Skip to content

DataShades/ckanext-search-tweaks

Repository files navigation

Tests

ckanext-search-tweaks

Set of tools providing control over search results, sorting, etc.

Requirements

Compatibility with core CKAN versions:

CKAN version Compatible?
2.8 and earlier no
2.9 yes

Installation

To install ckanext-search-tweaks:

  1. Activate your CKAN virtual environment, for example:

     . /usr/lib/ckan/default/bin/activate
    
  2. Install it on the virtualenv

     pip install ckanext-search-tweaks
    
  3. Add search_tweaks to the ckan.plugins setting in your CKAN config file (by default the config file is located at /etc/ckan/default/ckan.ini).

  4. Restart CKAN.

Usage

This extensions consists of multiple plugins. search_tweaks is the main (major) one, that must be enabled all the time. And depending on the set of secondary (minor) plugins, extra features and config options may be available. Bellow are listed all the plugins with their side effects.

Plugin Functionality
search_tweaks Allow all the other plugins to be enabled
search_tweaks_query_relevance Promote datasets that were visited most frequently for the current search query
search_tweaks_field_relevance Promote dataset depending on value of it's field
search_tweaks_spellcheck Provides "Did you mean?" feature

search_tweaks

Provides base functionality and essential pieces of logic used by all the other plugins. Must be enabled as long as at least one other plugin from this extension is enabled.

  • Switches search to edismax query parser if none was specified

  • Enables ckanext.search_tweaks.iterfaces.ISearchTweaks interface with the following methods:

      def get_search_boost_fn(self, search_params: dict[str, Any]) -> Optional[str]:
      	"""Returns optional boost function that will be applied to the search query.
      	"""
      	return None
    
      def get_extra_qf(self, search_params: dict[str, Any]) -> Optional[str]:
      	"""Return an additional fragment of the Solr's qf.
          This fragment will be appended to the current qf
      	"""
      	return None
    

CLI

ckan search-tweaks -
	Root of all the extension specific commands.
	Every command from minor plugins is registered under this section.

Config settings

# Rewrite the default value of the qf parameter sent to Solr
# (optional, default: value of ckan.lib.search.query.QUERY_FIELDS).
ckanext.search_tweaks.common.qf = title^5 text

# Search by misspelled queries.
# (optional, default: false).
ckanext.search_tweaks.common.fuzzy_search.enabled = on

# Maximum number of misspelled letters. Possible values are 1 and 2.
# (optional, default: 1).
ckanext.search_tweaks.common.fuzzy_search.distance = 2

# Use `boost` instead of `bf` when `edismax` query parser is active
# (optional, default: true).
ckanext.search_tweaks.common.prefer_boost = no

# MinimumShouldMatch used in queries
# (optional, default: 1).
ckanext.search_tweaks.common.mm = 2<-1 5<80%

# Keep original query when using fuzzy search, e.g. "(hello~2) OR (hello)" if true
# (optional, default: true).
ckanext.search_tweaks.common.fuzzy_search.keep_original

search_tweaks_query_relevance

Increase relevance of datasets for particular query depending on number of direct visits of the dataset after running this search. I.e, if user searches for something and then visits dataset B which is initially displayed in a third row of search results, eventually this dataset will be displayed on the second or even on the first row. This is implemented in three stages. On the first stage, statistics collected and stored inside storage(redis, by default) and then this statistics converted into numeric solr field via cronjob. Finally, Solr's boost function that scales number of visits and improves score for the given query is applied during search.

Following steps are required in order to configure this plugin:

  • Add field that will store statistics to schema.xml(query_relevance_ prefix can be changed via config option):

      <dynamicField name="query_relevance_*"  type="int" indexed="true" stored="true"/>
    
  • Configure a cronjob which will update search-index periodically:

      0 0 * * * ckan search-index rebuild
    

CLI

relevance query align - remove old data from storage. Actual result of this command depends
	on storage backend, that is controlled by config. At the momment, only `redis-daily` backend
	is affected by this command - all records older than `query_relevance.daily.age` days are removed.

relevance query export - export statistics as CSV.

relevance query import - import statistics from CSV. Note, records that are already in storage but
	are not listed in CSV won't be removed. It must be done manually

Config settings

# Which backend to use in order to collect information about dataset
# relevance for the particular search query. Possible values are:
# "redis-permanent", "redis-daily"
# (optional, default: redis-daily).
ckanext.search_tweaks.query_relevance.backend = redis-permanent

