Normalized and Unnormalized C/C++ AST Grammars

For C and C++ programs:

• ASTs represent the source text after preprocessing (though the existence of C preprocessor macros may be determined).
• There is one unnormalized AST per compilation unit.
• Normalized ASTs are available for a smaller set of syntactic constructs than unnormalized ASTs, which are fully complete.
• There can (in general) exist an optional normalized AST associated with the compilation unit, but there is no one, singly-rooted tree for the normalized ASTs; there is instead a forest of different abstract syntax trees, roughly one AST per program point.
• The unnormalized ASTs for C and C++ programs may also be thought of as a forest of trees, as the rough correspondence between ASTs and program points holds for unnormalized trees also.

CodeSonar generates two kinds of ASTs for C/C++ programs. We refer to these as the two AST families.

For each AST family we provide full details of each AST class along with two indexes: an alphabetical index to facilitate lookup by class name, and a tree index whose structure reflects that of the class inheritance hierarchy.

Family	Purpose	AST Family ; AST Class representation in API (see API note)
		C++	Python	C
unnormalized ASTs	An AST that has not been affected by normalization. Unnormalized ASTs are truer to the source code but are also more complex and unwieldy than normalized ASTs. full unnormalized AST class details tree index to unnormalized AST classes alphabetical index to unnormalized AST classes	cs::ast_family::C_UNNORMALIZED; cs::ast_class::UC_*	cs.ast_family.C_UNNORMALIZED; cs.ast_class.UC_*	csaf_c_unnormalized; csac_uc_*
normalized ASTs	An abstract syntax tree that represents program points that CodeSonar has normalized. Normalized ASTs are simpler and easier to learn than unnormalized ASTs, while still giving all the semantics of the program. Normalized ASTs are preferable for most applications because your scripts have fewer cases to handle. full normalized AST class details tree index to normalized AST classes alphabetical index to normalized AST classes normalized AST examples	cs::ast_family::C_NORMALIZED; cs::ast_class::NC_*	cs.ast_family.C_NORMALIZED; cs.ast_class.NC_*	csaf_c_normalized; csac_nc_*

The C/C++ AST families share a number of helper types.

• helper types for C/C++ ASTs

C and C++ ASTs Associated With Program Points

Program points of some kinds have associated ASTs, as shown in the following table.

normAST	"X" if program points of this kind have an associated normalized C/C++ AST, "." otherwise.
unnormASTs	"X" if program points of this kind have one or more associated unnormalized C/C++ AST, "." otherwise.

Kind	normAST	unnormASTs
actual-in	X	X
actual-out	X	X
auxiliary	.	.
body	.	X
call-site	.	X
control-point	X	X
declaration	.	X
entry	.	X
exceptional-exit	X	X
exceptional-return	X	X
exit	.	X
expression	X	X
formal-in	.	X
formal-out	.	X
global-actual-in	.	.
global-actual-out	.	.
global-formal-in	.	.
global-formal-out	.	.
indirect-call	X	X
jump	X	X
label	X	X
normal-exit	X	X
normal-return	X	X
phi	.	.
pi	.	.
return	X	X
switch-case	X	X
unavailable	.	X
unknown	.	.
variable-initialization	X	X