]> git.openstreetmap.org Git - nominatim.git/log
nominatim.git
3 years agointroduce a separate token type for partials
Sarah Hoffmann [Tue, 13 Jul 2021 14:54:51 +0000 (16:54 +0200)]
introduce a separate token type for partials

This means that the leading space can be removed as a partial
word indicator.

3 years agoMerge pull request #2393 from lonvia/fix-flake8-issues
Sarah Hoffmann [Tue, 13 Jul 2021 14:46:12 +0000 (16:46 +0200)]
Merge pull request #2393 from lonvia/fix-flake8-issues

Fix flake8 issues

3 years agouse psycopg's SQL quoting where possible
Sarah Hoffmann [Mon, 12 Jul 2021 20:05:22 +0000 (22:05 +0200)]
use psycopg's SQL quoting where possible

Use the SQL formatting supplied with psycopg whenever the
query needs to be put together from snippets.

3 years agoadd helper function for execute_values
Sarah Hoffmann [Mon, 12 Jul 2021 19:08:20 +0000 (21:08 +0200)]
add helper function for execute_values

Make psycopg2's convenience function accessible through
the cursor.

3 years agoprovide wrapper function for DROP TABLE
Sarah Hoffmann [Mon, 12 Jul 2021 18:32:46 +0000 (20:32 +0200)]
provide wrapper function for DROP TABLE

Use psycopg2 formatting to ensure correct quoting.

3 years agomore formatting fixes
Sarah Hoffmann [Mon, 12 Jul 2021 15:45:42 +0000 (17:45 +0200)]
more formatting fixes

Found by flake8.

3 years agoMerge pull request #2391 from lonvia/fix-sonar-issues
Sarah Hoffmann [Mon, 12 Jul 2021 15:14:59 +0000 (17:14 +0200)]
Merge pull request #2391 from lonvia/fix-sonar-issues

Fix bugs and code smells found by Sonarqube

3 years agofactor out connection reset code
Sarah Hoffmann [Mon, 12 Jul 2021 12:58:44 +0000 (14:58 +0200)]
factor out connection reset code

3 years agosimplify analyse function
Sarah Hoffmann [Mon, 12 Jul 2021 12:47:50 +0000 (14:47 +0200)]
simplify analyse function

3 years agosplit up variant computation for better readability
Sarah Hoffmann [Mon, 12 Jul 2021 12:43:50 +0000 (14:43 +0200)]
split up variant computation for better readability

3 years agoreorganise process_place function
Sarah Hoffmann [Mon, 12 Jul 2021 09:53:25 +0000 (11:53 +0200)]
reorganise process_place function

Move address processing into its own function as it is
rather extensive.

3 years agosimplify website setup code
Sarah Hoffmann [Mon, 12 Jul 2021 09:41:05 +0000 (11:41 +0200)]
simplify website setup code

Use formaat strings and move variable quoting code into extra
function.

3 years agoavoid repeated patterns for table name
Sarah Hoffmann [Mon, 12 Jul 2021 09:33:09 +0000 (11:33 +0200)]
avoid repeated patterns for table name

3 years agosimplify if statements
Sarah Hoffmann [Sun, 11 Jul 2021 22:16:25 +0000 (00:16 +0200)]
simplify if statements

3 years agoconvert single case switch to if statement
Sarah Hoffmann [Sun, 11 Jul 2021 21:48:16 +0000 (23:48 +0200)]
convert single case switch to if statement

3 years agoavoid local variable assignment
Sarah Hoffmann [Sun, 11 Jul 2021 21:22:16 +0000 (23:22 +0200)]
avoid local variable assignment

3 years agofix more missing braces on one-liners
Sarah Hoffmann [Sun, 11 Jul 2021 18:21:12 +0000 (20:21 +0200)]
fix more missing braces on one-liners

3 years agoremove dead code
Sarah Hoffmann [Sun, 11 Jul 2021 18:14:25 +0000 (20:14 +0200)]
remove dead code

3 years agodo not intermix params with and without default
Sarah Hoffmann [Sun, 11 Jul 2021 18:10:13 +0000 (20:10 +0200)]
do not intermix params with and without default

3 years agodirectly return data in function
Sarah Hoffmann [Sun, 11 Jul 2021 17:24:04 +0000 (19:24 +0200)]
directly return data in function

The temporary variable is not necessary.

