Merge remote-tracking branch 'origin/topic/jsiwek/type-alias-introspection'

* origin/topic/jsiwek/type-alias-introspection:
  Add enum_names() BIF to return names of an enum type's values
  Add type_aliases() BIF for introspecting type-names of types/values
  Change Type::type_aliases map to store IntrusivePtr
  Fix lookup_ID() BIF to return enum values

NEWS

@@ -148,6 +148,9 @@ Deprecated Functionality
 - Marked the Continuation.h and PacketDumper.h files as deprecated. The code
   contained within them is unused by Zeek.
 
+- ``Type::GetAliases()`` and ``Type::AddAlias()`` are deprecated, use
+  ``Type::Aliases()`` and ``Type::RegisterAlias()``.
+
 Zeek 3.2.0
 ==========
 

src/Type.cc

@@ -1348,13 +1348,11 @@ void EnumType::CheckAndAddName(const string& module_name, const char* name,
 	if ( vals.find(val) == vals.end() )
 		vals[val] = make_intrusive<EnumVal>(IntrusivePtr{NewRef{}, this}, val);
 
-	set<Type*> types = Type::GetAliases(GetName());
+	const auto& types = Type::Aliases(GetName());
-	set<Type*>::const_iterator it;
 
-	for ( it = types.begin(); it != types.end(); ++it )
+	for ( const auto& t : types )
-		if ( *it != this )
+		if ( t.get() != this )
-			(*it)->AsEnumType()->AddNameInternal(module_name, name, val,
+			t->AsEnumType()->AddNameInternal(module_name, name, val, is_export);
-							     is_export);
 	}
 
 void EnumType::AddNameInternal(const string& module_name, const char* name,

src/Type.h

@@ -270,13 +270,64 @@ public:
 	void SetName(const std::string& arg_name) { name = arg_name; }
 	const std::string& GetName() const { return name; }
 
-	typedef std::map<std::string, std::set<Type*> > TypeAliasMap;
+	struct TypePtrComparer {
+		bool operator()(const TypePtr& a, const TypePtr& b) const
+			{ return a.get() < b.get(); }
+	};
+	using TypePtrSet = std::set<TypePtr, TypePtrComparer>;
+	using TypeAliasMap = std::map<std::string, TypePtrSet, std::less<>>;
 
+	/**
+	 * Returns a mapping of type-name to all other type names declared as
+	 * an alias to it.
+	 */
+	static const TypeAliasMap& GetAliasMap()
+		{ return type_aliases; }
+
+	/**
+	 * Returns true if the given type name has any declared aliases
+	 */
+	static bool HasAliases(std::string_view type_name)
+		{ return Type::type_aliases.find(type_name) != Type::type_aliases.end(); }
+
+	/**
+	 * Returns the set of all type names declared as an aliases to the given
+	 * type name.  A static empty set is returned if there are no aliases.
+	 */
+	static const TypePtrSet& Aliases(std::string_view type_name)
+		{
+		static TypePtrSet empty;
+		auto it = Type::type_aliases.find(type_name);
+		return it == Type::type_aliases.end() ? empty : it->second;
+		}
+
+	[[deprecated("Remove in v4.1. Use zeek::Type::Aliases() instead.")]]
 	static std::set<Type*> GetAliases(const std::string& type_name)
-		{ return Type::type_aliases[type_name]; }
+		{
+		std::set<Type*> rval;
+		for ( const auto& t : Type::type_aliases[type_name] )
+			rval.emplace(t.get());
+		return rval;
+		}
 
