WDL/has_strings.h

   1 #ifndef _WDL_HASSTRINGS_H_
   2 #define _WDL_HASSTRINGS_H
   3
   4 #ifndef WDL_HASSTRINGS_EXPORT
   5 #define WDL_HASSTRINGS_EXPORT
   6 #endif
   7
   8 static bool hasStrings_isNonWordChar(int c)
   9 {
  10   // treat as whitespace when searching for " foo "
  11   switch (c)
  12   {
  13     case ' ':
  14     case '\t':
  15     case '.':
  16     case '/':
  17     case '\\':
  18
  19       return true;
  20
  21     default:
  22       return false;
  23   }
  24 }
  25
  26 WDL_HASSTRINGS_EXPORT bool WDL_hasStrings(const char *name, const LineParser *lp)
  27 {
  28   if (!lp) return true;
  29   const int ntok = lp->getnumtokens();
  30   if (ntok<1) return true;
  31
  32   char stack[1024]; // &1=not bit, 0x10 = ignoring subscopes, &2= state when 0x10 set
  33   int stacktop = 0;
  34   stack[0]=0;
  35
  36   int x,ln;
  37   const int strlen_name = (int)strlen(name);
  38   char matched_local=-1; // -1 = first eval for scope, 0=did not pass scope, 1=OK, 2=ignore rest of scope
  39   for (x = 0; x < ntok; x ++)
  40   {
  41     const char *n=lp->gettoken_str(x);
  42
  43     if (n[0] == '(' && !n[1] && !lp->gettoken_quotingchar(x))
  44     {
  45       if (!(matched_local&1))
  46       {
  47         stack[stacktop] |= matched_local | 0x10;
  48         matched_local=2; // ignore subscope
  49       }
  50       else
  51       {
  52         matched_local = -1; // new scope
  53       }
  54
  55       if (stacktop < (int)sizeof(stack) - 1) stack[++stacktop] = 0;
  56     }
  57     else if (stacktop && n[0] == ')' && !n[1] && !lp->gettoken_quotingchar(x))
  58     {
  59       if (stack[--stacktop]&0x10)
  60       {
  61         // restore state
  62         matched_local = stack[stacktop]&2;
  63       }
  64       else
  65       {
  66         matched_local = (matched_local != 0 ? 1 : 0) ^ (stack[stacktop]&1);
  67       }
  68       stack[stacktop] = 0;
  69     }
  70     else if (matched_local != 2 && !strcmp(n,"OR"))
  71     {
  72       matched_local = (matched_local > 0) ? 2 : -1;
  73       stack[stacktop] = 0;
  74     }
  75     else if (matched_local&1) // matches 1, -1
  76     {
  77       if (!strcmp(n,"NOT"))
  78       {
  79         stack[stacktop]^=1;
  80       }
  81       else if ((ln=(int)strlen(n))>0)
  82       {
  83         int lt=strlen_name;
  84         const char *t=name;
  85         // ^foo -- string starts (or follows \1 separator with) foo
  86         // foo$ -- string ends with foo (or is immediately followed by \1 separator)
  87         // " foo ", "foo ", " foo" include end of string/start of string has whitespace
  88         int wc_left = 0; // 1=require \1 or start of string, 2=require space or \1 or start
  89         int wc_right = 0; // 1=require \1 or \0, 2 = require space or \1 or \0
  90         // perhaps wc_left/wc_right of 2 should also match non-alnum characters in addition to space?
  91         if (ln>1)
  92         {
  93           switch (*n)
  94           {
  95             case ' ':
  96               if (*++n != ' ') wc_left=2;
  97               // else { multiple characters of whitespace = literal whitespace search (two spaces requires a single space, etc) }
  98
  99               ln--;
 100             break;
 101             case '^':
 102               ln--;
 103               n++;
 104               wc_left=1;
 105             break;
 106           }
 107         }
 108         if (ln>1)
 109         {
 110           switch (n[ln-1])
 111           {
 112             case ' ':
 113               if (n[--ln - 1] != ' ') wc_right=2;
 114               // else { multiple characters of whitespace = literal whitespace search (two spaces requires a single space, etc) }
 115             break;
 116             case '$':
 117               ln--;
 118               wc_right++;
 119             break;
 120           }
 121         }
 122         if (!wc_left && !wc_right && *n)
 123         {
 124           switch (lp->gettoken_quotingchar(x))
 125           {
 126             case '\'':
 127             case '"':
 128               { // if a quoted string has no whitespace in it, treat as whole word search
 129                 const char *p = n;
 130                 while (*p && *p != ' ' && *p != '\t') p++;
 131                 if (!*p) wc_left=wc_right=2;
 132               }
 133             break;
 134           }
 135
 136         }
 137
 138         if (wc_left>0)
 139         {
 140           unsigned char lastchar = 1;
 141           while (lt>=ln)
 142           {
 143             if ((lastchar < 2 || (wc_left>1 && hasStrings_isNonWordChar(lastchar))) && !strnicmp(t,n,ln))
 144             {
 145               if (wc_right == 0) break;
 146               const unsigned char nc=((const unsigned char*)t)[ln];
 147               if (nc < 2 || (wc_right > 1 && hasStrings_isNonWordChar(nc))) break;
 148             }
 149             lastchar = *(unsigned char*)t++;
 150             lt--;
 151           }
 152         }
 153         else
 154         {
 155           while (lt>=ln)
 156           {
 157             if (!strnicmp(t,n,ln))
 158             {
 159               if (wc_right == 0) break;
 160               const unsigned char nc=((const unsigned char*)t)[ln];
 161               if (nc < 2 || (wc_right > 1 && hasStrings_isNonWordChar(nc))) break;
 162             }
 163             t++;
 164             lt--;
 165           }
 166         }
 167
 168         matched_local = ((lt-ln)>=0) ^ (stack[stacktop]&1);
 169         stack[stacktop]=0;
 170       }
 171     }
 172   }
 173   while (stacktop > 0)
 174   {
 175     if (stack[--stacktop] & 0x10) matched_local=stack[stacktop]&2;
 176     else matched_local = (matched_local > 0 ? 1 : 0) ^ (stack[stacktop]&1);
 177   }
 178
 179   return matched_local!=0;
 180 }
 181
 182 WDL_HASSTRINGS_EXPORT bool WDL_makeSearchFilter(const char *flt, LineParser *lp)
 183 {
 184   if (!lp || !flt || !*flt) return false;
 185
 186 #ifdef WDL_LINEPARSER_HAS_LINEPARSERINT
 187   if (lp->parse_ex(flt,true,false,true)) // allow unterminated quotes
 188 #else
 189   if (lp->parse_ex(flt,true,false))
 190 #endif
 191   {
 192     lp->set_one_token(flt); // failed parsing search string, search as a single token
 193   }
 194
 195   return lp->getnumtokens()>0;
 196 }
 197
 198 #endif