tdlight-team/tdlight
tdlight-team/tdlight
1
1
Fork 0
Code Issues Pull Requests Projects Releases Wiki Activity

Optimize fix_url.

Browse Source
This commit is contained in:
levlam 2022-08-20 13:44:55 +03:00
parent 0649b7e2fa
commit 4543c485a7
2 changed files with 56 additions and 42 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

65
td/telegram/MessageEntity.cpp
View File

@ -1181,11 +1181,13 @@ static Slice fix_url(Slice str) {
}
domain.truncate(domain.rfind(':'));
if (domain.size() == 12 && (domain[0] == 't' || domain[0] == 'T')) {
string domain_lower = domain.str();
to_lower_inplace(domain_lower);
if (domain_lower == "teiegram.org") {
return Slice();
}
}
int32 balance[3] = {0, 0, 0};
size_t path_pos;
@ -1220,42 +1222,44 @@ static Slice fix_url(Slice str) {
}
full_url.remove_suffix(path.size() - path_pos);
vector<Slice> domain_parts = full_split(domain, '.');
if (domain_parts.size() <= 1) {
return Slice();
}
bool is_ipv4 = domain_parts.size() == 4;
size_t prev = 0;
size_t domain_part_count = 0;
bool has_non_digit = false;
for (auto &part : domain_parts) {
if (part.empty() || part.size() >= 64) {
bool is_ipv4 = true;
for (size_t i = 0; i <= domain.size(); i++) {
if (i == domain.size() || domain[i] == '.') {
auto part_size = i - prev;
if (part_size == 0 || part_size >= 64 || domain[i - 1] == '-') {
return Slice();
}
if (part.back() == '-') {
return Slice();
}
if (!has_non_digit) {
if (part.size() > 3) {
if (is_ipv4) {
if (part_size > 3) {
is_ipv4 = false;
}
for (auto c : part) {
if (!is_digit(c)) {
if (part_size == 3 &&
(domain[prev] >= '3' || (domain[prev] == '2' && (domain[prev + 1] >= '6' ||
(domain[prev + 1] == '5' && domain[prev + 2] >= '6'))))) {
is_ipv4 = false;
}
if (domain[prev] == '0' && part_size >= 2) {
is_ipv4 = false;
}
}
domain_part_count++;
if (i != domain.size()) {
prev = i + 1;
}
} else if (!is_digit(domain[i])) {
is_ipv4 = false;
has_non_digit = true;
}
}
if (part.size() == 3 &&
(part[0] >= '3' || (part[0] == '2' && (part[1] >= '6' || (part[1] == '5' && part[2] >= '6'))))) {
is_ipv4 = false;
}
if (part[0] == '0' && part.size() >= 2) {
is_ipv4 = false;
}
}
if (domain_part_count == 1) {
return Slice();
}
if (is_ipv4) {
if (is_ipv4 && domain_part_count == 4) {
return full_url;
}
@ -1263,7 +1267,7 @@ static Slice fix_url(Slice str) {
return Slice();
}
auto tld = domain_parts.back();
auto tld = domain.substr(prev);
if (utf8_length(tld) <= 1) {
return Slice();
}
@ -1290,9 +1294,14 @@ static Slice fix_url(Slice str) {
}
}
domain_parts.pop_back();
if (domain_parts.back().find('_') < domain_parts.back().size()) {
CHECK(prev > 0);
prev--;
while (prev-- > 0) {
if (domain[prev] == '_') {
return Slice();
} else if (domain[prev] == '.') {
break;
}
}
return full_url;

7
test/message_entities.cpp
View File

@ -562,7 +562,8 @@ TEST(MessageEntities, url) {
check_url("http://ÀТеСт.МоСкВач", {"http://ÀТеСт.МоСкВач"});
check_url("ÀÁ.com. ÀÁ.com.", {"ÀÁ.com", "ÀÁ.com"});
check_url("ÀÁ.com,ÀÁ.com.", {"ÀÁ.com", "ÀÁ.com"});
check_url("teiegram.org", {});
check_url("teiegram.org/test", {});
check_url("TeiegraM.org/test", {});
check_url("http://test.google.com/?q=abc()}[]def", {"http://test.google.com/?q=abc()"});
check_url("http://test.google.com/?q=abc([{)]}def", {"http://test.google.com/?q=abc([{)]}def"});
check_url("http://test.google.com/?q=abc(){}]def", {"http://test.google.com/?q=abc(){}"});
@ -713,6 +714,10 @@ TEST(MessageEntities, url) {
check_url("a:b#@gmail.com", {});
check_url("a!:b@gmail.com", {"a!:b@gmail.com"});
check_url("a:b!@gmail.com", {"a:b!@gmail.com"});
check_url("http://test_.com", {});
check_url("test_.com", {});
check_url("_test.com", {});
check_url("_.test.com", {"_.test.com"});
}
static void check_fix_formatted_text(td::string str, td::vector<td::MessageEntity> entities,