+	/**
+	 * Registers a new type alias.
+	 * @param type_name  the name of the type to register a new alias for.
+	 * @param type  the associated alias type of *type_name*.
+	 * @return  true if the alias is now registered or false if the alias was
+	 * already previously registered.
+	 */
+	static bool RegisterAlias(std::string_view type_name, TypePtr type)
+		{
+		auto it = Type::type_aliases.find(type_name);
+		if ( it == Type::type_aliases.end() )
+			it = Type::type_aliases.emplace(std::string{type_name}, TypePtrSet{}).first;
+		return it->second.emplace(std::move(type)).second;
+		}
+
+	[[deprecated("Remove in v4.1. Use zeek::Type::RegisterAlias().")]]
 	static void AddAlias(const std::string &type_name, Type* type)
-		{ Type::type_aliases[type_name].insert(type); }
+		{ Type::type_aliases[type_name].insert({NewRef{}, type}); }
 
 protected:
 	Type() = default;

src/Var.cc

@@ -382,10 +382,10 @@ void add_type(ID* id, TypePtr t, std::unique_ptr<std::vector<AttrPtr>> attr)
 		// Clone the type to preserve type name aliasing.
 		tnew = t->ShallowClone();
 
-	Type::AddAlias(new_type_name, tnew.get());
+	Type::RegisterAlias(new_type_name, tnew);
 
 	if ( new_type_name != old_type_name && ! old_type_name.empty() )
-		Type::AddAlias(old_type_name, tnew.get());
+		Type::RegisterAlias(old_type_name, tnew);
 
 	tnew->SetName(id->Name());
 

src/parse.y

@@ -171,7 +171,7 @@ static void parser_redef_enum (zeek::detail::ID *id)
 static void extend_record(zeek::detail::ID* id, std::unique_ptr<zeek::type_decl_list> fields,
                           std::unique_ptr<std::vector<zeek::detail::AttrPtr>> attrs)
 	{
-	std::set<zeek::Type*> types = zeek::Type::GetAliases(id->Name());
+	const auto& types = zeek::Type::Aliases(id->Name());
 
 	if ( types.empty() )
 		{
@@ -189,7 +189,7 @@ static void extend_record(zeek::detail::ID* id, std::unique_ptr<zeek::type_decl_
 				break;
 				}
 
-	for ( auto t : types )
+	for ( const auto& t : types )
 		{
 		auto error = t->AsRecordType()->AddFields(*fields, add_log_attr);
 

src/zeek.bif

@@ -1853,6 +1853,125 @@ function type_name%(t: any%): string
 	return zeek::make_intrusive<zeek::StringVal>(s);
 	%}
 
+%%{
+static std::map<std::string, std::set<std::string>> build_complete_alias_map()
+	{
+	std::map<std::string, std::set<std::string>> rval;
+
+	for ( const auto& [name, alias_types] : zeek::Type::GetAliasMap() )
+		{
+		rval[name].emplace(name);
+
+		for ( const auto& alias_type : alias_types )
+			{
+			const auto& alias_name = alias_type->GetName();
+			rval[name].emplace(alias_name);
+			rval[alias_name].emplace(name);
+			}
+		}
+
+	for ( const auto& [name, aliases] : rval )
+		{
+		std::map<std::string, bool> more;
+
+		for ( const auto& alias_name : aliases )
+			if ( alias_name != name )
+				more.emplace(alias_name, false);
+
+		for ( ; ; )
+			{
+			auto prev_size = more.size();
+
+			for ( auto& [alias_name, checked] : more )
+				{
+				if ( checked ) continue;
+
+				for ( const auto& alias_alias : rval[alias_name] )
+					more.emplace(alias_alias, false);
+
+				checked = true;
+				}
+
+			if ( prev_size == more.size() )
+				break;
+			}
+
+		for ( const auto& [alias_name, checked] : more )
+			rval[name].emplace(alias_name);
+		}
+
+	return rval;
+	}
+%%}
+
+## Returns all type name aliases of a value or type.
+##
+## x: An arbitrary value or type.
+##
+## Returns: The set of all type name aliases of *x* (or the type of *x*
+##          if it's a value instead of a type).  For primitive values
+##          and types like :zeek:type:`string` or :zeek:type:`count`,
+##          this returns an empty set.  For types with user-defined
+##          names like :zeek:type:`record` or :zeek:type:`enum`, the
+##          returned set contains the original user-defined name for the
+##          type along with all aliases.  For other compound types, like
+##          :zeek:type:`table`, the returned set is empty unless
+##          explicitly requesting aliases for a user-defined type alias
+##          or a value that was explicitly created using a type alias
+##          (as opposed to originating from an "anonymous" constructor
+##          or initializer for that compound type).
+function type_aliases%(x: any%): string_set
+	%{
+	auto rval = make_intrusive<zeek::TableVal>(zeek::id::string_set);
+
+	auto t = x->GetType();
+
+	if ( t->Tag() == TYPE_TYPE )
+		t = t->AsTypeType()->GetType();
+
+	const auto& tname = t->GetName();
+
+	if ( tname.empty() )
+		return rval;
+
+	static auto all_aliases = build_complete_alias_map();
+	auto it = all_aliases.find(tname);
+
+	if ( it == all_aliases.end() )
+		return rval;
+
+	for ( const auto& name : it->second )
+		rval->Assign(make_intrusive<zeek::StringVal>(name), nullptr);
+
+	return rval;
+	%}
+
+## Returns all value names associated with an enum type.
+##
+## et: An enum type.
+##
+## Returns: All enum value names associated with enum type *et*.
+##          If *et* is not an enum type, an empty set is returned.
+function enum_names%(et: any%): string_set
+	%{
+	auto rval = make_intrusive<zeek::TableVal>(zeek::id::string_set);
+
+	if ( et->GetType()->Tag() != TYPE_TYPE )
+		return rval;
+
+	const auto& t = et->GetType()->AsTypeType()->GetType();
+
+	if ( t->Tag() != TYPE_ENUM )
+		return rval;
+
+	auto enum_type = t->AsEnumType();
+
+	for ( const auto& [name, i] : enum_type->Names() )
+		rval->Assign(make_intrusive<zeek::StringVal>(name), nullptr);
+
+	return rval;
+	%}
+
 ## Returns: list of command-line arguments (``argv``) used to run Zeek.
 function zeek_args%(%): string_vec
 	%{
@@ -1983,6 +2102,13 @@ function lookup_ID%(id: string%) : any
 	if ( ! i )
 		return zeek::make_intrusive<zeek::StringVal>("<unknown id>");
 
