pg_query 2.0.3 → 5.1.0

data/lib/pg_query/fingerprint.rb

@@ -39,6 +39,11 @@ module PgQuery
       subhash = FingerprintSubHash.new
 
       if val.is_a?(Google::Protobuf::RepeatedField)
+        # For lists that have exactly one untyped node, just output the parent field (if needed) and return
+        if val.length == 1 && val[0].is_a?(Node) && val[0].node.nil?
+          hash.update(parent_field_name) if need_to_write_name
+          return
+        end
         fingerprint_list(val, subhash, parent_node_name, parent_field_name)
       elsif val.is_a?(List)
         fingerprint_list(val.items, subhash, parent_node_name, parent_field_name)
@@ -55,23 +60,20 @@ module PgQuery
     end
 
     def ignored_node_type?(node)
-      [A_Const, Alias, ParamRef, SetToDefault, IntList, OidList, Null].include?(node.class) ||
+      [A_Const, Alias, ParamRef, SetToDefault, IntList, OidList].include?(node.class) ||
         node.is_a?(TypeCast) && (node.arg.node == :a_const || node.arg.node == :param_ref)
     end
 
     def node_protobuf_field_name_to_json_name(node_class, field)
-      # Use protobuf descriptor once json_name support is fixed: https://github.com/protocolbuffers/protobuf/pull/8356
-      # node_class.descriptor.find { |d| d.name == field.to_s }.json_name
-      INTERNAL_PROTO_FIELD_NAME_TO_JSON_NAME.fetch([node_class, field])
+      node_class.descriptor.find { |d| d.name == field.to_s }.json_name
     end
 
     def fingerprint_node(node, hash, parent_node_name = nil, parent_field_name = nil) # rubocop:disable Metrics/CyclomaticComplexity
       return if ignored_node_type?(node)
 
       if node.is_a?(Node)
-        return if node.node.nil?
-        node_val = node[node.node.to_s]
-        unless ignored_node_type?(node_val)
+        node_val = node.inner
+        unless node_val.nil? || ignored_node_type?(node_val)
           unless node_val.is_a?(List)
             postgres_node_name = node_protobuf_field_name_to_json_name(node.class, node.node)
             hash.update(postgres_node_name)
@@ -92,12 +94,18 @@ module PgQuery
         when 'location'
           next
         when 'name'
-          next if [PrepareStmt, ExecuteStmt, DeallocateStmt].include?(node.class)
+          next if [PrepareStmt, ExecuteStmt, DeallocateStmt, FunctionParameter].include?(node.class)
           next if node.is_a?(ResTarget) && parent_node_name == 'SelectStmt' && parent_field_name == 'targetList'
-        when 'gid', 'options', 'savepoint_name'
+        when 'gid', 'savepoint_name'
           next if node.is_a?(TransactionStmt)
+        when 'options'
+          next if [TransactionStmt, CreateFunctionStmt].include?(node.class)
         when 'portalname'
           next if [DeclareCursorStmt, FetchStmt, ClosePortalStmt].include?(node.class)
+        when 'conditionname'
+          next if [ListenStmt, UnlistenStmt, NotifyStmt].include?(node.class)
+        when 'args'
+          next if node.is_a?(DoStmt)
         when 'relname'
           next if node.is_a?(RangeVar) && node.relpersistence == 't'
           if node.is_a?(RangeVar)
@@ -113,6 +121,10 @@ module PgQuery
             fingerprint_value(:AEXPR_OP, hash, postgres_node_name, postgres_field_name, true)
             next
           end
+        # libpg_query still outputs `str` parts when print a string node. Here we override that to
+        # the expected field name of `sval`.
+        when 'sval', 'fval', 'bsval'
+          postgres_field_name = 'str' if node.is_a?(String) || node.is_a?(BitString) || node.is_a?(Float)
         end
 
         fingerprint_value(val, hash, postgres_node_name, postgres_field_name, true)

