wildmatch.c

   1 /*
   2 **  Do shell-style pattern matching for ?, \, [], and * characters.
   3 **  It is 8bit clean.
   4 **
   5 **  Written by Rich $alz, mirror!rs, Wed Nov 26 19:03:17 EST 1986.
   6 **  Rich $alz is now <rsalz@bbn.com>.
   7 **
   8 **  Modified by Wayne Davison to special-case '/' matching, to make '**'
   9 **  work differently than '*', and to fix the character-class code.
  10 */
  11
  12 #include "cache.h"
  13 #include "wildmatch.h"
  14
  15 typedef unsigned char uchar;
  16
  17 /* What character marks an inverted character class? */
  18 #define NEGATE_CLASS    '!'
  19 #define NEGATE_CLASS2   '^'
  20
  21 #define FALSE 0
  22 #define TRUE 1
  23
  24 #define NOMATCH 1
  25 #define MATCH 0
  26 #define ABORT_ALL -1
  27 #define ABORT_TO_STARSTAR -2
  28
  29 #define CC_EQ(class, len, litmatch) ((len) == sizeof (litmatch)-1 \
  30                                     && *(class) == *(litmatch) \
  31                                     && strncmp((char*)class, litmatch, len) == 0)
  32
  33 #if defined STDC_HEADERS || !defined isascii
  34 # define ISASCII(c) 1
  35 #else
  36 # define ISASCII(c) isascii(c)
  37 #endif
  38
  39 #ifdef isblank
  40 # define ISBLANK(c) (ISASCII(c) && isblank(c))
  41 #else
  42 # define ISBLANK(c) ((c) == ' ' || (c) == '\t')
  43 #endif
  44
  45 #ifdef isgraph
  46 # define ISGRAPH(c) (ISASCII(c) && isgraph(c))
  47 #else
  48 # define ISGRAPH(c) (ISASCII(c) && isprint(c) && !isspace(c))
  49 #endif
  50
  51 #define ISPRINT(c) (ISASCII(c) && isprint(c))
  52 #define ISDIGIT(c) (ISASCII(c) && isdigit(c))
  53 #define ISALNUM(c) (ISASCII(c) && isalnum(c))
  54 #define ISALPHA(c) (ISASCII(c) && isalpha(c))
  55 #define ISCNTRL(c) (ISASCII(c) && iscntrl(c))
  56 #define ISLOWER(c) (ISASCII(c) && islower(c))
  57 #define ISPUNCT(c) (ISASCII(c) && ispunct(c))
  58 #define ISSPACE(c) (ISASCII(c) && isspace(c))
  59 #define ISUPPER(c) (ISASCII(c) && isupper(c))
  60 #define ISXDIGIT(c) (ISASCII(c) && isxdigit(c))
  61
  62 static int force_lower_case = 0;
  63
  64 /* Match pattern "p" against "text" */
  65 static int dowild(const uchar *p, const uchar *text)
  66 {
  67         uchar p_ch;
  68
  69         for ( ; (p_ch = *p) != '\0'; text++, p++) {
  70                 int matched, special;
  71                 uchar t_ch, prev_ch;
  72                 if ((t_ch = *text) == '\0' && p_ch != '*')
  73                         return ABORT_ALL;
  74                 if (force_lower_case && ISUPPER(t_ch))
  75                         t_ch = tolower(t_ch);
  76                 switch (p_ch) {
  77                 case '\\':
  78                         /* Literal match with following character.  Note that the test
  79                          * in "default" handles the p[1] == '\0' failure case. */
  80                         p_ch = *++p;
  81                         /* FALLTHROUGH */
  82                 default:
  83                         if (t_ch != p_ch)
  84                                 return NOMATCH;
  85                         continue;
  86                 case '?':
  87                         /* Match anything but '/'. */
  88                         if (t_ch == '/')
  89                                 return NOMATCH;
  90                         continue;
  91                 case '*':
  92                         if (*++p == '*') {
  93                                 while (*++p == '*') {}
  94                                 special = TRUE;
  95                         } else
  96                                 special = FALSE;
  97                         if (*p == '\0') {
  98                                 /* Trailing "**" matches everything.  Trailing "*" matches
  99                                  * only if there are no more slash characters. */
 100                                 if (!special) {
 101                                         if (strchr((char*)text, '/') != NULL)
 102                                                 return NOMATCH;
 103                                 }
 104                                 return MATCH;
 105                         }
 106                         while (1) {
 107                                 if (t_ch == '\0')
 108                                         break;
 109                                 if ((matched = dowild(p, text)) != NOMATCH) {
 110                                         if (!special || matched != ABORT_TO_STARSTAR)
 111                                                 return matched;
 112                                 } else if (!special && t_ch == '/')
 113                                         return ABORT_TO_STARSTAR;
 114                                 t_ch = *++text;
 115                         }
 116                         return ABORT_ALL;
 117                 case '[':
 118                         p_ch = *++p;
 119 #ifdef NEGATE_CLASS2
 120                         if (p_ch == NEGATE_CLASS2)
 121                                 p_ch = NEGATE_CLASS;
 122 #endif
 123                         /* Assign literal TRUE/FALSE because of "matched" comparison. */
 124                         special = p_ch == NEGATE_CLASS? TRUE : FALSE;
 125                         if (special) {
 126                                 /* Inverted character class. */
 127                                 p_ch = *++p;
 128                         }
 129                         prev_ch = 0;
 130                         matched = FALSE;
 131                         do {
 132                                 if (!p_ch)
 133                                         return ABORT_ALL;
 134                                 if (p_ch == '\\') {
 135                                         p_ch = *++p;
 136                                         if (!p_ch)
 137                                                 return ABORT_ALL;
