Initial commit

lib/text_to_tsquery.rb

@@ -0,0 +1,99 @@
+# frozen_string_literal: true
+
+# transforms "english like" text queries into a tsquery operation
+# https://www.postgresql.org/docs/9.5/textsearch-controls.html#TEXTSEARCH-PARSING-QUERIES
+class TextToTsquery
+  attr_reader :text
+
+  def initialize(text, wildcard: true)
+    @text = text.to_s.strip
+    @wildcard = wildcard
+    @exact_matches = []
+    validate!
+  end
+
+  EXACT_WORD_CHAR = '§'.freeze
+
+  def tsquery
+    @tsquery = @text
+    strip_exact_words
+    remove_duplicated_spaces
+    transform_or_into_operator
+    transform_and_into_operator
+    strip_spaces_from_parenthesis
+    transform_remaining_spaces_into_and_operator
+    transform_keywords
+    join_operators_with_and
+    remove_partial_match_from_not_keywords
+    add_exact_words
+    @tsquery
+  end
+
+  def validate!
+    parenthesis_error unless self.class.valid_search_parenthesis?(@text)
+  end
+
+  def self.valid_search_parenthesis?(text)
+    text.split('').reduce(0) do |acc, char|
+      return false if acc < 0
+
+      if char == '('
+        acc + 1
+      elsif char == ')'
+        acc - 1
+      else
+        acc
+      end
+    end.zero?
+  end
+
+  def parenthesis_error
+    raise ArgumentError, "incorrect number/order of parenthesis in search query: '#{@text}'"
+  end
+
+  def strip_exact_words
+    @exact_matches << Regexp.last_match(1) while @tsquery.sub!(/"(.*?)"/, EXACT_WORD_CHAR)
+  end
+
+  def remove_duplicated_spaces
+    @tsquery = @tsquery.gsub(/\s+/, ' ')
+  end
+
+  # transforms or/OR/|/|| into | operator
+  def transform_or_into_operator
+    @tsquery = @tsquery.gsub(/ ((or|\|+) )+/i, '|').gsub(/ *\|+ */, '|')
+  end
+
+  # transforms and/AND/&/&& into & operator
+  def transform_and_into_operator
+    @tsquery = @tsquery.gsub(/ ((and|\&+) )+/i, '&')
+  end
+
+  def strip_spaces_from_parenthesis
+    @tsquery = @tsquery.gsub(/ *([()]) */, '\1')
+  end
+
+  def transform_remaining_spaces_into_and_operator
+    @tsquery = @tsquery.tr(' ', '&')
+  end
+
+  # adds :* for partial match of words
+  def transform_keywords
+    keyword = @wildcard ? '\1:*' : '\1:'
+    @tsquery = @tsquery.gsub(/([^#{EXACT_WORD_CHAR}|&!())]+)/, keyword)
+  end
+
+  # adds & between operations
+  def join_operators_with_and
+    @tsquery = @tsquery.gsub(/:(\**)\!/, ':\1&!').gsub(/:(\**)\(/, ':\1&(').gsub(/\&+/, '&')
+  end
+
+  # removes partial match from NOT operations
+  def remove_partial_match_from_not_keywords
+    @tsquery = @tsquery.gsub(/\!([^|&!())]+):\**/, '!\1')
+  end
+
+  def add_exact_words
+    @exact_matches.each { |phrase| @tsquery = @tsquery.sub(EXACT_WORD_CHAR, "'#{phrase}'") }
+  end
+end