Fri, 9 Oct 2020 20:28:06 +0000 (22:28 +0200)]
adapt address levels for admin boundaries in Indonesia

4 years agoMerge pull request #2002 from lonvia/analyse-indexing-script
Sarah Hoffmann [Fri, 9 Oct 2020 18:23:51 +0000 (20:23 +0200)]
Merge pull request #2002 from lonvia/analyse-indexing-script

Add script for detailed explaing of indexing trigger

4 years agoadd script for detailed explaing of indexing trigger
Sarah Hoffmann [Fri, 9 Oct 2020 15:20:22 +0000 (17:20 +0200)]
add script for detailed explaing of indexing trigger

4 years agoMerge pull request #1996 from lonvia/remove-postcode-search-structured
Sarah Hoffmann [Tue, 6 Oct 2020 15:04:22 +0000 (17:04 +0200)]
Merge pull request #1996 from lonvia/remove-postcode-search-structured

Restrict postcode searches to postcode in first token

4 years agorestrict postcode searches to postcode in first token
Sarah Hoffmann [Tue, 6 Oct 2020 12:00:43 +0000 (14:00 +0200)]
restrict postcode searches to postcode in first token

In structured queries we should only assume that it is
a postcode search when only the postcode and optionally
the country is given. If any other term is present, it
is better to avoid the search for postcode as it yields
too many bad searches. Given that the terms in a structured
query are ordered, this means that the postcode must be
the first token just like in the unstructured query.

Fixes #1988.

4 years agoRevert "adapt tests to rank_search removal"
Sarah Hoffmann [Tue, 6 Oct 2020 11:59:50 +0000 (13:59 +0200)]
Revert "adapt tests to rank_search removal"

This reverts commit 2a717da8508ed697bfe6f5eded4e397054570a7a.

4 years agoMerge pull request #1995 from lonvia/update-osm2pgsql
Sarah Hoffmann [Mon, 5 Oct 2020 16:14:29 +0000 (18:14 +0200)]
Merge pull request #1995 from lonvia/update-osm2pgsql

update to latest osm2pgsql version

4 years agoupdate to latest osm2pgsql version
Sarah Hoffmann [Mon, 5 Oct 2020 15:11:13 +0000 (17:11 +0200)]
update to latest osm2pgsql version

The latest version of osm2pgsql no longer creates indexes on
the members of planet_osm_rels. So we do that ourselves.
Given that we only need to access associated street relations,
the index can become quite a bit smaller.

4 years agoMerge pull request #1986 from mtmail/document-drop-idx_placex_geometry_reverse_lookup...
Sarah Hoffmann [Mon, 5 Oct 2020 14:18:46 +0000 (16:18 +0200)]
Merge pull request #1986 from mtmail/document-drop-idx_placex_geometry_reverse_lookupPoint

migration guide: idx_placex_geometry_reverse_lookupPoint can be dropped

4 years agoMerge pull request #1993 from mtmail/country-name-eSwatini
Sarah Hoffmann [Mon, 5 Oct 2020 14:17:54 +0000 (16:17 +0200)]
Merge pull request #1993 from mtmail/country-name-eSwatini

country names: Swaziland => eSwatini

4 years agocountry names: Swaziland => eSwatini
marc tobias [Sun, 4 Oct 2020 11:38:15 +0000 (13:38 +0200)]
country names: Swaziland => eSwatini

4 years agomigration guide: idx_placex_geometry_reverse_lookupPoint can be dropped
marc tobias [Wed, 30 Sep 2020 12:28:20 +0000 (14:28 +0200)]
migration guide: idx_placex_geometry_reverse_lookupPoint can be dropped

4 years agoMerge pull request #1984 from lonvia/remove-reverse-index
Sarah Hoffmann [Wed, 30 Sep 2020 10:02:34 +0000 (12:02 +0200)]
Merge pull request #1984 from lonvia/remove-reverse-index

Remove unused idx_placex_geometry_reverse_lookupPoint

4 years agoremove removed index from database check
Sarah Hoffmann [Wed, 30 Sep 2020 09:33:15 +0000 (11:33 +0200)]
remove removed index from database check

4 years agoremove unused idx_placex_geometry_reverse_lookupPoint
Sarah Hoffmann [Wed, 30 Sep 2020 07:21:35 +0000 (09:21 +0200)]
remove unused idx_placex_geometry_reverse_lookupPoint

The index has been unused ever since the query using it was
changed two years ago. Given that it has not been missed much,
drop it completely here.

