C++: Support printing of global and namespace variables in PrintAST

Author: jketema

cpp/ql/lib/change-notes/2023-07-20-print-global-variables.md

@@ -0,0 +1,4 @@
+---
+category: majorAnalysis
+---
+* The `PrintAST` library now also prints global and namespace variables and their initializers.

cpp/ql/lib/semmle/code/cpp/Print.qll

@@ -6,7 +6,7 @@ private import PrintAST
  * that requests that function, or no `PrintASTConfiguration` exists.
  */
 private predicate shouldPrintDeclaration(Declaration decl) {
-  not decl instanceof Function
+  not (decl instanceof Function or decl instanceof GlobalOrNamespaceVariable)
   or
   not exists(PrintAstConfiguration c)
   or

cpp/ql/lib/semmle/code/cpp/PrintAST.qll

@@ -1,9 +1,9 @@
 /**
  * Provides queries to pretty-print a C++ AST as a graph.
  *
- * By default, this will print the AST for all functions in the database. To change this behavior,
- * extend `PrintASTConfiguration` and override `shouldPrintDeclaration` to hold for only the
- * declarations you wish to view the AST for.
+ * By default, this will print the AST for all functions and global and namespace variables in
+ * the database. To change this behavior, extend `PrintASTConfiguration` and override
+ * `shouldPrintDeclaration` to hold for only the declarations you wish to view the AST for.
  */
 
 import cpp
@@ -22,16 +22,15 @@ class PrintAstConfiguration extends TPrintAstConfiguration {
 
   /**
    * Holds if the AST for `decl` should be printed. By default, holds for all
-   * functions. Currently, does not support any other declaration types.
+   * functions and global and namespace variables. Currently, does not support any
+   * other declaration types.
    */
   predicate shouldPrintDeclaration(Declaration decl) { any() }
 }
 
 private predicate shouldPrintDeclaration(Declaration decl) {
-  exists(PrintAstConfiguration config |
-    config.shouldPrintDeclaration(decl) and
-    decl instanceof Function
-  )
+  exists(PrintAstConfiguration config | config.shouldPrintDeclaration(decl)) and
+  (decl instanceof Function or decl instanceof GlobalOrNamespaceVariable)
 }
 
 bindingset[s]
@@ -72,7 +71,7 @@ private predicate locationSortKeys(Locatable ast, string file, int line, int col
   )
 }
 
-private Function getEnclosingFunction(Locatable ast) {
+private Declaration getEnclosingDeclaration(Locatable ast) {
   result = ast.(Expr).getEnclosingFunction()
   or
   result = ast.(Stmt).getEnclosingFunction()
@@ -81,6 +80,10 @@ private Function getEnclosingFunction(Locatable ast) {
   or
   result = ast.(Parameter).getFunction()
   or
+  result = ast.(Expr).getEnclosingDeclaration()
+  or
+  result = ast.(Initializer).getDeclaration()
+  or
   result = ast
 }
 
@@ -89,7 +92,7 @@ private Function getEnclosingFunction(Locatable ast) {
  * nodes for things like parameter lists and constructor init lists.
  */
 private newtype TPrintAstNode =
-  TAstNode(Locatable ast) { shouldPrintDeclaration(getEnclosingFunction(ast)) } or
+  TAstNode(Locatable ast) { shouldPrintDeclaration(getEnclosingDeclaration(ast)) } or
   TDeclarationEntryNode(DeclStmt stmt, DeclarationEntry entry) {
     // We create a unique node for each pair of (stmt, entry), to avoid having one node with
     // multiple parents due to extractor bug CPP-413.
@@ -161,10 +164,10 @@ class PrintAstNode extends TPrintAstNode {
 
   /**
    * Holds if this node should be printed in the output. By default, all nodes
-   * within a function are printed, but the query can override
-   * `PrintASTConfiguration.shouldPrintDeclaration` to filter the output.
+   * within functions and global and namespace variables are printed, but the query
+   * can override `PrintASTConfiguration.shouldPrintDeclaration` to filter the output.
    */
-  final predicate shouldPrint() { shouldPrintDeclaration(this.getEnclosingFunction()) }
+  final predicate shouldPrint() { shouldPrintDeclaration(this.getEnclosingDeclaration()) }
 
   /**
    * Gets the children of this node.
@@ -232,10 +235,15 @@ class PrintAstNode extends TPrintAstNode {
   abstract string getChildAccessorPredicateInternal(int childIndex);
 