3 years agoremove unnecessayly nested ifs
Sarah Hoffmann [Sun, 11 Jul 2021 17:11:37 +0000 (19:11 +0200)]
remove unnecessayly nested ifs

Found by Sonarqube.

3 years agoremove unused functions
Sarah Hoffmann [Sun, 11 Jul 2021 17:10:04 +0000 (19:10 +0200)]
remove unused functions

The functions were necessary for the transitory code
to Python and are no longer used.

3 years agoavoid multiple returns of same value
Sarah Hoffmann [Sun, 11 Jul 2021 16:23:42 +0000 (18:23 +0200)]
avoid multiple returns of same value

Found by Sonarqube.

3 years agoalways use brackets on if statements
Sarah Hoffmann [Sat, 10 Jul 2021 12:59:38 +0000 (14:59 +0200)]
always use brackets on if statements

This adds bracket around all one-line if statements that did
not have them yet.

3 years agoremove unused variables
Sarah Hoffmann [Fri, 9 Jul 2021 14:36:42 +0000 (16:36 +0200)]
remove unused variables

As reported by sonarqube.

3 years agofix bad use of echo in PHP output
Sarah Hoffmann [Fri, 9 Jul 2021 10:50:35 +0000 (12:50 +0200)]
fix bad use of echo in PHP output

3 years agoMerge pull request #2390 from lonvia/responsible-disclosure
Sarah Hoffmann [Fri, 9 Jul 2021 10:32:37 +0000 (12:32 +0200)]
Merge pull request #2390 from lonvia/responsible-disclosure

Add security issue disclosure policy

3 years agoadd security issue disclosure policy
Sarah Hoffmann [Fri, 9 Jul 2021 09:36:59 +0000 (11:36 +0200)]
add security issue disclosure policy

3 years agoMerge pull request #2384 from lonvia/actions-add-icu-tokenizer
Sarah Hoffmann [Wed, 7 Jul 2021 12:39:53 +0000 (14:39 +0200)]
Merge pull request #2384 from lonvia/actions-add-icu-tokenizer

CI: run tests on Ubuntu 18

3 years agoadd missing pyyaml requirement
Sarah Hoffmann [Tue, 6 Jul 2021 21:04:01 +0000 (23:04 +0200)]
add missing pyyaml requirement

3 years agoenable PHP 7.2 for Ubuntu 18 CI
Sarah Hoffmann [Tue, 6 Jul 2021 20:52:57 +0000 (22:52 +0200)]
enable PHP 7.2 for Ubuntu 18 CI

3 years agocannot use capture_output in subprocess.run
Sarah Hoffmann [Tue, 6 Jul 2021 14:10:18 +0000 (16:10 +0200)]
cannot use capture_output in subprocess.run

Only available since Python 3.7.

3 years agoremove default parameter for namedtuple
Sarah Hoffmann [Tue, 6 Jul 2021 07:54:11 +0000 (09:54 +0200)]
remove default parameter for namedtuple

This is only available in Python 3.7.

3 years agoCI: run tests on older Ubuntu version as well
Sarah Hoffmann [Mon, 5 Jul 2021 15:15:07 +0000 (17:15 +0200)]
CI: run tests on older Ubuntu version as well

3 years agoMerge pull request #2382 from lonvia/remove-json-config
Sarah Hoffmann [Mon, 5 Jul 2021 10:34:34 +0000 (12:34 +0200)]
Merge pull request #2382 from lonvia/remove-json-config

Remove outdated ICU tokenizer JSON config

3 years agoMerge pull request #2383 from lonvia/remove-more-names
Sarah Hoffmann [Mon, 5 Jul 2021 10:34:16 +0000 (12:34 +0200)]
Merge pull request #2383 from lonvia/remove-more-names

Exclude name:etymology and name:signed

3 years agoexclude name:etymology and name:signed
Sarah Hoffmann [Mon, 5 Jul 2021 09:04:16 +0000 (11:04 +0200)]
exclude name:etymology and name:signed

name:etymology contains a description of the name origin and is
thus more informative than search-worthy.

name:signed basically indicates that the feature does not have
a name.