 138                                         if (t_ch == p_ch)
 139                                                 matched = TRUE;
 140                                 } else if (p_ch == '-' && prev_ch && p[1] && p[1] != ']') {
 141                                         p_ch = *++p;
 142                                         if (p_ch == '\\') {
 143                                                 p_ch = *++p;
 144                                                 if (!p_ch)
 145                                                         return ABORT_ALL;
 146                                         }
 147                                         if (t_ch <= p_ch && t_ch >= prev_ch)
 148                                                 matched = TRUE;
 149                                         p_ch = 0; /* This makes "prev_ch" get set to 0. */
 150                                 } else if (p_ch == '[' && p[1] == ':') {
 151                                         const uchar *s;
 152                                         int i;
 153                                         for (s = p += 2; (p_ch = *p) && p_ch != ']'; p++) {} /*SHARED ITERATOR*/
 154                                         if (!p_ch)
 155                                                 return ABORT_ALL;
 156                                         i = p - s - 1;
 157                                         if (i < 0 || p[-1] != ':') {
 158                                                 /* Didn't find ":]", so treat like a normal set. */
 159                                                 p = s - 2;
 160                                                 p_ch = '[';
 161                                                 if (t_ch == p_ch)
 162                                                         matched = TRUE;
 163                                                 continue;
 164                                         }
 165                                         if (CC_EQ(s,i, "alnum")) {
 166                                                 if (ISALNUM(t_ch))
 167                                                         matched = TRUE;
 168                                         } else if (CC_EQ(s,i, "alpha")) {
 169                                                 if (ISALPHA(t_ch))
 170                                                         matched = TRUE;
 171                                         } else if (CC_EQ(s,i, "blank")) {
 172                                                 if (ISBLANK(t_ch))
 173                                                         matched = TRUE;
 174                                         } else if (CC_EQ(s,i, "cntrl")) {
 175                                                 if (ISCNTRL(t_ch))
 176                                                         matched = TRUE;
 177                                         } else if (CC_EQ(s,i, "digit")) {
 178                                                 if (ISDIGIT(t_ch))
 179                                                         matched = TRUE;
 180                                         } else if (CC_EQ(s,i, "graph")) {
 181                                                 if (ISGRAPH(t_ch))
 182                                                         matched = TRUE;
 183                                         } else if (CC_EQ(s,i, "lower")) {
 184                                                 if (ISLOWER(t_ch))
 185                                                         matched = TRUE;
 186                                         } else if (CC_EQ(s,i, "print")) {
 187                                                 if (ISPRINT(t_ch))
 188                                                         matched = TRUE;
 189                                         } else if (CC_EQ(s,i, "punct")) {
 190                                                 if (ISPUNCT(t_ch))
 191                                                         matched = TRUE;
 192                                         } else if (CC_EQ(s,i, "space")) {
 193                                                 if (ISSPACE(t_ch))
 194                                                         matched = TRUE;
 195                                         } else if (CC_EQ(s,i, "upper")) {
 196                                                 if (ISUPPER(t_ch))
 197                                                         matched = TRUE;
 198                                         } else if (CC_EQ(s,i, "xdigit")) {
 199                                                 if (ISXDIGIT(t_ch))
 200                                                         matched = TRUE;
 201                                         } else /* malformed [:class:] string */
 202                                                 return ABORT_ALL;
 203                                         p_ch = 0; /* This makes "prev_ch" get set to 0. */
 204                                 } else if (t_ch == p_ch)
 205                                         matched = TRUE;
 206                         } while (prev_ch = p_ch, (p_ch = *++p) != ']');
 207                         if (matched == special || t_ch == '/')
 208                                 return NOMATCH;
 209                         continue;
 210                 }
 211         }
 212
 213         return *text ? NOMATCH : MATCH;
 214 }
 215
 216 /* Match the "pattern" against the "text" string. */
 217 int wildmatch(const char *pattern, const char *text)
 218 {
 219         return dowild((const uchar*)pattern, (const uchar*)text);
 220 }
 221
 222 /* Match the "pattern" against the forced-to-lower-case "text" string. */
 223 int iwildmatch(const char *pattern, const char *text)
 224 {
 225         int ret;
 226         force_lower_case = 1;
 227         ret = dowild((const uchar*)pattern, (const uchar*)text);
 228         force_lower_case = 0;
 229         return ret;
 230 }