static gboolean ends_in_dotpc (const char *str) { int len = strlen (str); if (len > EXT_LEN && str[len - 3] == '.' && FOLD (str[len - 2]) == 'p' && FOLD (str[len - 1]) == 'c') return TRUE; else return FALSE; }
/* * Return LESS, GREATER, or EQUAL depending on how the string1 compares with * string2 (s1 ??? s2). * * o Matches up to len(s1) are EQUAL. * o Matches up to len(s2) are GREATER. * * Compare understands about the -f and -d flags, and treats comparisons * appropriately. * * The string "s1" is null terminated. The string s2 is '\n' terminated (or * "back" terminated). */ int compare(char *s1, char *s2, char *back) { int ch; for (; *s1 && s2 < back && *s2 != '\n'; ++s1, ++s2) { ch = *s2; if (fflag) ch = FOLD((unsigned char)ch); if (dflag) ch = DICT((unsigned char)ch); if (ch == NO_COMPARE) { ++s2; /* Ignore character in comparison. */ continue; } if (*s1 != ch) return (*s1 < ch ? LESS : GREATER); } return (*s1 ? GREATER : EQUAL); }
/* * Return LESS, GREATER, or EQUAL depending on how the string1 compares with * string2 (s1 ??? s2). * * o Matches up to len(s1) are EQUAL. * o Matches up to len(s2) are GREATER. * * Compare understands about the -f and -d flags, and treats comparisons * appropriately. * * The string "s1" is null terminated. The string s2 is '\n' terminated (or * "back" terminated). */ static int compare(char *s1, char *s2, uim_look_ctx *ctx) { int ch; unsigned char *back = (unsigned char *)ctx->back; int fflag = ctx->fflag, dflag = ctx->dflag; for (; (unsigned char)*s1 && (unsigned char *)s2 < back && *s2 != '\n'; ++s1, ++s2) { ch = (unsigned char)*s2; if (fflag) ch = FOLD(ch); if (dflag) ch = DICT(ch); if (ch == NO_COMPARE) { ++s2; /* Ignore character in comparison. */ continue; } if ((unsigned char)*s1 != ch) return ((unsigned char)*s1 < ch ? LESS : GREATER); } return ((unsigned char)*s1 ? GREATER : EQUAL); }
int uim_look(char *string, uim_look_ctx *ctx) { int ch; unsigned char *readp, *writep; int fflag = ctx->fflag, dflag = ctx->dflag; /* Reformat string to avoid doing it multiple times later. */ for (readp = writep = (unsigned char *)string; (ch = *readp++) != '\0';) { if (fflag) ch = FOLD(ch); if (dflag) ch = DICT(ch); if (ch != NO_COMPARE) *(writep++) = (unsigned char)ch; } *writep = '\0'; ctx->front = binary_search(string, ctx); ctx->front = linear_search(string, ctx); return (ctx->front ? 1 : 0); }
int look(unsigned char *string, unsigned char *front, unsigned char *back) { int ch; unsigned char *readp, *writep; /* Reformat string string to avoid doing it multiple times later. */ for (readp = writep = string; (ch = *readp++) != '\0';) { if (fflag) ch = FOLD(ch); if (dflag) ch = DICT(ch); if (ch != NO_COMPARE) *(writep++) = ch; } *writep = '\0'; front = binary_search(string, front, back); front = linear_search(string, front, back); if (front) print_from(string, front, back); return (front ? 0 : 1); }
/* Match STRING against the filename pattern PATTERN, returning zero if it matches, nonzero if not. */ int fnmatch(const char *pattern, const char *string, int flags) { register const char *p = pattern, *n = string; register char c; /* Note that this evaluates C many times. */ # define FOLD(c) ((flags & FNM_CASEFOLD) && ISUPPER (c) ? tolower (c) : (c)) while ((c = *p++) != '\0') { c = FOLD(c); switch (c) { case '?': if (*n == '\0') return FNM_NOMATCH; else if ((flags & FNM_FILE_NAME) && *n == '/') return FNM_NOMATCH; else if ((flags & FNM_PERIOD) && *n == '.' && (n == string || ((flags & FNM_FILE_NAME) && n[-1] == '/'))) return FNM_NOMATCH; break; case '\\': if (!(flags & FNM_NOESCAPE)) { c = *p++; if (c == '\0') /* Trailing \ loses. */ return FNM_NOMATCH; c = FOLD(c); } if (FOLD(*n) != c) return FNM_NOMATCH; break; case '*': if ((flags & FNM_PERIOD) && *n == '.' && (n == string || ((flags & FNM_FILE_NAME) && n[-1] == '/'))) return FNM_NOMATCH; for (c = *p++; c == '?' || c == '*'; c = *p++) { if ((flags & FNM_FILE_NAME) && *n == '/') /* A slash does not match a wildcard under FNM_FILE_NAME. */ return FNM_NOMATCH; else if (c == '?') { /* A ? needs to match one character. */ if (*n == '\0') /* There isn't another character; no match. */ return FNM_NOMATCH; else /* One character of the string is consumed in matching this ? wildcard, so *??? won't match if there are less than three characters. */ ++n; } } if (c == '\0') return 0; { char c1 = (!(flags & FNM_NOESCAPE) && c == '\\') ? *p : c; c1 = FOLD(c1); for (--p; *n != '\0'; ++n) if ((c == '[' || FOLD(*n) == c1) && fnmatch(p, n, flags & ~FNM_PERIOD) == 0) return 0; return FNM_NOMATCH; } case '[': { /* Nonzero if the sense of the character class is inverted. */ register int not; if (*n == '\0') return FNM_NOMATCH; if ((flags & FNM_PERIOD) && *n == '.' && (n == string || ((flags & FNM_FILE_NAME) && n[-1] == '/'))) return FNM_NOMATCH; not = (*p == '!' || *p == '^'); if (not) ++p; c = *p++; for (;;) { register char cstart = c, cend = c; if (!