3 years agoremove outdated ICU tokenizer JSON config
Sarah Hoffmann [Mon, 5 Jul 2021 09:01:35 +0000 (11:01 +0200)]
remove outdated ICU tokenizer JSON config

3 years agoMerge pull request #2371 from lonvia/increase-python-version
Sarah Hoffmann [Mon, 5 Jul 2021 08:32:38 +0000 (10:32 +0200)]
Merge pull request #2371 from lonvia/increase-python-version

Increase minimum required Python version to 3.6

3 years agoMerge pull request #2381 from lonvia/reorganise-abbreviations
Sarah Hoffmann [Mon, 5 Jul 2021 08:32:16 +0000 (10:32 +0200)]
Merge pull request #2381 from lonvia/reorganise-abbreviations

Reorganise abbreviation handling

3 years agoadd warning about experimental nature of ICU tokenizer
Sarah Hoffmann [Sun, 4 Jul 2021 08:44:58 +0000 (10:44 +0200)]
add warning about experimental nature of ICU tokenizer

3 years agolimit the number of variants that can be produced
Sarah Hoffmann [Fri, 2 Jul 2021 14:42:13 +0000 (16:42 +0200)]
limit the number of variants that can be produced

3 years agorestrict partial word counting to names of reasoanble length
Sarah Hoffmann [Fri, 2 Jul 2021 13:05:17 +0000 (15:05 +0200)]
restrict partial word counting to names of reasoanble length

The partial word count does not split names to save a bit of time.
The result is that it might enounter unreasonably long names
which in truth consist of multiple words. No accurate statistics
are needed so simply restrict the count to words shorter than
75 characters.

3 years agofix subsequent replacements
Sarah Hoffmann [Thu, 1 Jul 2021 15:56:23 +0000 (17:56 +0200)]
fix subsequent replacements

Two replacement words directly following each other did not
work as expected because each expects a space at the
beginning/end while there was only one space available.

Also forbit composing a word after a space was added in the
end by a previous replacement.

3 years agoleave ICU variant properties empty for now
Sarah Hoffmann [Wed, 30 Jun 2021 19:52:33 +0000 (21:52 +0200)]
leave ICU variant properties empty for now

Saving unused properties causes unnecessary duplicates.

3 years agoimport abbreviations from OSM Wiki
Sarah Hoffmann [Wed, 30 Jun 2021 19:37:29 +0000 (21:37 +0200)]
import abbreviations from OSM Wiki

Replaces the variant rules with a slightly cleaned-up
version of the abbreviation lists at
https://wiki.openstreetmap.org/wiki/Name_finder:Abbreviations

3 years agoimprove normalization
Sarah Hoffmann [Sat, 26 Jun 2021 17:38:08 +0000 (19:38 +0200)]
improve normalization

Make sure all special symbols are removed during normalization already.
Those won't be interpreted in any way because they are unlikely to be
searched for.

3 years agoonly consider partials in multi-words for initial count
Sarah Hoffmann [Sat, 26 Jun 2021 09:57:09 +0000 (11:57 +0200)]
only consider partials in multi-words for initial count

This ensures that it is less likely that we exclude meaningful
words like 'hauptstrasse' just because they are frequent.

3 years agoadd documentation for ICU tokenizer configuration
Sarah Hoffmann [Sat, 26 Jun 2021 08:13:33 +0000 (10:13 +0200)]
add documentation for ICU tokenizer configuration

3 years agoswitch to a more flexible variant description format
Sarah Hoffmann [Thu, 24 Jun 2021 18:02:07 +0000 (20:02 +0200)]
switch to a more flexible variant description format

The new format combines compound splitting and abbreviation.
It also allows to restrict rules to additional conditions
(like language or region). This latter ability is not used
yet.

3 years agouse yaml tag syntax to mark include files
Sarah Hoffmann [Sun, 20 Jun 2021 21:45:33 +0000 (23:45 +0200)]
use yaml tag syntax to mark include files

3 years agoadd dependency on datrie
Sarah Hoffmann [Tue, 15 Jun 2021 07:02:17 +0000 (09:02 +0200)]
add dependency on datrie

3 years agotests for composing decomposed suffixes
Sarah Hoffmann [Tue, 15 Jun 2021 06:59:03 +0000 (08:59 +0200)]
tests for composing decomposed suffixes

3 years agomake compund decomposition pure import feature
Sarah Hoffmann [Fri, 11 Jun 2021 08:03:31 +0000 (10:03 +0200)]
make compund decomposition pure import feature

Compound decomposition now creates a full name variant on
import just like abbreviations. This simplifies query time
normalization and opens a path for changing abbreviation
and compund decomposition lists for an existing database.

