Make directed callgraph include nodes with no edges

Note, if the callgraph was created from a particular entry
point then this will not change the output. It is only
when the callgraph is created for a goto_functionst that
the output will change - it will now include functions
that have no edges, i.e. that are not called by any other
functions and which don't call any other functions.

Author: Owen Jones

src/analyses/call_graph.cpp

@@ -41,8 +41,10 @@ call_grapht::call_grapht(
 {
   forall_goto_functions(f_it, goto_functions)
   {
-    const goto_programt &body=f_it->second.body;
-    add(f_it->first, body);
+    const irep_idt &function_name = f_it->first;
+    const goto_programt &body = f_it->second.body;
+    nodes.insert(function_name);
+    add(function_name, body);
   }
 }
 
@@ -84,12 +86,14 @@ call_grapht::call_grapht(
     const goto_programt &goto_program=
       goto_functions.function_map.at(function).body;
 
+    nodes.insert(function);
+
     forall_callsites(
       goto_program,
       [&](goto_programt::const_targett i_it, const irep_idt &callee)
       {
         add(function, callee, i_it);
-        if(graph.find(callee)==graph.end())
+        if(edges.find(callee)==edges.end())
           pending_stack.push(callee);
       }
     ); // NOLINT
@@ -129,7 +133,9 @@ void call_grapht::add(
   const irep_idt &caller,
   const irep_idt &callee)
 {
-  graph.insert(std::pair<irep_idt, irep_idt>(caller, callee));
+  edges.insert({caller, callee});
+  nodes.insert(caller);
+  nodes.insert(callee);
 }
 
 /// Add edge with optional callsite information
@@ -152,7 +158,8 @@ void call_grapht::add(
 call_grapht call_grapht::get_inverted() const
 {
   call_grapht result;
-  for(const auto &caller_callee : graph)
+  result.nodes = nodes;
+  for(const auto &caller_callee : edges)
     result.add(caller_callee.second, caller_callee.first);
   return result;
 }
@@ -197,7 +204,12 @@ call_grapht::directed_grapht call_grapht::get_directed_graph() const
   call_grapht::directed_grapht ret;
   function_indicest function_indices(ret);
 
-  for(const auto &edge : graph)
+  // To make sure we include unreachable functions we first create indices
+  // for all nodes in the graph
+  for(const irep_idt &function_name : nodes)
+    function_indices[function_name];
+
+  for(const auto &edge : edges)
   {
     auto a_index=function_indices[edge.first];
     auto b_index=function_indices[edge.second];
@@ -237,7 +249,7 @@ void call_grapht::output_dot(std::ostream &out) const
 {
   out << "digraph call_graph {\n";
 
-  for(const auto &edge : graph)
+  for(const auto &edge : edges)
   {
     out << "  \"" << edge.first << "\" -> "
         << "\"" << edge.second << "\" "
@@ -252,7 +264,7 @@ void call_grapht::output_dot(std::ostream &out) const
 
 void call_grapht::output(std::ostream &out) const
 {
-  for(const auto &edge : graph)
+  for(const auto &edge : edges)
   {
     out << edge.first << " -> " << edge.second << "\n";
     if(collect_callsites)
@@ -267,7 +279,7 @@ void call_grapht::output_xml(std::ostream &out) const
   if(collect_callsites)
     out << "<!-- XML call-graph representation does not document callsites yet."
       " If you need this, edit call_grapht::output_xml -->\n";
-  for(const auto &edge : graph)
+  for(const auto &edge : edges)
   {
     out << "<call_graph_edge caller=\"";
     xmlt::escape_attribute(id2string(edge.first), out);

src/analyses/call_graph.h

@@ -62,13 +62,17 @@ class call_grapht
   void output(std::ostream &out) const;
   void output_xml(std::ostream &out) const;
 
-  /// Type of the call graph. Note parallel edges (e.g. A having two callsites
-  /// both targeting B) result in multiple graph edges.
-  typedef std::multimap<irep_idt, irep_idt> grapht;
+  /// Type of the nodes in the call graph.
+  typedef std::unordered_set<irep_idt, irep_id_hash> nodest;
 
-  /// Type of a call graph edge in `grapht`
+  /// Type of the edges in the call graph. Note parallel edges (e.g. A having
+  /// two callsites both targeting B) result in multiple graph edges.
+  typedef std::multimap<irep_idt, irep_idt> edgest;
+
+  /// Type of a call graph edge in `edgest`
   typedef std::pair<irep_idt, irep_idt> edget;
 
+
   /// Type of a callsite stored in member `callsites`
   typedef goto_programt::const_targett locationt;
 
