This repository has been archived on 2020-05-25. You can view files and clone it, but cannot push or open issues or pull requests.
tdlib-fork/test/message_entities.cpp
levlam a4bd81b631 Fix test.
GitOrigin-RevId: 4e040bbe77f337c2a9fa094fbba9f71cf197fe8b
2019-10-03 18:30:43 +03:00

706 lines
32 KiB
C++
Raw Blame History

This file contains invisible Unicode characters

This file contains invisible Unicode characters that are indistinguishable to humans but may be processed differently by a computer. If you think that this is intentional, you can safely ignore this warning. Use the Escape button to reveal them.

This file contains Unicode characters that might be confused with other characters. If you think that this is intentional, you can safely ignore this warning. Use the Escape button to reveal them.

//
// Copyright Aliaksei Levin (levlam@telegram.org), Arseny Smirnov (arseny30@gmail.com) 2014-2019
//
// Distributed under the Boost Software License, Version 1.0. (See accompanying
// file LICENSE_1_0.txt or copy at http://www.boost.org/LICENSE_1_0.txt)
//
#include "td/telegram/MessageEntity.h"
#include "td/utils/format.h"
#include "td/utils/logging.h"
#include "td/utils/tests.h"
#include <algorithm>
REGISTER_TESTS(message_entities);
using namespace td;
static void check_mention(string str, std::vector<string> expected) {
auto result_slice = find_mentions(str);
std::vector<string> result;
for (auto &it : result_slice) {
result.push_back(it.str());
}
if (result != expected) {
LOG(FATAL) << tag("text", str) << tag("got", format::as_array(result))
<< tag("expected", format::as_array(expected));
}
}
TEST(MessageEntities, mention) {
check_mention("@mention", {"@mention"});
check_mention("@mention ", {"@mention"});
check_mention(" @mention", {"@mention"});
check_mention(" @mention ", {"@mention"});
check_mention("@abc @xyz @abc @xyz @xxx@yyy @ttt", {});
check_mention("@abcde @xyzxy @abcde @xyzxy @xxxxx@yyyyy @ttttt",
{"@abcde", "@xyzxy", "@abcde", "@xyzxy", "@xxxxx", "@ttttt"});
check_mention("no@mention", {});
check_mention("@n", {});
check_mention("@abcdefghijklmnopqrstuvwxyz123456", {"@abcdefghijklmnopqrstuvwxyz123456"});
check_mention("@abcdefghijklmnopqrstuvwxyz1234567", {});
check_mention("ะฝะตั‚@mention", {});
check_mention("@ya @gif @wiki @vid @bing @pic @bold @imdb @coub @like @vote @giff @cap ya cap @y @yar @bingg @bin",
{"@gif", "@wiki", "@vid", "@bing", "@pic", "@bold", "@imdb", "@coub", "@like", "@vote", "@bingg"});
};
static void check_bot_command(string str, std::vector<string> expected) {
auto result_slice = find_bot_commands(str);
std::vector<string> result;
for (auto &it : result_slice) {
result.push_back(it.str());
}
if (result != expected) {
LOG(FATAL) << tag("text", str) << tag("got", format::as_array(result))
<< tag("expected", format::as_array(expected));
}
}
TEST(MessageEntities, bot_command) {
// 1..64@3..32
check_bot_command("/abc", {"/abc"});
check_bot_command(" /abc", {"/abc"});
check_bot_command("/abc ", {"/abc"});
check_bot_command(" /abc ", {"/abc"});
check_bot_command("/a@abc", {"/a@abc"});
check_bot_command("/a@b", {});
check_bot_command("/@bfdsa", {});
check_bot_command("/test/", {});
