Merge remote-tracking branch 'origin/topic/jsiwek/983'

Closes #983.

* origin/topic/jsiwek/983:
  Add named constructor examples to docs.
  Allow named vector constructors. Addresses #983.
  Allow named table constructors.  Addresses #983.
  Improve set constructor argument coercion.
  Allow named set constructors. Addresses #983.
  Allow named record constructors. Addresses #983.

NEWS

@@ -46,6 +46,19 @@ New Functionality
   have changed their signatures to work with opaques types rather
   than global state as it was before.
 
+- The scripting language now supports a constructing sets, tables,
+  vectors, and records by name:
+
+        type MyRecordType: record {
+            c: count;
+            s: string &optional;
+        };
+
+        global r: MyRecordType = record($c = 7);
+
+        type MySet: set[MyRec];
+        global s = MySet([$c=1], [$c=2]);
+
 - Strings now support the subscript operator to extract individual
   characters and substrings (e.g., s[4], s[1,5]). The index expression
   can take up to two indices for the start and end index of the

aux/broctl

@@ -1 +1 @@
-Subproject commit 3389de4a6045451f66b6cd52074c746ec9be551e
+Subproject commit 0eca32b35d16a4d387f41976ab46360ee6ecaed8

doc/scripts/builtins.rst

@@ -246,6 +246,31 @@ The Bro scripting language supports the following built-in types.
             [5] = "five",
         };
 
+    A table constructor (equivalent to above example) can also be used
+    to create a table:
+
+    .. code:: bro
+
+        global t2: table[count] of string = table(
+            [11] = "eleven",
+            [5] = "five"
+        );
+
+    Table constructors can also be explicitly named by a type, which is
+    useful for when a more complex index type could otherwise be
+    ambiguous:
+
+    .. code:: bro
+
+        type MyRec: record {
+            a: count &optional;
+            b: count;
+        };
+
+        type MyTable: table[MyRec] of string;
+
+        global t3 = MyTable([[$b=5]] = "b5", [[$b=7]] = "b7");
+
     Accessing table elements if provided by enclosing values within square
     brackets (``[]``), for example:
 
@@ -308,6 +333,28 @@ The Bro scripting language supports the following built-in types.
     The types are explicitly shown in the example above, but they could
     have been left to type inference.
 
+    A set constructor (equivalent to above example) can also be used to
+    create a set:
+
+    .. code:: bro
+
+        global s3: set[port] = set(21/tcp, 23/tcp, 80/tcp, 443/tcp);
+
+    Set constructors can also be explicitly named by a type, which is
+    useful for when a more complex index type could otherwise be
+    ambiguous:
+
+    .. code:: bro
+
+        type MyRec: record {
+            a: count &optional;
+            b: count;
+        };
+
+        type MySet: set[MyRec];
+
+        global s4 = MySet([$b=1], [$b=2]);
+
     Set membership is tested with ``in``:
 
     .. code:: bro
@@ -349,6 +396,21 @@ The Bro scripting language supports the following built-in types.
 
         global v: vector of string = vector("one", "two", "three");
 
+    Vector constructors can also be explicitly named by a type, which
+    is useful for when a more complex yield type could otherwise be
+    ambiguous.
+
+    .. code:: bro
+
+        type MyRec: record {
+            a: count &optional;
+            b: count;
+        };
+
+        type MyVec: vector of MyRec;
+
+        global v2 = MyVec([$b=1], [$b=2], [$b=3]);
+
     Adding an element to a vector involves accessing/assigning it:
 
     .. code:: bro
@@ -402,6 +464,19 @@ The Bro scripting language supports the following built-in types.
         if ( r?$s )
             ...
 