(flags & FNM_NOESCAPE) && c == '\\') { if (*p == '\0') return FNM_NOMATCH; cstart = cend = *p++; } cstart = cend = FOLD(cstart); if (c == '\0') /* [ (unterminated) loses. */ return FNM_NOMATCH; c = *p++; c = FOLD(c); if ((flags & FNM_FILE_NAME) && c == '/') /* [/] can never match. */ return FNM_NOMATCH; if (c == '-' && *p != ']') { cend = *p++; if (!(flags & FNM_NOESCAPE) && cend == '\\') cend = *p++; if (cend == '\0') return FNM_NOMATCH; cend = FOLD(cend); c = *p++; } if (FOLD(*n) >= cstart && FOLD(*n) <= cend) goto matched; if (c == ']') break; } if (!not) return FNM_NOMATCH; break; matched:; /* Skip the rest of the [...] that already matched. */ while (c != ']') { if (c == '\0') /* [... (unterminated) loses. */ return FNM_NOMATCH; c = *p++; if (!(flags & FNM_NOESCAPE) && c == '\\') { if (*p == '\0') return FNM_NOMATCH; /* XXX 1003.2d11 is unclear if this is right. */ ++p; } } if (not) return FNM_NOMATCH; } break; default: if (c != FOLD(*n)) return FNM_NOMATCH; } ++n; } if (*n == '\0') return 0; if ((flags & FNM_LEADING_DIR) && *n == '/') /* The FNM_LEADING_DIR flag says that "foo*" matches "foobar/frobozz". */ return 0; return FNM_NOMATCH; # undef FOLD }
int CFileMatcher::FileNameMatch(LPCTSTR pattern, LPCTSTR string, BOOL fMatchedDot) { // In V7, changed all chars to unsigned chars // // In V8 we added support for "null" extensions by specifying a mask that ends with a "." (added fMatchedDot parameter) // That is, "*." specifies all files with no extension register const unsigned short *p = (unsigned short *)pattern; register const unsigned short *n = (unsigned short *)string; register unsigned short c; BOOL fNot = (*p == '!') ? TRUE : FALSE; if (fNot) p++; while ((c = *p++) != '\0') { c = FOLD(c); switch (c) { case '?': if (*n == '\0') { // Unlike the original version, '?' can match 0 occurences. return (fNot) ? FNM_NOMATCH : FNM_MATCH; } break; case '\\': //if (!(flags & FNM_NOESCAPE)) { c = *p++; c = FOLD(c); } if (FOLD(*n) != c) return (fNot) ? FNM_MATCH : FNM_NOMATCH; break; case '*': // Added in V2002 - Check for *.* // If we are at a terminating *.*, ignore the final ".*" by advancing the // mask pointer two positions. // This has the effect of matching "abc" if we search for "a*.*", whereas before, it didn't. if ((p[0] == '.') && (p[1] == '*') && (p[2] == 0)) p += 2; // At this point, p points to the character in the mask *after* the "*" for (c = *p++; ((c == '?') || (c == '*')); c = *p++, ++n) { // This seems to advance the pattern pointer past any following * or ? (eg, handles patterns like ** or *?)! // if ((c == '?') && (*n == '\0')) return (fNot) ? FNM_MATCH : FNM_NOMATCH; } if (c == '\0') return (fNot) ? FNM_NOMATCH : FNM_MATCH; // c is now the character after the '*', p points to the character after that // n points to the remainder of the string that is to be matched (in the case of "*.", n will point to the *entire* file name) { unsigned short c1 = (c == '\\') ? *p : c; BOOL fNoExtension = ((c == '.') && (*p == 0)) ? TRUE : FALSE; c1 = FOLD(c1); for (--p; *n != '\0'; ++n) { // If the character after the "*" is a "[" then we need to recursively call // FileNameMatch on the remainder of the pattern/string. // Similarly if the character after the "*" matches the current character in the string, // we recursively try to match the rest of the string // // Otherwise, we will keep going down the string until we find a match unsigned short s = FOLD(*n); if (fNoExtension) { // Our pattern ends in "*." - we are only interested to see if the string has an extension or not // We just keep going along the string until we either find a '.' or it terminates ASSERT(c1 == c); ASSERT(c1 == '.'); if (s == c1) { // We have found a dot - therefore, the file has an extension return (fNot) ? FNM_MATCH : FNM_NOMATCH; } } else if (((c == '[') || (s == c1)) && FileNameMatch((LPCTSTR)p, (LPCTSTR)n, fMatchedDot) == FNM_MATCH) return (fNot) ? FNM_NOMATCH : FNM_MATCH; // Set a flag to indicate that * has matched a '.' - we use this later when matching a trailing "." in a pattern if (s == '.') fMatchedDot = TRUE; } // We come here if what comes after the "*" does not match anything in the string (eg, searching for "*a" in "xyz") if (fNoExtension) { // If we are searching for "*." we come here if we didn't find an extension in the string - therefore we have a match! return (fNot) ? FNM_NOMATCH : FNM_MATCH; } return (fNot) ? FNM_MATCH : FNM_NOMATCH; } case '[': { // Nonzero if the sense of the character class is inverted. register int not; if (*n == '\0') return (fNot) ? FNM_MATCH : FNM_NOMATCH; not = (*p == '!' || *p == '^'); if (not) ++p; c = *p++; for (;;) { register unsigned short cstart = c, cend = c; if (c == '\\') cstart = cend = *p++; cstart = cend = FOLD(cstart); if (c == '\0') { // Unterminated return (fNot) ? FNM_MATCH : FNM_NOMATCH; } c = *p++; c = FOLD(c); if (c == '-' && *p != ']') { cend = *p++; if (cend == '\\') cend = *p++; if (cend == '\0') return (fNot) ? FNM_MATCH : FNM_NOMATCH; cend = FOLD(cend); c = *p++; } if (FOLD(*n) >= cstart && FOLD(*n) <= cend) goto matched; if (c == ']') break; } if (!not) return (fNot) ? FNM_MATCH : FNM_NOMATCH; break; matched:; // Skip the rest