3 years agocomplete tests for icu tokenizer
Sarah Hoffmann [Thu, 10 Jun 2021 15:18:23 +0000 (17:18 +0200)]
complete tests for icu tokenizer

3 years agofix full term token in special phrases
Sarah Hoffmann [Thu, 10 Jun 2021 08:28:46 +0000 (10:28 +0200)]
fix full term token in special phrases

3 years agocomplete tests for rule loader
Sarah Hoffmann [Thu, 10 Jun 2021 08:06:49 +0000 (10:06 +0200)]
complete tests for rule loader

3 years agocorrectly quote strings when copying in data
Sarah Hoffmann [Thu, 10 Jun 2021 07:36:43 +0000 (09:36 +0200)]
correctly quote strings when copying in data

Encapsulate the copy string in a class that ensures that
copy lines are written with correct quoting.

3 years agoupdate unit tests for adapted abbreviation code
Sarah Hoffmann [Wed, 9 Jun 2021 13:07:36 +0000 (15:07 +0200)]
update unit tests for adapted abbreviation code

3 years agoadd abbreviations from legacy tokenizer
Sarah Hoffmann [Wed, 9 Jun 2021 08:53:39 +0000 (10:53 +0200)]
add abbreviations from legacy tokenizer

These abbreviations are not a perfect fit anymore because
abbreviation replacement is now applied before transliteration.

3 years agoadapt tests for ICU tokenizer
Sarah Hoffmann [Sun, 6 Jun 2021 09:00:44 +0000 (11:00 +0200)]
adapt tests for ICU tokenizer

3 years agomove abbreviation computation into import phase
Sarah Hoffmann [Fri, 28 May 2021 20:06:13 +0000 (22:06 +0200)]
move abbreviation computation into import phase

This adds precomputation of abbreviated terms for names and removes
abbreviation of terms in the query. Basic import works but still
needs some thorough testing as well as speed improvements during
import.

New dependency for python library datrie.

3 years agoicu tokenizer: move transliteration rules in separate file
Sarah Hoffmann [Wed, 26 May 2021 18:50:34 +0000 (20:50 +0200)]
icu tokenizer: move transliteration rules in separate file

The tokenizer configuration has become difficult to handle
due to the additional manual transliteration rules. Allow
to have a separate rule file that is given to the ICU library
as is.

3 years agodocs: nominatim-ui should be installed from the release
Sarah Hoffmann [Sat, 3 Jul 2021 19:14:43 +0000 (21:14 +0200)]
docs: nominatim-ui should be installed from the release

The development version does not provide the pre-packaged
dist directory anymore.

3 years agoMerge pull request #2373 from lonvia/tweak-search-cost
Sarah Hoffmann [Sat, 26 Jun 2021 14:21:08 +0000 (16:21 +0200)]
Merge pull request #2373 from lonvia/tweak-search-cost

Further tweaking of search cost

3 years agoremove penalty for full words in address
Sarah Hoffmann [Sat, 26 Jun 2021 09:20:25 +0000 (11:20 +0200)]
remove penalty for full words in address

Now that mutli-word partials no longer exist, multi-word full
words need to be used to search in addresses and therefore no
longer should have a penalty.

Also changes the condition when a full word is included into
the address. It is no longer relevant if an equivalent partial
exists but only if the term consists of more than one word.

3 years agoadjust penalty for housenumber-in-name searches
Sarah Hoffmann [Sat, 26 Jun 2021 08:31:55 +0000 (10:31 +0200)]
adjust penalty for housenumber-in-name searches

When searching for house numbers in the name (for place-only
terms) then the same penalties need to apply as for the
regular house number search.

Change the code to first compute the penalties and then create
the new search variants.

3 years agoincrease minimum Python to 3.6
Sarah Hoffmann [Mon, 21 Jun 2021 14:32:54 +0000 (16:32 +0200)]
increase minimum Python to 3.6

Python 3.6 introduces formatted string literals and
flag enums as well as a much faster dict implementation.
These changes make the code so much simpler as to warrant
dropping Python 3.5 support.

Affected distributions are Ubuntu 16.04 and Debian Stretch.

