Fix corner case where struct attributes are declared from #include

The `struct link_map` definition of glibc includes an header which must
be expanded for correct output.  This is an corner-case which was already
acknowledged for enums, hence expand and refactor the logic to TagDecls.

Signed-off-by: Giuliano Belinassi <gbelinassi@suse.de>

Author: giulianobelinassi

libcextract/ArgvParser.cpp

@@ -58,6 +58,8 @@ static const char *UnsupportedGCCArgs[] = {
   "-Wno-stringop-truncation",
   "-Werror=designated-init",
   "-Wimplicit-fallthrough=",
+  "-Wendif-labels",
+  "-Wmissing-selector-name",
 };
 
 ArgvParser::ArgvParser(int argc, char **argv)

libcextract/PrettyPrint.cpp

@@ -20,12 +20,13 @@
 #include "LLVMMisc.hh"
 
 #include <clang/AST/Attr.h>
+#include <llvm/Support/Regex.h>
 
 /** Public methods.  */
 
 #define Out (*Out)
 
-void PrettyPrint::Print_Decl(Decl *decl)
+void PrettyPrint::Print_Decl(Decl *decl, bool keep_includes)
 {
   /* When handling C code, we need to check if given declaration is a function
      with body.  If yes, we can simply print the declaration, but otherwise
@@ -38,22 +39,22 @@ void PrettyPrint::Print_Decl(Decl *decl)
   if (f && f->hasBody() && f->isThisDeclarationADefinition()) {
     Print_Decl_Raw(f);
     Out << "\n\n";
-  } else if (e) {
-      decl->print(Out, PPolicy);
-      Out << ";\n\n";
   } else if (!e && t && t->getName() == "") {
     /* If the RecordType doesn't have a name, then don't print it.  Except when
        it is an empty named enum declaration, which in this case we must print
        because it contains declared constants.  */
   } else {
-    /* Structs and prototypes */
+    /* Structs, enums, and prototypes */
 
-    /** Check if we can get a partial declaration of `decl` rather than a full
-        declaration.  */
-    bool full_def_removed = false;
-    if (t && t->isCompleteDefinitionRequired() == false) {
-      /* We don't need the full defintion.  Hide the body for Print_Decl_Raw.  */
-      t->setCompleteDefinition(false);
+    /** Check if the output string has an include directive.  In that case, if we
+        don't want to keep the includes, we must print the ast dump rather than
+        the source text to get rid of the #include directive.
+        FIXME:  regexes are slow.  */
+    static llvm::Regex regex("# *include *(<|\")");
+
+    /* FIXME: Why isCompleteDefinitionRequired does not work for EnumDecls?  */
+    if (t && ((!e && t->isCompleteDefinitionRequired() == false)
+              || (!keep_includes && regex.match(Get_Source_Text(t->getSourceRange()))))) {
 
       /* FIXME: The Print_Decl_Raw class will attempt to write this declaration
          as the user wrote, that means WITH a body.  To avoid this, we set
@@ -63,7 +64,15 @@ void PrettyPrint::Print_Decl(Decl *decl)
          correct way of doing this would be update the source location to
          the correct range.  */
       t->setLocStart(t->getEndLoc());
+    }
 
+    /** Check if we can get a partial declaration of `decl` rather than a full
+        declaration.  */
+    bool full_def_removed = false;
+    /* FIXME: Why isCompleteDefinitionRequired does not work for EnumDecls?  */
+    if (t && !e && t->isCompleteDefinitionRequired() == false) {
+      /* We don't need the full defintion.  Hide the body for Print_Decl_Raw.  */
+      t->setCompleteDefinition(false);
       full_def_removed = true;
     }
 
@@ -613,7 +622,7 @@ void RecursivePrint::Print_Decl(Decl *decl)
         PrettyPrint::Print_RawComment(sm, comment);
       }
     }
-    PrettyPrint::Print_Decl(decl);
+    PrettyPrint::Print_Decl(decl, KeepIncludes);
   }
 }
 