+    Records can also be created using a constructor syntax:
+
+    .. code:: bro
+
+        global r2: MyRecordType = record($c = 7);
+
+    And the constructor can be explicitly named by type, too, which
+    is arguably more readable code:
+
+    .. code:: bro
+
+        global r3 = MyRecordType($c = 42);
+
 .. bro:type:: opaque
 
     A data type whose actual representation/implementation is

src/Expr.cc

@@ -3320,12 +3320,20 @@ bool HasFieldExpr::DoUnserialize(UnserialInfo* info)
 	return UNSERIALIZE(&not_used) && UNSERIALIZE_STR(&field_name, 0) && UNSERIALIZE(&field);
 	}
 
-RecordConstructorExpr::RecordConstructorExpr(ListExpr* constructor_list)
+RecordConstructorExpr::RecordConstructorExpr(ListExpr* constructor_list,
+					     BroType* arg_type)
 : UnaryExpr(EXPR_RECORD_CONSTRUCTOR, constructor_list)
 	{
 	if ( IsError() )
 		return;
 
+	if ( arg_type && arg_type->Tag() != TYPE_RECORD )
+		{
+		Error("bad record constructor type", arg_type);
+		SetError();
+		return;
+		}
+
 	// Spin through the list, which should be comprised of
 	// either record's or record-field-assign, and build up a
 	// record type to associate with this constructor.
@@ -3365,7 +3373,17 @@ RecordConstructorExpr::RecordConstructorExpr(ListExpr* constructor_list)
 			}
 		}
 
-	SetType(new RecordType(record_types));
+	ctor_type = new RecordType(record_types);
+
+	if ( arg_type )
+		SetType(arg_type->Ref());
+	else
+		SetType(ctor_type->Ref());
+	}
+
+RecordConstructorExpr::~RecordConstructorExpr()
+	{
+	Unref(ctor_type);
 	}
 
 Val* RecordConstructorExpr::InitVal(const BroType* t, Val* aggr) const
@@ -3391,7 +3409,7 @@ Val* RecordConstructorExpr::InitVal(const BroType* t, Val* aggr) const
 Val* RecordConstructorExpr::Fold(Val* v) const
 	{
 	ListVal* lv = v->AsListVal();
-	RecordType* rt = type->AsRecordType();
+	RecordType* rt = ctor_type->AsRecordType();
 
 	if ( lv->Length() != rt->NumFields() )
 		Internal("inconsistency evaluating record constructor");
@@ -3401,6 +3419,19 @@ Val* RecordConstructorExpr::Fold(Val* v) const
 	for ( int i = 0; i < lv->Length(); ++i )
 		rv->Assign(i, lv->Index(i)->Ref());
 
+	if ( ! same_type(rt, type) )
+		{
+		RecordVal* new_val = rv->CoerceTo(type->AsRecordType());
+
+		if ( new_val )
+			{
+			Unref(rv);
+			rv = new_val;
+			}
+		else
+			Internal("record constructor coercion failed");
+		}
+
 	return rv;
 	}
 
@@ -3416,37 +3447,89 @@ IMPLEMENT_SERIAL(RecordConstructorExpr, SER_RECORD_CONSTRUCTOR_EXPR);
 bool RecordConstructorExpr::DoSerialize(SerialInfo* info) const
 	{
 	DO_SERIALIZE(SER_RECORD_CONSTRUCTOR_EXPR, UnaryExpr);
+	SERIALIZE_OPTIONAL(ctor_type);
 	return true;
 	}
 
 bool RecordConstructorExpr::DoUnserialize(UnserialInfo* info)
 	{
 	DO_UNSERIALIZE(UnaryExpr);
+	BroType* t = 0;
+	UNSERIALIZE_OPTIONAL(t, RecordType::Unserialize(info));
+	ctor_type = t->AsRecordType();
 	return true;
 	}
 
 TableConstructorExpr::TableConstructorExpr(ListExpr* constructor_list,
-						attr_list* arg_attrs)
+					   attr_list* arg_attrs, BroType* arg_type)
 : UnaryExpr(EXPR_TABLE_CONSTRUCTOR, constructor_list)
 	{
 	if ( IsError() )
 		return;
 
