mirror of
https://github.com/Growstuff/growstuff.git
synced 2026-10-31 03:09:06 -04:00
Brenda Wallace
6332f26c58
Admin for crop job templates, a tidier crop wrangling page, and crop and template seeds
105 lines
4.1 KiB
Ruby
105 lines
4.1 KiB
Ruby
# frozen_string_literal: true
|
|
|
|
# Crop search, done in Postgres. A crop matches when a word of its name, an
|
|
# alternate name or a scientific name starts with the term, or when one of them
|
|
# is similar to it (pg_trgm), which is what forgives typos and plurals.
|
|
#
|
|
# Names are compared lower-cased with accents folded, so "jalapeno" finds
|
|
# "jalape\u00f1o". Each name table has a stored `normalised_name` column holding
|
|
# that (see db/migrate/20261021000000_add_crop_name_search_indexes.rb), with a
|
|
# trigram index that the queries below use. The folding is done with
|
|
# translate() because unaccent() is not immutable. ACCENTED and PLAIN must match
|
|
# the migration, and the term is folded the same way by `normalise`.
|
|
#
|
|
# Similarity uses the `<%` operator so that the indexes apply. It compares
|
|
# against pg_trgm.word_similarity_threshold, which config/database.yml sets to
|
|
# MIN_SIMILARITY for every connection.
|
|
module SearchCrops
|
|
extend ActiveSupport::Concern
|
|
|
|
MIN_SIMILARITY = 0.35
|
|
NAME_WEIGHT = 5
|
|
PLANTED_BOOST = 5
|
|
|
|
ACCENTED = %w(
|
|
àáâãäåāçćčèéêëēė
|
|
ęìíîïīįłñńòóôõöø
|
|
ōùúûüūýÿžźż
|
|
).join.freeze
|
|
PLAIN = "#{'a' * 7}ccc#{'e' * 7}#{'i' * 6}lnn#{'o' * 7}#{'u' * 5}yyzzz".freeze
|
|
|
|
def self.normalise(text)
|
|
text.to_s.downcase.tr(ACCENTED, PLAIN)
|
|
end
|
|
|
|
# [table, crop id column, weight]: a crop's own name counts for more than its other names.
|
|
NAME_SOURCES = [['crops', 'id', NAME_WEIGHT], ['alternate_names', 'crop_id', 1], ['scientific_names', 'crop_id', 1]].freeze
|
|
|
|
included do
|
|
# Only the search queries read it.
|
|
self.ignored_columns += %w(normalised_name)
|
|
|
|
# Crops of any status whose name, an alternate name or a scientific name
|
|
# matches the term, with the match's score as `matches.score`.
|
|
scope :matching_name, lambda { |term|
|
|
joins("INNER JOIN (#{SearchCrops.matches_sql(term.to_s.squish)}) matches ON matches.crop_id = crops.id")
|
|
}
|
|
|
|
# For crops from `matching_name`: the crop named exactly the term first, then
|
|
# by the search's own rank (see rank_sql).
|
|
scope :best_match_first, lambda { |term, member: nil|
|
|
exact = sanitize_sql_array(['lower(crops.name) = ?', SearchCrops.normalise(term.to_s.squish)])
|
|
order(Arel.sql("#{exact} DESC"), Arel.sql("#{SearchCrops.rank_sql(member)} DESC"))
|
|
}
|
|
|
|
# Approved crops matching the term, best first. A blank term matches them all.
|
|
# Crops with more plantings rank higher, and so do crops the member has planted.
|
|
scope :search_by_name, lambda { |term, member: nil|
|
|
term = term.to_s.squish
|
|
next approved.order(Arel.sql(SearchCrops.popularity_sql)) if term.blank?
|
|
|
|
approved
|
|
.matching_name(term)
|
|
.order(Arel.sql("#{SearchCrops.rank_sql(member)} DESC"), :name)
|
|
}
|
|
end
|
|
|
|
def self.matches_sql(term)
|
|
normalised = normalise(term)
|
|
like = ApplicationRecord.sanitize_sql_like(normalised)
|
|
binds = { term: normalised, contains: "%#{like}%", starts: "#{like}%", word: "% #{like}%", hyphen: "%-#{like}%", min: MIN_SIMILARITY }
|
|
|
|
selects = NAME_SOURCES.map do |table, crop_id, weight|
|
|
name = "#{table}.normalised_name"
|
|
<<~SQL.squish
|
|
SELECT #{crop_id} AS crop_id, #{weight} AS weight,
|
|
CASE WHEN #{name} LIKE :starts OR #{name} LIKE :word OR #{name} LIKE :hyphen
|
|
THEN 1.0 ELSE word_similarity(:term, #{name}) END AS score
|
|
FROM #{table}
|
|
WHERE #{name} LIKE :contains OR :term <% #{name}
|
|
SQL
|
|
end
|
|
|
|
ApplicationRecord.sanitize_sql_array([<<~SQL.squish, binds])
|
|
SELECT crop_id, MAX(weight * score) AS score
|
|
FROM (#{selects.join(' UNION ALL ')}) scored
|
|
WHERE score >= :min
|
|
GROUP BY crop_id
|
|
SQL
|
|
end
|
|
|
|
def self.rank_sql(member)
|
|
sql = "matches.score * (1 + LN(1 + crops.plantings_count))"
|
|
return sql unless member
|
|
|
|
own = ApplicationRecord.sanitize_sql_array(
|
|
["(SELECT COUNT(*) FROM plantings WHERE plantings.crop_id = crops.id AND plantings.owner_id = ?)", member.id]
|
|
)
|
|
"#{sql} * (1 + #{PLANTED_BOOST} * LN(1 + #{own}))"
|
|
end
|
|
|
|
def self.popularity_sql
|
|
"(1 + LN(1 + crops.plantings_count)) * (1 + LN(1 + crops.harvests_count)) DESC, crops.name"
|
|
end
|
|
end
|