index-pack: fix compilation with NO_PTHREADS
[gitweb.git] / wildmatch.c
index 6d992d314abccf675c2b3bc0e3285ee7c6abe1b7..f91ba99f32c047e5f3238668ae83de647ab92df2 100644 (file)
@@ -18,14 +18,6 @@ typedef unsigned char uchar;
 #define NEGATE_CLASS   '!'
 #define NEGATE_CLASS2  '^'
 
-#define FALSE 0
-#define TRUE 1
-
-#define NOMATCH 1
-#define MATCH 0
-#define ABORT_ALL -1
-#define ABORT_TO_STARSTAR -2
-
 #define CC_EQ(class, len, litmatch) ((len) == sizeof (litmatch)-1 \
                                    && *(class) == *(litmatch) \
                                    && strncmp((char*)class, litmatch, len) == 0)
@@ -59,20 +51,21 @@ typedef unsigned char uchar;
 #define ISUPPER(c) (ISASCII(c) && isupper(c))
 #define ISXDIGIT(c) (ISASCII(c) && isxdigit(c))
 
-static int force_lower_case = 0;
-
 /* Match pattern "p" against "text" */
-static int dowild(const uchar *p, const uchar *text)
+static int dowild(const uchar *p, const uchar *text, unsigned int flags)
 {
        uchar p_ch;
+       const uchar *pattern = p;
 