of the [...] that already matched. while (c != ']') { if (c == '\0') { // Unterminated return (fNot) ? FNM_MATCH : FNM_NOMATCH; } c = *p++; if (c == '\\') ++p; // XXX 1003.2d11 is unclear if this is right. } if (not) return (fNot) ? FNM_MATCH : FNM_NOMATCH; } break; case '.': { // '.' case added in V8 to handle a "null' extension if (*p == 0) { // Our pattern ends with a '.' - we will treat this as specifying a file with "no" extension // Note that in Windows the names "ABC" and "ABC." are the same. // // We check fMatchedDot to see if we have previously matched a dot when using "*" // For example, if we filter on *a., we will get a match for "b.a" because the "*" will match "b." // However, we do not want this to match "b.a"! if ((*n == '\0') && !fMatchedDot) return (fNot) ? FNM_NOMATCH : FNM_MATCH; return (fNot) ? FNM_MATCH : FNM_NOMATCH; } // If the pattern doesn't *end* in a '.' we need to fall through to the default processing - ie, don't put a "break" here! } default: if (c != FOLD(*n)) return (fNot) ? FNM_MATCH : FNM_NOMATCH; } ++n; } if (*n == '\0') return (fNot) ? FNM_NOMATCH : FNM_MATCH; return (fNot) ? FNM_MATCH : FNM_NOMATCH; }
// Perform match static bool domatch(const char *pattern,const char *string,FXuint flags){ register const char *p=pattern; register const char *q=string; register const char *s; register char c,cs,ce,cc,neg; register int level; while((c=*p++)!='\0'){ switch(c){ case '?': if(*q=='\0') return false; if((flags&FILEMATCH_FILE_NAME) && ISPATHSEP(*q)) return false; if((flags&FILEMATCH_PERIOD) && (*q=='.') && ((q==string) || ((flags&FILEMATCH_FILE_NAME) && ISPATHSEP(*(q-1))))) return false; q++; break; case '*': c=*p; while(c=='*') c=*++p; if((flags&FILEMATCH_PERIOD) && (*q=='.') && ((q==string) || ((flags&FILEMATCH_FILE_NAME) && ISPATHSEP(*(q-1))))) return false; if(c=='\0'){ // Optimize for case of trailing '*' if(flags&FILEMATCH_FILE_NAME){ for(s=q; *s; s++){ if(ISPATHSEP(*s)) return false; } } return 1; } while(!domatch(p,q,flags&~FILEMATCH_PERIOD)){ if((flags&FILEMATCH_FILE_NAME) && ISPATHSEP(*q)) return false; if(*q++=='\0') return false; } return 1; case '[': if(*q=='\0') return false; if((flags&FILEMATCH_PERIOD) && (*q=='.') && ((q==string) || ((flags&FILEMATCH_FILE_NAME) && ISPATHSEP(*(q-1))))) return false; cc=FOLD(*q); neg=((*p=='!') || (*p=='^')); if(neg) p++; c=*p++; do{ if(c=='\\' && !(flags&FILEMATCH_NOESCAPE)) c=*p++; cs=ce=FOLD(c); if(c=='\0') return false; c=*p++; c=FOLD(c); if((flags&FILEMATCH_FILE_NAME) && ISPATHSEP(c)) return false; if(c=='-' && *p!=']'){ c = *p++; if(c=='\\' && !(flags&FILEMATCH_NOESCAPE)) c=*p++; if(c=='\0') return false; ce=FOLD(c); c=*p++; } if(((FXuchar)cs)<=((FXuchar)cc) && ((FXuchar)cc)<=((FXuchar)ce)) goto match; } while(c!=']'); if(!neg) return false; q++; break; match: while(c!=']'){ if(c=='\0') return false; c=*p++; if(c=='\\' && !(flags&FILEMATCH_NOESCAPE)) p++; } if(neg) return false; q++; break; case '(': nxt: if(domatch(p,q,flags)) return true; for(level=0; *p && 0<=level; ){ switch(*p++){ case '\\': if(*p) p++; break; case '(': level++; break; case ')': level--; break; case '|': case ',': if (level==0) goto nxt; } } return false; case ')': break; case '|': case ',': for(level=0; *p && 0<=level; ){ switch(*p++){ case '\\': if(*p) p++; break; case '(': level++; break; case ')': level--; break; } } break; case '\\': if(*p && !(flags&FILEMATCH_NOESCAPE)) c=*p++; // Trailing escape represents itself default: if(FOLD(c)!=FOLD(*q)) return false; q++; break; } } return (*q=='\0') || (ISPATHSEP(*q) && (flags&FILEMATCH_LEADING_DIR)); }
/* Match STRING against the filename pattern PATTERN, returning zero if it matches, nonzero if not. */ int fnmatch (const char *pattern,const char *string,int flags ) { register const char *p = pattern, *n = string; register unsigned char c; /* Note that this evalutes C many times. */ #define FOLD(c) ((flags & FNM_CASEFOLD) && isupper (c) ? tolower (c) : (c)) #ifdef _WIN32_ORTE flags |= FNM_CASEFOLD; #endif while ((c = *p++) != '\0') { c = FOLD (c); switch (c) { case '?': if (*n == '\0') return FNM_NOMATCH; else if ((flags & FNM_FILE_NAME) && *n == '/') return FNM_NOMATCH; else if ((flags & FNM_PERIOD) && *n == '.' && (n == string || ((flags & FNM_FILE_NAME) && n[-1] == '/'))) return FNM_NOMATCH; break; case '\\': if (!(flags & FNM_NOESCAPE)) { c = *p++; c = FOLD (c); } if (FOLD ((unsigned char)*n) != c) return FNM_NOMATCH; break; case '*': if ((flags & FNM_PERIOD) && *n == '.' && (n == string || ((flags & FNM_FILE_NAME) && n[-1] == '/'))) return FNM_NOMATCH; for (c = *p++; c == '?' || c == '*'; c = *p++, ++n) if (((flags & FNM_FILE_NAME) && *n == '/') || (c == '?' && *n == '\0')) return FNM_NOMATCH; if (c == '\0') return 0; { unsigned char c1 = (!(flags & FNM_NOESCAPE) && c == '\\') ? *p : c; c1 = FOLD (c1); for (--p; *n != '\0'; ++n) if ((c == '[' || FOLD ((unsigned char)*n) == c1) && fnmatch (p, n, flags & ~FNM_PERIOD) == 0) return 0; return FNM_NOMATCH; } case '[': { /* Nonzero if the sense of the character class is inverted. */ register int not; if (*n == '\0') return FNM_NOMATCH; if ((flags & FNM_PERIOD) && *n == '.' && (n == string || ((flags & FNM_FILE_NAME) && n[-1] == '/'))) return FNM_NOMATCH; not = (*p == '!' || *p == '^'); if (not) ++p; c = *p++; for (;;) { register unsigned char cstart = c, cend = c; if (!