4 years agoMerge pull request #1982 from lonvia/more-rank-search-removal
Sarah Hoffmann [Sat, 26 Sep 2020 09:08:20 +0000 (11:08 +0200)]
Merge pull request #1982 from lonvia/more-rank-search-removal

More rank search removal

4 years agoadapt tests to rank_search removal
Sarah Hoffmann [Sat, 26 Sep 2020 07:10:37 +0000 (09:10 +0200)]
adapt tests to rank_search removal

4 years agoRemove more rank_search usage from address computation
Sarah Hoffmann [Fri, 25 Sep 2020 15:50:36 +0000 (17:50 +0200)]
Remove more rank_search usage from address computation

Fixes #1904.

4 years agoMerge pull request #1980 from lonvia/add-descriptive-term-for-address-rank-24
Sarah Hoffmann [Fri, 25 Sep 2020 14:56:35 +0000 (16:56 +0200)]
Merge pull request #1980 from lonvia/add-descriptive-term-for-address-rank-24

add descriptive term for address rank 24

4 years agoadd descriptive term for address rank 24
Sarah Hoffmann [Fri, 25 Sep 2020 14:02:17 +0000 (16:02 +0200)]
add descriptive term for address rank 24

With that term we have terms for all ranks, so that no generic
'administrative' term will show up in the address details anymore.

4 years agoMerge pull request #1975 from lonvia/simplify-parent-assignment-for-unlisted-places
Sarah Hoffmann [Wed, 23 Sep 2020 17:10:42 +0000 (19:10 +0200)]
Merge pull request #1975 from lonvia/simplify-parent-assignment-for-unlisted-places

Use closest containing place area for parent of unlisted addr:place

4 years agoMerge pull request #1970 from lonvia/remove-duplicate-geometry-check
Sarah Hoffmann [Wed, 23 Sep 2020 16:51:35 +0000 (18:51 +0200)]
Merge pull request #1970 from lonvia/remove-duplicate-geometry-check

Switch recursive updates to using rank_search

4 years agouse rank_address when invalidating containing objects
Sarah Hoffmann [Tue, 22 Sep 2020 14:34:28 +0000 (16:34 +0200)]
use rank_address when invalidating containing objects

Only rank_address is now relevant for determining if a place
could be part of an address.

4 years agoremove ST_Covers check when also testing for ST_Intersects
Sarah Hoffmann [Tue, 22 Sep 2020 13:51:04 +0000 (15:51 +0200)]
remove ST_Covers check when also testing for ST_Intersects

Using both is slightly problematic because they have different
ways to use the index. Newer versions of Postgis exhibit a
query planner issue when both functions appear together.
As ST_Intersects includes ST_Covers, simply remove the latter.

4 years agouse closest containing place for unlisted addr:place
Sarah Hoffmann [Wed, 23 Sep 2020 15:33:42 +0000 (17:33 +0200)]
use closest containing place for unlisted addr:place

We can't use getNearFeatures() to determine the parent of a
place with an unlisted addr:place because this function
returns place nodes that are potentially outside the area
of interest. Doing the complete address computation is too
expensive, so simply use the area with the largest rank that
contains the feature instead.

