20858: Enable 'exists' filter to use index
[arvados.git] / services / api / lib / record_filters.rb
index f8898d63c90de2169fc8d18b53d40f68171ae945..e51223254f7d21a34b5910e2f68a59abee4c22cb 100644 (file)
@@ -47,9 +47,10 @@ module RecordFilters
         raise ArgumentError.new("Invalid operator '#{operator}' (#{operator.class}) in filter")
       end
 
+      operator = operator.downcase
       cond_out = []
 
-      if attrs_in == 'any' && (operator.casecmp('ilike').zero? || operator.casecmp('like').zero?) && (operand.is_a? String) && operand.match('^[%].*[%]$')
+      if attrs_in == 'any' && (operator == 'ilike' || operator == 'like') && (operand.is_a? String) && operand.match('^[%].*[%]$')
         # Trigram index search
         cond_out << model_class.full_text_trgm + " #{operator} ?"
         param_out << operand
@@ -85,9 +86,9 @@ module RecordFilters
           end
 
           # jsonb search
-          case operator.downcase
+          case operator
           when '=', '!='
-            not_in = if operator.downcase == "!=" then "NOT " else "" end
+            not_in = if operator == "!=" then "NOT " else "" end
             cond_out << "#{not_in}(#{attr_table_name}.#{attr} @> ?::jsonb)"
             param_out << SafeJSON.dump({proppath => operand})
           when 'in'
@@ -120,9 +121,9 @@ module RecordFilters
             end
           when 'exists'
             if operand == true
-              cond_out << "jsonb_exists(#{attr_table_name}.#{attr}, ?)"
+              cond_out << "jsonb_exists_inline_op(#{attr_table_name}.#{attr}, ?)"
             elsif operand == false
-              cond_out << "(NOT jsonb_exists(#{attr_table_name}.#{attr}, ?)) OR #{attr_table_name}.#{attr} is NULL"
+              cond_out << "(NOT jsonb_exists_inline_op(#{attr_table_name}.#{attr}, ?)) OR #{attr_table_name}.#{attr} is NULL"
             else
               raise ArgumentError.new("Invalid operand '#{operand}' for '#{operator}' must be true or false")
             end
@@ -134,21 +135,39 @@ module RecordFilters
           else
             raise ArgumentError.new("Invalid operator for subproperty search '#{operator}'")
           end
-        elsif operator.downcase == "exists"
-          if col.type != :jsonb
+        elsif operator == "exists"
+          if col.nil? or col.type != :jsonb
             raise ArgumentError.new("Invalid attribute '#{attr}' for operator '#{operator}' in filter")
           end
 
-          cond_out << "jsonb_exists(#{attr_table_name}.#{attr}, ?)"
+          cond_out << "jsonb_exists_inline_op(#{attr_table_name}.#{attr}, ?)"
           param_out << operand
+        elsif expr = /^ *\( *(\w+) *(<=?|>=?|=) *(\w+) *\) *$/.match(attr)
+          if operator != '=' || ![true,"true"].index(operand)
+            raise ArgumentError.new("Invalid expression filter '#{attr}': subsequent elements must be [\"=\", true]")
+          end
+          operator = expr[2]
+          attr1, attr2 = expr[1], expr[3]
+          allowed = attr_model_class.searchable_columns(operator)
+          [attr1, attr2].each do |tok|
+            if !allowed.index(tok)
+              raise ArgumentError.new("Invalid attribute in expression: '#{tok}'")
+            end
+            col = attr_model_class.columns.select { |c| c.name == tok }.first
+            if col.type != :integer
+              raise ArgumentError.new("Non-numeric attribute in expression: '#{tok}'")
+            end
+          end
+          cond_out << "#{attr1} #{operator} #{attr2}"
         else
-          if !attr_model_class.searchable_columns(operator).index attr
+          if !attr_model_class.searchable_columns(operator).index(attr) &&
+             !(col.andand.type == :jsonb && ['contains', '=', '<>', '!='].index(operator))
             raise ArgumentError.new("Invalid attribute '#{attr}' in filter")
           end
+          attr_type = attr_model_class.attribute_column(attr).type
 
