Fixed problem with duplicate named groups in regular expressions

Author: nickgammon

regexp.cpp

@@ -4,6 +4,9 @@
 #include "MUSHclient.h"
 #include "doc.h"
 #include "dialogs\RegexpProblemDlg.h"
+#include "pcre\config.h"
+#include "pcre\pcre.h"
+#include "pcre\pcre_internal.h"
 
 #ifdef _DEBUG
 #define new DEBUG_NEW
@@ -146,4 +149,70 @@ pcre * program;
   dlg.m_iColumn = erroroffset + 1;
   dlg.DoModal ();
   return false;   // bad
-  }
+  }
+
+
+  // returns a named wildcard
+string t_regexp::GetWildcard (const string& sName) const
+  {
+  int iNumber = PCRE_ERROR_NOSUBSTRING;
+  if (IsStringNumber (sName))
+    iNumber = atoi (sName.c_str ());
+  else
+    {
+    if (m_program == NULL)
+      iNumber = PCRE_ERROR_NOSUBSTRING;
+    else
+      {
+      /* now do named subpatterns  */
+      int namecount;
+      pcre_fullinfo(m_program, m_extra, PCRE_INFO_NAMECOUNT, &namecount);
+
+
+      if (namecount > 0)
+        {
+        unsigned char *name_table;
+        int name_entry_size;
+        unsigned char *tabptr;
+        int ncapt;
+        int jchanged;
+        pcre_fullinfo(m_program, m_extra, PCRE_INFO_CAPTURECOUNT, &ncapt);
+        pcre_fullinfo(m_program, m_extra, PCRE_INFO_NAMETABLE, &name_table);
+        pcre_fullinfo(m_program, m_extra, PCRE_INFO_NAMEENTRYSIZE, &name_entry_size);
+        pcre_fullinfo(m_program, m_extra, PCRE_INFO_JCHANGED, &jchanged);
+        tabptr = name_table;
+        set<string> found_strings;
+        for (int i = 0; i < namecount; i++, tabptr += name_entry_size) 
+          {
+          int n = (tabptr[0] << 8) | tabptr[1];
+          const unsigned char * name = tabptr + 2;
+          // if duplicates were possible then ...
+          if (jchanged)
+            {
+            // this code is to ensure that we don't find a match (eg. mob = Kobold)
+            // and then if duplicates were allowed, replace Kobold with false.
+
+            string sName = (LPCTSTR) name;
+
+            // for duplicate names, see if we already added this name
+            if (found_strings.find (sName) != found_strings.end ())
+              {
+              // do not replace if this one is out of range or empty
+              if (n < 0 || n > m_iCount || GetWildcard (n) == "")
+                continue;
+              } // end of duplicate
+            else
+              found_strings.insert (sName);
+            }
+
+          if (n >= 0 && n <= m_iCount) 
+            iNumber = n;
+      
+          }   // end of wildcard loop
+        } // end of having named wildcards
+
+      } // end of program not NULL
+
+    }   // end of wanting a named wildcard
+  return GetWildcard (iNumber);
+  } // end of t_regexp::GetWildcard

regexp.h

@@ -74,20 +74,7 @@ class t_regexp
     };
 
   // returns a named wildcard
-  string GetWildcard (const string& sName) const
-    {
-    int iNumber;
-    if (IsStringNumber (sName))
-      iNumber = atoi (sName.c_str ());
-    else
-      {
-      if (m_program == NULL)
-        iNumber = PCRE_ERROR_NOSUBSTRING;
-      else
-        iNumber = njg_get_first_set (m_program, sName.c_str (), &m_vOffsets [0]);
-      }
-    return GetWildcard (iNumber);
-    }
+  string GetWildcard (const string& sName) const;
 
   };
 

scripting/lua_scripting.cpp

@@ -546,16 +546,18 @@ bool CScriptEngine::ExecuteLua (DISPID & dispid,  // dispatch ID, will be set to
     pcre_fullinfo(regexp->m_program, regexp->m_extra, PCRE_INFO_NAMECOUNT, &namecount);
     if (namecount > 0)
       {
+      int jchanged;
       pcre_fullinfo(regexp->m_program, regexp->m_extra, PCRE_INFO_NAMETABLE, &name_table);
       pcre_fullinfo(regexp->m_program, regexp->m_extra, PCRE_INFO_NAMEENTRYSIZE, &name_entry_size);
+      pcre_fullinfo(regexp->m_program, regexp->m_extra, PCRE_INFO_JCHANGED, &jchanged);
       tabptr = name_table;
       set<string> found_strings;
       for (i = 0; i < namecount; i++, tabptr += name_entry_size) 
         {
         int n = (tabptr[0] << 8) | tabptr[1];
         const unsigned char * name = tabptr + 2;
         // if duplicates were possible then ...
-        if ((regexp->m_program->options & (PCRE_DUPNAMES | PCRE_JCHANGED)) != 0)
+        if (jchanged)
           {
           // this code is to ensure that we don't find a match (eg. mob = Kobold)
           // and then if duplicates were allowed, replace Kobold with false.
@@ -566,7 +568,7 @@ bool CScriptEngine::ExecuteLua (DISPID & dispid,  // dispatch ID, will be set to
           if (found_strings.find (sName) != found_strings.end ())
             {
             // do not replace if this one is out of range
-            if (n < 0 || n > ncapt)
+            if (n < 0 || n > ncapt || regexp->GetWildcard (n) == "")
               continue;
             } // end of duplicate
           else