int glob_url(URLGlob** glob, char* url, int *urlnum, FILE *error)
{
/*
* We can deal with any-size, just make a buffer with the same length
* as the specified URL!
*/
URLGlob *glob_expand;
int amount;
char *glob_buffer;
GlobCode res;
*glob = NULL;
glob_buffer = malloc(strlen(url) + 1);
if(!glob_buffer)
return CURLE_OUT_OF_MEMORY;
glob_expand = calloc(1, sizeof(URLGlob));
if(!glob_expand) {
free(glob_buffer);
return CURLE_OUT_OF_MEMORY;
}
glob_expand->size = 0;
glob_expand->urllen = strlen(url);
glob_expand->glob_buffer = glob_buffer;
glob_expand->beenhere = 0;
res = glob_word(glob_expand, url, 1, &amount);
if(!res)
*urlnum = amount;
else {
if(error && glob_expand->errormsg[0]) {
/* send error description to the error-stream */
fprintf(error, "curl: (%d) [globbing] %s",
(res == GLOB_NO_MEM) ? CURLE_OUT_OF_MEMORY : CURLE_URL_MALFORMAT,
glob_expand->errormsg);
}
/* it failed, we cleanup */
free(glob_buffer);
free(glob_expand);
*urlnum = 1;
return (res == GLOB_NO_MEM) ? CURLE_OUT_OF_MEMORY : CURLE_URL_MALFORMAT;
}
*glob = glob_expand;
return CURLE_OK;
}
static GlobCode glob_set(URLGlob *glob, char *pattern,
size_t pos, int *amount)
{
/* processes a set expression with the point behind the opening '{'
','-separated elements are collected until the next closing '}'
*/
bool done = FALSE;
char* buf = glob->glob_buffer;
URLPattern *pat;
pat = (URLPattern*)&glob->pattern[glob->size / 2];
/* patterns 0,1,2,... correspond to size=1,3,5,... */
pat->type = UPTSet;
pat->content.Set.size = 0;
pat->content.Set.ptr_s = 0;
pat->content.Set.elements = (char**)malloc(0);
++glob->size;
while (!done) {
bool skip;
switch (*pattern) {
case '\0': /* URL ended while set was still open */
snprintf(glob->errormsg, sizeof(glob->errormsg),
"unmatched brace at pos %d\n", (int)pos);
return GLOB_ERROR;
case '{':
case '[': /* no nested expressions at this time */
snprintf(glob->errormsg, sizeof(glob->errormsg),
"nested braces not supported at pos %d\n", (int)pos);
return GLOB_ERROR;
case ',':
case '}': /* set element completed */
*buf = '\0';
pat->content.Set.elements =
realloc(pat->content.Set.elements,
(pat->content.Set.size + 1) * sizeof(char*));
if (!pat->content.Set.elements) {
snprintf(glob->errormsg, sizeof(glob->errormsg), "out of memory");
return GLOB_ERROR;
}
pat->content.Set.elements[pat->content.Set.size] =
strdup(glob->glob_buffer);
++pat->content.Set.size;
if (*pattern == '}') {
/* entire set pattern completed */
int wordamount;
/* always check for a literal (may be "") between patterns */
if(GLOB_ERROR == glob_word(glob, ++pattern, ++pos, &wordamount))
wordamount=1;
*amount = pat->content.Set.size * wordamount;
done = TRUE;
continue;
}
buf = glob->glob_buffer;
++pattern;
++pos;
break;
case ']': /* illegal closing bracket */
snprintf(glob->errormsg, sizeof(glob->errormsg),
"illegal pattern at pos %d\n", (int)pos);
return GLOB_ERROR;
case '\\': /* escaped character, skip '\' */
switch(pattern[1]) {
case '[':
case ']':
case '{':
case '}':
case ',':
skip = TRUE;
break;
default:
skip = FALSE;
break;
}
if(skip) {
if (*(buf+1) == '\0') { /* but no escaping of '\0'! */
snprintf(glob->errormsg, sizeof(glob->errormsg),
"illegal pattern at pos %d\n", (int)pos);
return GLOB_ERROR;
}
++pattern;
++pos;
}
/* intentional fallthrough */
default:
*buf++ = *pattern++; /* copy character to set element */
++pos;
}
}
return GLOB_OK;
}
static GlobCode glob_range(URLGlob *glob, char *pattern,
size_t pos, int *amount)
{
/* processes a range expression with the point behind the opening '['
- char range: e.g. "a-z]", "B-Q]"
- num range: e.g. "0-9]", "17-2000]"
- num range with leading zeros: e.g. "001-999]"
expression is checked for well-formedness and collected until the next ']'
*/
URLPattern *pat;
char *c;
int wordamount=1;
char sep;
char sep2;
int step;
int rc;
pat = (URLPattern*)&glob->pattern[glob->size / 2];
/* patterns 0,1,2,... correspond to size=1,3,5,... */
++glob->size;
if (ISALPHA(*pattern)) { /* character range detected */
char min_c;
char max_c;
pat->type = UPTCharRange;
rc = sscanf(pattern, "%c-%c%c%d%c", &min_c, &max_c, &sep, &step, &sep2);
if ((rc < 3) || (min_c >= max_c) || ((max_c - min_c) > ('z' - 'a'))) {
/* the pattern is not well-formed */
snprintf(glob->errormsg, sizeof(glob->errormsg),
"errpr: bad range specification after pos %d\n", pos);
return GLOB_ERROR;
}
/* check the (first) separating character */
if((sep != ']') && (sep != ':')) {
snprintf(glob->errormsg, sizeof(glob->errormsg),
"error: unsupported character (%c) after range at pos %d\n",
sep, pos);
return GLOB_ERROR;
}
/* if there was a ":[num]" thing, use that as step or else use 1 */