   /**
-   * Gets the `Function` that contains this node.
+   * Gets the `Declaration` that contains this node.
    */
-  private Function getEnclosingFunction() {
-    result = this.getParent*().(FunctionNode).getFunction()
+  private Declaration getEnclosingDeclaration() { result = this.getParent*().getDeclaration() }
+
+  /**
+   * Gets the `Declaration` this node represents.
+   */
+  private Declaration getDeclaration() {
+    result = this.(AstNode).getAst() and shouldPrintDeclaration(result)
   }
 }
 
@@ -574,16 +582,53 @@ class DestructorDestructionsNode extends PrintAstNode, TDestructorDestructionsNo
   final Destructor getDestructor() { result = dtor }
 }
 
+abstract private class FunctionOrGlobalOrNamespaceVariableNode extends AstNode {
+  override string toString() { result = qlClass(ast) + getIdentityString(ast) }
+
+  private int getOrder() {
+    this =
+      rank[result](FunctionOrGlobalOrNamespaceVariableNode node, Declaration decl, string file,
+        int line, int column |
+        node.getAst() = decl and
+        locationSortKeys(decl, file, line, column)
+      |
+        node order by file, line, column, getIdentityString(decl)
+      )
+  }
+
+  override string getProperty(string key) {
+    result = super.getProperty(key)
+    or
+    key = "semmle.order" and result = this.getOrder().toString()
+  }
+}
+
+/**
+ * A node representing a `GlobalOrNamespaceVariable`.
+ */
+class GlobalOrNamespaceVariableNode extends FunctionOrGlobalOrNamespaceVariableNode {
+  GlobalOrNamespaceVariable var;
+
+  GlobalOrNamespaceVariableNode() { var = ast }
+
+  override PrintAstNode getChildInternal(int childIndex) {
+    childIndex = 0 and
+    result.(AstNode).getAst() = var.getInitializer()
+  }
+
+  override string getChildAccessorPredicateInternal(int childIndex) {
+    childIndex = 0 and result = "getInitializer()"
+  }
+}
+
 /**
  * A node representing a `Function`.
  */
-class FunctionNode extends AstNode {
+class FunctionNode extends FunctionOrGlobalOrNamespaceVariableNode {
   Function func;
 
   FunctionNode() { func = ast }
 
-  override string toString() { result = qlClass(func) + getIdentityString(func) }
-
   override PrintAstNode getChildInternal(int childIndex) {
     childIndex = 0 and
     result.(ParametersNode).getFunction() = func
@@ -607,31 +652,10 @@ class FunctionNode extends AstNode {
     or
     childIndex = 3 and result = "<destructions>"
   }
-
-  private int getOrder() {
-    this =
-      rank[result](FunctionNode node, Function function, string file, int line, int column |
-        node.getAst() = function and
-        locationSortKeys(function, file, line, column)
-      |
-        node order by file, line, column, getIdentityString(function)
-      )
-  }
-
-  override string getProperty(string key) {
-    result = super.getProperty(key)
-    or
-    key = "semmle.order" and result = this.getOrder().toString()
-  }
-
-  /**
-   * Gets the `Function` this node represents.
-   */
-  final Function getFunction() { result = func }
 }
 
 private string getChildAccessorWithoutConversions(Locatable parent, Element child) {
-  shouldPrintDeclaration(getEnclosingFunction(parent)) and
+  shouldPrintDeclaration(getEnclosingDeclaration(parent)) and
   (
     exists(Stmt s | s = parent |
       namedStmtChildPredicates(s, child, result)
@@ -650,7 +674,7 @@ private string getChildAccessorWithoutConversions(Locatable parent, Element chil
 }
 
 private predicate namedStmtChildPredicates(Locatable s, Element e, string pred) {
-  shouldPrintDeclaration(getEnclosingFunction(s)) and
+  shouldPrintDeclaration(getEnclosingDeclaration(s)) and
   (
     exists(int n | s.(BlockStmt).getStmt(n) = e and pred = "getStmt(" + n + ")")
     or
@@ -738,7 +762,7 @@ private predicate namedStmtChildPredicates(Locatable s, Element e, string pred)
 }
 
 private predicate namedExprChildPredicates(Expr expr, Element ele, string pred) {
-  shouldPrintDeclaration(expr.getEnclosingFunction()) and
+  shouldPrintDeclaration(expr.getEnclosingDeclaration()) and
   (
     expr.(Access).getTarget() = ele and pred = "getTarget()"
     or

