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 CC_EQ(class, len, litmatch) ((len) == sizeof (litmatch)-1 \
  22                                     && *(class) == *(litmatch) \
  23                                     && strncmp((char*)class, litmatch, len) == 0)
  24
  25 #if defined STDC_HEADERS || !defined isascii
  26 # define ISASCII(c) 1
  27 #else
  28 # define ISASCII(c) isascii(c)
  29 #endif
  30
  31 #ifdef isblank
  32 # define ISBLANK(c) (ISASCII(c) && isblank(c))
  33 #else
  34 # define ISBLANK(c) ((c) == ' ' || (c) == '\t')
  35 #endif
  36
  37 #ifdef isgraph
  38 # define ISGRAPH(c) (ISASCII(c) && isgraph(c))
  39 #else
  40 # define ISGRAPH(c) (ISASCII(c) && isprint(c) && !isspace(c))
  41 #endif
  42
  43 #define ISPRINT(c) (ISASCII(c) && isprint(c))
  44 #define ISDIGIT(c) (ISASCII(c) && isdigit(c))
  45 #define ISALNUM(c) (ISASCII(c) && isalnum(c))
  46 #define ISALPHA(c) (ISASCII(c) && isalpha(c))
  47 #define ISCNTRL(c) (ISASCII(c) && iscntrl(c))
  48 #define ISLOWER(c) (ISASCII(c) && islower(c))
  49 #define ISPUNCT(c) (ISASCII(c) && ispunct(c))
  50 #define ISSPACE(c) (ISASCII(c) && isspace(c))
  51 #define ISUPPER(c) (ISASCII(c) && isupper(c))
  52 #define ISXDIGIT(c) (ISASCII(c) && isxdigit(c))
  53
  54 /* Match pattern "p" against "text" */
  55 static int dowild(const uchar *p, const uchar *text, unsigned int flags)
  56 {
  57         uchar p_ch;
  58         const uchar *pattern = p;
  59
  60         for ( ; (p_ch = *p) != '\0'; text++, p++) {
  61                 int matched, match_slash, negated;
  62                 uchar t_ch, prev_ch;
  63                 if ((t_ch = *text) == '\0' && p_ch != '*')
  64                         return WM_ABORT_ALL;
  65                 if ((flags & WM_CASEFOLD) && ISUPPER(t_ch))
  66                         t_ch = tolower(t_ch);
  67                 if ((flags & WM_CASEFOLD) && ISUPPER(p_ch))
  68                         p_ch = tolower(p_ch);
  69                 switch (p_ch) {
  70                 case '\\':
  71                         /* Literal match with following character.  Note that the test
  72                          * in "default" handles the p[1] == '\0' failure case. */
  73                         p_ch = *++p;
  74                         /* FALLTHROUGH */
  75                 default:
  76                         if (t_ch != p_ch)
  77                                 return WM_NOMATCH;
  78                         continue;
  79                 case '?':
  80                         /* Match anything but '/'. */
  81                         if ((flags & WM_PATHNAME) && t_ch == '/')
  82                                 return WM_NOMATCH;
  83                         continue;
  84                 case '*':
  85                         if (*++p == '*') {
  86                                 const uchar *prev_p = p - 2;
  87                                 while (*++p == '*') {}
  88                                 if (!(flags & WM_PATHNAME))
  89                                         /* without WM_PATHNAME, '*' == '**' */
  90                                         match_slash = 1;
  91                                 else if ((prev_p < pattern || *prev_p == '/') &&
  92                                     (*p == '\0' || *p == '/' ||
  93                                      (p[0] == '\\' && p[1] == '/'))) {
  94                                         /*
  95                                          * Assuming we already match 'foo/' and are at
  96                                          * <star star slash>, just assume it matches
  97                                          * nothing and go ahead match the rest of the
  98                                          * pattern with the remaining string. This
  99                                          * helps make foo/<*><*>/bar (<> because
 100                                          * otherwise it breaks C comment syntax) match
 101                                          * both foo/bar and foo/a/bar.