data/lib/pg_query/node.rb

@@ -1,27 +1,32 @@
 module PgQuery
   # Patch the auto-generated generic node type with additional convenience functions
   class Node
+    def self.inner_class_to_name(klass)
+      @inner_class_to_name ||= descriptor.lookup_oneof('node').to_h { |f| [f.subtype.msgclass, f.name.to_sym] }
+      @inner_class_to_name[klass]
+    end
+
+    def inner
+      self[node.to_s]
+    end
+
+    def inner=(submsg)
+      name = self.class.inner_class_to_name(submsg.class)
+      public_send("#{name}=", submsg)
+    end
+
     def inspect
-      node ? format('<PgQuery::Node: %s: %s>', node, public_send(node).inspect) : '<PgQuery::Node>'
+      node ? format('<PgQuery::Node: %s: %s>', node, inner.inspect) : '<PgQuery::Node>'
     end
 
     # Make it easier to initialize nodes from a given node child object
     def self.from(node_field_val)
-      # This needs to match libpg_query naming for the Node message field names
-      # (see "underscore" method in libpg_query's scripts/generate_protobuf_and_funcs.rb)
-      node_field_name = node_field_val.class.name.split('::').last
-      node_field_name.gsub!(/^([A-Z\d])([A-Z][a-z])/, '\1__\2')
-      node_field_name.gsub!(/([A-Z\d]+[a-z]+)([A-Z][a-z])/, '\1_\2')
-      node_field_name.gsub!(/([a-z\d])([A-Z])/, '\1_\2')
-      node_field_name.tr!('-', '_')
-      node_field_name.downcase!
-
-      PgQuery::Node.new(node_field_name => node_field_val)
+      PgQuery::Node.new(inner_class_to_name(node_field_val.class) => node_field_val)
     end
 
     # Make it easier to initialize value nodes
-    def self.from_string(str)
-      PgQuery::Node.new(string: PgQuery::String.new(str: str))
+    def self.from_string(sval)
+      PgQuery::Node.new(string: PgQuery::String.new(sval: sval))
     end
 
     def self.from_integer(ival)

data/lib/pg_query/param_refs.rb

@@ -29,7 +29,7 @@ module PgQuery
             location = typeloc
           end
 
-          results << { 'location' => location, 'length' => length, 'typename' => t.names.map { |n| n.string.str } }
+          results << { 'location' => location, 'length' => length, 'typename' => t.names.map { |n| n.string.sval } }
         end
       end
 

data/lib/pg_query/parse.rb

@@ -3,7 +3,13 @@ module PgQuery
     result, stderr = parse_protobuf(query)
 
     begin
-      result = PgQuery::ParseResult.decode(result)
+      result = if PgQuery::ParseResult.method(:decode).arity == 1
+                 PgQuery::ParseResult.decode(result)
+               elsif PgQuery::ParseResult.method(:decode).arity == -1
+                 PgQuery::ParseResult.decode(result, recursion_limit: 1_000)
+               else
+                 raise ArgumentError, 'Unsupported protobuf Ruby API'
+               end
     rescue Google::Protobuf::ParseError => e
       raise PgQuery::ParseError.new(format('Failed to parse tree: %s', e.message), __FILE__, __LINE__, -1)
     end
@@ -29,6 +35,7 @@ module PgQuery
       @tables = nil
       @aliases = nil
       @cte_names = nil
+      @functions = nil
     end
 
     def dup_tree
@@ -51,27 +58,52 @@ module PgQuery
       tables_with_details.select { |t| t[:type] == :ddl }.map { |t| t[:name] }.uniq
     end
 
+    # Returns function names, ignoring their argument types. This may be insufficient
+    # if you need to disambiguate two functions with the same name but different argument
+    # types.
+    def functions
+      functions_with_details.map { |f| f[:function] }.uniq
+    end
+
+    def ddl_functions
+      functions_with_details.select { |f| f[:type] == :ddl }.map { |f| f[:function] }.uniq
+    end
+
+    def call_functions
+      functions_with_details.select { |f| f[:type] == :call }.map { |f| f[:function] }.uniq
+    end
+
     def cte_names
-      load_tables_and_aliases! if @cte_names.nil?
+      load_objects! if @cte_names.nil?
       @cte_names
     end
 
     def aliases
-      load_tables_and_aliases! if @aliases.nil?
+      load_objects! if @aliases.nil?
       @aliases
     end
 
     def tables_with_details
-      load_tables_and_aliases! if @tables.nil?
+      load_objects! if @tables.nil?
       @tables
     end
 