cpp/ql/test/examples/expressions/PrintAST.expected

@@ -1308,6 +1308,11 @@ union_etc.cpp:
 #    6|             Type = [IntType] int
 #    6|             ValueCategory = prvalue(load)
 #    6|     getStmt(1): [ReturnStmt] return ...
+#    7| [GlobalVariable] S s
+#    7|   getInitializer(): [Initializer] initializer for s
+#    7|     getExpr(): [ConstructorCall] call to S
+#    7|         Type = [VoidType] void
+#    7|         ValueCategory = prvalue
 #    9| [CopyAssignmentOperator] C& C::operator=(C const&)
 #    9|   <params>: 
 #-----|     getParameter(0): [Parameter] (unnamed parameter 0)
@@ -1332,6 +1337,7 @@ union_etc.cpp:
 #   12|   <params>: 
 #-----|     getParameter(0): [Parameter] (unnamed parameter 0)
 #-----|         Type = [RValueReferenceType] U &&
+#   14| [GlobalVariable] C c
 #   16| [CopyAssignmentOperator] U& U::operator=(U const&)
 #   16|   <params>: 
 #-----|     getParameter(0): [Parameter] (unnamed parameter 0)
@@ -1356,6 +1362,7 @@ union_etc.cpp:
 #   18|   <params>: 
 #-----|     getParameter(0): [Parameter] (unnamed parameter 0)
 #-----|         Type = [RValueReferenceType] C &&
+#   20| [GlobalVariable] U u
 #   22| [TopLevelFunction] int foo()
 #   22|   <params>: 
 #   22|   getEntryPoint(): [BlockStmt] { ... }