pat->content.CharRange.step =
((sep == ':') && (rc == 5) && (sep2 == ']'))?step:1;
pat->content.CharRange.ptr_c = pat->content.CharRange.min_c = min_c;
pat->content.CharRange.max_c = max_c;
}
else if (ISDIGIT(*pattern)) { /* numeric range detected */
int min_n;
int max_n;
pat->type = UPTNumRange;
pat->content.NumRange.padlength = 0;
rc = sscanf(pattern, "%d-%d%c%d%c", &min_n, &max_n, &sep, &step, &sep2);
if ((rc < 2) || (min_n >= max_n)) {
/* the pattern is not well-formed */
snprintf(glob->errormsg, sizeof(glob->errormsg),
"error: bad range specification after pos %d\n", pos);
return GLOB_ERROR;
}
pat->content.NumRange.ptr_n = pat->content.NumRange.min_n = min_n;
pat->content.NumRange.max_n = max_n;
/* if there was a ":[num]" thing, use that as step or else use 1 */
pat->content.NumRange.step =
((sep == ':') && (rc == 5) && (sep2 == ']'))?step:1;
if (*pattern == '0') { /* leading zero specified */
c = pattern;
while (ISDIGIT(*c)) {
c++;
++pat->content.NumRange.padlength; /* padding length is set for all
instances of this pattern */
}
}
}
else {
snprintf(glob->errormsg, sizeof(glob->errormsg),
"illegal character in range specification at pos %d\n", pos);
return GLOB_ERROR;
}
c = (char*)strchr(pattern, ']'); /* continue after next ']' */
if(c)
c++;
else {
snprintf(glob->errormsg, sizeof(glob->errormsg), "missing ']'");
return GLOB_ERROR; /* missing ']' */
}
/* always check for a literal (may be "") between patterns */
if(GLOB_ERROR == glob_word(glob, c, pos + (c - pattern), &wordamount))
wordamount = 1;
if(pat->type == UPTCharRange)
*amount = (pat->content.CharRange.max_c -
pat->content.CharRange.min_c + 1) *
wordamount;
else
*amount = (pat->content.NumRange.max_n -
pat->content.NumRange.min_n + 1) * wordamount;
return GLOB_OK;
}
static GlobCode glob_set(URLGlob *glob, char *pattern,
size_t pos, int *amount)
{
/* processes a set expression with the point behind the opening '{'
','-separated elements are collected until the next closing '}'
*/
URLPattern *pat;
GlobCode res;
bool done = false;
char* buf = glob->glob_buffer;
pat = &glob->pattern[glob->size / 2];
/* patterns 0,1,2,... correspond to size=1,3,5,... */
pat->type = UPTSet;
pat->content.Set.size = 0;
pat->content.Set.ptr_s = 0;
pat->content.Set.elements = NULL;
++glob->size;
while(!done) {
switch (*pattern) {
case '\0': /* URL ended while set was still open */
snprintf(glob->errormsg, sizeof(glob->errormsg),
"unmatched brace at pos %zu\n", pos);
return GLOB_ERROR;
case '{':
case '[': /* no nested expressions at this time */
snprintf(glob->errormsg, sizeof(glob->errormsg),
"nested braces not supported at pos %zu\n", pos);
return GLOB_ERROR;
case ',':
case '}': /* set element completed */
*buf = '\0';
if(pat->content.Set.elements) {
char **new_arr = realloc(pat->content.Set.elements,
(pat->content.Set.size + 1) * sizeof(char*));
if(!new_arr) {
short elem;
for(elem = 0; elem < pat->content.Set.size; elem++)
free(pat->content.Set.elements[elem]);
free(pat->content.Set.elements);
pat->content.Set.ptr_s = 0;
pat->content.Set.size = 0;
}
pat->content.Set.elements = new_arr;
}
else
pat->content.Set.elements = malloc(sizeof(char*));
if(!pat->content.Set.elements) {
snprintf(glob->errormsg, sizeof(glob->errormsg), "out of memory\n");
return GLOB_NO_MEM;
}
pat->content.Set.elements[pat->content.Set.size] =
strdup(glob->glob_buffer);
if(!pat->content.Set.elements[pat->content.Set.size]) {
short elem;
for(elem = 0; elem < pat->content.Set.size; elem++)
free(pat->content.Set.elements[elem]);
free(pat->content.Set.elements);
pat->content.Set.ptr_s = 0;
pat->content.Set.size = 0;
snprintf(glob->errormsg, sizeof(glob->errormsg), "out of memory\n");
return GLOB_NO_MEM;
}
++pat->content.Set.size;
if(*pattern == '}') {
/* entire set pattern completed */
int wordamount;
/* always check for a literal (may be "") between patterns */
res = glob_word(glob, ++pattern, ++pos, &wordamount);
if(res) {
short elem;
for(elem = 0; elem < pat->content.Set.size; elem++)
free(pat->content.Set.elements[elem]);
free(pat->content.Set.elements);
pat->content.Set.ptr_s = 0;
pat->content.Set.size = 0;
return res;
}
*amount = pat->content.Set.size * wordamount;
done = true;
continue;
}
buf = glob->glob_buffer;
++pattern;
++pos;
break;
case ']': /* illegal closing bracket */
snprintf(glob->errormsg, sizeof(glob->errormsg),
"illegal pattern at pos %zu\n", pos);
return GLOB_ERROR;
case '\\': /* escaped character, skip '\' */
if(pattern[1]) {
++pattern;
++pos;
}
/* intentional fallthrough */
default:
*buf++ = *pattern++; /* copy character to set element */
++pos;
}
}
return GLOB_OK;
}
