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);
}
static int _print_psl_entries_dafsa_binary(const char *fname, const _psl_vector_t *v)
{
FILE *fp;
int ret = 0, it, rc;
char cmd[256];
if ((fp = fopen("in.tmp", "w"))) {
for (it = 0; it < v->cur; it++) {
_psl_entry_t *e = _vector_get(v, it);
unsigned char *s = (unsigned char *)e->label_buf;
/* search for non-ASCII label and skip it */
while (*s && *s < 128) s++;
if (*s) continue;
fprintf(fp, "%s, %X\n", e->label_buf, (int) (e->flags & 0x0F));
}
fclose(fp);
} else {
fprintf(stderr, "Failed to write open 'in.tmp'\n");
return 3;
}
snprintf(cmd, sizeof(cmd), MAKE_DAFSA " --binary in.tmp %s", fname);
if ((rc = system(cmd))) {
fprintf(stderr, "Failed to execute '%s' (%d)\n", cmd, rc);
ret = 2;
}
unlink("in.tmp");
return ret;
}
static void _print_psl_entries(FILE *fpout, const _psl_vector_t *v, const char *varname)
{
int it;
#ifdef BUILTIN_GENERATOR_LIBICU
do {
UVersionInfo version_info;
char version[U_MAX_VERSION_STRING_LENGTH];
u_getVersion(version_info);
u_versionToString(version_info, version);
fprintf(fpout, "/* automatically generated by psl2c (punycode generated with libicu/%s) */\n", version);
} while (0);
#elif defined(BUILTIN_GENERATOR_LIBIDN2)
fprintf(fpout, "/* automatically generated by psl2c (punycode generated with libidn2/%s) */\n", idn2_check_version(NULL));
#elif defined(BUILTIN_GENERATOR_LIBIDN)
fprintf(fpout, "/* automatically generated by psl2c (punycode generated with libidn/%s) */\n", stringprep_check_version(NULL));
#else
fprintf(fpout, "/* automatically generated by psl2c (without punycode support) */\n");
#endif
fprintf(fpout, "static _psl_entry_t %s[] = {\n", varname);
for (it = 0; it < v->cur; it++) {
_psl_entry_t *e = _vector_get(v, it);
fprintf(fpout, "\t{ \"%s\", NULL, %hd, %d, %d },\n",
e->label_buf, e->length, (int) e->nlabels, (int) e->flags);
}
fprintf(fpout, "};\n");
}
void *_vector_push(struct vector *v, size_t elemsize)
{
if (v->count == v->allocated_count) {
if (!vector_grow(v)) {
return NULL;
}
}
vector_resize(v, v->count+1);
return _vector_get(v, elemsize, v->count-1);
}
static void _print_psl_entries_dafsa(FILE *fpout, const _psl_vector_t *v)
{
FILE *fp;
int it;
#ifdef BUILTIN_GENERATOR_LIBICU
do {
UVersionInfo version_info;
char version[U_MAX_VERSION_STRING_LENGTH];
u_getVersion(version_info);
u_versionToString(version_info, version);
fprintf(fpout, "/* automatically generated by psl2c (punycode generated with libicu/%s) */\n", version);
} while (0);
#elif defined(BUILTIN_GENERATOR_LIBIDN2)
fprintf(fpout, "/* automatically generated by psl2c (punycode generated with libidn2/%s) */\n", idn2_check_version(NULL));
#elif defined(BUILTIN_GENERATOR_LIBIDN)
fprintf(fpout, "/* automatically generated by psl2c (punycode generated with libidn/%s) */\n", stringprep_check_version(NULL));
#else
fprintf(fpout, "/* automatically generated by psl2c (punycode generated internally) */\n");
#endif
if ((fp = fopen("in.tmp", "w"))) {
for (it = 0; it < v->cur; it++) {
_psl_entry_t *e = _vector_get(v, it);
unsigned char *s = (unsigned char *)e->label_buf;
/* search for non-ASCII label and skip it */
while (*s && *s < 128) s++;
if (*s) continue;
fprintf(fp, "%s, %X\n", e->label_buf, (int) (e->flags & 0x0F));
}
fclose(fp);
}
if ((it = system(MAKE_DAFSA " in.tmp out.tmp")))
fprintf(stderr, "Failed to execute " MAKE_DAFSA "\n");
if ((fp = fopen("out.tmp", "r"))) {
char buf[256];
while (fgets(buf, sizeof(buf), fp))
fputs(buf, fpout);
fclose(fp);
}
unlink("in.tmp");
unlink("out.tmp");
}
static void _add_punycode_if_needed(_psl_vector_t *v, _psl_entry_t *e)
