aboutsummaryrefslogtreecommitdiffstats
path: root/lib/curl_fnmatch.c
diff options
context:
space:
mode:
Diffstat (limited to 'lib/curl_fnmatch.c')
-rw-r--r--lib/curl_fnmatch.c327
1 files changed, 131 insertions, 196 deletions
diff --git a/lib/curl_fnmatch.c b/lib/curl_fnmatch.c
index f33bba1f..0179a4f7 100644
--- a/lib/curl_fnmatch.c
+++ b/lib/curl_fnmatch.c
@@ -47,14 +47,7 @@
#define CURLFNM_UPPER (CURLFNM_CHARSET_LEN + 10)
typedef enum {
- CURLFNM_LOOP_DEFAULT = 0,
- CURLFNM_LOOP_BACKSLASH
-} loop_state;
-
-typedef enum {
CURLFNM_SCHS_DEFAULT = 0,
- CURLFNM_SCHS_MAYRANGE,
- CURLFNM_SCHS_MAYRANGE2,
CURLFNM_SCHS_RIGHTBR,
CURLFNM_SCHS_RIGHTBRLEFTBR
} setcharset_state;
@@ -64,6 +57,13 @@ typedef enum {
CURLFNM_PKW_DDOT
} parsekey_state;
+typedef enum {
+ CCLASS_OTHER = 0,
+ CCLASS_DIGIT,
+ CCLASS_UPPER,
+ CCLASS_LOWER
+} char_class;
+
#define SETCHARSET_OK 1
#define SETCHARSET_FAIL 0
@@ -81,12 +81,12 @@ static int parsekeyword(unsigned char **pattern, unsigned char *charset)
return SETCHARSET_FAIL;
switch(state) {
case CURLFNM_PKW_INIT:
- if(ISALPHA(c) && ISLOWER(c))
+ if(ISLOWER(c))
keyword[i] = c;
else if(c == ':')
state = CURLFNM_PKW_DDOT;
else
- return 0;
+ return SETCHARSET_FAIL;
break;
case CURLFNM_PKW_DDOT:
if(c == ']')
@@ -123,14 +123,48 @@ static int parsekeyword(unsigned char **pattern, unsigned char *charset)
return SETCHARSET_OK;
}
+/* Return the character class. */
+static char_class charclass(unsigned char c)
+{
+ if(ISUPPER(c))
+ return CCLASS_UPPER;
+ if(ISLOWER(c))
+ return CCLASS_LOWER;
+ if(ISDIGIT(c))
+ return CCLASS_DIGIT;
+ return CCLASS_OTHER;
+}
+
+/* Include a character or a range in set. */
+static void setcharorrange(unsigned char **pp, unsigned char *charset)
+{
+ unsigned char *p = (*pp)++;
+ unsigned char c = *p++;
+
+ charset[c] = 1;
+ if(ISALNUM(c) && *p++ == '-') {
+ char_class cc = charclass(c);
+ unsigned char endrange = *p++;
+
+ if(endrange == '\\')
+ endrange = *p++;
+ if(endrange >= c && charclass(endrange) == cc) {
+ while(c++ != endrange)
+ if(charclass(c) == cc) /* Chars in class may be not consecutive. */
+ charset[c] = 1;
+ *pp = p;
+ }
+ }
+}
+
/* returns 1 (true) if pattern is OK, 0 if is bad ("p" is pattern pointer) */
static int setcharset(unsigned char **p, unsigned char *charset)
{
setcharset_state state = CURLFNM_SCHS_DEFAULT;
- unsigned char rangestart = 0;
- unsigned char lastchar = 0;
bool something_found = FALSE;
unsigned char c;
+
+ memset(charset, 0, CURLFNM_CHSET_SIZE);
for(;;) {
c = **p;
if(!c)
@@ -138,14 +172,7 @@ static int setcharset(unsigned char **p, unsigned char *charset)
switch(state) {
case CURLFNM_SCHS_DEFAULT:
- if(ISALNUM(c)) { /* ASCII value */
- rangestart = c;
- charset[c] = 1;
- (*p)++;
- state = CURLFNM_SCHS_MAYRANGE;
- something_found = TRUE;
- }
- else if(c == ']') {
+ if(c == ']') {
if(something_found)
return SETCHARSET_OK;
something_found = TRUE;
@@ -154,26 +181,16 @@ static int setcharset(unsigned char **p, unsigned char *charset)
(*p)++;
}
else if(c == '[') {
- char c2 = *((*p) + 1);
- if(c2 == ':') { /* there has to be a keyword */
- (*p) += 2;
- if(parsekeyword(p, charset)) {
- state = CURLFNM_SCHS_DEFAULT;
- }
- else
- return SETCHARSET_FAIL;
- }
+ unsigned char *pp = *p + 1;
+
+ if(*pp++ == ':' && parsekeyword(&pp, charset))