 102                                          */
 103                                         if (p[0] == '/' &&
 104                                             dowild(p + 1, text, flags) == WM_MATCH)
 105                                                 return WM_MATCH;
 106                                         match_slash = 1;
 107                                 } else
 108                                         return WM_ABORT_MALFORMED;
 109                         } else
 110                                 /* without WM_PATHNAME, '*' == '**' */
 111                                 match_slash = flags & WM_PATHNAME ? 0 : 1;
 112                         if (*p == '\0') {
 113                                 /* Trailing "**" matches everything.  Trailing "*" matches
 114                                  * only if there are no more slash characters. */
 115                                 if (!match_slash) {
 116                                         if (strchr((char*)text, '/') != NULL)
 117                                                 return WM_NOMATCH;
 118                                 }
 119                                 return WM_MATCH;
 120                         } else if (!match_slash && *p == '/') {
 121                                 /*
 122                                  * _one_ asterisk followed by a slash
 123                                  * with WM_PATHNAME matches the next
 124                                  * directory
 125                                  */
 126                                 const char *slash = strchr((char*)text, '/');
 127                                 if (!slash)
 128                                         return WM_NOMATCH;
 129                                 text = (const uchar*)slash;
 130                                 /* the slash is consumed by the top-level for loop */
 131                                 break;
 132                         }
 133                         while (1) {
 134                                 if (t_ch == '\0')
 135                                         break;
 136                                 if ((matched = dowild(p, text, flags)) != WM_NOMATCH) {
 137                                         if (!match_slash || matched != WM_ABORT_TO_STARSTAR)
 138                                                 return matched;
 139                                 } else if (!match_slash && t_ch == '/')
 140                                         return WM_ABORT_TO_STARSTAR;
 141                                 t_ch = *++text;
 142                         }
 143                         return WM_ABORT_ALL;
 144                 case '[':
 145                         p_ch = *++p;
 146 #ifdef NEGATE_CLASS2
 147                         if (p_ch == NEGATE_CLASS2)
 148                                 p_ch = NEGATE_CLASS;
 149 #endif
 150                         /* Assign literal 1/0 because of "matched" comparison. */
 151                         negated = p_ch == NEGATE_CLASS ? 1 : 0;
 152                         if (negated) {
 153                                 /* Inverted character class. */
 154                                 p_ch = *++p;
 155                         }
 156                         prev_ch = 0;
 157                         matched = 0;
 158                         do {
 159                                 if (!p_ch)
 160                                         return WM_ABORT_ALL;
 161                                 if (p_ch == '\\') {
 162                                         p_ch = *++p;
 163                                         if (!p_ch)
 164                                                 return WM_ABORT_ALL;
 165                                         if (t_ch == p_ch)
 166                                                 matched = 1;
 167                                 } else if (p_ch == '-' && prev_ch && p[1] && p[1] != ']') {
 168                                         p_ch = *++p;
 169                                         if (p_ch == '\\') {
 170                                                 p_ch = *++p;
 171                                                 if (!p_ch)
 172                                                         return WM_ABORT_ALL;
 173                                         }
 174                                         if (t_ch <= p_ch && t_ch >= prev_ch)
 175                                                 matched = 1;
 176                                         p_ch = 0; /* This makes "prev_ch" get set to 0. */
 177                                 } else if (p_ch == '[' && p[1] == ':') {
 178                                         const uchar *s;
 179                                         int i;
 180                                         for (s = p += 2; (p_ch = *p) && p_ch != ']'; p++) {} /*SHARED ITERATOR*/
 181                                         if (!p_ch)
 182                                                 return WM_ABORT_ALL;
 183                                         i = p - s - 1;
 184                                         if (i < 0 || p[-1] != ':') {
 185                                                 /* Didn't find ":]", so treat like a normal set. */
 186                                                 p = s - 2;
 187                                                 p_ch = '[';
 188                                                 if (t_ch == p_ch)
 189                                                         matched = 1;
 190                                                 continue;
 191                                         }
 192                                         if (CC_EQ(s,i, "alnum")) {
 193                                                 if (ISALNUM(t_ch))
 194                                                         matched = 1;
 195                                         } else if (CC_EQ(s,i, "alpha")) {
 196                                                 if (ISALPHA(t_ch))
 197                                                         matched = 1;
 198                                         } else if (CC_EQ(s,i, "blank")) {
 199                                                 if (ISBLANK(t_ch))
 200                                                         matched = 1;
 201                                         } else if (CC_EQ(s,i, "cntrl")) {
 202                                                 if (ISCNTRL(t_ch))
 203                                                         matched = 1;
 204                                         } else if (CC_EQ(s,i, "digit")) {
 205                                                 if (ISDIGIT(t_ch))
 206                                                         matched = 1;
 207                                         } else if (CC_EQ(s,i, "graph")) {
 208                                                 if (ISGRAPH(t_ch))
 209                                                         matched = 1;
 210                                         } else if (CC_EQ(s,i, "lower")) {
 211                                                 if (ISLOWER(t_ch))
 212                                                         matched = 1;
 213                                         } else if (CC_EQ(s,i, "print")) {
 214                                                 if (ISPRINT(t_ch))
 215                                                         matched = 1;
 216                                         } else if (CC_EQ(s,i, "punct")) {
 217                                                 if (ISPUNCT(t_ch))
 218                                                         matched = 1;
 219                                         } else if (CC_EQ(s,i, "space")) {
 220                                                 if (ISSPACE(t_ch))
 221                                                         matched = 1;
 222                                         } else if (CC_EQ(s,i, "upper")) {
 223                                                 if (ISUPPER(t_ch))
 224                                                         matched = 1;
 225                                         } else if (CC_EQ(s,i, "xdigit")) {
 226                                                 if (ISXDIGIT(t_ch))
 227                                                         matched = 1;
 228                                         } else /* malformed [:class:] string */
 229                                                 return WM_ABORT_ALL;
 230                                         p_ch = 0; /* This makes "prev_ch" get set to 0. */
 231                                 } else if (t_ch == p_ch)
 232                                         matched = 1;
 233                         } while (prev_ch = p_ch, (p_ch = *++p) != ']');
 234                         if (matched == negated ||
 235                             ((flags & WM_PATHNAME) && t_ch == '/'))
 236                                 return WM_NOMATCH;
 237                         continue;
 238                 }
 239         }
 240
 241         return *text ? WM_NOMATCH : WM_MATCH;
 242 }
 243
 244 /* Match the "pattern" against the "text" string. */
 245 int wildmatch(const char *pattern, const char *text,
 246               unsigned int flags, struct wildopts *wo)
 247 {
 248         return dowild((const uchar*)pattern, (const uchar*)text, flags);
 249 }