pg_fulltext 0.2.2 → 1.0.0

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: b8701d60187af1ad0460c88d00294d05e4a2b844276ad8b3986b29babc41be15
-  data.tar.gz: 3618abfbe086adfef098af5c2c7672d1021bf86839cd8503c76caa5961ccae6b
+  metadata.gz: 5561eb4580813730f611fe05d6bbf17e6a152f09d79e91f4c90f3e754407c184
+  data.tar.gz: 1843ebf4389d67f22666787d2203d869a365e953d4fa9705db6efbfacdc28428
 SHA512:
-  metadata.gz: 9ed5a212569e382a8b40a6dcc67f1f24a3af1adcb5dee59d28721db890a7da93dffa16afec5b79dfcd53b19d525369171a24078b9586389f677df7883a8b2e56
-  data.tar.gz: ca268b23a69304788491dad1f79680b91d56714cadd45b175aeed9773a02088c9ffc4a83704f831d7c8d755a4b48b537e905614142dfa6c26c511babb9d517cf
+  metadata.gz: 9852ff9a4843ff90aa94ef5ec1500621d293d29bdd0ea68846d82435be174e2b4f7f4286765830c1a4b66e3cd416386581fbcb971a95eed5711c3574ab11b30b
+  data.tar.gz: bed84f7603d1b0f926ce21ccb81c0202103fe3e25966a0729735bad4979ba1b1d3605c571cb380dcd7f3e68ad5ae1573a8dfb710747b0f87835dd578010cc650

data/README.md

@@ -33,37 +33,17 @@ end
 You can then use the `search` method (configurable via the first parameter of the `add_search_scope` method):
 
 ```ruby
-MyModel.search('foo bar "include this phrase" !butnotthis !"and and also not this"')
+MyModel.search('foo bar "include this phrase" -butnotthis -"and and also not this phrase"')
 ```
 
 The defaults for this include support for negation, phrases, phrase negation, and prefix searches, but those can be 
 configured per the following options:
 
-| Option            | Default | Description |
-| ----------------  | :------ | :----------- |
-| `tsvector_column` | `tsv`   | If you have a different column containing your tsvector, specify it here. |
-| `search_type`     | `nil`   | Your PostgreSQL probably defaults to `'english'`, but set this to match the tsvector you've generated. |
-| `order`           | `true`  | Whether or not the `order` method should be applied against the generated `rank` for the fulltext query. If you just care about returning matches and not their respective rank, set this to `false`. |
-| `prefix`          | `true`  | Default search will match partial words as well as whole words. Set this to `false` if only whole words should be matched. |
-| `reorder`         | `false` | If you already have `order` set on this relation, it will take precedence over the fulltext `rank`. `reorder` will call clear, effectively clearing the existing order and applying `rank`. |
-| `any_word`        | `false` | Default search uses the `&` operator, ensuring that all terms are matched in the query.  If you want to match _any_ term in the query, set this to `true`. |
-| `ignore_accents`  | `false` | By default, search queries with accents will be sent through as-is. Setting this to `true` will `unaccent()` the query, which helps match `tsv` columns that have also been unaccented.  Alternatively, you can have your `tsv` column be a combination of both, and this option will be unnecesary. Requires the `unaccent` Postgres extension. |
-
-## Standalone Configuration
-
-There's not much, here, but the `PgFulltext::Query.to_tsquery_string` method will generate a nice `tsvector`-compatible 
-string for you to use as you wish.  
-
-Something this should do the trick:
-
-```ruby
-db = PG.connect(dbname: 'mydb')
-search_string = 'foo bar "include this phrase" !butnotthis !"and and also not this"'
-tsv_query = db.escape_string(PgFulltext::Query.to_tsquery_string(search_string))
-sql = <<~SQL
-  SELECT *
-  FROM my_model
-  WHERE tsv @@ to_tsquery('portuguese', '#{tsv_query}')
-SQL
-db.exec(sql)
-```
+| Option            | Default  | Description |
+| ----------------  | :------- | :----------- |
+| `tsvector_column` | `tsv`    | If you have a different column containing your tsvector, specify it here. |
+| `search_type`     | `simple` | Your PostgreSQL probably defaults to `'english'`, but the default for this option is `simple` to ensure the most predictable behavior. IMPORTANT: Set this to match the tsvector you've generated. |
+| `order`           | `true`   | Whether or not the `order` method should be applied against the generated `rank` for the fulltext query. If you just care about returning matches and not their respective rank, set this to `false`. |
+| `prefix`          | `true`   | Default search will match partial words as well as whole words. Set this to `false` if only whole words should be matched. |
+| `reorder`         | `false`  | If you already have `order` set on this relation, it will take precedence over the fulltext `rank`. `reorder` will call clear, effectively clearing the existing order and applying `rank`. |
+| `ignore_accents`  | `false`  | By default, search queries with accents will be sent through as-is. Setting this to `true` will `unaccent()` the query, which helps match `tsv` columns that have also been unaccented.  Alternatively, you can have your `tsv` column be a combination of both, and this option will be unnecesary. Requires the `unaccent` Postgres extension. |