+    def functions_with_details
+      load_objects! if @functions.nil?
+      @functions
+    end
+
     protected
 
-    def load_tables_and_aliases! # rubocop:disable Metrics/CyclomaticComplexity
+    # Parses the query and finds table and function references
+    #
+    # Note we use ".to_ary" on arrays from the Protobuf library before
+    # passing them to concat, because of https://bugs.ruby-lang.org/issues/18140
+    def load_objects! # rubocop:disable Metrics/CyclomaticComplexity
       @tables = [] # types: select, dml, ddl
       @cte_names = []
       @aliases = {}
+      @functions = [] # types: call, ddl
 
       statements = @tree.stmts.dup.to_a.map(&:stmt)
       from_clause_items = [] # types: select, dml, ddl
@@ -86,22 +118,18 @@ module PgQuery
           # The following statement types do not modify tables and are added to from_clause_items
           # (and subsequently @tables)
           when :select_stmt
-            subselect_items.concat(statement.select_stmt.target_list)
+            subselect_items.concat(statement.select_stmt.target_list.to_ary)
             subselect_items << statement.select_stmt.where_clause if statement.select_stmt.where_clause
             subselect_items.concat(statement.select_stmt.sort_clause.collect { |h| h.sort_by.node })
-            subselect_items.concat(statement.select_stmt.group_clause)
+            subselect_items.concat(statement.select_stmt.group_clause.to_ary)
             subselect_items << statement.select_stmt.having_clause if statement.select_stmt.having_clause
 
             case statement.select_stmt.op
             when :SETOP_NONE
               (statement.select_stmt.from_clause || []).each do |item|
-                if item.node == :range_subselect
-                  statements << item.range_subselect.subquery
-                else
-                  from_clause_items << { item: item, type: :select }
-                end
+                from_clause_items << { item: item, type: :select }
               end
-            when :SETOP_UNION
+            when :SETOP_UNION, :SETOP_EXCEPT, :SETOP_INTERSECT
               statements << PgQuery::Node.new(select_stmt: statement.select_stmt.larg) if statement.select_stmt.larg
               statements << PgQuery::Node.new(select_stmt: statement.select_stmt.rarg) if statement.select_stmt.rarg
             end
@@ -111,16 +139,32 @@ module PgQuery
               @cte_names.concat(cte_names)
               statements.concat(cte_statements)
             end
+
+            if statement.select_stmt.into_clause
+              from_clause_items << { item: PgQuery::Node.new(range_var: statement.select_stmt.into_clause.rel), type: :ddl }
+            end
           # The following statements modify the contents of a table
           when :insert_stmt, :update_stmt, :delete_stmt
-            value = statement.public_send(statement.node)
+            value = statement.inner
             from_clause_items << { item: PgQuery::Node.new(range_var: value.relation), type: :dml }
             statements << value.select_stmt if statement.node == :insert_stmt && value.select_stmt
 
-            subselect_items.concat(statement.update_stmt.target_list) if statement.node == :update_stmt
+            if statement.node == :update_stmt
+              value.from_clause.each do |item|
+                from_clause_items << { item: item, type: :select }
+              end
+              subselect_items.concat(statement.update_stmt.target_list.to_ary)
+            end
+
             subselect_items << statement.update_stmt.where_clause if statement.node == :update_stmt && statement.update_stmt.where_clause
             subselect_items << statement.delete_stmt.where_clause if statement.node == :delete_stmt && statement.delete_stmt.where_clause
 