4 years agoMerge pull request #1974 from lonvia/show-unknown-addr-place
Sarah Hoffmann [Wed, 23 Sep 2020 13:26:12 +0000 (15:26 +0200)]
Merge pull request #1974 from lonvia/show-unknown-addr-place

add unknown addr:place to address output

4 years agoadd unknown addr:place to address output
Sarah Hoffmann [Wed, 23 Sep 2020 09:55:18 +0000 (11:55 +0200)]
add unknown addr:place to address output

When a POI has no addr:street but an addr:place that is not
contained in the name list of the parent place, then remember
this situation and merge the content of addr:place into the
address output.

We don't need to care about translations in this case because
it is obvious that no object with translations exists if the
parent isn't the object named in addr:place.

4 years agoMerge pull request #1971 from lonvia/drop-support-for-isin
Sarah Hoffmann [Wed, 23 Sep 2020 07:20:35 +0000 (09:20 +0200)]
Merge pull request #1971 from lonvia/drop-support-for-isin

Drop support for is_in tag

4 years agoMerge pull request #1972 from lonvia/exclude-unnamed-highway-areas
Sarah Hoffmann [Wed, 23 Sep 2020 07:20:16 +0000 (09:20 +0200)]
Merge pull request #1972 from lonvia/exclude-unnamed-highway-areas

Exclude unnamed highway areas

4 years agoexclude unnamed highway areas
Sarah Hoffmann [Tue, 22 Sep 2020 11:27:05 +0000 (13:27 +0200)]
exclude unnamed highway areas

These are used to mark large paved areas. Sometimes they exists
together with named regular streets. In such cases the unnamed
area may overshadow the actual street when computing the address
parent. As unnamed highways are not very useful anyway, we
simply remove them from the database.

4 years agoremove test for is_in
Sarah Hoffmann [Tue, 22 Sep 2020 19:36:49 +0000 (21:36 +0200)]
remove test for is_in

4 years agodrop suport for is_in tag
Sarah Hoffmann [Sun, 13 Sep 2020 08:53:25 +0000 (10:53 +0200)]
drop suport for is_in tag

4 years agoMerge pull request #1965 from lonvia/make-addr-tags-searchable
Sarah Hoffmann [Tue, 22 Sep 2020 18:21:06 +0000 (20:21 +0200)]
Merge pull request #1965 from lonvia/make-addr-tags-searchable

Make addresses searchable by their addr: tags

4 years agoMerge pull request #1966 from lonvia/remove-dead-code
Sarah Hoffmann [Mon, 21 Sep 2020 08:59:58 +0000 (10:59 +0200)]
Merge pull request #1966 from lonvia/remove-dead-code

remove dead code

4 years agoremove dead code
Sarah Hoffmann [Mon, 21 Sep 2020 08:39:14 +0000 (10:39 +0200)]
remove dead code

The SQL query has moved into the addTokensFromDB() funtion.

4 years agoalways bind addr:place to place instead of street
Sarah Hoffmann [Thu, 3 Sep 2020 19:42:00 +0000 (21:42 +0200)]
always bind addr:place to place instead of street

If an addr:place is given but no addr:street tag, then bind
the rank 30 object always to a <=25 object, even when there
is none found with the same name.

4 years agomerge addr tags into search_name table
Sarah Hoffmann [Thu, 3 Sep 2020 08:38:33 +0000 (10:38 +0200)]
merge addr tags into search_name table

When a place of rank 30 has addr tags that are not covered by the
search terms of the parent, add a separate entry for the POI in
the search_name table that includes the addr tags. We can only
do that with named places. For POIs without a name the housenumber
is used as name. If that is not available either, searching still
won't work.

4 years agoMerge pull request #1964 from lonvia/remove-postcodes-with-colon
Sarah Hoffmann [Sat, 19 Sep 2020 15:47:43 +0000 (17:47 +0200)]
Merge pull request #1964 from lonvia/remove-postcodes-with-colon

ignore postcodes with colons

4 years agoignore postcodes with colons
Sarah Hoffmann [Sat, 19 Sep 2020 15:23:40 +0000 (17:23 +0200)]
ignore postcodes with colons

Colons are used as a delimiter in tiger:left and tiger:right tags
when multiple postcodes are given. Ignore those. This was already
done in the postcode update script. This changes just makes the
two places consistent where postcodes are added.

