xml_parsert: construct with message handler

This both avoids an object of static lifetime as well as it fixes the
(transitive) use of the deprecated messaget() constructor. Avoid global
objects in the lexer (as side-effect making it reentrant) as
initialisation is required. The parser continues to have global state,
so guard against reentrant use.

src/xmllang/parser.y

@@ -3,12 +3,12 @@
 
 #include "xml_parser.h"
 
-int yyxmllex();
-extern char *yyxmltext;
+int yyxmllex(void *);
+char *yyxmlget_text(void *);
 
-int yyxmlerror(const std::string &error)
+int yyxmlerror(xml_parsert &xml_parser, void *scanner, const std::string &error)
 {
-  xml_parser.parse_error(error, yyxmltext);
+  xml_parser.parse_error(error, yyxmlget_text(scanner));
   return 0;
 }
 
@@ -26,6 +26,10 @@ int yyxmlerror(const std::string &error)
 #endif
 %}
 
+%parse-param {xml_parsert &xml_parser}
+%parse-param {void *scanner}
+%lex-param {void *scanner}
+
 %union {char *s;}
 
 %token STARTXMLDECL

src/xmllang/scanner.l

@@ -3,6 +3,9 @@
 %option noinput
 %option nounistd
 %option never-interactive
+%option noyywrap
+%option reentrant
+%option extra-type="xml_parsert *"
 
 %{
 
@@ -19,7 +22,7 @@
 #include "xml_parser.h"
 #include "xml_y.tab.h"
 
-#define PARSER xml_parser
+#define PARSER (*yyextra)
 
 //static int keep;  /* To store start condition */
 
@@ -87,9 +90,5 @@ string     \"([^"&]|{esc})*\"|\'([^'&]|{esc})*\'
 <DTD>.            {/* skip */}
 <DTD>\]{close}    {BEGIN(INITIAL); /* skip */}
 
-.                 { yyxmlerror("unexpected character"); }
+.                 { yyxmlerror(*yyextra, yyscanner, "unexpected character"); }
 {nl}              {/* skip, must be an extra one at EOF */;}
-
-%%
-
-int yywrap() { return 1; }

src/xmllang/xml_parser.cpp

@@ -10,7 +10,20 @@ Author: Daniel Kroening, [email protected]
 
 #include <fstream>
 
-xml_parsert xml_parser;
+int xml_parsert::instance_count = 0;
+
+int yyxmllex_init_extra(xml_parsert *, void **);
+int yyxmllex_destroy(void *);
+int yyxmlparse(xml_parsert &, void *);
+
+bool xml_parsert::parse()
+{
+  void *scanner;
+  yyxmllex_init_extra(this, &scanner);
+  bool parse_fail = yyxmlparse(*this, scanner) != 0;
+  yyxmllex_destroy(scanner);
+  return parse_fail;
+}
 
 // 'do it all' function
 bool parse_xml(
@@ -19,19 +32,16 @@ bool parse_xml(
   message_handlert &message_handler,
   xmlt &dest)
 {
-  xml_parser.clear();
+  xml_parsert xml_parser{message_handler};
+
   xml_parser.set_file(filename);
   xml_parser.in=&in;
-  xml_parser.log.set_message_handler(message_handler);
 
-  bool result=yyxmlparse()!=0;
+  bool result = xml_parser.parse();
 
   // save result
   xml_parser.parse_tree.element.swap(dest);
 
-  // save some memory
-  xml_parser.clear();
-
   return result;
 }
 

src/xmllang/xml_parser.h

@@ -14,11 +14,25 @@ Author: Daniel Kroening, [email protected]
 
 #include "xml_parse_tree.h"
 
-int yyxmlparse();
-
 class xml_parsert:public parsert
 {
 public:
+  explicit xml_parsert(message_handlert &message_handler)
+    : parsert(message_handler)
+  {
+    // Simplistic check that we don't attempt to do reentrant parsing as the
+    // Bison-generated parser has global state.
+    PRECONDITION(++instance_count == 1);
+    stack.push_back(&parse_tree.element);
+  }
+
+  xml_parsert(const xml_parsert &) = delete;
+
+  ~xml_parsert() override
+  {
+    --instance_count;
+  }
+
   xml_parse_treet parse_tree;
 
   std::list<xmlt *> stack;
@@ -28,29 +42,30 @@ class xml_parsert:public parsert
     return *stack.back();
   }
 
-  virtual bool parse()
-  {
-    return yyxmlparse()!=0;
-  }
+  bool parse() override;
 
   void new_level()
   {
     current().elements.push_back(xmlt());
     stack.push_back(&current().elements.back());
   }
 
-  virtual void clear()
+  /// Clears the parser state. May be removed in future as there should not be a
+  /// need to re-use an existing parser object.
+  void clear() override
   {
     parse_tree.clear();
     // set up stack
     stack.clear();
     stack.push_back(&parse_tree.element);
+    parsert::clear();
   }
-};
 
-extern xml_parsert xml_parser;
+protected:
+  static int instance_count;
+};
 
-int yyxmlerror(const std::string &error);
+int yyxmlerror(xml_parsert &, void *, const std::string &);
 
 // 'do it all' functions
 bool parse_xml(