static void _add_punycode_if_needed(_psl_vector_t *v)
{
int it, n;
/* do not use 'it < v->cur' since v->cur is changed by _vector_add() ! */
for (it = 0, n = v->cur; it < n; it++) {
_psl_entry_t *e = _vector_get(v, it);
if (_str_needs_encoding(e->label_buf)) {
_psl_entry_t suffix, *suffixp;
char lookupname[64] = "";
/* this is much slower than the libidn2 API but should have no license issues */
FILE *pp;
char cmd[16 + sizeof(e->label_buf)];
snprintf(cmd, sizeof(cmd), "idn2 '%s'", e->label_buf);
if ((pp = popen(cmd, "r"))) {
if (fscanf(pp, "%63s", lookupname) >= 1 && strcmp(e->label_buf, lookupname)) {
/* fprintf(stderr, "idn2 '%s' -> '%s'\n", e->label_buf, lookupname); */
_suffix_init(&suffix, lookupname, strlen(lookupname));
suffix.wildcard = e->wildcard;
suffixp = _vector_get(v, _vector_add(v, &suffix));
suffixp->label = suffixp->label_buf; /* set label to changed address */
}
pclose(pp);
} else
fprintf(stderr, "Failed to call popen(%s, \"r\")\n", cmd);
}
}
_vector_sort(v);
}
void @TYPE@_vector_select_unique(@TYPE@_vector_type * vector) {
@TYPE@_vector_assert_writable( vector );
{
@TYPE@_vector_type * copy = @TYPE@_vector_alloc_copy( vector );
@TYPE@_vector_sort( copy );
@TYPE@_vector_reset( vector );
{
int i;
@TYPE@ previous_value = @TYPE@_vector_iget( copy , 0);
@TYPE@_vector_append( vector , previous_value);
for (i=1; i < copy->size; i++) {
@TYPE@ value = @TYPE@_vector_iget( copy , i );
if (value != previous_value)
@TYPE@_vector_append( vector , value);
previous_value = value;
}
}
@TYPE@_vector_free( copy );
}
}
/**
* psl_load_fp:
* @fp: FILE pointer
*
* This function loads the public suffixes from a FILE pointer.
* To free the allocated resources, call psl_free().
*
* The suffixes are expected to be lowercase UTF-8 encoded if they are international.
*
* Returns: Pointer to a PSL context or %NULL on failure.
*
* Since: 0.1
*/
psl_ctx_t *psl_load_fp(FILE *fp)
{
psl_ctx_t *psl;
_psl_entry_t suffix, *suffixp;
char buf[256], *linep, *p;
#ifdef WITH_LIBICU
UIDNA *idna;
UErrorCode status = 0;
#endif
if (!fp)
return NULL;
if (!(psl = calloc(1, sizeof(psl_ctx_t))))
return NULL;
#ifdef WITH_LIBICU
idna = uidna_openUTS46(UIDNA_USE_STD3_RULES, &status);
#endif
/*
* as of 02.11.2012, the list at http://publicsuffix.org/list/ contains ~6000 rules and 40 exceptions.
* as of 19.02.2014, the list at http://publicsuffix.org/list/ contains ~6500 rules and 19 exceptions.
*/
psl->suffixes = _vector_alloc(8*1024, _suffix_compare);
psl->suffix_exceptions = _vector_alloc(64, _suffix_compare);
while ((linep = fgets(buf, sizeof(buf), fp))) {
while (isspace(*linep)) linep++; /* ignore leading whitespace */
if (!*linep) continue; /* skip empty lines */
if (*linep == '/' && linep[1] == '/')
continue; /* skip comments */
/* parse suffix rule */
for (p = linep; *linep && !isspace(*linep);) linep++;
*linep = 0;
if (*p == '!') {
/* add to exceptions */
if (_suffix_init(&suffix, p + 1, linep - p - 1) == 0) {
suffixp = _vector_get(psl->suffix_exceptions, _vector_add(psl->suffix_exceptions, &suffix));
suffixp->label = suffixp->label_buf; /* set label to changed address */
#ifdef WITH_LIBICU
_add_punycode_if_needed(idna, psl->suffix_exceptions, suffixp);
#elif defined(WITH_LIBIDN2) || defined(WITH_LIBIDN)
_add_punycode_if_needed(psl->suffix_exceptions, suffixp);
#endif
}
} else {
/* add to suffixes */
if (_suffix_init(&suffix, p, linep - p) == 0) {
suffixp = _vector_get(psl->suffixes, _vector_add(psl->suffixes, &suffix));
suffixp->label = suffixp->label_buf; /* set label to changed address */
#ifdef WITH_LIBICU
_add_punycode_if_needed(idna, psl->suffixes, suffixp);
#elif defined(WITH_LIBIDN2) || defined(WITH_LIBIDN)
_add_punycode_if_needed(psl->suffixes, suffixp);
#endif
}
}
}
_vector_sort(psl->suffix_exceptions);
_vector_sort(psl->suffixes);
#ifdef WITH_LIBICU
if (idna)
uidna_close(idna);
#endif
return psl;
}