-	if ( constructor_list->Exprs().length() == 0 )
-		SetType(new TableType(new TypeList(base_type(TYPE_ANY)), 0));
+	if ( arg_type )
+		{
+		if ( ! arg_type->IsTable() )
+			{
+			Error("bad table constructor type", arg_type);
+			SetError();
+			return;
+			}
+
+		SetType(arg_type->Ref());
+		}
 	else
 		{
-		SetType(init_type(constructor_list));
+		if ( constructor_list->Exprs().length() == 0 )
+			SetType(new TableType(new TypeList(base_type(TYPE_ANY)), 0));
+		else
+			{
+			SetType(init_type(constructor_list));
 
-		if ( ! type )
-			SetError();
+			if ( ! type )
+				SetError();
 
-		else if ( type->Tag() != TYPE_TABLE ||
-			  type->AsTableType()->IsSet() )
-			SetError("values in table(...) constructor do not specify a table");
+			else if ( type->Tag() != TYPE_TABLE ||
+				  type->AsTableType()->IsSet() )
+				SetError("values in table(...) constructor do not specify a table");
+			}
 		}
 
 	attrs = arg_attrs ? new Attributes(arg_attrs, type, false) : 0;
+
+	type_list* indices = type->AsTableType()->Indices()->Types();
+	const expr_list& cle = constructor_list->Exprs();
+
+	// check and promote all index expressions in ctor list
+	loop_over_list(cle, i)
+		{
+		if ( cle[i]->Tag() != EXPR_ASSIGN )
+			continue;
+
+		Expr* idx_expr = cle[i]->AsAssignExpr()->Op1();
+
+		if ( idx_expr->Tag() != EXPR_LIST )
+			continue;
+
+		expr_list& idx_exprs = idx_expr->AsListExpr()->Exprs();
+
+		if ( idx_exprs.length() != indices->length() )
+			continue;
+
+		loop_over_list(idx_exprs, j)
+			{
+			Expr* idx = idx_exprs[j];
+
+			if ( check_and_promote_expr(idx, (*indices)[j]) )
+				{
+				if ( idx != idx_exprs[j] )
+					idx_exprs.replace(j, idx);
+				continue;
+				}
+
+			ExprError("inconsistent types in table constructor");
+			}
+		}
 	}
 
 Val* TableConstructorExpr::Eval(Frame* f) const
@@ -3502,16 +3585,30 @@ bool TableConstructorExpr::DoUnserialize(UnserialInfo* info)
 	}
 
 SetConstructorExpr::SetConstructorExpr(ListExpr* constructor_list,
-					attr_list* arg_attrs)
+				       attr_list* arg_attrs, BroType* arg_type)
 : UnaryExpr(EXPR_SET_CONSTRUCTOR, constructor_list)
 	{
 	if ( IsError() )
 		return;
 
-	if ( constructor_list->Exprs().length() == 0 )
-		SetType(new ::SetType(new TypeList(base_type(TYPE_ANY)), 0));
+	if ( arg_type )
+		{
+		if ( ! arg_type->IsSet() )
+			{
+			Error("bad set constructor type", arg_type);
+			SetError();
+			return;
+			}
+
+		SetType(arg_type->Ref());
+		}
 	else
-		SetType(init_type(constructor_list));
+		{
+		if ( constructor_list->Exprs().length() == 0 )
+			SetType(new ::SetType(new TypeList(base_type(TYPE_ANY)), 0));
+		else
+			SetType(init_type(constructor_list));
+		}
 
 	if ( ! type )
 		SetError();
@@ -3520,6 +3617,37 @@ SetConstructorExpr::SetConstructorExpr(ListExpr* constructor_list,
 		SetError("values in set(...) constructor do not specify a set");
 
 	attrs = arg_attrs ? new Attributes(arg_attrs, type, false) : 0;
+
+	type_list* indices = type->AsTableType()->Indices()->Types();
+	expr_list& cle = constructor_list->Exprs();
+
+	if ( indices->length() == 1 )
+		{
+		if ( ! check_and_promote_exprs_to_type(constructor_list,
+		                                       (*indices)[0]) )
+			ExprError("inconsistent type in set constructor");
+		}
+
+	else if ( indices->length() > 1 )
+		{
+		// Check/promote each expression in composite index.
+		loop_over_list(cle, i)
+			{
+			Expr* ce = cle[i];
+			ListExpr* le = ce->AsListExpr();
+
+			if ( ce->Tag() == EXPR_LIST &&
+			     check_and_promote_exprs(le, type->AsTableType()->Indices()) )
+				{
+				if ( le != cle[i] )
+					cle.replace(i, le);
+
+				continue;
+				}
+
+			ExprError("inconsistent types in set constructor");
+			}
+		}
 	}
 