3 years agomake sure old data gets deleted on place type change
Sarah Hoffmann [Fri, 18 Jun 2021 08:58:41 +0000 (10:58 +0200)]
make sure old data gets deleted on place type change

When changing from some other place type to place=postcode
make sure that the old place type entry in the place table
is deleted.

3 years agoupdate postcode in place if it already exists
Sarah Hoffmann [Thu, 17 Jun 2021 22:28:10 +0000 (00:28 +0200)]
update postcode in place if it already exists

3 years agoMerge pull request #2369 from lonvia/exclude-poi-from-housenumber-search
Sarah Hoffmann [Thu, 17 Jun 2021 13:30:05 +0000 (15:30 +0200)]
Merge pull request #2369 from lonvia/exclude-poi-from-housenumber-search

Do not return POIs when dropping house number in query

3 years agodo not return POIs when dropping house number in query
Sarah Hoffmann [Thu, 17 Jun 2021 10:05:33 +0000 (12:05 +0200)]
do not return POIs when dropping house number in query

We've previously added searching through rank 30 in a house
number search to enable searches for house number+name.
This had the unintended side effect that rank 30 objects
are also returned in s search that dropped the house number
from the query. This is wrong because POIs cannot function
as a parent to a house number.

This fix drops all rank 30 objects from the results for a
house number search if they do not match the requested house
number.

3 years agoMerge pull request #2360 from AntoJvlt/postcodes-place-table
Sarah Hoffmann [Wed, 16 Jun 2021 09:45:07 +0000 (11:45 +0200)]
Merge pull request #2360 from AntoJvlt/postcodes-place-table

Use place instead of placex to compute postcodes

3 years agoImproved performance of the postcodes query and some code cleaning
AntoJvlt [Sat, 12 Jun 2021 13:46:08 +0000 (15:46 +0200)]
Improved performance of the postcodes query and some code cleaning

3 years agoAlways delete old placex entry for type=postcode when inserting a new one into the...
AntoJvlt [Sat, 12 Jun 2021 13:35:51 +0000 (15:35 +0200)]
Always delete old placex entry for type=postcode when inserting a new one into the place table

3 years agoHandle postcode type change in place insert trigger
AntoJvlt [Wed, 9 Jun 2021 07:24:25 +0000 (09:24 +0200)]
Handle postcode type change in place insert trigger

3 years agoClean and update tests for postcodes
AntoJvlt [Tue, 8 Jun 2021 20:39:04 +0000 (22:39 +0200)]
Clean and update tests for postcodes

3 years agoUse place_exists() into can_compute() for postcodes
AntoJvlt [Tue, 8 Jun 2021 07:33:10 +0000 (09:33 +0200)]
Use place_exists() into can_compute() for postcodes

3 years agoUpdate tests for postcodes
AntoJvlt [Mon, 7 Jun 2021 13:02:53 +0000 (15:02 +0200)]
Update tests for postcodes

3 years agoUse place instead of placex to compute postcodes
AntoJvlt [Fri, 4 Jun 2021 19:26:13 +0000 (21:26 +0200)]
Use place instead of placex to compute postcodes

3 years agodo not fail CI on codecov errors
Sarah Hoffmann [Tue, 8 Jun 2021 08:42:14 +0000 (10:42 +0200)]
do not fail CI on codecov errors

The CodeCove upload depends on unreliable external code.

3 years agoMerge pull request #2359 from lonvia/switch-bdd-tests-to-api-search
Sarah Hoffmann [Sun, 6 Jun 2021 16:29:51 +0000 (18:29 +0200)]
Merge pull request #2359 from lonvia/switch-bdd-tests-to-api-search

Remove deprecated commandline query function

3 years agoremove deprecated query interface
Sarah Hoffmann [Sun, 6 Jun 2021 13:28:21 +0000 (15:28 +0200)]
remove deprecated query interface

Searches can now be done via the thin API wrapper.