        for ( ; (p_ch = *p) != '\0'; text++, p++) {
-               int matched, special;
+               int matched, match_slash, negated;
                uchar t_ch, prev_ch;
                if ((t_ch = *text) == '\0' && p_ch != '*')
-                       return ABORT_ALL;
-               if (force_lower_case && ISUPPER(t_ch))
+                       return WM_ABORT_ALL;
+               if ((flags & WM_CASEFOLD) && ISUPPER(t_ch))
                        t_ch = tolower(t_ch);
+               if ((flags & WM_CASEFOLD) && ISUPPER(p_ch))
+                       p_ch = tolower(p_ch);
                switch (p_ch) {
                case '\\':
                        /* Literal match with following character.  Note that the test
@@ -81,150 +74,206 @@ static int dowild(const uchar *p, const uchar *text)
                        /* FALLTHROUGH */
                default:
                        if (t_ch != p_ch)
-                               return NOMATCH;
+                               return WM_NOMATCH;
                        continue;
                case '?':
                        /* Match anything but '/'. */
-                       if (t_ch == '/')
-                               return NOMATCH;
+                       if ((flags & WM_PATHNAME) && t_ch == '/')
+                               return WM_NOMATCH;
                        continue;
                case '*':
                        if (*++p == '*') {
+                               const uchar *prev_p = p - 2;
                                while (*++p == '*') {}
-                               special = TRUE;
+                               if (!(flags & WM_PATHNAME))
+                                       /* without WM_PATHNAME, '*' == '**' */
+                                       match_slash = 1;
+                               else if ((prev_p < pattern || *prev_p == '/') &&
+                                   (*p == '\0' || *p == '/' ||
+                                    (p[0] == '\\' && p[1] == '/'))) {
+                                       /*
+                                        * Assuming we already match 'foo/' and are at
+                                        * <star star slash>, just assume it matches
+                                        * nothing and go ahead match the rest of the
+                                        * pattern with the remaining string. This
+                                        * helps make foo/<*><*>/bar (<> because
+                                        * otherwise it breaks C comment syntax) match
+                                        * both foo/bar and foo/a/bar.
+                                        */
+                                       if (p[0] == '/' &&
+                                           dowild(p + 1, text, flags) == WM_MATCH)
+                                               return WM_MATCH;
+                                       match_slash = 1;
+                               } else
+                                       return WM_ABORT_MALFORMED;
                        } else
-                               special = FALSE;
+                               /* without WM_PATHNAME, '*' == '**' */
+                               match_slash = flags & WM_PATHNAME ? 0 : 1;
                        if (*p == '\0') {
                                /* Trailing "**" matches everything.  Trailing "*" matches
                                 * only if there are no more slash characters. */
-                               if (!special) {
+                               if (!match_slash) {
                                        if (strchr((char*)text, '/') != NULL)
-                                               return NOMATCH;
+                                               return WM_NOMATCH;
                                }
-                               return MATCH;
+                               return WM_MATCH;
+                       } else if (!match_slash && *p == '/') {
+                               /*
+                                * _one_ asterisk followed by a slash
+                                * with WM_PATHNAME matches the next
+                                * directory
+                                */
+                               const char *slash = strchr((char*)text, '/');
+                               if (!slash)
+                                       return WM_NOMATCH;
+                               text = (const uchar*)slash;
+                               /* the slash is consumed by the top-level for loop */
+                               break;
                        }
                        while (1) {
                                if (t_ch == '\0')
                                        break;
-                               if ((matched = dowild(p, text)) != NOMATCH) {
-                                       if (!special || matched != ABORT_TO_STARSTAR)
+                               /*
+                                * Try to advance faster when an asterisk is
+                                * followed by a literal. We know in this case
+                                * that the the string before the literal
+                                * must belong to "*".
+                                * If match_slash is false, do not look past
+                                * the first slash as it cannot belong to '*'.
+                                */
+                               if (!is_glob_special(*p)) {
+                                       p_ch = *p;
+                                       if ((flags & WM_CASEFOLD) && ISUPPER(p_ch))
+                                               p_ch = tolower(p_ch);
+                                       while ((t_ch = *text) != '\0' &&
+                                              (match_slash || t_ch != '/')) {
+                                               if ((flags & WM_CASEFOLD) && ISUPPER(t_ch))
+                                                       t_ch = tolower(t_ch);
+                                               if (t_ch == p_ch)
+                                                       break;
+                                               text++;
+                                       }
+                                       if (t_ch != p_ch)
+                                               return WM_NOMATCH;
+                               }
+                               if ((matched = dowild(p, text, flags)) != WM_NOMATCH) {
+                                       if (!match_slash || matched != WM_ABORT_TO_STARSTAR)
                                                return matched;
-                               } else if (!special && t_ch == '/')
-                                       return ABORT_TO_STARSTAR;
+                               } else if (!match_slash && t_ch == '/')
+                                       return WM_ABORT_TO_STARSTAR;
                                t_ch = *++text;
                        }
-                       return ABORT_ALL;
+                       return WM_ABORT_ALL;
                case '[':
                        p_ch = *++p;
 #ifdef NEGATE_CLASS2
                        if (p_ch == NEGATE_CLASS2)
                                p_ch = NEGATE_CLASS;
 #endif
-                       /* Assign literal TRUE/FALSE because of "matched" comparison. */
-                       special = p_ch == NEGATE_CLASS? TRUE : FALSE;
-                       if (special) {
+                       /* Assign literal 1/0 because of "matched" comparison. */
+                       negated = p_ch == NEGATE_CLASS ? 1 : 0;
+                       if (negated) {
                                /* Inverted character class. */
                                p_ch = *++p;
                        }
                        prev_ch = 0;
-                       matched = FALSE;
+                       matched = 0;
                        do {
                                if (!p_ch)
-                                       return ABORT_ALL;