(flags & FNM_NOESCAPE) && c == '\\') cstart = cend = *p++; cstart = cend = FOLD (cstart); if (c == '\0') /* [ (unterminated) loses. */ return FNM_NOMATCH; c = *p++; c = FOLD (c); if ((flags & FNM_FILE_NAME) && c == '/') /* [/] can never match. */ return FNM_NOMATCH; if (c == '-' && *p != ']') { cend = *p++; if (!(flags & FNM_NOESCAPE) && cend == '\\') cend = *p++; if (cend == '\0') return FNM_NOMATCH; cend = FOLD (cend); c = *p++; } if (FOLD ((unsigned char)*n) >= cstart && FOLD ((unsigned char)*n) <= cend) goto matched; if (c == ']') break; } if (!not) return FNM_NOMATCH; break; matched:; /* Skip the rest of the [...] that already matched. */ while (c != ']') { if (c == '\0') /* [... (unterminated) loses. */ return FNM_NOMATCH; c = *p++; if (!(flags & FNM_NOESCAPE) && c == '\\') /* XXX 1003.2d11 is unclear if this is right. */ ++p; } if (not) return FNM_NOMATCH; } break; default: if (c != FOLD ((unsigned char)*n)) return FNM_NOMATCH; } // switch (c) ++n; } if (*n == '\0') return 0; if ((flags & FNM_LEADING_DIR) && *n == '/') /* The FNM_LEADING_DIR flag says that "foo*" matches "foobar/frobozz". */ return 0; return FNM_NOMATCH; }
static int internal_function FCT (const CHAR *pattern, const CHAR *string, const CHAR *string_end, bool no_leading_period, int flags) { register const CHAR *p = pattern, *n = string; register UCHAR c; #ifdef _LIBC # if WIDE_CHAR_VERSION const char *collseq = (const char *) _NL_CURRENT(LC_COLLATE, _NL_COLLATE_COLLSEQWC); # else const UCHAR *collseq = (const UCHAR *) _NL_CURRENT(LC_COLLATE, _NL_COLLATE_COLLSEQMB); # endif #endif while ((c = *p++) != L_('\0')) { bool new_no_leading_period = false; c = FOLD (c); switch (c) { case L_('?'): if (__builtin_expect (flags & FNM_EXTMATCH, 0) && *p == '(') { int res; res = EXT (c, p, n, string_end, no_leading_period, flags); if (res != -1) return res; } if (n == string_end) return FNM_NOMATCH; else if (*n == L_('/') && (flags & FNM_FILE_NAME)) return FNM_NOMATCH; else if (*n == L_('.') && no_leading_period) return FNM_NOMATCH; break; case L_('\\'): if (!(flags & FNM_NOESCAPE)) { c = *p++; if (c == L_('\0')) /* Trailing \ loses. */ return FNM_NOMATCH; c = FOLD (c); } if (n == string_end || FOLD ((UCHAR) *n) != c) return FNM_NOMATCH; break; case L_('*'): if (__builtin_expect (flags & FNM_EXTMATCH, 0) && *p == '(') { int res; res = EXT (c, p, n, string_end, no_leading_period, flags); if (res != -1) return res; } if (n != string_end && *n == L_('.') && no_leading_period) return FNM_NOMATCH; for (c = *p++; c == L_('?') || c == L_('*'); c = *p++) { if (*p == L_('(') && (flags & FNM_EXTMATCH) != 0) { const CHAR *endp = END (p); if (endp != p) { /* This is a pattern. Skip over it. */ p = endp; continue; } } if (c == L_('?')) { /* A ? needs to match one character. */ if (n == string_end) /* There isn't another character; no match. */ return FNM_NOMATCH; else if (*n == L_('/') && __builtin_expect (flags & FNM_FILE_NAME, 0)) /* A slash does not match a wildcard under FNM_FILE_NAME. */ return FNM_NOMATCH; else /* One character of the string is consumed in matching this ? wildcard, so *??? won't match if there are less than three characters. */ ++n; } } if (c == L_('\0')) /* The wildcard(s) is/are the last element of the pattern. If the name is a file name and contains another slash this means it cannot match, unless the FNM_LEADING_DIR flag is set. */ { int result = (flags & FNM_FILE_NAME) == 0 ? 0 : FNM_NOMATCH; if (flags & FNM_FILE_NAME) { if (flags & FNM_LEADING_DIR) result = 0; else { if (MEMCHR (n, L_('/'), string_end - n) == NULL) result = 0; } } return result; } else { const CHAR *endp; endp = MEMCHR (n, (flags & FNM_FILE_NAME) ? L_('/') : L_('\0'), string_end - n); if (endp == NULL) endp = string_end; if (c == L_('[') || (__builtin_expect (flags & FNM_EXTMATCH, 0) != 0 && (c == L_('@') || c == L_('+') || c == L_('!')) && *p == L_('('))) { int flags2 = ((flags & FNM_FILE_NAME) ? flags : (flags & ~FNM_PERIOD)); bool no_leading_period2 = no_leading_period; for (--p; n < endp; ++n, no_leading_period2 = false) if (FCT (p, n, string_end, no_leading_period2, flags2) == 0) return 0; } else if (c == L_('/') && (flags & FNM_FILE_NAME)) { while (n < string_end && *n != L_('/')) ++n; if (n < string_end && *n == L_('/') && (FCT (p, n + 1, string_end, flags & FNM_PERIOD, flags) == 0)) return 0; } else { int flags2 = ((flags & FNM_FILE_NAME) ? flags : (flags & ~FNM_PERIOD)); int no_leading_period2 = no_leading_period; if (c == L_('\\') && !(flags & FNM_NOESCAPE)) c = *p; c = FOLD (c); for (--p; n < endp; ++n, no_leading_period2 = false) if (FOLD ((UCHAR) *n) == c && (FCT (p, n, string_end, no_leading_period2, flags2) == 0)) return 0; } } /* If we come here no match is possible with the wildcard. */ return FNM_NOMATCH; case L_('['): { /* Nonzero if the sense of the character class is inverted. */ register bool not; CHAR cold; UCHAR fn; if (posixly_correct == 0) posixly_correct = getenv ("POSIXLY_CORRECT") != NULL ? 