4 years agoUpdate broken links in documentation
Ben Antony [Thu, 17 Sep 2020 13:11:01 +0000 (15:11 +0200)]
Update broken links in documentation

4 years agoMerge pull request #1963 from lonvia/remove-postcodes-from-search-index
Sarah Hoffmann [Fri, 18 Sep 2020 20:41:24 +0000 (22:41 +0200)]
Merge pull request #1963 from lonvia/remove-postcodes-from-search-index

Remove postcodes from search index

4 years agoremove special casing for rank 25 postcodes
Sarah Hoffmann [Fri, 18 Sep 2020 14:18:02 +0000 (16:18 +0200)]
remove special casing for rank 25 postcodes

They can be computed like any other place.

4 years agouse same label for all types of postcode in address
Sarah Hoffmann [Fri, 18 Sep 2020 14:17:30 +0000 (16:17 +0200)]
use same label for all types of postcode in address

4 years agoremove postcodes entirely from indexing
Sarah Hoffmann [Fri, 18 Sep 2020 13:09:35 +0000 (15:09 +0200)]
remove postcodes entirely from indexing

place=postcode places are artificial places that collect addr:postcode
points for aggration. They should neither show up in the address nor
be searchable. That means that there is no need to index them at all.
Only let boundary=postal_code through which define correct areas for
postcodes.

4 years agopostal boundary may be imported without name
Sarah Hoffmann [Fri, 18 Sep 2020 09:08:47 +0000 (11:08 +0200)]
postal boundary may be imported without name

Postal boundaries usually just have the postcode tag set and are
therefore officially 'nameless'. We want to have them as
boundary=postal_code anyways in order to distiguish them from postcode
points inherited from addr: tags.

4 years agoMerge pull request #1961 from lonvia/set-place-type-for-result-in-address
Sarah Hoffmann [Thu, 17 Sep 2020 19:23:40 +0000 (21:23 +0200)]
Merge pull request #1961 from lonvia/set-place-type-for-result-in-address

Use place type of for result object in address parts

4 years agoMerge pull request #1960 from lonvia/fix-postcodes-duplicated-by-normalization
Sarah Hoffmann [Thu, 17 Sep 2020 19:23:23 +0000 (21:23 +0200)]
Merge pull request #1960 from lonvia/fix-postcodes-duplicated-by-normalization

Make sure that all postcodes have an entry in the word table

4 years agoMerge pull request #1957 from lonvia/docs-separate-out-deployment
Sarah Hoffmann [Thu, 17 Sep 2020 19:22:48 +0000 (21:22 +0200)]
Merge pull request #1957 from lonvia/docs-separate-out-deployment

Restructure vagrant scripts and installation documentation

4 years agoMerge pull request #1962 from mtmail/travis-ci-without-webserver
Sarah Hoffmann [Thu, 17 Sep 2020 18:38:17 +0000 (20:38 +0200)]
Merge pull request #1962 from mtmail/travis-ci-without-webserver

travis-ci: we dont need Apache installed

4 years agotravis-ci: we dont need Apache installed
marc tobias [Thu, 17 Sep 2020 16:21:16 +0000 (18:21 +0200)]
travis-ci: we dont need Apache installed

4 years agouse place type of for result object in address parts
Sarah Hoffmann [Thu, 17 Sep 2020 16:17:01 +0000 (18:17 +0200)]
use place type of for result object in address parts

Boundaries shound derive the address part type from the
linked place if possible. This was already implemented
for the address objects but not for the address information
from the address itself.

Fixes #1949.

4 years agomake sure that all postcodes have an entry in word
Sarah Hoffmann [Thu, 17 Sep 2020 15:11:22 +0000 (17:11 +0200)]
make sure that all postcodes have an entry in word

It may happen that two different postcodes normalize to exactly
the same token. In that case we still need two different entries
in the word table. Token lookup will then make sure that the correct
one is choosen.

