Switch from an recursive algorithm for computing the word sets
to an iterative one that benefits from caching intermediate
results. This considerably reduces the amount of memory needed,
so that the depth restriction can be dropped. To ensure that
the number of word sets remains manageable, only sets up to
a certain length are accepted and only a certain number of
total word sets. If word sets need to be dropped, we drop
the ones with more words per word set first.
To further reduce the number of potential word sets, the valid
tokens are looked up first and then only word sets containing
valid tokens are computed.
Fixes#1403, #1404 and #654.
The owner should never be restored, the table should be owned
by the caller instead. Non-existing indexes and similar only
started to throw a warning with Postgresql 9.4 and later, so
ignore them explicitly there.
The same result may be found with different result ranks
in the same search loop when housenumber or postcode are
part of the name or address. In this case we need to keep
the result with the lower result rank.
Fixes#1264.
By starting with the lowest address level, when collecting
administrative boundaries by level, there is a better chance
to actually get the boundary when the capital of the
administrative boundary is tagged with the level of the
boundary as well.
This is just a heuristics adaption to make the best out of
the imprecise admin_level definition for place nodes.
Fixes#1261.
The initial search and address rank is saved in a table
that is set up from a json configuration file. Ranks may
be assigned on a country level according to class and
type of the object. Special handling that depends on the
geometry or OSM type is still hard-coded in placex insert.
The new default config file mimicks the current assignment
as close as possible. A couple of exceptions have been
removed, most notably the exception for Irish townlands.
- factor out runWithEnv
- require explicit connect() call to avoid rechecking for oDB
(more for readability than for speed)
- clean DSNInfo of empty strings and simplify check for entries
Switch to functions server_version_num and postgis_lib_version
which both only return the version string, so that no elaborate
string parsing is necessary anymore. The version string could
become especially cumbersome in pre-release versions.
Now that we have result ranking, we can keep the street results
for housenumber searches and reuse them in the next group round
if required. Also fixes an issue where postcode and housenumber
are in the query and one of them is wrong.
Fixes#1200.