{
char *lookupname = NULL;
int rc;
uint8_t *lower, resbuf[256];
size_t len = sizeof(resbuf) - 1; /* leave space for additional \0 byte */
if (_str_is_ascii(e->label_buf))
return;
/* we need a conversion to lowercase */
lower = u8_tolower((uint8_t *)e->label_buf, u8_strlen((uint8_t *)e->label_buf), 0, UNINORM_NFKC, resbuf, &len);
if (!lower) {
/* fprintf(stderr, "u8_tolower(%s) failed (%d)\n", e->label_buf, errno); */
return;
}
/* u8_tolower() does not terminate the result string */
if (lower == resbuf) {
lower[len]=0;
} else {
uint8_t *tmp = lower;
lower = (uint8_t *)strndup((char *)lower, len);
free(tmp);
}
if ((rc = idn2_lookup_u8(lower, (uint8_t **)&lookupname, 0)) == IDN2_OK) {
if (strcmp(e->label_buf, lookupname)) {
_psl_entry_t suffix, *suffixp;
/* fprintf(stderr, "libidn '%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 */
} /* else ignore */
} /* else
fprintf(stderr, "toASCII(%s) failed (%d): %s\n", lower, rc, idn2_strerror(rc)); */
if (lower != resbuf)
free(lower);
}
static void _add_punycode_if_needed(_psl_vector_t *v, _psl_entry_t *e)
{
char *lookupname = NULL;
int rc;
if (_str_is_ascii(e->label_buf))
return;
/* idna_to_ascii_8z() automatically converts UTF-8 to lowercase */
if ((rc = idna_to_ascii_8z(e->label_buf, &lookupname, IDNA_USE_STD3_ASCII_RULES)) == IDNA_SUCCESS) {
if (strcmp(e->label_buf, lookupname)) {
_psl_entry_t suffix, *suffixp;
/* fprintf(stderr, "libidn '%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 */
} /* else ignore */
} /* else
fprintf(_(stderr, "toASCII failed (%d): %s\n"), rc, idna_strerror(rc)); */
}
static void _add_punycode_if_needed(UIDNA *idna, _psl_vector_t *v, _psl_entry_t *e)
{
if (_str_is_ascii(e->label_buf))
return;
/* IDNA2008 UTS#46 punycode conversion */
if (idna) {
char lookupname[128] = "";
UErrorCode status = 0;
UIDNAInfo info = UIDNA_INFO_INITIALIZER;
UChar utf16_dst[128], utf16_src[128];
int32_t utf16_src_length;
u_strFromUTF8(utf16_src, sizeof(utf16_src)/sizeof(utf16_src[0]), &utf16_src_length, e->label_buf, -1, &status);
if (U_SUCCESS(status)) {
int32_t dst_length = uidna_nameToASCII(idna, utf16_src, utf16_src_length, utf16_dst, sizeof(utf16_dst)/sizeof(utf16_dst[0]), &info, &status);
if (U_SUCCESS(status)) {
u_strToUTF8(lookupname, sizeof(lookupname), NULL, utf16_dst, dst_length, &status);
if (U_SUCCESS(status)) {
if (strcmp(e->label_buf, lookupname)) {
_psl_entry_t suffix, *suffixp;
/* fprintf(stderr, "libicu '%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 */
} /* else ignore */
} /* else
fprintf(stderr, "Failed to convert UTF-16 to UTF-8 (status %d)\n", status); */
} /* else
fprintf(stderr, "Failed to convert to ASCII (status %d)\n", status); */
} /* else
fprintf(stderr, "Failed to convert UTF-8 to UTF-16 (status %d)\n", status); */
}
}
static int _check_psl(const psl_ctx_t *psl)
{
int it, pos, err = 0;
/* check if plain suffix also appears in exceptions */
for (it = 0; it < psl->suffixes->cur; it++) {
_psl_entry_t *e = _vector_get(psl->suffixes, it);
if (!e->wildcard && _vector_find(psl->suffix_exceptions, e) >= 0) {
fprintf(stderr, "Found entry '%s' also in exceptions\n", e->label);
err = 1;
}
}
/* check if exception also appears in suffix list as plain entry */
for (it = 0; it < psl->suffix_exceptions->cur; it++) {
_psl_entry_t *e2, *e = _vector_get(psl->suffix_exceptions, it);
if ((e2 = _vector_get(psl->suffixes, pos = _vector_find(psl->suffixes, e)))) {
if (!e2->wildcard) {
fprintf(stderr, "Found exception '!%s' also as suffix\n", e->label);
err = 1;
}
/* Two same domains in a row are allowed: wildcard and non-wildcard.
