int main(int argc, const char *const *argv)
{
int mode = 1;
const char *const *arg, *psl_file = NULL, *cookie_domain = NULL;
psl_ctx_t *psl = (psl_ctx_t *) psl_builtin();
/* set current locale according to the environment variables */
#include <locale.h>
setlocale(LC_ALL, "");
for (arg = argv + 1; arg < argv + argc; arg++) {
if (!strncmp(*arg, "--", 2)) {
if (!strcmp(*arg, "--is-public-suffix"))
mode = 1;
else if (!strcmp(*arg, "--print-unreg-domain"))
mode = 2;
else if (!strcmp(*arg, "--print-reg-domain"))
mode = 3;
else if (!strcmp(*arg, "--print-info"))
mode = 99;
else if (!strcmp(*arg, "--is-cookie-domain-acceptable") && arg < argv + argc - 1) {
mode = 4;
cookie_domain = *(++arg);
}
else if (!strcmp(*arg, "--use-builtin-data")) {
psl_free(psl);
if (psl_file) {
fprintf(stderr, "Dropped data from %s\n", psl_file);
psl_file = NULL;
}
if (!(psl = (psl_ctx_t *) psl_builtin()))
printf("No builtin PSL data available\n");
}
else if (!strcmp(*arg, "--load-psl-file") && arg < argv + argc - 1) {
psl_free(psl);
if (psl_file) {
fprintf(stderr, "Dropped data from %s\n", psl_file);
psl_file = NULL;
}
if (!(psl = psl_load_file(psl_file = *(++arg)))) {
fprintf(stderr, "Failed to load PSL data from %s\n\n", psl_file);
psl_file = NULL;
}
}
else if (!strcmp(*arg, "--help")) {
fprintf(stdout, "`psl' explores the Public Suffix List\n\n");
usage(0, stdout);
}
else if (!strcmp(*arg, "--version")) {
printf("psl %s (0x%06x)\n", PACKAGE_VERSION, psl_check_version_number(0));
printf("libpsl %s\n", psl_get_version());
printf("\n");
printf("Copyright (C) 2014-2015 Tim Ruehsen\n");
printf("License: MIT\n");
exit(0);
}
else if (!strcmp(*arg, "--")) {
arg++;
break;
}
else {
fprintf(stderr, "Unknown option '%s'\n", *arg);
usage(1, stderr);
}
} else
break;
}
if (mode != 99) {
if (!psl) {
fprintf(stderr, "No PSL data available - aborting\n");
exit(2);
}
if (arg >= argv + argc) {
char buf[256], *domain, *lower;
size_t len;
psl_error_t rc;
/* read URLs from STDIN */
while (fgets(buf, sizeof(buf), stdin)) {
for (domain = buf; isspace(*domain); domain++); /* skip leading spaces */
if (*domain == '#' || !*domain) continue; /* skip empty lines and comments */
for (len = strlen(domain); len && isspace(domain[len - 1]); len--); /* skip trailing spaces */
domain[len] = 0;
if ((rc = psl_str_to_utf8lower(domain, NULL, NULL, &lower)) != PSL_SUCCESS)
fprintf(stderr, "%s: Failed to convert to lowercase UTF-8 (%d)\n", domain, rc);
else if (mode == 1)
printf("%s: %d (%s)\n", domain, psl_is_public_suffix(psl, lower), lower);
else if (mode == 2)
printf("%s: %s\n", domain, psl_unregistrable_domain(psl, lower));
else if (mode == 3)
printf("%s: %s\n", domain, psl_registrable_domain(psl, lower));
else if (mode == 4) {
char *cookie_domain_lower;
if ((rc = psl_str_to_utf8lower(domain, NULL, NULL, &cookie_domain_lower)) != PSL_SUCCESS)
fprintf(stderr, "%s: Failed to convert cookie domain '%s' to lowercase UTF-8 (%d)\n", domain, cookie_domain, rc);
else
printf("%s: %d\n", domain, psl_is_cookie_domain_acceptable(psl, lower, cookie_domain));
free(cookie_domain_lower);
}
free(lower);
}
psl_free(psl);
exit(0);
}
}
if (mode == 1) {
for (; arg < argv + argc; arg++)
printf("%s: %d\n", *arg, psl_is_public_suffix(psl, *arg));
}
else if (mode == 2) {
for (; arg < argv + argc; arg++)
printf("%s: %s\n", *arg, psl_unregistrable_domain(psl, *arg));
}
else if (mode == 3) {
for (; arg < argv + argc; arg++)
printf("%s: %s\n", *arg, psl_registrable_domain(psl, *arg));
}
else if (mode == 4) {
for (; arg < argv + argc; arg++)
printf("%s: %d\n", *arg, psl_is_cookie_domain_acceptable(psl, *arg, cookie_domain));
}
else if (mode == 99) {
if (psl && psl != psl_builtin()) {
printf("suffixes: %d\n", psl_suffix_count(psl));
printf("exceptions: %d\n", psl_suffix_exception_count(psl));
