Removed limit of 1000 wildcards in a regular expression

Author: nickgammon

regexp.cpp

@@ -42,18 +42,23 @@ pcre_extra * extra;
   re->m_extra = extra;
   re->m_iExecutionError = 0; // no error now
 
+  // inspired by a suggestion by Twisol (to remove a hard-coded limit on the number of wildcards)
+  int capturecount = 0;
+  // how many captures did we get?
+  pcre_fullinfo(program, NULL, PCRE_INFO_CAPTURECOUNT, &capturecount);
+  // allocate memory for them
+  re->m_vOffsets.resize ((capturecount + 1) * 3);  // add 1 for the whole expression
+
   return re;
   }
 
-#define MAX_PCRE_WILDCARDS 1000
-
 int regexec(register t_regexp *prog, 
             register const char *string,
             const int start_offset)
   {
 int options = App.m_bRegexpMatchEmpty ? 0 : PCRE_NOTEMPTY;    // don't match on an empty string
 int count;
-static int offsets [MAX_PCRE_WILDCARDS * 3];  // hopefully we won't recurse and crash ;)
+
 
 LARGE_INTEGER start, 
               finish;
@@ -67,7 +72,7 @@ LARGE_INTEGER start,
 
   pcre_callout = NULL;
   count = pcre_exec(prog->m_program, prog->m_extra, string, strlen (string),
-                    start_offset, options, offsets, NUMITEMS (offsets));
+                    start_offset, options, &prog->m_vOffsets [0], prog->m_vOffsets.size ());
 
   if (App.m_iCounterFrequency)
     {
@@ -83,6 +88,8 @@ LARGE_INTEGER start,
     {
     pcre_free (prog->m_program);
     prog->m_program = NULL;
+    pcre_free (prog->m_extra);
+    prog->m_extra = NULL;
     prog->m_iExecutionError = count; // remember reason
     }
 
@@ -99,9 +106,6 @@ LARGE_INTEGER start,
 
   prog->m_sTarget = string;  // for extracting wildcards
   prog->m_iCount = count;    // ditto
-  prog->m_vOffsets.clear ();
-  // only need first 2/3 of offsets
-  copy (offsets, &offsets [count * 2], back_inserter (prog->m_vOffsets));
 
   return true; // match
   }

scripting/lua_scripting.cpp

@@ -468,15 +468,10 @@ bool CScriptEngine::ExecuteLua (DISPID & dispid,  // dispatch ID, will be set to
     paramCount++;   // we have one more parameter to the call
     pcre_fullinfo(regexp->m_program, regexp->m_extra, PCRE_INFO_CAPTURECOUNT, &ncapt);
 
-    int iTot = regexp->m_vOffsets.size ();    // how many did we actually get?
-
     for (i = 0; i <= ncapt; i++) 
       {
       int j = i * 2;
-      if (j < iTot)    // save if available, otherwise push a false value
         lua_pushlstring(L, regexp->m_sTarget.c_str () + regexp->m_vOffsets[j], regexp->m_vOffsets[j + 1] - regexp->m_vOffsets[j]);
-      else
-        lua_pushboolean (L, 0);
       lua_rawseti (L, -2, i);
     }
     // now add item 0 - the whole matching line
@@ -508,7 +503,7 @@ 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 || j >= iTot)
+            if (n < 0 || n > ncapt)
               continue;
             } // end of duplicate
           else
@@ -517,12 +512,7 @@ bool CScriptEngine::ExecuteLua (DISPID & dispid,  // dispatch ID, will be set to
 
         lua_pushstring (L, (LPCTSTR) name);
         if (n >= 0 && n <= ncapt) 
-          {
-          if (j < iTot)    // save if available, otherwise push a false value
-            lua_pushlstring(L, regexp->m_sTarget.c_str () + regexp->m_vOffsets[j], regexp->m_vOffsets[j + 1] - regexp->m_vOffsets[j]);
-          else
-            lua_pushboolean(L, 0);
-        }
+          lua_pushlstring(L, regexp->m_sTarget.c_str () + regexp->m_vOffsets[j], regexp->m_vOffsets[j + 1] - regexp->m_vOffsets[j]);
         else
           lua_pushnil (L);  /* n out of range */
         lua_settable (L, -3);