Fixes #1953.

4 years agoMerge pull request #1945 from mtmail/travis-ubuntu-20
Sarah Hoffmann [Thu, 17 Sep 2020 14:20:19 +0000 (16:20 +0200)]
Merge pull request #1945 from mtmail/travis-ubuntu-20

Upgrade Travis-CI from Ubuntu 18 to 20

4 years agodocs: fix formatting
Sarah Hoffmann [Thu, 17 Sep 2020 08:16:25 +0000 (10:16 +0200)]
docs: fix formatting

4 years agorestructure developer's manual
Sarah Hoffmann [Thu, 17 Sep 2020 07:54:46 +0000 (09:54 +0200)]
restructure developer's manual

Add a section on setting up the development environment which now
also includes the former chapter on recreating the documentation.
Move the README from test/ into the manual as the new Testing
chapter.

4 years agomake CentOS 8 the default vagrant script
Sarah Hoffmann [Wed, 16 Sep 2020 15:34:36 +0000 (17:34 +0200)]
make CentOS 8 the default vagrant script

This puts it in line with the Ubunutu scripts.

4 years agoadapt Ubunut-20 vagrant file to triple webserver config
Sarah Hoffmann [Wed, 16 Sep 2020 14:27:34 +0000 (16:27 +0200)]
adapt Ubunut-20 vagrant file to triple webserver config

4 years agorestructure webserver setup in ubuntu 18 script
Sarah Hoffmann [Wed, 16 Sep 2020 09:19:38 +0000 (11:19 +0200)]
restructure webserver setup in ubuntu 18 script

Unify the two vagrant scripts for Ubuntu 18. The script can now
be run in three modes: no webserver, with apache, with nginx.
The default mode is to not install any webserver at all. This is
normally sufficient when just developping.

The commit also switches from bento to generic boxes and adds config
for running with a libvirt provider. You need an NFS deamon for
synchronized folders.

4 years agoadd 'make serve-global'
Sarah Hoffmann [Wed, 16 Sep 2020 09:15:55 +0000 (11:15 +0200)]
add 'make serve-global'

This runs the PHP development server in a mode where it listens
globally. This is needed when running inside vagrant and port-forwarding
to the host machine.

4 years agoremove host from default website URL
Sarah Hoffmann [Wed, 16 Sep 2020 09:13:51 +0000 (11:13 +0200)]
remove host from default website URL

Just assume that Nominatim runs under the root URL. This is a
more versatile base that also makes 'make serve' work out of the
box.

4 years agodocs: move webserver installation into separate chapter
Sarah Hoffmann [Tue, 15 Sep 2020 21:51:25 +0000 (23:51 +0200)]
docs: move webserver installation into separate chapter

4 years agoUpgrade Travis-CI from Ubuntu 18 to 20
marc tobias [Sun, 13 Sep 2020 22:28:53 +0000 (00:28 +0200)]
Upgrade Travis-CI from Ubuntu 18 to 20

4 years agoMerge pull request #1943 from mtmail/no-get-magic-quotes
Sarah Hoffmann [Mon, 14 Sep 2020 06:47:49 +0000 (08:47 +0200)]
Merge pull request #1943 from mtmail/no-get-magic-quotes

starting PHP 5.4 get_magic_quotes_gpc() returns false, no need to check

4 years agostarting PHP 5.4 get_magic_quotes_gpc() returns false, no need to check
marc tobias [Sun, 13 Sep 2020 22:45:22 +0000 (00:45 +0200)]
starting PHP 5.4 get_magic_quotes_gpc() returns false, no need to check

4 years agoMerge pull request #1940 from mtmail/north-macedonia
Sarah Hoffmann [Sat, 12 Sep 2020 14:15:22 +0000 (16:15 +0200)]
Merge pull request #1940 from mtmail/north-macedonia

country_name Macedonia => North Macedonia

4 years agocountry_name Macedonia => North Macedonia
marc tobias [Mon, 7 Sep 2020 15:24:06 +0000 (17:24 +0200)]
country_name Macedonia => North Macedonia