+ *p = pp;
else {
charset[c] = 1;
(*p)++;
}
something_found = TRUE;
}
- else if(c == '?' || c == '*') {
- something_found = TRUE;
- charset[c] = 1;
- (*p)++;
- }
else if(c == '^' || c == '!') {
if(!something_found) {
if(charset[CURLFNM_NEGATE]) {
@@ -189,82 +206,17 @@ static int setcharset(unsigned char **p, unsigned char *charset)
}
else if(c == '\\') {
c = *(++(*p));
- if(ISPRINT((c))) {
- something_found = TRUE;
- state = CURLFNM_SCHS_MAYRANGE;
- charset[c] = 1;
- rangestart = c;
- (*p)++;
- }
+ if(c)
+ setcharorrange(p, charset);
else
- return SETCHARSET_FAIL;
+ charset['\\'] = 1;
+ something_found = TRUE;
}
else {
- charset[c] = 1;
- (*p)++;
+ setcharorrange(p, charset);
something_found = TRUE;
}
break;
- case CURLFNM_SCHS_MAYRANGE:
- if(c == '-') {
- charset[c] = 1;
- (*p)++;
- lastchar = '-';
- state = CURLFNM_SCHS_MAYRANGE2;
- }
- else if(c == '[') {
- state = CURLFNM_SCHS_DEFAULT;
- }
- else if(ISALNUM(c)) {
- charset[c] = 1;
- (*p)++;
- }
- else if(c == '\\') {
- c = *(++(*p));
- if(ISPRINT(c)) {
- charset[c] = 1;
- (*p)++;
- }
- else
- return SETCHARSET_FAIL;
- }
- else if(c == ']') {
- return SETCHARSET_OK;
- }
- else
- return SETCHARSET_FAIL;
- break;
- case CURLFNM_SCHS_MAYRANGE2:
- if(c == ']') {
- return SETCHARSET_OK;
- }
- else if(c == '\\') {
- c = *(++(*p));
- if(ISPRINT(c)) {
- charset[c] = 1;
- state = CURLFNM_SCHS_DEFAULT;
- (*p)++;
- }
- else
- return SETCHARSET_FAIL;
- }
- else if(c >= rangestart) {
- if((ISLOWER(c) && ISLOWER(rangestart)) ||
- (ISDIGIT(c) && ISDIGIT(rangestart)) ||
- (ISUPPER(c) && ISUPPER(rangestart))) {
- charset[lastchar] = 0;
- rangestart++;
- while(rangestart++ <= c)
- charset[rangestart-1] = 1;
- (*p)++;
- state = CURLFNM_SCHS_DEFAULT;
- }
- else
- return SETCHARSET_FAIL;
- }
- else
- return SETCHARSET_FAIL;
- break;
case CURLFNM_SCHS_RIGHTBR:
if(c == '[') {
state = CURLFNM_SCHS_RIGHTBRLEFTBR;
@@ -286,14 +238,11 @@ static int setcharset(unsigned char **p, unsigned char *charset)
goto fail;
break;
case CURLFNM_SCHS_RIGHTBRLEFTBR:
- if(c == ']') {
+ if(c == ']')
return SETCHARSET_OK;
- }
- else {
- state = CURLFNM_SCHS_DEFAULT;
- charset[c] = 1;
- (*p)++;
- }
+ state = CURLFNM_SCHS_DEFAULT;
+ charset[c] = 1;
+ (*p)++;
break;
}
}
@@ -304,107 +253,93 @@ fail:
static int loop(const unsigned char *pattern, const unsigned char *string,
int maxstars)
{
- loop_state state = CURLFNM_LOOP_DEFAULT;
unsigned char *p = (unsigned char *)pattern;
unsigned char *s = (unsigned char *)string;
unsigned char charset[CURLFNM_CHSET_SIZE] = { 0 };
- int rc = 0;
for(;;) {
- switch(state) {
- case CURLFNM_LOOP_DEFAULT:
- if(*p == '*') {
- if(!maxstars)
- return CURL_FNMATCH_NOMATCH;
- while(*(p + 1) == '*') /* eliminate multiple stars */
- p++;
- if(*s == '\0' && *(p + 1) == '\0')
- return CURL_FNMATCH_MATCH;
- rc = loop(p + 1, s, maxstars - 1); /* *.txt matches .txt <=>
- .txt matches .txt */
- if(rc == CURL_FNMATCH_MATCH)
+ unsigned char *pp;
+
+ switch(*p) {
+ case '*':
+ if(!maxstars)
+ return CURL_FNMATCH_NOMATCH;
+ /* Regroup consecutive stars and question marks. This can be done because
+ '*?*?*' can be expressed as '??*'. */
+ for(;;) {
+ if(*++p == '\0')
return CURL_FNMATCH_MATCH;
- if(*s) /* let the star eat up one character */
- s++;
- else
- return CURL_FNMATCH_NOMATCH;