-          case operator.downcase
+          case operator
           when '=', '<', '<=', '>', '>=', '!=', 'like', 'ilike'
-            attr_type = attr_model_class.attribute_column(attr).type
             operator = '<>' if operator == '!='
             if operand.is_a? String
               if attr_type == :boolean
@@ -162,8 +181,8 @@ module RecordFilters
                 when '0', 'f', 'false', 'n', 'no'
                   operand = false
                 else
-                  raise ArgumentError("Invalid operand '#{operand}' for " \
-                                      "boolean attribute '#{attr}'")
+                  raise ArgumentError.new("Invalid operand '#{operand}' for " \
+                                          "boolean attribute '#{attr}'")
                 end
               end
               if operator == '<>'
@@ -187,6 +206,10 @@ module RecordFilters
               cond_out << "#{attr_table_name}.#{attr} #{operator} ?"
               param_out << operand
             elsif (attr_type == :integer)
+              if !operand.is_a?(Integer) || operand.bit_length > 64
+                raise ArgumentError.new("Invalid operand '#{operand}' "\
+                                        "for integer attribute '#{attr}'")
+              end
               cond_out << "#{attr_table_name}.#{attr} #{operator} ?"
               param_out << operand
             else
@@ -194,17 +217,24 @@ module RecordFilters
                                       "for '#{operator}' operator in filters")
             end
           when 'in', 'not in'
-            if operand.is_a? Array
-              cond_out << "#{attr_table_name}.#{attr} #{operator} (?)"
-              param_out << operand
-              if operator == 'not in' and not operand.include?(nil)
-                # explicitly allow NULL
-                cond_out[-1] = "(#{cond_out[-1]} OR #{attr_table_name}.#{attr} IS NULL)"
-              end
-            else
+            if !operand.is_a? Array
               raise ArgumentError.new("Invalid operand type '#{operand.class}' "\
                                       "for '#{operator}' operator in filters")
             end
+            if attr_type == :integer
+              operand.each do |el|
+                if !el.is_a?(Integer) || el.bit_length > 64
+                  raise ArgumentError.new("Invalid element '#{el}' in array "\
+                                          "for integer attribute '#{attr}'")
+                end
+              end
+            end
+            cond_out << "#{attr_table_name}.#{attr} #{operator} (?)"
+            param_out << operand
+            if operator == 'not in' and not operand.include?(nil)
+              # explicitly allow NULL
+              cond_out[-1] = "(#{cond_out[-1]} OR #{attr_table_name}.#{attr} IS NULL)"
+            end
           when 'is_a'
             operand = [operand] unless operand.is_a? Array
             cond = []
@@ -227,6 +257,31 @@ module RecordFilters
               end
             end
             cond_out << cond.join(' OR ')
+          when 'contains'
+            if col.andand.type != :jsonb
+              raise ArgumentError.new("Invalid attribute '#{attr}' for '#{operator}' operator")
+            end
+            if operand == []
+              raise ArgumentError.new("Invalid operand '#{operand.inspect}' for '#{operator}' operator")
+            end
+            operand = [operand] unless operand.is_a? Array
+            operand.each do |op|
+              if !op.is_a?(String)
+                raise ArgumentError.new("Invalid element #{operand.inspect} in operand for #{operator.inspect} operator (operand must be a string or array of strings)")
+              end
+            end
+            # We use jsonb_exists_all_inline_op(a,b) instead of "a ?&
+            # b" because the pg gem thinks "?" is a bind var.
+            #
+            # See note in migration
+            # 20230815160000_jsonb_exists_functions about _inline_op
+            # functions.
+            #
+            # We use string interpolation instead of param_out
+            # because the pg gem flattens param_out / doesn't support
+            # passing arrays as bind vars.
+            q = operand.map { |s| ActiveRecord::Base.connection.quote(s) }.join(',')
+            cond_out << "jsonb_exists_all_inline_op(#{attr_table_name}.#{attr}, array[#{q}])"
           else
             raise ArgumentError.new("Invalid operator '#{operator}'")
           end