data/lib/pg_fulltext/active_record.rb

@@ -19,11 +19,10 @@ module PgFulltext
         relation,
         query,
         tsvector_column: :tsv,
-        search_type: nil,
+        search_type: :simple,
         order: true,
         prefix: true,
         reorder: false,
-        any_word: false,
         ignore_accents: false
       )
         serial = SecureRandom.hex(4)
@@ -36,7 +35,6 @@ module PgFulltext
           query,
           tsvector_column: tsvector_column,
           search_type: search_type,
-          any_word: any_word,
           prefix: prefix,
           ignore_accents: ignore_accents,
         )
@@ -58,16 +56,15 @@ module PgFulltext
         query,
         tsvector_column: :tsv,
         search_type: nil,
-        any_word: false,
         prefix: true,
         ignore_accents: false
       )
-        tsquery_string_quoted = connection.quote(PgFulltext::Query.to_tsquery_string(query, operator: any_word ? '|' : '&', prefix: prefix))
+        tsquery_string_quoted = connection.quote(query)
         tsquery_string_quoted = "unaccent(#{tsquery_string_quoted})" if ignore_accents
         column_quoted = connection.quote_column_name(tsvector_column)
         fqc_quoted = "#{quoted_table_name}.#{column_quoted}"
-        tsquery = "to_tsquery(#{"#{connection.quote search_type}, " if search_type.present?}#{tsquery_string_quoted})"
-
+        tsquery = "websearch_to_tsquery(#{"#{connection.quote search_type}, " if search_type.present?}#{tsquery_string_quoted})"
+        tsquery = "regexp_replace(#{tsquery}::text, '''([a-z0-9\\-_@.]+)''', '''\\1'':*', 'g')::tsquery" if prefix
         relation
           .select(:id, "ts_rank_cd(#{fqc_quoted}, #{tsquery}) AS rank")
           .where("#{fqc_quoted} @@ #{tsquery}")

data/lib/pg_fulltext.rb

@@ -1,4 +1,3 @@
 module PgFulltext; end
 
 require 'pg_fulltext/active_record'
-require 'pg_fulltext/query'

metadata

@@ -1,14 +1,14 @@
 --- !ruby/object:Gem::Specification
 name: pg_fulltext
 version: !ruby/object:Gem::Version
-  version: 0.2.2
+  version: 1.0.0
 platform: ruby
 authors:
 - Adam Robertson
 autorequire:
 bindir: bin
 cert_chain: []
-date: 2021-11-22 00:00:00.000000000 Z
+date: 2023-07-27 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: activerecord
@@ -90,7 +90,6 @@ files:
 - Rakefile
 - lib/pg_fulltext.rb
 - lib/pg_fulltext/active_record.rb
-- lib/pg_fulltext/query.rb
 homepage: https://github.com/arcreative/pg_fulltext
 licenses:
 - MIT
@@ -112,7 +111,7 @@ required_rubygems_version: !ruby/object:Gem::Requirement
     - !ruby/object:Gem::Version
       version: '0'
 requirements: []
-rubygems_version: 3.1.6
+rubygems_version: 3.2.33
 signing_key:
 specification_version: 4
 summary: PostgreSQL fulltext search

data/lib/pg_fulltext/query.rb

@@ -1,58 +0,0 @@
-require 'rltk/lexer'
-
-module PgFulltext
-  module Query
-    class Lexer < RLTK::Lexer
-      rule(/!+/)
-      rule(/"+/)
-
-      rule(/(!?)[0-9\p{L}!]+/) { |v| [:WORD, v] }
-      rule(/"[\p{L}\s!]+"/) { |v| [:PHRASE, v[1..-2]] }
-      rule(/!"[\p{L}\s!]+"/) { |v| [:NOT_PHRASE, v[2..-2]] }
-
-      rule(/\s+/)
-      rule(/[^\p{L}\s"]+/)
-    end
-
-    def self.to_tsquery_string(query, prefix: true, operator: '&')
-      query = normalize_query(query)
-
-      terms = []
-      Lexer.lex(query).each do |token|
-        if token.type == :WORD
-          terms << format_term(token.value, prefix: prefix)
-        elsif %i[PHRASE NOT_PHRASE].include?(token.type)
-          phrase_terms = Lexer.lex(token.value).map do |phrase_term|
-            phrase_term.value.nil? ? nil : format_term(phrase_term.value, prefix: prefix)
-          end.compact
-          terms << "#{'!' if token.type == :NOT_PHRASE}(#{phrase_terms.join(' <-> ')})"
-        end
-      end
-
-      terms.join(" #{operator} ")
-    end
-
-    private
-
-    def self.normalize_query(query)
-      query
-        .gsub(/[.,]/, ' ')            # Replace all periods and commas with spaces (reasonable delimiters)
-        .gsub(/[^\s\p{L}0-9"!]/, '')  # Remove all non-unicode, whitespace, numbers, quotes ("), and bangs (!)
-        .gsub(/\s+/, ' ')             # Replace repeat whitespace occurrences with single spaces
-        .strip                        # Strip space from beginning and end of line
-    end
-
-    def self.format_term(term, prefix: true)
-      # Remove any ! that's not at the beginning of the term, as it will break the query
-      term.gsub!(/(?<!^)!/, '')
-
-      # Add the prefix if prefix is set
-      "#{term}#{':*' if prefix}"
-    end
-
-    def self.reject_falsy(terms)
-      false_values = [nil, '', '"', '!', ':*', '":*', '!:*']
-      terms.reject { |v| false_values.include?(v) }
-    end
-  end
-end