4 years agoMerge pull request #1936 from lonvia/tweeking-of-ranks
Sarah Hoffmann [Tue, 1 Sep 2020 19:23:28 +0000 (21:23 +0200)]
Merge pull request #1936 from lonvia/tweeking-of-ranks

More fine tuning of default rank assignments

4 years agoadapt tests to ranking changes
Sarah Hoffmann [Tue, 1 Sep 2020 16:03:17 +0000 (18:03 +0200)]
adapt tests to ranking changes

4 years agotweak size of large POIs
Sarah Hoffmann [Tue, 1 Sep 2020 16:00:40 +0000 (18:00 +0200)]
tweak size of large POIs

Further reduce the size from which on POIs are no longer bound
to streets but only to larger objects. The point of reference,
of what a largest POI could be that is still bound is JFK airport.

4 years agoaddress rank adjustment for addressable boundaries only
Sarah Hoffmann [Tue, 1 Sep 2020 15:59:26 +0000 (17:59 +0200)]
address rank adjustment for addressable boundaries only

Only administrative boundaries with an address rank need
to be adjusted. Otherwise just handle them like any other
object.

4 years agoimprove fallback ranking
Sarah Hoffmann [Tue, 1 Sep 2020 15:55:40 +0000 (17:55 +0200)]
improve fallback ranking

Boundaries and places now always get a rank < 26 to make sure that
they do not parent to a street. Skip boundary=place completely
because they will be covered throught the secondary place tag.

4 years agoMerge pull request #1934 from lonvia/fix-deletion-of-large-highway-areas
Sarah Hoffmann [Fri, 28 Aug 2020 08:08:42 +0000 (10:08 +0200)]
Merge pull request #1934 from lonvia/fix-deletion-of-large-highway-areas

Do not block deletion of large highway areas

4 years agodo not block deletion of large highway areas
Sarah Hoffmann [Fri, 28 Aug 2020 07:49:21 +0000 (09:49 +0200)]
do not block deletion of large highway areas

Deletion of areas should only e blocked for addressable features.
Streets and POIs do not have a large impact on updates.

4 years agoplace lookup: filter places that have no details
Sarah Hoffmann [Thu, 27 Aug 2020 07:33:21 +0000 (09:33 +0200)]
place lookup: filter places that have no details

In rare cases search_name might have entries for places for
which we do not return details, in particular for linkees.
Need to remove those entries in the result list before returning
the details.

Fixes #1932.

4 years agoMerge pull request #1931 from lonvia/stable-sort-for-results
Sarah Hoffmann [Wed, 26 Aug 2020 18:52:17 +0000 (20:52 +0200)]
Merge pull request #1931 from lonvia/stable-sort-for-results

Reranking of results must be stable

4 years agoensure that ordering by importance is stable
Sarah Hoffmann [Wed, 26 Aug 2020 15:42:43 +0000 (17:42 +0200)]
ensure that ordering by importance is stable

The initial search results retrieved from the database already come
preordered, either by importnace or by distance. We want to keep
that order if all other things are equal.

4 years agoPlaceLookup should return results in input order
Sarah Hoffmann [Wed, 26 Aug 2020 15:15:11 +0000 (17:15 +0200)]
PlaceLookup should return results in input order

4 years agoMerge pull request #1930 from lonvia/add-support-for-squares
Sarah Hoffmann [Wed, 26 Aug 2020 13:11:45 +0000 (15:11 +0200)]
Merge pull request #1930 from lonvia/add-support-for-squares

Add support for place=square

4 years agoadd support for place=square
Sarah Hoffmann [Wed, 26 Aug 2020 10:12:52 +0000 (12:12 +0200)]
add support for place=square

Squares are now addressable (on address level 25) and thus can
be attached to a house number via addr:place. Needed to increase
the rank range for matching up addr:place to 25.