1 : -1; if (n == string_end) return FNM_NOMATCH; if (*n == L_('.') && no_leading_period) return FNM_NOMATCH; if (*n == L_('/') && (flags & FNM_FILE_NAME)) /* `/' cannot be matched. */ return FNM_NOMATCH; not = (*p == L_('!') || (posixly_correct < 0 && *p == L_('^'))); if (not) ++p; fn = FOLD ((UCHAR) *n); c = *p++; for (;;) { if (!(flags & FNM_NOESCAPE) && c == L_('\\')) { if (*p == L_('\0')) return FNM_NOMATCH; c = FOLD ((UCHAR) *p); ++p; if (c == fn) goto matched; } else if (c == L_('[') && *p == L_(':')) { /* Leave room for the null. */ CHAR str[CHAR_CLASS_MAX_LENGTH + 1]; size_t c1 = 0; #if defined _LIBC || WIDE_CHAR_SUPPORT wctype_t wt; #endif const CHAR *startp = p; for (;;) { if (c1 == CHAR_CLASS_MAX_LENGTH) /* The name is too long and therefore the pattern is ill-formed. */ return FNM_NOMATCH; c = *++p; if (c == L_(':') && p[1] == L_(']')) { p += 2; break; } if (c < L_('a') || c >= L_('z')) { /* This cannot possibly be a character class name. Match it as a normal range. */ p = startp; c = L_('['); goto normal_bracket; } str[c1++] = c; } str[c1] = L_('\0'); #if defined _LIBC || WIDE_CHAR_SUPPORT wt = IS_CHAR_CLASS (str); if (wt == 0) /* Invalid character class name. */ return FNM_NOMATCH; # if defined _LIBC && ! WIDE_CHAR_VERSION /* The following code is glibc specific but does there a good job in speeding up the code since we can avoid the btowc() call. */ if (_ISCTYPE ((UCHAR) *n, wt)) goto matched; # else if (ISWCTYPE (BTOWC ((UCHAR) *n), wt)) goto matched; # endif #else if ((STREQ (str, L_("alnum")) && ISALNUM ((UCHAR) *n)) || (STREQ (str, L_("alpha")) && ISALPHA ((UCHAR) *n)) || (STREQ (str, L_("blank")) && ISBLANK ((UCHAR) *n)) || (STREQ (str, L_("cntrl")) && ISCNTRL ((UCHAR) *n)) || (STREQ (str, L_("digit")) && ISDIGIT ((UCHAR) *n)) || (STREQ (str, L_("graph")) && ISGRAPH ((UCHAR) *n)) || (STREQ (str, L_("lower")) && ISLOWER ((UCHAR) *n)) || (STREQ (str, L_("print")) && ISPRINT ((UCHAR) *n)) || (STREQ (str, L_("punct")) && ISPUNCT ((UCHAR) *n)) || (STREQ (str, L_("space")) && ISSPACE ((UCHAR) *n)) || (STREQ (str, L_("upper")) && ISUPPER ((UCHAR) *n)) || (STREQ (str, L_("xdigit")) && ISXDIGIT ((UCHAR) *n))) goto matched; #endif c = *p++; } #ifdef _LIBC else if (c == L_('[') && *p == L_('=')) { UCHAR str[1]; uint32_t nrules = _NL_CURRENT_WORD (LC_COLLATE, _NL_COLLATE_NRULES); const CHAR *startp = p; c = *++p; if (c == L_('\0')) { p = startp; c = L_('['); goto normal_bracket; } str[0] = c; c = *++p; if (c != L_('=') || p[1] != L_(']')) { p = startp; c = L_('['); goto normal_bracket; } p += 2; if (nrules == 0) { if ((UCHAR) *n == str[0]) goto matched; } else { const int32_t *table; # if WIDE_CHAR_VERSION const int32_t *weights; const int32_t *extra; # else const unsigned char *weights; const unsigned char *extra; # endif const int32_t *indirect; int32_t idx; const UCHAR *cp = (const UCHAR *) str; /* This #include defines a local function! */ # if WIDE_CHAR_VERSION # include <locale/weightwc.h> # else # include <locale/weight.h> # endif # if WIDE_CHAR_VERSION table = (const int32_t *) _NL_CURRENT (LC_COLLATE, _NL_COLLATE_TABLEWC); weights = (const int32_t *) _NL_CURRENT (LC_COLLATE, _NL_COLLATE_WEIGHTWC); extra = (const int32_t *) _NL_CURRENT (LC_COLLATE, _NL_COLLATE_EXTRAWC); indirect = (const int32_t *) _NL_CURRENT (LC_COLLATE, _NL_COLLATE_INDIRECTWC); # else table = (const int32_t *) _NL_CURRENT (LC_COLLATE, _NL_COLLATE_TABLEMB); weights = (const unsigned char *) _NL_CURRENT (LC_COLLATE, _NL_COLLATE_WEIGHTMB); extra = (const unsigned char *) _NL_CURRENT (LC_COLLATE, _NL_COLLATE_EXTRAMB); indirect = (const int32_t *) _NL_CURRENT (LC_COLLATE, _NL_COLLATE_INDIRECTMB); # endif idx = findidx (&cp); if (idx != 0) { /* We found a table entry. Now see whether the character we are currently at has the same equivalance class value. */ int len = weights[idx]; int32_t idx2; const UCHAR *np = (const UCHAR *) n; idx2 = findidx (&np); if (idx2 != 0 && len == weights[idx2]) { int cnt = 0; while (cnt < len && (weights[idx + 1 + cnt] == weights[idx2 + 1 + cnt])) ++cnt; if (cnt == len) goto matched; } } } c = *p++; } #endif else if (c == L_('\0')) /* [ (unterminated) loses. */ return FNM_NOMATCH; else { bool is_range = false; #ifdef _LIBC bool is_seqval = false; if (c == L_('[') && *p == L_('.')) { uint32_t nrules = _NL_CURRENT_WORD (LC_COLLATE, _NL_COLLATE_NRULES); const CHAR *startp = p; size_t c1 = 0; while (1) { c = *++p; if (c == L_('.') && p[1] == L_(']')) { p += 2; break; } if (c == '\0') return FNM_NOMATCH; ++c1; } /* We have to handling the symbols differently in ranges since then the collation sequence is important. */ is_range = *p == L_('-') && p[1] != L_('\0'); if (nrules == 0) { /* There are no names defined in the collation data. Therefore we only accept the trivial names consisting of the character itself. */ if (c1 != 1) return