+            if statement.node == :delete_stmt
+              statement.delete_stmt.using_clause.each do |using_clause|
+                from_clause_items << { item: using_clause, type: :select }
+              end
+            end
+
             if value.with_clause
               cte_statements, cte_names = statements_and_cte_names_for_with_clause(value.with_clause)
               @cte_names.concat(cte_names)
@@ -131,7 +175,12 @@ module PgQuery
             statements << statement.copy_stmt.query
           # The following statement types are DDL (changing table structure)
           when :alter_table_stmt
-            from_clause_items << { item: PgQuery::Node.new(range_var: statement.alter_table_stmt.relation), type: :ddl }
+            case statement.alter_table_stmt.objtype
+            when :OBJECT_INDEX # Index # rubocop:disable Lint/EmptyWhen
+              # ignore `ALTER INDEX index_name`
+            else
+              from_clause_items << { item: PgQuery::Node.new(range_var: statement.alter_table_stmt.relation), type: :ddl }
+            end
           when :create_stmt
             from_clause_items << { item: PgQuery::Node.new(range_var: statement.create_stmt.relation), type: :ddl }
           when :create_table_as_stmt
@@ -146,6 +195,11 @@ module PgQuery
             statements << statement.view_stmt.query
           when :index_stmt
             from_clause_items << { item: PgQuery::Node.new(range_var: statement.index_stmt.relation), type: :ddl }
+            statement.index_stmt.index_params.each do |p|
+              next if p.index_elem.expr.nil?
+              subselect_items << p.index_elem.expr
+            end
+            subselect_items << statement.index_stmt.where_clause if statement.index_stmt.where_clause
           when :create_trig_stmt
             from_clause_items << { item: PgQuery::Node.new(range_var: statement.create_trig_stmt.relation), type: :ddl }
           when :rule_stmt
@@ -158,9 +212,9 @@ module PgQuery
             objects = statement.drop_stmt.objects.map do |obj|
               case obj.node
               when :list
-                obj.list.items.map { |obj2| obj2.string.str if obj2.node == :string }
+                obj.list.items.map { |obj2| obj2.string.sval if obj2.node == :string }
               when :string
-                obj.string.str
+                obj.string.sval
               end
             end
             case statement.drop_stmt.remove_type
@@ -168,6 +222,10 @@ module PgQuery
               @tables += objects.map { |r| { name: r.join('.'), type: :ddl } }
             when :OBJECT_RULE, :OBJECT_TRIGGER
               @tables += objects.map { |r| { name: r[0..-2].join('.'), type: :ddl } }
+            when :OBJECT_FUNCTION
+              # Only one function can be dropped in a statement
+              obj = statement.drop_stmt.objects[0].object_with_args
+              @functions << { function: obj.objname.map { |f| f.string.sval }.join('.'), type: :ddl }
             end
           when :grant_stmt
             objects = statement.grant_stmt.objects
@@ -184,12 +242,30 @@ module PgQuery
           # The following are other statements that don't fit into query/DML/DDL
           when :explain_stmt
             statements << statement.explain_stmt.query
+          when :create_function_stmt
+            @functions << {
+              function: statement.create_function_stmt.funcname.map { |f| f.string.sval }.join('.'),
+              type: :ddl
+            }
+          when :rename_stmt
+            if statement.rename_stmt.rename_type == :OBJECT_FUNCTION
+              original_name = statement.rename_stmt.object.object_with_args.objname.map { |f| f.string.sval }.join('.')
+              new_name = statement.rename_stmt.newname
+              @functions += [
+                { function: original_name, type: :ddl },
+                { function: new_name, type: :ddl }
+              ]
+            end
+          when :prepare_stmt
+            statements << statement.prepare_stmt.query
           end
         end
 