+	if ( i->IsEnumConst() )
+		{
+		auto et = i->GetType()->AsEnumType();
+		auto idx = et->Lookup(detail::GLOBAL_MODULE_NAME, i->Name());
+		return et->GetEnumVal(idx);
+		}
+
 	if ( ! i->GetVal() )
 		return zeek::make_intrusive<zeek::StringVal>("<no ID value>");
 

testing/btest/Baseline/bifs.enum_names/out

@@ -0,0 +1,13 @@
+### BTest baseline data generated by btest-diff. Do not edit. Use "btest -U/-u" to update. Requires BTest >= 0.63.
+{
+RED,
+GREEN,
+PURPLE,
+BLUE
+}
+{
+RED,
+GREEN,
+PURPLE,
+BLUE
+}

testing/btest/Baseline/bifs.lookup_ID/out

@@ -1,5 +1,7 @@
+### BTest baseline data generated by btest-diff. Do not edit. Use "btest -U/-u" to update. Requires BTest >= 0.63.
 zeek test
 <unknown id>
 <unknown id>
 <unknown id>
+GREEN
 event()

testing/btest/Baseline/bifs.type_aliases/out

@@ -0,0 +1,22 @@
+### BTest baseline data generated by btest-diff. Do not edit. Use "btest -U/-u" to update. Requires BTest >= 0.63.
+type aliases for 'RED enum val': ColorAlias Color
+type aliases for 'Color enum type': ColorAlias Color
+type aliases for 'MyRec val': MyRec MyRecAlias
+type aliases for 'MyRecAlias val': MyRec MyRecAlias
+type aliases for 'MyRec type': MyRec MyRecAlias
+type aliases for 'MyRecalias type': MyRec MyRecAlias
+type aliases for 'MyString val': it's just a 'string'
+type aliases for 'MyString type': MyString AnotherString
+type aliases for 'MyOtherString type': MyOtherString
+type aliases for 'AnotherString type': MyString AnotherString
+type aliases for 'string literal value': it's just a 'string'
+type aliases for 'count literal value': it's just a 'count'
+type aliases for 'MyTable value': MyTable2 MyTable3 MyTable MyTable4
+type aliases for 'MyTable2 value': MyTable2 MyTable3 MyTable MyTable4
+type aliases for 'MyTable3 value': MyTable2 MyTable3 MyTable MyTable4
+type aliases for 'MyTable4 value': MyTable2 MyTable3 MyTable MyTable4
+type aliases for 'MyTable type': MyTable2 MyTable3 MyTable MyTable4
+type aliases for 'MyTable2 type': MyTable2 MyTable3 MyTable MyTable4
+type aliases for 'MyTable3 type': MyTable2 MyTable3 MyTable MyTable4
+type aliases for 'MyTable4 type': MyTable2 MyTable3 MyTable MyTable4
+type aliases for 'table value': it's just a 'table[count] of string'