- }
- else if(*p == '?') {
- if(ISPRINT(*s)) {
- s++;
- p++;
+ if(*p == '?') {
+ if(!*s++)
+ return CURL_FNMATCH_NOMATCH;
}
- else if(*s == '\0')
- return CURL_FNMATCH_NOMATCH;
- else
- return CURL_FNMATCH_FAIL; /* cannot deal with other character */
+ else if(*p != '*')
+ break;
}
- else if(*p == '\0') {
- if(*s == '\0')
+ /* Skip string characters until we find a match with pattern suffix. */
+ for(maxstars--; *s; s++) {
+ if(loop(p, s, maxstars) == CURL_FNMATCH_MATCH)
return CURL_FNMATCH_MATCH;
- return CURL_FNMATCH_NOMATCH;
}
- else if(*p == '\\') {
- state = CURLFNM_LOOP_BACKSLASH;
+ return CURL_FNMATCH_NOMATCH;
+ case '?':
+ if(!*s)
+ return CURL_FNMATCH_NOMATCH;
+ s++;
+ p++;
+ break;
+ case '\0':
+ return *s? CURL_FNMATCH_NOMATCH: CURL_FNMATCH_MATCH;
+ case '\\':
+ if(p[1])
p++;
- }
- else if(*p == '[') {
- unsigned char *pp = p + 1; /* cannot handle with pointer to register */
- if(setcharset(&pp, charset)) {
- int found = FALSE;
- if(charset[(unsigned int)*s])
- found = TRUE;
- else if(charset[CURLFNM_ALNUM])
- found = ISALNUM(*s);
- else if(charset[CURLFNM_ALPHA])
- found = ISALPHA(*s);
- else if(charset[CURLFNM_DIGIT])
- found = ISDIGIT(*s);
- else if(charset[CURLFNM_XDIGIT])
- found = ISXDIGIT(*s);
- else if(charset[CURLFNM_PRINT])
- found = ISPRINT(*s);
- else if(charset[CURLFNM_SPACE])
- found = ISSPACE(*s);
- else if(charset[CURLFNM_UPPER])
- found = ISUPPER(*s);
- else if(charset[CURLFNM_LOWER])
- found = ISLOWER(*s);
- else if(charset[CURLFNM_BLANK])
- found = ISBLANK(*s);
- else if(charset[CURLFNM_GRAPH])
- found = ISGRAPH(*s);
+ if(*s++ != *p++)
+ return CURL_FNMATCH_NOMATCH;
+ break;
+ case '[':
+ pp = p + 1; /* Copy in case of syntax error in set. */
+ if(setcharset(&pp, charset)) {
+ int found = FALSE;
+ if(!*s)
+ return CURL_FNMATCH_NOMATCH;
+ if(charset[(unsigned int)*s])
+ found = TRUE;
+ else if(charset[CURLFNM_ALNUM])
+ found = ISALNUM(*s);
+ else if(charset[CURLFNM_ALPHA])
+ found = ISALPHA(*s);
+ else if(charset[CURLFNM_DIGIT])
+ found = ISDIGIT(*s);
+ else if(charset[CURLFNM_XDIGIT])
+ found = ISXDIGIT(*s);
+ else if(charset[CURLFNM_PRINT])
+ found = ISPRINT(*s);
+ else if(charset[CURLFNM_SPACE])
+ found = ISSPACE(*s);
+ else if(charset[CURLFNM_UPPER])
+ found = ISUPPER(*s);
+ else if(charset[CURLFNM_LOWER])
+ found = ISLOWER(*s);
+ else if(charset[CURLFNM_BLANK])
+ found = ISBLANK(*s);
+ else if(charset[CURLFNM_GRAPH])
+ found = ISGRAPH(*s);
- if(charset[CURLFNM_NEGATE])
- found = !found;
+ if(charset[CURLFNM_NEGATE])
+ found = !found;
- if(found) {
- p = pp + 1;
- if(*s)
- /* don't advance if we're matching on an empty string */
- s++;
- memset(charset, 0, CURLFNM_CHSET_SIZE);
- }
- else
- return CURL_FNMATCH_NOMATCH;
- }
- else
- return CURL_FNMATCH_FAIL;
- }
- else {
- if(*p++ != *s++)
+ if(!found)
return CURL_FNMATCH_NOMATCH;
+ p = pp + 1;
+ s++;
+ break;
}
- break;
- case CURLFNM_LOOP_BACKSLASH:
- if(ISPRINT(*p)) {
- if(*p++ == *s++)
- state = CURLFNM_LOOP_DEFAULT;
- else
- return CURL_FNMATCH_NOMATCH;
- }
- else
- return CURL_FNMATCH_FAIL;
+
+ /* Syntax error in set: this must be taken as a regular character. */
+ /* FALLTHROUGH */
+ default:
+ if(*p++ != *s++)
+ return CURL_FNMATCH_NOMATCH;
break;
}
}
generated by cgit v1.2.3 (git 2.25.1) at 2024-05-30 03:45:43 +0000