         next_item = subselect_items.shift
         if next_item
           case next_item.node
+          when :list
+            subselect_items += next_item.list.items.to_ary
           when :a_expr
             %w[lexpr rexpr].each do |side|
               elem = next_item.a_expr.public_send(side)
@@ -201,47 +277,62 @@ module PgQuery
               end
             end
           when :bool_expr
-            subselect_items.concat(next_item.bool_expr.args)
+            subselect_items.concat(next_item.bool_expr.args.to_ary)
+          when :coalesce_expr
+            subselect_items.concat(next_item.coalesce_expr.args.to_ary)
+          when :min_max_expr
+            subselect_items.concat(next_item.min_max_expr.args.to_ary)
           when :res_target
             subselect_items << next_item.res_target.val
           when :sub_link
             statements << next_item.sub_link.subselect
+          when :func_call
+            subselect_items.concat(next_item.func_call.args.to_ary)
+            @functions << {
+              function: next_item.func_call.funcname.map { |f| f.string.sval }.join('.'),
+              type: :call
+            }
+          when :case_expr
+            subselect_items.concat(next_item.case_expr.args.map { |arg| arg.case_when.expr })
+            subselect_items.concat(next_item.case_expr.args.map { |arg| arg.case_when.result })
+            subselect_items << next_item.case_expr.defresult
+          when :type_cast
+            subselect_items << next_item.type_cast.arg
           end
         end
 
-        break if subselect_items.empty? && statements.empty?
-      end
-
-      loop do
         next_item = from_clause_items.shift
-        break unless next_item && next_item[:item]
-
-        case next_item[:item].node
-        when :join_expr
-          from_clause_items << { item: next_item[:item].join_expr.larg, type: next_item[:type] }
-          from_clause_items << { item: next_item[:item].join_expr.rarg, type: next_item[:type] }
-        when :row_expr
-          from_clause_items += next_item[:item].row_expr.args.map { |a| { item: a, type: next_item[:type] } }
-        when :range_var
-          rangevar = next_item[:item].range_var
-          next if rangevar.schemaname.empty? && @cte_names.include?(rangevar.relname)
-
-          table = [rangevar.schemaname, rangevar.relname].reject { |s| s.nil? || s.empty? }.join('.')
-          @tables << {
-            name: table,
-            type: next_item[:type],
-            location: rangevar.location,
-            schemaname: (rangevar.schemaname unless rangevar.schemaname.empty?),
-            relname: rangevar.relname,
-            inh: rangevar.inh
-          }
-          @aliases[rangevar.alias.aliasname] = table if rangevar.alias
-        when :range_subselect
-          from_clause_items << { item: next_item[:item].range_subselect.subquery, type: next_item[:type] }
-        when :select_stmt
-          from_clause = next_item[:item].select_stmt.from_clause
-          from_clause_items += from_clause.map { |r| { item: r, type: next_item[:type] } } if from_clause
+        if next_item && next_item[:item]
+          case next_item[:item].node
+          when :join_expr
+            from_clause_items << { item: next_item[:item].join_expr.larg, type: next_item[:type] }
+            from_clause_items << { item: next_item[:item].join_expr.rarg, type: next_item[:type] }
+            subselect_items << next_item[:item].join_expr.quals
+          when :row_expr
+            from_clause_items += next_item[:item].row_expr.args.map { |a| { item: a, type: next_item[:type] } }
+          when :range_var
+            rangevar = next_item[:item].range_var
+            next if rangevar.schemaname.empty? && @cte_names.include?(rangevar.relname)
+
+            table = [rangevar.schemaname, rangevar.relname].reject { |s| s.nil? || s.empty? }.join('.')
+            @tables << {
+              name: table,
+              type: next_item[:type],
+              location: rangevar.location,
+              schemaname: (rangevar.schemaname unless rangevar.schemaname.empty?),
+              relname: rangevar.relname,
+              inh: rangevar.inh,
+              relpersistence: rangevar.relpersistence
+            }
+            @aliases[rangevar.alias.aliasname] = table if rangevar.alias
+          when :range_subselect
+            statements << next_item[:item].range_subselect.subquery
+          when :range_function
+            subselect_items += next_item[:item].range_function.functions
+          end
         end
+
+        break if subselect_items.empty? && statements.empty? && from_clause_items.empty?
       end
 
       @tables.uniq!