printf("wildcards: %d\n", psl_suffix_wildcard_count(psl));
}
psl_free(psl);
psl = (psl_ctx_t *) psl_builtin();
if (psl) {
printf("builtin suffixes: %d\n", psl_suffix_count(psl));
printf("builtin exceptions: %d\n", psl_suffix_exception_count(psl));
printf("builtin wildcards: %d\n", psl_suffix_wildcard_count(psl));
printf("builtin filename: %s\n", psl_builtin_filename());
printf("builtin file time: %ld (%s)\n", psl_builtin_file_time(), time2str(psl_builtin_file_time()));
printf("builtin SHA1 file hash: %s\n", psl_builtin_sha1sum());
printf("builtin outdated: %d\n", psl_builtin_outdated());
} else
printf("No builtin PSL data available\n");
}
psl_free(psl);
return 0;
}
static void test_psl(void)
{
/* punycode generation: idn ?? */
/* octal code generation: echo -n "??" | od -b */
static const struct test_data {
const char
*domain;
int
result;
} test_data[] = {
{ "www.example.com", 0 },
{ "com.ar", 1 },
{ "www.com.ar", 0 },
{ "cc.ar.us", 1 },
{ ".cc.ar.us", 1 },
{ "www.cc.ar.us", 0 },
{ "www.ck", 0 }, /* exception from *.ck */
{ "abc.www.ck", 0 },
{ "xxx.ck", 1 },
{ "www.xxx.ck", 0 },
{ "\345\225\206\346\240\207", 1 }, /* xn--czr694b or ?? */
{ "www.\345\225\206\346\240\207", 0 },
/* some special test follow ('name' and 'forgot.his.name' are public, but e.g. his.name is not) */
{ "name", 1 },
{ ".name", 1 },
{ "his.name", 0 },
{ ".his.name", 0 },
{ "forgot.his.name", 1 },
{ ".forgot.his.name", 1 },
{ "whoever.his.name", 0 },
{ "whoever.forgot.his.name", 0 },
{ ".", 1 }, /* special case */
{ "", 1 }, /* special case */
{ NULL, 1 }, /* special case */
{ "adfhoweirh", 1 }, /* unknown TLD */
};
unsigned it;
int result, ver;
psl_ctx_t *psl;
psl = psl_load_file(PSL_FILE);
printf("loaded %d suffixes and %d exceptions\n", psl_suffix_count(psl), psl_suffix_exception_count(psl));
for (it = 0; it < countof(test_data); it++) {
const struct test_data *t = &test_data[it];
result = psl_is_public_suffix(psl, t->domain);
if (result == t->result) {
ok++;
} else {
failed++;
printf("psl_is_public_suffix(%s)=%d (expected %d)\n", t->domain, result, t->result);
}
}
/* do some checks to cover more code paths in libpsl */
psl_is_public_suffix(NULL, "xxx");
if ((ver = psl_check_version_number(0)) == 0) {
printf("psl_check_version_number(0) is 0\n");
failed++;
} else {
if (((result = psl_check_version_number(ver)) != ver)) {
printf("psl_check_version_number(%06X) is %06X\n", ver, result);
failed++;
}
if (((result = psl_check_version_number(ver - 1)) != 0)) {
printf("psl_check_version_number(%06X) is %06X\n", ver - 1, result);
failed++;
}
if (((result = psl_check_version_number(ver + 1)) != ver)) {
printf("psl_check_version_number(%06X) is %06X\n", ver, result);
failed++;
}
}
psl_str_to_utf8lower("www.example.com", "utf-8", "en", NULL);
psl_str_to_utf8lower(NULL, "utf-8", "en", NULL);
{
char *lower = NULL;
psl_str_to_utf8lower("www.example.com", NULL, "de", &lower);
free(lower); lower = NULL;
psl_str_to_utf8lower("\374bel.de", NULL, "de", &lower);
free(lower); lower = NULL;
psl_str_to_utf8lower("\374bel.de", "iso-8859-1", NULL, &lower);
free(lower); lower = NULL;
psl_str_to_utf8lower(NULL, "utf-8", "en", &lower);
free(lower); lower = NULL;
}
psl_get_version();
psl_dist_filename();
psl_builtin_filename();
psl_builtin_outdated();
psl_builtin_file_time();
psl_builtin_sha1sum();
psl_suffix_wildcard_count(NULL);
psl_suffix_wildcard_count(psl);
psl_suffix_wildcard_count(psl_builtin());
psl_suffix_count(NULL);
psl_suffix_exception_count(NULL);
psl_load_file(NULL);
psl_load_fp(NULL);
psl_registrable_domain(NULL, "");
psl_registrable_domain(psl, NULL);
psl_registrable_domain(psl, "www.example.com");
psl_unregistrable_domain(NULL, "");
psl_unregistrable_domain(psl, NULL);
psl_is_public_suffix2(NULL, "", PSL_TYPE_ANY);
psl_is_public_suffix2(psl, NULL, PSL_TYPE_ANY);
psl_free(psl);
}