FNM_NOMATCH; if (!is_range && *n == startp[1]) goto matched; cold = startp[1]; c = *p++; } else { int32_t table_size; const int32_t *symb_table; # ifdef WIDE_CHAR_VERSION char str[c1]; size_t strcnt; # else # define str (startp + 1) # endif const unsigned char *extra; int32_t idx; int32_t elem; int32_t second; int32_t hash; # ifdef WIDE_CHAR_VERSION /* We have to convert the name to a single-byte string. This is possible since the names consist of ASCII characters and the internal representation is UCS4. */ for (strcnt = 0; strcnt < c1; ++strcnt) str[strcnt] = startp[1 + strcnt]; # endif table_size = _NL_CURRENT_WORD (LC_COLLATE, _NL_COLLATE_SYMB_HASH_SIZEMB); symb_table = (const int32_t *) _NL_CURRENT (LC_COLLATE, _NL_COLLATE_SYMB_TABLEMB); extra = (const unsigned char *) _NL_CURRENT (LC_COLLATE, _NL_COLLATE_SYMB_EXTRAMB); /* Locate the character in the hashing table. */ hash = elem_hash (str, c1); idx = 0; elem = hash % table_size; second = hash % (table_size - 2); while (symb_table[2 * elem] != 0) { /* First compare the hashing value. */ if (symb_table[2 * elem] == hash && c1 == extra[symb_table[2 * elem + 1]] && memcmp (str, &extra[symb_table[2 * elem + 1] + 1], c1) == 0) { /* Yep, this is the entry. */ idx = symb_table[2 * elem + 1]; idx += 1 + extra[idx]; break; } /* Next entry. */ elem += second; } if (symb_table[2 * elem] != 0) { /* Compare the byte sequence but only if this is not part of a range. */ # ifdef WIDE_CHAR_VERSION int32_t *wextra; idx += 1 + extra[idx]; /* Adjust for the alignment. */ idx = (idx + 3) & ~3; wextra = (int32_t *) &extra[idx + 4]; # endif if (! is_range) { # ifdef WIDE_CHAR_VERSION for (c1 = 0; (int32_t) c1 < wextra[idx]; ++c1) if (n[c1] != wextra[1 + c1]) break; if ((int32_t) c1 == wextra[idx]) goto matched; # else for (c1 = 0; c1 < extra[idx]; ++c1) if (n[c1] != extra[1 + c1]) break; if (c1 == extra[idx]) goto matched; # endif } /* Get the collation sequence value. */ is_seqval = true; # ifdef WIDE_CHAR_VERSION cold = wextra[1 + wextra[idx]]; # else /* Adjust for the alignment. */ idx += 1 + extra[idx]; idx = (idx + 3) & ~4; cold = *((int32_t *) &extra[idx]); # endif c = *p++; } else if (c1 == 1) { /* No valid character. Match it as a single byte. */ if (!is_range && *n == str[0]) goto matched; cold = str[0]; c = *p++; } else return FNM_NOMATCH; } } else # undef str #endif { c = FOLD (c); normal_bracket: /* We have to handling the symbols differently in ranges since then the collation sequence is important. */ is_range = (*p == L_('-') && p[1] != L_('\0') && p[1] != L_(']')); if (!is_range && c == fn) goto matched; cold = c; c = *p++; } if (c == L_('-') && *p != L_(']')) { #if _LIBC /* We have to find the collation sequence value for C. Collation sequence is nothing we can regularly access. The sequence value is defined by the order in which the definitions of the collation values for the various characters appear in the source file. A strange concept, nowhere documented. */ uint32_t fcollseq; uint32_t lcollseq; UCHAR cend = *p++; # ifdef WIDE_CHAR_VERSION /* Search in the `names' array for the characters. */ fcollseq = __collseq_table_lookup (collseq, fn); if (fcollseq == ~((uint32_t) 0)) /* XXX We don't know anything about the character we are supposed to match. This means we are failing. */ goto range_not_matched; if (is_seqval) lcollseq = cold; else lcollseq = __collseq_table_lookup (collseq, cold); # else fcollseq = collseq[fn]; lcollseq = is_seqval ? cold : collseq[(UCHAR) cold]; # endif is_seqval = false; if (cend == L_('[') && *p == L_('.')) { uint32_t nrules = _NL_CURRENT_WORD (LC_COLLATE, _NL_COLLATE_NRULES); const CHAR *startp = p; size_t c1 = 0; while (1) { c = *++p; if (c == L_('.') && p[1] == L_(']')) { p += 2; break; } if (c == '\0') return FNM_NOMATCH; ++c1; } if (nrules == 0) { /* There are no names defined in the collation data. Therefore we only accept the trivial names consisting of the character itself. */ if (c1 != 1) return FNM_NOMATCH; cend = startp[1]; } else { int32_t table_size; const int32_t *symb_table; # ifdef WIDE_CHAR_VERSION char str[c1]; size_t strcnt; # else # define str (startp + 1) # endif const unsigned char *extra; int32_t idx; int32_t elem; int32_t second; int32_t hash; # ifdef WIDE_CHAR_VERSION /* We have to convert the name to a single-byte string. This is possible since the names consist of ASCII characters and the internal representation is UCS4. */ for (strcnt = 0; strcnt < c1; ++strcnt) str[strcnt] = startp[1 + strcnt]; # endif table_size = _NL_CURRENT_WORD (LC_COLLATE, _NL_COLLATE_SYMB_HASH_SIZEMB); symb_table = (const int32_t *) _NL_CURRENT (LC_COLLATE, _NL_COLLATE_SYMB_TABLEMB); extra = (const unsigned char *) _NL_CURRENT (LC_COLLATE, _NL_COLLATE_SYMB_EXTRAMB); /* Locate the character in the hashing table. */ hash = elem_hash (str, c1); idx = 0; elem = hash % table_size; second = hash % (table_size - 2); while (symb_table[2 * elem] != 0) { /* First compare the hashing value. */ if (symb_table[2 * elem] == hash && (c1 == extra[symb_table[2 * elem + 1]]) && memcmp (str, &extra[symb_table[2 * elem + 1] + 1], c1) == 0) { /* Yep, this is the entry. */ idx = symb_table[2 * elem + 1]; idx += 1 + extra[idx]; break; } /* Next entry. */ elem += second; } if (symb_table[2 * elem] != 0) { /* Compare the byte sequence but only if this is not part of a range. */ # ifdef WIDE_CHAR_VERSION int32_t *wextra; idx += 1 + extra[idx]; /* Adjust for the alignment. */ idx = (idx + 3) & ~4; wextra = (int32_t *) &extra[idx + 4]; # endif /* Get the collation sequence value. */ is_seqval = true; # ifdef WIDE_CHAR_VERSION cend = wextra[1 + wextra[idx]]; # else /* Adjust for the alignment. */ idx += 1 + extra[idx]; idx = (idx + 3) & ~4; cend = *((int32_t *) &extra[idx]); # endif } else if (symb_table[2 * elem] != 0 && c1 == 1) { cend = str[0]; c = *p++; } else return FNM_NOMATCH; } # undef str } else { if (!(flags & FNM_NOESCAPE) && cend == L_('\\')) cend = *p++; if (cend == L_('\0')) return FNM_NOMATCH; cend = FOLD (cend); } /* XXX It is not entirely clear to me how to handle characters which are not mentioned in the collation specification. */ if ( # ifdef WIDE_CHAR_VERSION lcollseq == 0xffffffff || # endif lcollseq <= fcollseq) { /* We have to look at the upper bound. */ uint32_t hcollseq; if (is_seqval) hcollseq = cend; else { # ifdef WIDE_CHAR_VERSION hcollseq = __collseq_table_lookup (collseq, cend); if (hcollseq == ~((uint32_t) 0)) { /* Hum, no information about the upper bound. The matching succeeds if the lower bound is matched exactly. */ if (lcollseq != fcollseq) goto range_not_matched; goto matched; } # else hcollseq = collseq[cend]; # endif } if (lcollseq <= hcollseq && fcollseq <= hcollseq) goto matched; } # ifdef WIDE_CHAR_VERSION range_not_matched: # endif #else /* We use a boring value comparison of the character values. This is better than comparing using `strcoll' since the latter would have surprising and sometimes fatal consequences. */ UCHAR cend = *p++; if (!(flags & FNM_NOESCAPE) && cend == L_('\\')) cend = *p++; if (cend == L_('\0')) return FNM_NOMATCH; /* It is a range. */ if (cold <= fn && fn <= cend) goto matched; #endif c = *p++; } } if (c == L_(']')) break; } if (!not) return FNM_NOMATCH; break; matched: /* Skip the rest of the [...] that already matched. */ do { ignore_next: c = *p++; if (c == L_('\0')) /* [... (unterminated) loses. */ return FNM_NOMATCH; if (!(flags & FNM_NOESCAPE) && c == L_('\\')) { if (*p == L_('\0')) return FNM_NOMATCH; /* XXX 1003.2d11 is unclear if this is right. */ ++p; } else if (c == L_('[') && *p == L_(':')) { int c1 = 0; const CHAR *startp = p; while (1) { c = *++p; if (++c1 == CHAR_CLASS_MAX_LENGTH) return FNM_NOMATCH; if (*p == L_(':') && p[1] == L_(']')) break; if (c < L_('a') || c >= L_('z')) { p = startp; goto ignore_next; } } p += 2; c = *p++; } else if (c == L_('[') && *p == L_('=')) { c = *++p; if (c == L_('\0')) return FNM_NOMATCH; c = *++p; if (c != L_('=') || p[1] != L_(']')) return FNM_NOMATCH; p += 2; c = *p++; } else if (c == L_('[') && *p == L_('.')) { ++p; while (1) { c = *++p; if (c == '\0') return FNM_NOMATCH; if (*p == L_('.') && p[1] == L_(']')) break; } p += 2; c = *p++; } } while (c != L_(']')); if (not) return FNM_NOMATCH; } break; case L_('+'): case L_('@'): case L_('!'): if (__builtin_expect (flags & FNM_EXTMATCH, 0) && *p == '(') { int res; res = EXT (c, p, n, string_end, no_leading_period, flags); if (res != -1) return res; } goto normal_match; case L_('/'): if (NO_LEADING_PERIOD (flags)) { if (n == string_end || c != (UCHAR) *n) return FNM_NOMATCH; new_no_leading_period = true; break; } /* FALLTHROUGH */ default: normal_match: if (n == string_end || c != FOLD ((UCHAR) *n)) return FNM_NOMATCH; } no_leading_period = new_no_leading_period; ++n; } if (n == string_end) return 0; if ((flags & FNM_LEADING_DIR) && n != string_end && *n == L_('/')) /* The FNM_LEADING_DIR flag says that "foo*" matches "foobar/frobozz". */ return 0; return FNM_NOMATCH; }
static int internal_function internal_fnmatch (const char *pattern, const char *string, int no_leading_period, int flags) { register const char *p = pattern, *n = string; register unsigned char c; /* Note that this evaluates C many times. */ # ifdef _LIBC # define FOLD(c) ((flags & FNM_CASEFOLD) ? tolower (c) : (c)) # else # define FOLD(c) ((flags & FNM_CASEFOLD) && ISUPPER (c) ? tolower (c) : (c)) # endif while ((c = *p++) != '\0') { c = FOLD (c); switch (c) { case '?': if (*n == '\0') return FNM_NOMATCH; else if (*n == '/' && (flags & FNM_FILE_NAME)) return FNM_NOMATCH; else if (*n == '.' && no_leading_period && (n == string || (n[-1] == '/' && (flags & FNM_FILE_NAME)))) return FNM_NOMATCH; break; case '\\': if (!