testing/btest/bifs/enum_names.zeek

@@ -0,0 +1,10 @@
+# @TEST-EXEC: zeek -b %INPUT >out
+# @TEST-EXEC: btest-diff out
+
+type Color: enum { RED, GREEN, BLUE };
+type ColorAlias: Color;
+
+redef enum Color += { PURPLE };
+
+print enum_names(Color);
+print enum_names(ColorAlias);

testing/btest/bifs/lookup_ID.zeek

@@ -4,6 +4,8 @@
 
 global a = "zeek test";
 
+type Color: enum { RED, GREEN, BLUE };
+
 event zeek_init()
 	{
 	local b = "local value";
@@ -12,5 +14,6 @@ event zeek_init()
 	print lookup_ID("");
 	print lookup_ID("xyz");
 	print lookup_ID("b");
+	print lookup_ID("GREEN");
 	print type_name( lookup_ID("zeek_init") );
 	}

testing/btest/bifs/type_aliases.zeek

@@ -0,0 +1,60 @@
+# @TEST-EXEC: zeek -b %INPUT >out
+# @TEST-EXEC: btest-diff out
+
+type MyRec: record {
+	a: bool &optional;
+};
+
+type Color: enum { RED, GREEN, BLUE };
+type ColorAlias: Color;
+
+type MyRecAlias: MyRec;
+type MyString: string;
+type MyOtherString: string;
+type AnotherString: MyString;
+type MyTable: table[count] of string;
+type MyTable2: MyTable;
+type MyTable3: MyTable2;
+type MyTable4: MyTable3;
+
+function type_alias_list(label: string, x: any): string
+	{
+	local rval = fmt("type aliases for '%s':", label);
+	local aliases = type_aliases(x);
+
+	if ( |aliases| == 0 )
+		rval += fmt(" it's just a '%s'", type_name(x));
+	else
+		for ( a in aliases )
+			rval += fmt(" %s", a);
+
+	return rval;
+	}
+
+print type_alias_list("RED enum val", RED);
+print type_alias_list("Color enum type", Color);
+
+print type_alias_list("MyRec val", MyRec());
+print type_alias_list("MyRecAlias val", MyRecAlias());
+
+print type_alias_list("MyRec type", MyRec);
+print type_alias_list("MyRecalias type", MyRecAlias);
+
+local mys: MyString = "hi";
+print type_alias_list("MyString val", mys);
+print type_alias_list("MyString type", MyString);
+print type_alias_list("MyOtherString type", MyOtherString);
+print type_alias_list("AnotherString type", AnotherString);
+
+print type_alias_list("string literal value", "test");
+print type_alias_list("count literal value", 7);
+
+print type_alias_list("MyTable value", MyTable([1] = "one", [2] = "two"));
+print type_alias_list("MyTable2 value", MyTable2([1] = "one", [2] = "two"));
+print type_alias_list("MyTable3 value", MyTable3([1] = "one", [2] = "two"));
+print type_alias_list("MyTable4 value", MyTable4([1] = "one", [2] = "two"));
+print type_alias_list("MyTable type", MyTable);
+print type_alias_list("MyTable2 type", MyTable2);
+print type_alias_list("MyTable3 type", MyTable3);
+print type_alias_list("MyTable4 type", MyTable4);
+print type_alias_list("table value", table([1] = "one", [2] = "two"));