cpp/ql/test/library-tests/ir/ir/PrintAST.expected

@@ -8529,6 +8529,11 @@ ir.cpp:
 # 1035|               Type = [Struct] EmptyStruct
 # 1035|               ValueCategory = prvalue
 # 1036|     getStmt(1): [ReturnStmt] return ...
+# 1038| [GlobalVariable] (lambda [] type at line 1038, col. 12) lam
+# 1038|   getInitializer(): [Initializer] initializer for lam
+# 1038|     getExpr(): [LambdaExpression] [...](...){...}
+# 1038|         Type = [Closure] decltype([...](...){...})
+# 1038|         ValueCategory = prvalue
 # 1038| [CopyAssignmentOperator] (lambda [] type at line 1038, col. 12)& (lambda [] type at line 1038, col. 12)::operator=((lambda [] type at line 1038, col. 12) const&)
 # 1038|   <params>: 
 #-----|     getParameter(0): [Parameter] (unnamed parameter 0)
@@ -13976,6 +13981,54 @@ ir.cpp:
 # 1815|                 Type = [IntType] int
 # 1815|                 ValueCategory = prvalue(load)
 # 1817|     getStmt(8): [ReturnStmt] return ...
+# 1821| [GlobalVariable] int global_2
+# 1821|   getInitializer(): [Initializer] initializer for global_2
+# 1821|     getExpr(): [Literal] 1
+# 1821|         Type = [IntType] int
+# 1821|         Value = [Literal] 1
+# 1821|         ValueCategory = prvalue
+# 1823| [GlobalVariable] int const global_3
+# 1823|   getInitializer(): [Initializer] initializer for global_3
+# 1823|     getExpr(): [Literal] 2
+# 1823|         Type = [IntType] int
+# 1823|         Value = [Literal] 2
+# 1823|         ValueCategory = prvalue
+# 1825| [GlobalVariable] constructor_only global_4
+# 1825|   getInitializer(): [Initializer] initializer for global_4
+# 1825|     getExpr(): [ConstructorCall] call to constructor_only
+# 1825|         Type = [VoidType] void
+# 1825|         ValueCategory = prvalue
+# 1825|       getArgument(0): [Literal] 1
+# 1825|           Type = [IntType] int
+# 1825|           Value = [Literal] 1
+# 1825|           ValueCategory = prvalue
+# 1827| [GlobalVariable] constructor_only global_5
+# 1827|   getInitializer(): [Initializer] initializer for global_5
+# 1827|     getExpr(): [ConstructorCall] call to constructor_only
+# 1827|         Type = [VoidType] void
+# 1827|         ValueCategory = prvalue
+# 1827|       getArgument(0): [Literal] 2
+# 1827|           Type = [IntType] int
+# 1827|           Value = [Literal] 2
+# 1827|           ValueCategory = prvalue
+# 1829| [GlobalVariable] char* global_string
+# 1829|   getInitializer(): [Initializer] initializer for global_string
+# 1829|     getExpr(): global string
+# 1829|         Type = [ArrayType] const char[14]
+# 1829|         Value = [StringLiteral] "global string"
+# 1829|         ValueCategory = lvalue
+# 1829|     getExpr().getFullyConverted(): [CStyleCast] (char *)...
+# 1829|         Conversion = [PointerConversion] pointer conversion
+# 1829|         Type = [CharPointerType] char *
+# 1829|         ValueCategory = prvalue
+# 1829|       getExpr(): [ArrayToPointerConversion] array to pointer conversion
+# 1829|           Type = [PointerType] const char *
+# 1829|           ValueCategory = prvalue
+# 1831| [GlobalVariable] int global_6
+# 1831|   getInitializer(): [Initializer] initializer for global_6
+# 1831|     getExpr(): [VariableAccess] global_2
+# 1831|         Type = [IntType] int
+# 1831|         ValueCategory = prvalue(load)
 # 1834| [CopyAssignmentOperator] block_assignment::A& block_assignment::A::operator=(block_assignment::A const&)
 # 1834|   <params>: 
 #-----|     getParameter(0): [Parameter] (unnamed parameter 0)
@@ -14377,6 +14430,23 @@ ir.cpp:
 # 1885|             Type = [ClassTemplateInstantiation,Struct] Bar2<int>
 # 1885|             ValueCategory = lvalue
 # 1886|     getStmt(2): [ReturnStmt] return ...
+# 1889| [GlobalVariable] char global_template<char>
+# 1889|   getInitializer(): [Initializer] initializer for global_template
+# 1889|     getExpr(): [Literal] 42
+# 1889|         Type = [IntType] int
+# 1889|         Value = [Literal] 42
+# 1889|         ValueCategory = prvalue
+# 1889|     getExpr().getFullyConverted(): [CStyleCast] (char)...
+# 1889|         Conversion = [IntegralConversion] integral conversion
+# 1889|         Type = [PlainCharType] char
+# 1889|         Value = [CStyleCast] 42
+# 1889|         ValueCategory = prvalue
+# 1889| [GlobalVariable] int global_template<int>
+# 1889|   getInitializer(): [Initializer] initializer for global_template
+# 1889|     getExpr(): [Literal] 42
+# 1889|         Type = [IntType] int
+# 1889|         Value = [Literal] 42
+# 1889|         ValueCategory = prvalue
 # 1891| [TopLevelFunction] int test_global_template_int()
 # 1891|   <params>: 
 # 1891|   getEntryPoint(): [BlockStmt] { ... }
@@ -15170,6 +15240,40 @@ struct_init.cpp:
 #    4|   <params>: 
 #-----|     getParameter(0): [Parameter] (unnamed parameter 0)
 #-----|         Type = [RValueReferenceType] Info &&
+#    9| [GlobalVariable] Info infos_in_file[]
+#    9|   getInitializer(): [Initializer] initializer for infos_in_file
+#    9|     getExpr(): [ArrayAggregateLiteral] {...}
+#    9|         Type = [ArrayType] Info[2]
+#    9|         ValueCategory = prvalue
+#   10|       getAnElementExpr(0): [ClassAggregateLiteral] {...}
+#   10|           Type = [Struct] Info
+#   10|           ValueCategory = prvalue
+#   10|         getAFieldExpr(name): 1
+#   10|             Type = [ArrayType] const char[2]
+#   10|             Value = [StringLiteral] "1"
+#   10|             ValueCategory = lvalue
+#   10|         getAFieldExpr(handler): [FunctionAccess] handler1
+#   10|             Type = [FunctionPointerType] ..(*)(..)
+#   10|             ValueCategory = prvalue(load)
+#   10|         getAFieldExpr(name).getFullyConverted(): [ArrayToPointerConversion] array to pointer conversion
+#   10|             Type = [PointerType] const char *
+#   10|             ValueCategory = prvalue
+#   11|       getAnElementExpr(1): [ClassAggregateLiteral] {...}
+#   11|           Type = [Struct] Info
+#   11|           ValueCategory = prvalue
+#   11|         getAFieldExpr(name): 3
+#   11|             Type = [ArrayType] const char[2]
+#   11|             Value = [StringLiteral] "3"
+#   11|             ValueCategory = lvalue
+#   11|         getAFieldExpr(handler): [AddressOfExpr] & ...
+#   11|             Type = [FunctionPointerType] ..(*)(..)
+#   11|             ValueCategory = prvalue
+#   11|           getOperand(): [FunctionAccess] handler2
+#   11|               Type = [RoutineType] ..()(..)
+#   11|               ValueCategory = lvalue
+#   11|         getAFieldExpr(name).getFullyConverted(): [ArrayToPointerConversion] array to pointer conversion
+#   11|             Type = [PointerType] const char *
+#   11|             ValueCategory = prvalue
 #   16| [TopLevelFunction] void let_info_escape(Info*)
 #   16|   <params>: 
 #   16|     getParameter(0): [Parameter] info