(flags & FNM_NOESCAPE)) { c = *p++; if (c == '\0') /* Trailing \ loses. */ return FNM_NOMATCH; c = FOLD (c); } if (FOLD ((unsigned char) *n) != c) return FNM_NOMATCH; break; case '*': if (*n == '.' && no_leading_period && (n == string || (n[-1] == '/' && (flags & FNM_FILE_NAME)))) return FNM_NOMATCH; for (c = *p++; c == '?' || c == '*'; c = *p++) { if (*n == '/' && (flags & FNM_FILE_NAME)) /* A slash does not match a wildcard under FNM_FILE_NAME. */ return FNM_NOMATCH; else if (c == '?') { /* A ? needs to match one character. */ if (*n == '\0') /* There isn't another character; no match. */ return FNM_NOMATCH; else /* One character of the string is consumed in matching this ? wildcard, so *??? won't match if there are less than three characters. */ ++n; } } if (c == '\0') /* The wildcard(s) is/are the last element of the pattern. If the name is a file name and contains another slash this does mean it cannot match. */ return ((flags & FNM_FILE_NAME) && strchr (n, '/') != NULL ? FNM_NOMATCH : 0); else { const char *endp; endp = __strchrnul (n, (flags & FNM_FILE_NAME) ? '/' : '\0'); if (c == '[') { int flags2 = ((flags & FNM_FILE_NAME) ? flags : (flags & ~FNM_PERIOD)); for (--p; n < endp; ++n) if (internal_fnmatch (p, n, (no_leading_period && (n == string || (n[-1] == '/' && (flags & FNM_FILE_NAME)))), flags2) == 0) return 0; } else if (c == '/' && (flags & FNM_FILE_NAME)) { while (*n != '\0' && *n != '/') ++n; if (*n == '/' && (internal_fnmatch (p, n + 1, flags & FNM_PERIOD, flags) == 0)) return 0; } else { int flags2 = ((flags & FNM_FILE_NAME) ? flags : (flags & ~FNM_PERIOD)); if (c == '\\' && !(flags & FNM_NOESCAPE)) c = *p; c = FOLD (c); for (--p; n < endp; ++n) if (FOLD ((unsigned char) *n) == c && (internal_fnmatch (p, n, (no_leading_period && (n == string || (n[-1] == '/' && (flags & FNM_FILE_NAME)))), flags2) == 0)) return 0; } } /* If we come here no match is possible with the wildcard. */ return FNM_NOMATCH; case '[': { /* Nonzero if the sense of the character class is inverted. */ static int posixly_correct; register int not; char cold; if (posixly_correct == 0) posixly_correct = getenv ("POSIXLY_CORRECT") != NULL ? 1 : -1; if (*n == '\0') return FNM_NOMATCH; if (*n == '.' && no_leading_period && (n == string || (n[-1] == '/' && (flags & FNM_FILE_NAME)))) return FNM_NOMATCH; if (*n == '/' && (flags & FNM_FILE_NAME)) /* `/' cannot be matched. */ return FNM_NOMATCH; not = (*p == '!' || (posixly_correct < 0 && *p == '^')); if (not) ++p; c = *p++; for (;;) { unsigned char fn = FOLD ((unsigned char) *n); if (!(flags & FNM_NOESCAPE) && c == '\\') { if (*p == '\0') return FNM_NOMATCH; c = FOLD ((unsigned char) *p); ++p; if (c == fn) goto matched; } else if (c == '[' && *p == ':') { /* Leave room for the null. */ char str[CHAR_CLASS_MAX_LENGTH + 1]; size_t c1 = 0; # if defined _LIBC || (defined HAVE_WCTYPE_H && defined HAVE_WCHAR_H) wctype_t wt; # endif const char *startp = p; for (;;) { if (c1 > CHAR_CLASS_MAX_LENGTH) /* The name is too long and therefore the pattern is ill-formed. */ return FNM_NOMATCH; c = *++p; if (c == ':' && p[1] == ']') { p += 2; break; } if (c < 'a' || c >= 'z') { /* This cannot possibly be a character class name. Match it as a normal range. */ p = startp; c = '['; goto normal_bracket; } str[c1++] = c; } str[c1] = '\0'; # if defined _LIBC || (defined HAVE_WCTYPE_H && defined HAVE_WCHAR_H) wt = IS_CHAR_CLASS (str); if (wt == 0) /* Invalid character class name. */ return FNM_NOMATCH; if (__iswctype (__btowc ((unsigned char) *n), wt)) goto matched; # else if ((STREQ (str, "alnum") && ISALNUM ((unsigned char) *n)) || (STREQ (str, "alpha") && ISALPHA ((unsigned char) *n)) || (STREQ (str, "blank") && ISBLANK ((unsigned char) *n)) || (STREQ (str, "cntrl") && ISCNTRL ((unsigned char) *n)) || (STREQ (str, "digit") && ISDIGIT ((unsigned char) *n)) || (STREQ (str, "graph") && ISGRAPH ((unsigned char) *n)) || (STREQ (str, "lower") && ISLOWER ((unsigned char) *n)) || (STREQ (str, "print") && ISPRINT ((unsigned char) *n)) || (STREQ (str, "punct") && ISPUNCT ((unsigned char) *n)) || (STREQ (str, "space") && ISSPACE ((unsigned char) *n)) || (STREQ (str, "upper") && ISUPPER ((unsigned char) *n)) || (STREQ (str, "xdigit") && ISXDIGIT ((unsigned char) *n))) goto matched; # endif } else if (c == '\0') /* [ (unterminated) loses. */ return FNM_NOMATCH; else { normal_bracket: if (FOLD (c) == fn) goto matched; cold = c; c = *p++; if (c == '-' && *p != ']') { /* It is a range. */ unsigned char cend = *p++; if (!(flags & FNM_NOESCAPE) && cend == '\\') cend = *p++; if (cend == '\0') return FNM_NOMATCH; if (cold <= fn && fn <= FOLD (cend)) goto matched; c = *p++; } } if (c == ']') break; } if (!not) return FNM_NOMATCH; break; matched: /* Skip the rest of the [...] that already matched. */ while (c != ']') { if (c == '\0') /* [... (unterminated) loses. */ return FNM_NOMATCH; c = *p++; if (!(flags & FNM_NOESCAPE) && c == '\\') { if (*p == '\0') return FNM_NOMATCH; /* XXX 1003.2d11 is unclear if this is right. */ ++p; } else if (c == '[' && *p == ':') { do if (*++p == '\0') return FNM_NOMATCH; while (*p != ':' || p[1] == ']'); p += 2; c = *p; } } if (not) return FNM_NOMATCH; } break; default: if (c != FOLD ((unsigned char) *n)) return FNM_NOMATCH; } ++n; } if (*n == '\0') return 0; if ((flags & FNM_LEADING_DIR) && *n == '/') /* The FNM_LEADING_DIR flag says that "foo*" matches "foobar/frobozz". */ return 0; return FNM_NOMATCH; # undef FOLD }