# How long(in days) information about dataset visits will be stored in order to
# update relevance of dataset in search query.
# (optional, default: 90).
ckanext.search_tweaks.query_relevance.daily.age = 90

# Solr boost function with $field placeholder that will be replaced by
# the correspoinding field name
# (optional, default: "scale(def($field,0),1,1.2)").
ckanext.search_tweaks.query_relevance.boost_function = recip($field,1,1000,1000)

# Prefix of the numeric field defined in Solr schema. This field will hold
# dataset's relevance for the given query.
# (optional, default: query_relevance_).
ckanext.search_tweaks.query_relevance.field_prefix = custom_score_

search_tweaks_field_relevance

Increases the relevance of a dataset depending on value of its numeric field. For now it's impossible to promote dataset using field with textual type.

No magic here either, this plugin allows you to specify Solr's boost function that will be used during all the searches. One can achieve exactly the same result using ISearchTweaks.get_search_boost_fn. But I expect this option to be used often, so there is a possibility to update relevance without any extra line of code.

Config settings

# Solr boost function for static numeric field
# (optional, default: None).
ckanext.search_tweaks.field_relevance.boost_function = pow(promoted_level,2)

# Field with dataset promotion level
# (optional, default: promotion_level).
ckanext.search_tweaks.field_relevance.blueprint.promotion.field_name = promotion

# Register pacakge promotion route
# (optional, default: False).
ckanext.search_tweaks.field_relevance.blueprint.promotion.enabled = true

Auth functions

search_tweaks_field_relevance_promote: access package promotion route. Calls `package_update` by default.

search_tweaks_spellcheck

Exposes search suggestions from the Solr's spellcheck component to CKAN templates. This plugin doesn't do much and mainly relies on the Solr's built-in functionality. Thus you have to make a lot of changes inside Solr in order to use it:

  • solrconfig.xml. Configure spellcheck component. Search for <searchComponent name="spellcheck" class="solr.SpellCheckComponent"> section and add the following item under it:

      <lst name="spellchecker">
      	<str name="name">did_you_mean</str>
      	<str name="field">did_you_mean</str>
      	<str name="buildOnCommit">false</str>
      </lst>
    
  • Add cron job that will update suggestions dictionary periodically:

      ckan search-tweaks spellcheck rebuild
    
  • solrconfig.xml. Add spellcheck component to the search handler (<requestHandler name="/select" class="solr.SearchHandler">):

      <arr name="last-components">
      	<str>spellcheck</str>
      </arr>
    
  • Define spellcheck field in the schema. If you want to use an existing field(text for example), change <str name="field">did_you_mean</str> value inside solrconfig.xml to the name of the selected field instead.

      <field name="did_you_mean" type="textgen" indexed="true" multiValued="true" />
    
  • Note: skip if you've decided to use an existing field in the previous step.
    Copy meaningfull values into this field:

      <copyField source="title" dest="did_you_mean"/>
      <copyField source="notes" dest="did_you_mean"/>
      <copyField source="res_name" dest="did_you_mean"/>
      <copyField source="res_description" dest="did_you_mean"/>
      <copyField source="extras_*" dest="did_you_mean"/>
    

After that you have to restart Solr service and rebuild search index:

ckan search-index rebuild

Now you can use spellcheck_did_you_mean template helper that returns better search query when available instead of the current one. Consider including search_tweaks/did_you_mean.html fragment under search form.

Config settings

# Do not show suggestions that have fewer results than current query
# (optional, default: true).
ckanext.search_tweaks.spellcheck.more_results_only = off

# How many different suggestions you expect to see for query
# (optional, default: 1).
ckanext.search_tweaks.spellcheck.max_suggestions = 3

CLI

spellcheck rebuild - rebuild/reload spellcheck dictionary.

Developer installation

To install ckanext-search-tweaks for development, activate your CKAN virtualenv and do:

git clone https://github.com/DataShades/ckanext-search-tweaks.git
cd ckanext-search-tweaks
python setup.py develop
pip install -r dev-requirements.txt

Tests

Apart from the default configuration for CKAN testing, you have to create ckan_search_tweaks Solr's core, replace its schema with ckanext/search_tweaks/tests/schema.xml and make changes to solrconfig.xml that are required by search_tweaks_spellcheck.

To run the tests, do:

pytest --ckan-ini=test.ini ckanext/search_tweaks/tests

License

AGPL