 Val* SetConstructorExpr::Eval(Frame* f) const
@@ -3590,31 +3718,50 @@ bool SetConstructorExpr::DoUnserialize(UnserialInfo* info)
 	return true;
 	}
 
-VectorConstructorExpr::VectorConstructorExpr(ListExpr* constructor_list)
+VectorConstructorExpr::VectorConstructorExpr(ListExpr* constructor_list,
+					     BroType* arg_type)
 : UnaryExpr(EXPR_VECTOR_CONSTRUCTOR, constructor_list)
 	{
 	if ( IsError() )
 		return;
 
-	if ( constructor_list->Exprs().length() == 0 )
+	if ( arg_type )
 		{
-		// vector().
-		SetType(new ::VectorType(base_type(TYPE_ANY)));
-		return;
-		}
+		if ( arg_type->Tag() != TYPE_VECTOR )
+			{
+			Error("bad vector constructor type", arg_type);
+			SetError();
+			return;
+			}
 
-	BroType* t = merge_type_list(constructor_list);
-	if ( t )
-		{
-		SetType(new VectorType(t->Ref()));
-
-		if ( ! check_and_promote_exprs_to_type(constructor_list, t) )
-			ExprError("inconsistent types in vector constructor");
-
-		Unref(t);
+		SetType(arg_type->Ref());
 		}
 	else
-		SetError();
+		{
+		if ( constructor_list->Exprs().length() == 0 )
+			{
+			// vector().
+			SetType(new ::VectorType(base_type(TYPE_ANY)));
+			return;
+			}
+
+		BroType* t = merge_type_list(constructor_list);
+
+		if ( t )
+			{
+			SetType(new VectorType(t->Ref()));
+			Unref(t);
+			}
+		else
+			{
+			SetError();
+			return;
+			}
+		}
+
+	if ( ! check_and_promote_exprs_to_type(constructor_list,
+					       type->AsVectorType()->YieldType()) )
+		ExprError("inconsistent types in vector constructor");
 	}
 
 Val* VectorConstructorExpr::Eval(Frame* f) const

src/Expr.h

@@ -57,6 +57,8 @@ extern const char* expr_name(BroExprTag t);
 class Stmt;
 class Frame;
 class ListExpr;
+class NameExpr;
+class AssignExpr;
 class CallExpr;
 class EventExpr;
 
@@ -159,12 +161,37 @@ public:
 		CHECK_TAG(tag, EXPR_LIST, "ExprVal::AsListExpr", expr_name)
 		return (const ListExpr*) this;
 		}
+
 	ListExpr* AsListExpr()
 		{
 		CHECK_TAG(tag, EXPR_LIST, "ExprVal::AsListExpr", expr_name)
 		return (ListExpr*) this;
 		}
 
+	const NameExpr* AsNameExpr() const
+		{
+		CHECK_TAG(tag, EXPR_NAME, "ExprVal::AsNameExpr", expr_name)
+		return (const NameExpr*) this;
+		}
+
+	NameExpr* AsNameExpr()
+		{
+		CHECK_TAG(tag, EXPR_NAME, "ExprVal::AsNameExpr", expr_name)
+		return (NameExpr*) this;
+		}
+
+	const AssignExpr* AsAssignExpr() const
+		{
+		CHECK_TAG(tag, EXPR_ASSIGN, "ExprVal::AsAssignExpr", expr_name)
+		return (const AssignExpr*) this;
+		}
+
+	AssignExpr* AsAssignExpr()
+		{
+		CHECK_TAG(tag, EXPR_ASSIGN, "ExprVal::AsAssignExpr", expr_name)
+		return (AssignExpr*) this;
+		}
+
 	void Describe(ODesc* d) const;
 
 	bool Serialize(SerialInfo* info) const;
@@ -729,7 +756,8 @@ protected:
 
 class RecordConstructorExpr : public UnaryExpr {
 public:
-	RecordConstructorExpr(ListExpr* constructor_list);
+	RecordConstructorExpr(ListExpr* constructor_list, BroType* arg_type = 0);
+	~RecordConstructorExpr();
 
 protected:
 	friend class Expr;
@@ -741,11 +769,14 @@ protected:
 	void ExprDescribe(ODesc* d) const;
 