3 years agoswitch BDD tests to always use search API
Sarah Hoffmann [Sun, 6 Jun 2021 13:27:52 +0000 (15:27 +0200)]
switch BDD tests to always use search API

3 years agoMerge pull request #2358 from AntoJvlt/documentation-update
Sarah Hoffmann [Fri, 4 Jun 2021 21:54:37 +0000 (23:54 +0200)]
Merge pull request #2358 from AntoJvlt/documentation-update

Update documentation

3 years agoUpdate documentation
AntoJvlt [Tue, 1 Jun 2021 15:02:45 +0000 (17:02 +0200)]
Update documentation

3 years agoMerge pull request #2357 from lonvia/legacy-tokenizer-fix-word-entries
Sarah Hoffmann [Wed, 2 Jun 2021 18:58:14 +0000 (20:58 +0200)]
Merge pull request #2357 from lonvia/legacy-tokenizer-fix-word-entries

Fix insertion of special terms and countries into word table

3 years agofix insertion of special terms and countries into word table
Sarah Hoffmann [Wed, 2 Jun 2021 15:37:27 +0000 (17:37 +0200)]
fix insertion of special terms and countries into word table

Special terms need to be prefixed by a space because they are
full terms.

For countries avoid duplicate entries of word tokens.

Adds tests for adding country terms.

3 years agoMerge pull request #2356 from lonvia/freeze-after-import
Sarah Hoffmann [Wed, 2 Jun 2021 14:25:26 +0000 (16:25 +0200)]
Merge pull request #2356 from lonvia/freeze-after-import

Call freeze after running and non-updateable import

3 years agodocs: reload SQL when migrating to 3.6
Sarah Hoffmann [Wed, 2 Jun 2021 14:11:29 +0000 (16:11 +0200)]
docs: reload SQL when migrating to 3.6

SQL functions must always be reloaded when updating the software.
All other updates included the instruction as part of some other
migration. From 3.7 on it will happen as part of the migration
command.

Fixes #2335.

3 years agocall freeze after running and non-updateable import
Sarah Hoffmann [Wed, 2 Jun 2021 09:08:48 +0000 (11:08 +0200)]
call freeze after running and non-updateable import

Some of the tables will have already been removed but
the tables for indexing are still there and should be
dropped.

3 years agocommit changes to replication log table
Sarah Hoffmann [Wed, 26 May 2021 09:47:08 +0000 (11:47 +0200)]
commit changes to replication log table

Fixes #2350.

3 years agoalways compute guessed postcode for POIs from centroid
Sarah Hoffmann [Wed, 26 May 2021 09:04:02 +0000 (11:04 +0200)]
always compute guessed postcode for POIs from centroid

When guessing postcodes from the area, only postcodes within
that area are accepted. For POIs that is usually not what we
want as the postcode would have to be within a house for
example.

Fixes #2301.

3 years agoMerge pull request #2349 from lonvia/fix-website-refresh
Sarah Hoffmann [Tue, 25 May 2021 18:43:44 +0000 (20:43 +0200)]
Merge pull request #2349 from lonvia/fix-website-refresh

Only initialise tokenizer for refresh functions where needed

3 years agoonly initialise tokenizer for refresh functions where needed
Sarah Hoffmann [Tue, 25 May 2021 17:16:22 +0000 (19:16 +0200)]
only initialise tokenizer for refresh functions where needed

Fixes #2347.

3 years agoMerge pull request #2346 from lonvia/words-vs-tokens
Sarah Hoffmann [Mon, 24 May 2021 15:41:38 +0000 (17:41 +0200)]
Merge pull request #2346 from lonvia/words-vs-tokens

Cleanup use of partial words in legacy tokenizers

3 years agoadd tests for new full name computation with ICU
Sarah Hoffmann [Mon, 24 May 2021 08:29:21 +0000 (10:29 +0200)]
add tests for new full name computation with ICU

3 years agoreorganize keyword creation for legacy tokenizer
Sarah Hoffmann [Sun, 23 May 2021 21:58:58 +0000 (23:58 +0200)]
reorganize keyword creation for legacy tokenizer

- only save partial words without internal spaces
- consider comma and semicolon a separator of full words
- consider parts before an opening bracket a full word
  (but not the part after the bracket)

Fixes #244.

3 years agouse make_keywords for place search terms also
Sarah Hoffmann [Sun, 23 May 2021 21:08:11 +0000 (23:08 +0200)]
use make_keywords for place search terms also

Ensures that place indeed uses the same search names as other
names.

3 years agoalways ignore multi term partials in search
Sarah Hoffmann [Sun, 23 May 2021 20:13:03 +0000 (22:13 +0200)]
always ignore multi term partials in search

Partial terms should only ever consist of one word. Ignore
any other, they are a leftover from inefficient word index
builts.