* Binary search find either of them, so also check previous and next entry. */
else if (pos > 0 && _suffix_compare(e, e2 = _vector_get(psl->suffixes, pos - 1)) == 0 && !e2->wildcard) {
fprintf(stderr, "Found exception '!%s' also as suffix\n", e->label);
err = 1;
}
else if (pos < psl->suffixes->cur - 1 && _suffix_compare(e, e2 = _vector_get(psl->suffixes, pos + 1)) == 0 && !e2->wildcard) {
fprintf(stderr, "Found exception '!%s' also as suffix\n", e->label);
err = 1;
}
}
}
/* check if non-wildcard entry is already covered by wildcard entry */
for (it = 0; it < psl->suffixes->cur; it++) {
const char *p;
_psl_entry_t *e = _vector_get(psl->suffixes, it);
if (e->nlabels > 1 && !e->wildcard && (p = strchr(e->label, '.'))) {
_psl_entry_t *e2, *e3, suffix;
suffix.label = p + 1;
suffix.length = strlen(p + 1);
suffix.nlabels = e->nlabels - 1;
e2 = _vector_get(psl->suffixes, pos = _vector_find(psl->suffixes, &suffix));
if (e2) {
if (e2->wildcard) {
fprintf(stderr, "Found superfluous '%s' already covered by '*.%s'\n", e->label, e2->label);
err = 1;
}
/* Two same domains in a row are allowed: wildcard and non-wildcard.
* Binary search find either of them, so also check previous and next entry. */
else if (pos > 0 && _suffix_compare(e2, e3 = _vector_get(psl->suffixes, pos - 1)) == 0 && e3->wildcard) {
fprintf(stderr, "Found superfluous '%s' already covered by '*.%s'\n", e->label, e2->label);
err = 1;
}
else if (pos < psl->suffixes->cur - 1 && _suffix_compare(e2, e3 = _vector_get(psl->suffixes, pos + 1)) == 0 && e3->wildcard) {
fprintf(stderr, "Found superfluous '%s' already covered by '*.%s'\n", e->label, e2->label);
err = 1;
}
}
}
}
return err;
}
extern void* vector_get(vector_t* v, size_t index) { return *(void**) _vector_get(v, index); }
/**
* 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;
}
static int _psl_is_public_suffix(const psl_ctx_t *psl, const char *domain)
{
_psl_entry_t suffix, *rule;
const char *p, *label_bak;
unsigned short length_bak;
/* this function should be called without leading dots, just make sure */
suffix.label = domain + (*domain == '.');
suffix.length = strlen(suffix.label);
suffix.wildcard = 0;
suffix.nlabels = 1;
for (p = suffix.label; *p; p++)
if (*p == '.')
suffix.nlabels++;
/* if domain has enough labels, it is public */
if (psl == &_builtin_psl)
rule = &suffixes[0];
else
rule = _vector_get(psl->suffixes, 0);
if (!rule || rule->nlabels < suffix.nlabels - 1)
return 0;
if (psl == &_builtin_psl)
rule = bsearch(&suffix, suffixes, countof(suffixes), sizeof(suffixes[0]), (int(*)(const void *, const void *))_suffix_compare);
else
rule = _vector_get(psl->suffixes, _vector_find(psl->suffixes, &suffix));
if (rule) {
/* definitely a match, no matter if the found rule is a wildcard or not */
return 1;
} else if (suffix.nlabels == 1) {
/* unknown TLD, this is the prevailing '*' match */
return 1;
}
label_bak = suffix.label;
length_bak = suffix.length;
if ((suffix.label = strchr(suffix.label, '.'))) {
suffix.label++;
suffix.length = strlen(suffix.label);
suffix.nlabels--;
if (psl == &_builtin_psl)
rule = bsearch(&suffix, suffixes, countof(suffixes), sizeof(suffixes[0]), (int(*)(const void *, const void *))_suffix_compare);
else
rule = _vector_get(psl->suffixes, _vector_find(psl->suffixes, &suffix));
if (rule) {
if (rule->wildcard) {
/* now that we matched a wildcard, we have to check for an exception */
suffix.label = label_bak;
suffix.length = length_bak;
suffix.nlabels++;
if (psl == &_builtin_psl) {
if (bsearch(&suffix, suffix_exceptions, countof(suffix_exceptions), sizeof(suffix_exceptions[0]), (int(*)(const void *, const void *))_suffix_compare))
return 0; /* found an exception, so 'domain' is not a public suffix */
} else {
if (_vector_get(psl->suffix_exceptions, _vector_find(psl->suffix_exceptions, &suffix)) != 0)
return 0; /* found an exception, so 'domain' is not a public suffix */
}
return 1;
}
}
}
return 0;
}