 	DECLARE_SERIAL(RecordConstructorExpr);
+
+	RecordType* ctor_type; // type inferred from the ctor expression list args
 };
 
 class TableConstructorExpr : public UnaryExpr {
 public:
-	TableConstructorExpr(ListExpr* constructor_list, attr_list* attrs);
+	TableConstructorExpr(ListExpr* constructor_list, attr_list* attrs,
+	                     BroType* arg_type = 0);
 	~TableConstructorExpr()	{ Unref(attrs); }
 
 	Attributes* Attrs() { return attrs; }
@@ -767,7 +798,8 @@ protected:
 
 class SetConstructorExpr : public UnaryExpr {
 public:
-	SetConstructorExpr(ListExpr* constructor_list, attr_list* attrs);
+	SetConstructorExpr(ListExpr* constructor_list, attr_list* attrs,
+	                   BroType* arg_type = 0);
 	~SetConstructorExpr()	{ Unref(attrs); }
 
 	Attributes* Attrs() { return attrs; }
@@ -789,7 +821,7 @@ protected:
 
 class VectorConstructorExpr : public UnaryExpr {
 public:
-	VectorConstructorExpr(ListExpr* constructor_list);
+	VectorConstructorExpr(ListExpr* constructor_list, BroType* arg_type = 0);
 
 	Val* Eval(Frame* f) const;
 

src/Type.h

@@ -217,6 +217,11 @@ public:
 		return tag == TYPE_TABLE && (YieldType() == 0);
 		}
 
+	int IsTable() const
+		{
+		return tag == TYPE_TABLE && (YieldType() != 0);
+		}
+
 	BroType* Ref()		{ ::Ref(this); return this; }
 
 	virtual void Describe(ODesc* d) const;

src/parse.y

@@ -522,10 +522,52 @@ expr:
 			$$ = new VectorConstructorExpr($3);
 			}
 
-	|	expr '(' opt_expr_list ')'
+	|	expr '('
 			{
-			set_location(@1, @4);
-			$$ = new CallExpr($1, $3, in_hook > 0);
+			if ( $1->Tag() == EXPR_NAME && $1->Type()->IsTable() )
+				++in_init;
+			}
+
+		opt_expr_list
+			{
+			if ( $1->Tag() == EXPR_NAME && $1->Type()->IsTable() )
+				--in_init;
+			}
+
+		')'
+			{
+			set_location(@1, @6);
+
+			BroType* ctor_type = 0;
+
+			if ( $1->Tag() == EXPR_NAME &&
+			     (ctor_type = $1->AsNameExpr()->Id()->AsType()) )
+				{
+				switch ( ctor_type->Tag() ) {
+				case TYPE_RECORD:
+					$$ = new RecordConstructorExpr($4, ctor_type);
+					break;
+
+				case TYPE_TABLE:
+					if ( ctor_type->IsTable() )
+						$$ = new TableConstructorExpr($4, 0, ctor_type);
+					else
+						$$ = new SetConstructorExpr($4, 0, ctor_type);
+
+					break;
+
+				case TYPE_VECTOR:
+					$$ = new VectorConstructorExpr($4, ctor_type);
+					break;
+
+				default:
+					$1->Error("constructor type not implemented");
+					YYERROR;
+				}
+				}
+
+			else
+				$$ = new CallExpr($1, $4, in_hook > 0);
 			}
 
 	|	TOK_HOOK { ++in_hook; } expr

testing/btest/Baseline/language.named-record-ctors/out

@@ -0,0 +1,2 @@
+[min=<uninitialized>, max=2]
+[min=7, max=42]