@@ -656,13 +665,7 @@ void RecursivePrint::Print(void)
         decl = (*ASTIterator).AsDecl;
         Print_Decl(decl);
         ++ASTIterator;
-
-        /* Skip to the end of the Declaration.  */
-        if (!dyn_cast<EnumDecl>(decl)) {
-          /* EnumDecls are handled somewhat differently: we dump the AST, not
-             what the user wrote.  */
-          ASTIterator.Skip_Until(decl->getEndLoc());
-        }
+        ASTIterator.Skip_Until(decl->getEndLoc());
         break;
 
       case TopLevelASTIterator::ReturnType::TYPE_PREPROCESSED_ENTITY:

libcextract/PrettyPrint.hh

@@ -38,7 +38,7 @@ class PrettyPrint
 {
   public:
   /** Print a Decl node into ostream `Out`.  */
-  static void Print_Decl(Decl *decl);
+  static void Print_Decl(Decl *decl, bool keep_includes = false);
 
   /** Print Decl node as is, without any kind of processing.  */
   static void Print_Decl_Raw(Decl *decl);

testsuite/small/enum-11.c

@@ -0,0 +1,14 @@
+/* { dg-options "-DCE_EXTRACT_FUNCTIONS=f -DCE_NO_EXTERNALIZATION" }*/
+
+enum consts {
+  CST0 = 0,
+  CST1,
+#include "enums.h"
+};
+
+enum consts f(void)
+{
+  return CST2;
+}
+
+/* { dg-final { scan-tree-dump "CST2," } } */

testsuite/small/enum-6.c

@@ -16,4 +16,4 @@ int f()
 
 /* { dg-final { scan-tree-dump "ANOTHER_CONST = 1 << CONST0" } } */
 /* { dg-final { scan-tree-dump "return ANOTHER_CONST;" } } */
-/* { dg-final { scan-tree-dump "enum *{\n *CONST0" } } */
+/* { dg-final { scan-tree-dump "enum\n? *{\n? *CONST0" } } */

testsuite/small/enum-8.c

@@ -32,6 +32,6 @@ int f() {
 }
 
 /* { dg-final { scan-tree-dump "TRACE_EVENT_FL_TRACEPOINT_BIT," } } */
-/* { dg-final { scan-tree-dump "TRACE_EVENT_FL_TRACEPOINT *= \(1 << TRACE_EVENT_FL_TRACEPOINT_BIT\)," } } */
+/* { dg-final { scan-tree-dump "TRACE_EVENT_FL_TRACEPOINT( |\t)*= \(1 << TRACE_EVENT_FL_TRACEPOINT_BIT\)," } } */
 /* { dg-final { scan-tree-dump "#define MACRO TRACE_EVENT_FL_TRACEPOINT" } } */
 /* { dg-final { scan-tree-dump "return 1 \+ MACRO;" } } */

testsuite/small/enums.h

@@ -0,0 +1,2 @@
+CST2,
+CST3,

testsuite/small/fields.h

@@ -0,0 +1,2 @@
+int field2;
+int field3;

testsuite/small/record-12.c

@@ -0,0 +1,15 @@
+/* { dg-options "-DCE_EXTRACT_FUNCTIONS=f -DCE_NO_EXTERNALIZATION" }*/
+
+struct record {
+  int field1;
+#   include    "fields.h"
+  int field4;
+};
+
+int f(struct record *r)
+{
+  return r->field2;
+}
+
+/* { dg-final { scan-tree-dump "struct record {" } } */
+/* { dg-final { scan-tree-dump "int field2;" } } */

testsuite/small/typedef-6.c

@@ -21,8 +21,8 @@ int f(struct A *a) {
   return a->y;
 }
 
-/* { dg-final { scan-tree-dump "enum CONSTANT {\n *CONST1,\n *CONST2\n};" } } */
-/* { dg-final { scan-tree-dump "enum CONSTANT2 {\n *AAAA,\n *BBBB\n};" } } */
+/* { dg-final { scan-tree-dump "enum CONSTANT {\n *CONST1,\n *CONST2,?\n};" } } */
+/* { dg-final { scan-tree-dump "enum CONSTANT2 {\n *AAAA,\n *BBBB,?\n};" } } */
 /* { dg-final { scan-tree-dump "typedef int \(\*GEN_SESSION_CB\)\(enum CONSTANT \*, enum CONSTANT2 \*\);" } } */
 /* { dg-final { scan-tree-dump "struct A {\n *GEN_SESSION_CB a;\n *int y;\n};" } } */
 /* { dg-final { scan-tree-dump "int f" } } */