+                                       return WM_ABORT_ALL;
                                if (p_ch == '\\') {
                                        p_ch = *++p;
                                        if (!p_ch)
-                                               return ABORT_ALL;
+                                               return WM_ABORT_ALL;
                                        if (t_ch == p_ch)
-                                               matched = TRUE;
+                                               matched = 1;
                                } else if (p_ch == '-' && prev_ch && p[1] && p[1] != ']') {
                                        p_ch = *++p;
                                        if (p_ch == '\\') {
                                                p_ch = *++p;
                                                if (!p_ch)
-                                                       return ABORT_ALL;
+                                                       return WM_ABORT_ALL;
                                        }
                                        if (t_ch <= p_ch && t_ch >= prev_ch)
-                                               matched = TRUE;
+                                               matched = 1;
+                                       else if ((flags & WM_CASEFOLD) && ISLOWER(t_ch)) {
+                                               uchar t_ch_upper = toupper(t_ch);
+                                               if (t_ch_upper <= p_ch && t_ch_upper >= prev_ch)
+                                                       matched = 1;
+                                       }
                                        p_ch = 0; /* This makes "prev_ch" get set to 0. */
                                } else if (p_ch == '[' && p[1] == ':') {
                                        const uchar *s;
                                        int i;
                                        for (s = p += 2; (p_ch = *p) && p_ch != ']'; p++) {} /*SHARED ITERATOR*/
                                        if (!p_ch)
-                                               return ABORT_ALL;
+                                               return WM_ABORT_ALL;
                                        i = p - s - 1;
                                        if (i < 0 || p[-1] != ':') {
                                                /* Didn't find ":]", so treat like a normal set. */
                                                p = s - 2;
                                                p_ch = '[';
                                                if (t_ch == p_ch)
-                                                       matched = TRUE;
+                                                       matched = 1;
                                                continue;
                                        }
                                        if (CC_EQ(s,i, "alnum")) {
                                                if (ISALNUM(t_ch))
-                                                       matched = TRUE;
+                                                       matched = 1;
                                        } else if (CC_EQ(s,i, "alpha")) {
                                                if (ISALPHA(t_ch))
-                                                       matched = TRUE;
+                                                       matched = 1;
                                        } else if (CC_EQ(s,i, "blank")) {
                                                if (ISBLANK(t_ch))
-                                                       matched = TRUE;
+                                                       matched = 1;
                                        } else if (CC_EQ(s,i, "cntrl")) {
                                                if (ISCNTRL(t_ch))
-                                                       matched = TRUE;
+                                                       matched = 1;
                                        } else if (CC_EQ(s,i, "digit")) {
                                                if (ISDIGIT(t_ch))
-                                                       matched = TRUE;
+                                                       matched = 1;
                                        } else if (CC_EQ(s,i, "graph")) {
                                                if (ISGRAPH(t_ch))
-                                                       matched = TRUE;
+                                                       matched = 1;
                                        } else if (CC_EQ(s,i, "lower")) {
                                                if (ISLOWER(t_ch))
-                                                       matched = TRUE;
+                                                       matched = 1;
                                        } else if (CC_EQ(s,i, "print")) {
                                                if (ISPRINT(t_ch))
-                                                       matched = TRUE;
+                                                       matched = 1;
                                        } else if (CC_EQ(s,i, "punct")) {
                                                if (ISPUNCT(t_ch))
-                                                       matched = TRUE;
+                                                       matched = 1;
                                        } else if (CC_EQ(s,i, "space")) {
                                                if (ISSPACE(t_ch))
-                                                       matched = TRUE;
+                                                       matched = 1;
                                        } else if (CC_EQ(s,i, "upper")) {
                                                if (ISUPPER(t_ch))
-                                                       matched = TRUE;
+                                                       matched = 1;
+                                               else if ((flags & WM_CASEFOLD) && ISLOWER(t_ch))
+                                                       matched = 1;
                                        } else if (CC_EQ(s,i, "xdigit")) {
                                                if (ISXDIGIT(t_ch))
-                                                       matched = TRUE;
+                                                       matched = 1;
                                        } else /* malformed [:class:] string */
-                                               return ABORT_ALL;
+                                               return WM_ABORT_ALL;
                                        p_ch = 0; /* This makes "prev_ch" get set to 0. */
                                } else if (t_ch == p_ch)
-                                       matched = TRUE;
+                                       matched = 1;
                        } while (prev_ch = p_ch, (p_ch = *++p) != ']');
-                       if (matched == special || t_ch == '/')
-                               return NOMATCH;
+                       if (matched == negated ||
+                           ((flags & WM_PATHNAME) && t_ch == '/'))
+                               return WM_NOMATCH;
                        continue;
                }
        }
 
-       return *text ? NOMATCH : MATCH;
+       return *text ? WM_NOMATCH : WM_MATCH;
 }
 
 /* Match the "pattern" against the "text" string. */
-int wildmatch(const char *pattern, const char *text)
-{
-       return dowild((const uchar*)pattern, (const uchar*)text);
-}
-
-/* Match the "pattern" against the forced-to-lower-case "text" string. */
-int iwildmatch(const char *pattern, const char *text)
+int wildmatch(const char *pattern, const char *text,
+             unsigned int flags, struct wildopts *wo)
 {
-       int ret;
-       force_lower_case = 1;
-       ret = dowild((const uchar*)pattern, (const uchar*)text);
-       force_lower_case = 0;
-       return ret;
+       return dowild((const uchar*)pattern, (const uchar*)text, flags);
 }