testing/btest/Baseline/language.named-set-ctors/out

@@ -0,0 +1,13 @@
+{
+1,
+5,
+3
+}
+{
+[min=<uninitialized>, max=5],
+[min=<uninitialized>, max=2]
+}
+{
+[test, 1] ,
+[cool, 2] 
+}

testing/btest/Baseline/language.named-table-ctors/out

@@ -0,0 +1,19 @@
+{
+[1] = one,
+[5] = five,
+[3] = three
+}
+{
+[[min=<uninitialized>, max=5]] = max5,
+[[min=<uninitialized>, max=2]] = max2
+}
+{
+[test, 1] = test1,
+[cool, 2] = cool2
+}
+{
+[two] = 2.0,
+[one] = 1.0,
+[three] = 3.0
+}
+0

testing/btest/Baseline/language.named-vector-ctors/out

@@ -0,0 +1,3 @@
+[one, two, three]
+[1.0, 2.0, 3.0]
+[[min=<uninitialized>, max=1], [min=<uninitialized>, max=2], [min=<uninitialized>, max=3]]

testing/btest/language/named-record-ctors.bro

@@ -0,0 +1,12 @@
+# @TEST-EXEC: bro -b %INPUT >out
+# @TEST-EXEC: btest-diff out
+
+type MyRec: record {
+	min: count &optional;
+	max: count;
+};
+
+local myrec: MyRec = MyRec($max=2);
+print myrec;
+myrec = MyRec($min=7, $max=42);
+print myrec;

testing/btest/language/named-set-ctors.bro

@@ -0,0 +1,19 @@
+# @TEST-EXEC: bro -b %INPUT >out
+# @TEST-EXEC: btest-diff out
+
+type MyRec: record {
+	min: count &optional;
+	max: count;
+};
+
+type FooSet: set[count];
+type FooSetRec: set[MyRec];
+type FooSetComp: set[string, count];
+
+global myset: FooSet = FooSet(1, 5, 3);
+global mysetrec: FooSetRec = FooSetRec([$max=5], [$max=2]);
+global mysetcomp: FooSetComp = FooSetComp(["test", 1], ["cool", 2]);
+
+print myset;
+print mysetrec;
+print mysetcomp;

testing/btest/language/named-table-ctors.bro

@@ -0,0 +1,24 @@
+# @TEST-EXEC: bro -b %INPUT >out
+# @TEST-EXEC: btest-diff out
+
+type MyRec: record {
+	min: count &optional;
+	max: count;
+};
+
+type FooTable: table[count] of string;
+type FooTableRec: table[MyRec] of string;
+type FooTableComp: table[string, count] of string;
+type FooTableY: table[string] of double;
+
+global mytable: FooTable = FooTable([1] = "one", [5] = "five", [3] = "three");
+global mytablerec: FooTableRec = FooTableRec([[$max=5]] = "max5", [[$max=2]] = "max2");
+global mytablecomp: FooTableComp = FooTableComp(["test", 1] = "test1", ["cool", 
+2] = "cool2");
+global mytabley: FooTableY = FooTableY(["one"] = 1, ["two"] = 2, ["three"] = 3) &default=0;
+
+print mytable;
+print mytablerec;
+print mytablecomp;
+print mytabley;
+print mytabley["test"];

testing/btest/language/named-vector-ctors.bro

@@ -0,0 +1,19 @@
+# @TEST-EXEC: bro -b %INPUT >out
+# @TEST-EXEC: btest-diff out
+
+type MyRec: record {
+	min: count &optional;
+	max: count;
+};
+
+type FooVector: vector of string;
+type FooVectorD: vector of double;
+type FooVectorRec: vector of MyRec;
+
+global myvec: FooVector = FooVector("one", "two", "three");
+global myvecd: FooVectorD = FooVectorD(1, 2, 3);
+global myvecrec: FooVectorRec = FooVectorRec([$max=1], [$max=2], [$max=3]);
+
+print myvec;
+print myvecd;
+print myvecrec;