@@ -84,7 +88,8 @@ class call_grapht
   /// backward compatibility; use `get_directed_graph()` to get a generic
   /// directed graph representation that provides more graph algorithms
   /// (shortest path, SCCs and so on).
-  grapht graph;
+  edgest edges;
+  nodest nodes;
 
   /// Map from call-graph edges to a set of callsites that make the given call.
   callsitest callsites;

unit/analyses/call_graph.cpp

@@ -64,6 +64,7 @@ SCENARIO("call_graph",
     // }
     // void C() { }
     // void D() { }
+    // void E() { }
 
     goto_modelt goto_model;
     code_typet void_function_type;
@@ -101,6 +102,8 @@ SCENARIO("call_graph",
       create_void_function_symbol("C", code_skipt()));
     goto_model.symbol_table.add(
       create_void_function_symbol("D", code_skipt()));
+    goto_model.symbol_table.add(
+      create_void_function_symbol("E", code_skipt()));
 
     stream_message_handlert msg(std::cout);
     goto_convert(goto_model, msg);
@@ -111,7 +114,7 @@ SCENARIO("call_graph",
     {
       THEN("We expect A -> { A, B, B }, B -> { C, D }")
       {
-        const auto &check_graph=call_graph_from_goto_functions.graph;
+        const auto &check_graph=call_graph_from_goto_functions.edges;
         REQUIRE(check_graph.size()==5);
         REQUIRE(multimap_key_matches(check_graph, "A", {"A", "B", "B"}));
         REQUIRE(multimap_key_matches(check_graph, "B", {"C", "D"}));
@@ -128,7 +131,7 @@ SCENARIO("call_graph",
         call_graph_from_goto_functions.get_inverted();
       THEN("We expect A -> { A }, B -> { A, A }, C -> { B }, D -> { B }")
       {
-        const auto &check_graph=inverse_call_graph_from_goto_functions.graph;
+        const auto &check_graph=inverse_call_graph_from_goto_functions.edges;
         REQUIRE(check_graph.size()==5);
         REQUIRE(multimap_key_matches(check_graph, "A", {"A"}));
         REQUIRE(multimap_key_matches(check_graph, "B", {"A", "A"}));
@@ -143,9 +146,9 @@ SCENARIO("call_graph",
       THEN("We expect two callsites for the A -> B edge, one for all others")
       {
         const auto &check_callsites=call_graph_from_goto_functions.callsites;
-        for(const auto &edge : call_graph_from_goto_functions.graph)
+        for(const auto &edge : call_graph_from_goto_functions.edges)
         {
-          if(edge==call_grapht::grapht::value_type("A", "B"))
+          if(edge==call_grapht::edgest::value_type("A", "B"))
             REQUIRE(check_callsites.at(edge).size()==2);
           else
             REQUIRE(check_callsites.at(edge).size()==1);
@@ -167,7 +170,7 @@ SCENARIO("call_graph",
         call_grapht::create_from_root_function(goto_model, "B", false);
       THEN("We expect only B -> C and B -> D in the resulting graph")
       {
-        const auto &check_graph=call_graph_from_b.graph;
+        const auto &check_graph=call_graph_from_b.edges;
         REQUIRE(check_graph.size()==2);
         REQUIRE(multimap_key_matches(check_graph, "B", {"C", "D"}));
       }
@@ -183,6 +186,11 @@ SCENARIO("call_graph",
       for(node_indext i=0; i<exported.size(); ++i)
         nodes_by_name[exported[i].function]=i;
 
+      THEN("We expect 5 nodes")
+      {
+        REQUIRE(exported.size() == 5);
+      }
+
       THEN("We expect edges A -> { A, B }, B -> { C, D }")
       {
         // Note that means the extra A -> B edge has gone away (the grapht
@@ -228,6 +236,14 @@ SCENARIO("call_graph",
         REQUIRE(predecessors.count("B"));
         REQUIRE(predecessors.count("D"));
       }
+
+      THEN("We expect {E} to be able to reach E")
+      {
+        std::set<irep_idt> predecessors =
+          get_reaching_functions(exported, "E");
+        REQUIRE(predecessors.size() == 1);
+        REQUIRE(predecessors.count("E"));
+      }
     }
 
     WHEN("The call graph, with call sites, is exported as a grapht")
@@ -241,6 +257,11 @@ SCENARIO("call_graph",
       for(node_indext i=0; i<exported.size(); ++i)
         nodes_by_name[exported[i].function]=i;
 
+      THEN("We expect 5 nodes")
+      {
+        REQUIRE(exported.size() == 5);
+      }
+
       THEN("We expect edges A -> { A, B }, B -> { C, D }")
       {
         // Note that means the extra A -> B edge has gone away (the grapht