}
static void check_hashtag(string str, std::vector<string> expected) {
auto result_slice = find_hashtags(str);
std::vector<string> result;
for (auto &it : result_slice) {
result.push_back(it.str());
}
if (result != expected) {
LOG(FATAL) << tag("text", str) << tag("got", format::as_array(result))
<< tag("expected", format::as_array(expected));
}
}
TEST(MessageEntities, hashtag) {
check_hashtag("", {});
check_hashtag("#", {});
check_hashtag("##", {});
check_hashtag("###", {});
check_hashtag("#a", {"#a"});
check_hashtag(" #a", {"#a"});
check_hashtag("#a ", {"#a"});
check_hashtag(" #ั ", {"#ั"});
check_hashtag(" ั#a ", {});
check_hashtag(" #a# ", {});
check_hashtag(" #123 ", {});
check_hashtag(" #123a ", {"#123a"});
check_hashtag(" #a123 ", {"#a123"});
check_hashtag(" #123a# ", {});
check_hashtag(" #" + string(300, '1'), {});
check_hashtag(" #" + string(256, '1'), {});
check_hashtag(" #" + string(256, '1') + "a ", {});
check_hashtag(" #" + string(255, '1') + "a", {"#" + string(255, '1') + "a"});
check_hashtag(" #" + string(255, '1') + "ะฏ", {"#" + string(255, '1') + "ะฏ"});
check_hashtag(" #" + string(255, '1') + "a" + string(255, 'b') + "# ", {});
check_hashtag("#a#b #c #d", {"#c", "#d"});
check_hashtag("#test", {"#test"});
check_hashtag(u8"\U0001F604\U0001F604\U0001F604\U0001F604 \U0001F604\U0001F604\U0001F604#" + string(200, '1') +
"ะžะžะž" + string(200, '2'),
{"#" + string(200, '1') + "ะžะžะž" + string(53, '2')});
check_hashtag(u8"#a\u2122", {"#a"});
}
static void check_cashtag(string str, std::vector<string> expected) {
auto result_slice = find_cashtags(str);
std::vector<string> result;
for (auto &it : result_slice) {
result.push_back(it.str());
}
if (result != expected) {
LOG(FATAL) << tag("text", str) << tag("got", format::as_array(result))
<< tag("expected", format::as_array(expected));
}
}
TEST(MessageEntities, cashtag) {
check_cashtag("", {});
check_cashtag("$", {});
check_cashtag("$$", {});
check_cashtag("$$$", {});
check_cashtag("$a", {});
check_cashtag(" $a", {});
check_cashtag("$a ", {});
check_cashtag(" $ั ", {});
check_cashtag("$ab", {});
check_cashtag("$abc", {});
check_cashtag("$", {});
check_cashtag("$A", {});
check_cashtag("$AB", {});
check_cashtag("$ะBC", {});
check_cashtag("$ะะ’ะก", {});
check_cashtag("$ABC", {"$ABC"});
check_cashtag("$ABCD", {"$ABCD"});
check_cashtag("$ABCDE", {"$ABCDE"});
check_cashtag("$ABCDEF", {"$ABCDEF"});
check_cashtag("$ABCDEFG", {"$ABCDEFG"});
check_cashtag("$ABCDEFGH", {"$ABCDEFGH"});
check_cashtag("$ABCDEFGHJ", {});
check_cashtag("$ABCDEFGH1", {});
check_cashtag(" $XYZ", {"$XYZ"});
check_cashtag("$XYZ ", {"$XYZ"});
check_cashtag(" $XYZ ", {"$XYZ"});
check_cashtag(" $$XYZ ", {});
check_cashtag(" $XYZ$ ", {});
check_cashtag(" $ABC1 ", {});
check_cashtag(" $1ABC ", {});
check_cashtag(" 1$ABC ", {});
check_cashtag(" ะ$ABC ", {});
check_cashtag("$ABC$DEF $GHI $KLM", {"$GHI", "$KLM"});
check_cashtag("$TEST", {"$TEST"});
check_cashtag(u8"$ABC\u2122", {"$ABC"});
check_cashtag(u8"\u2122$ABC", {"$ABC"});
check_cashtag(u8"\u2122$ABC\u2122", {"$ABC"});
}
static void check_is_email_address(string str, bool expected) {
bool result = is_email_address(str);
LOG_IF(FATAL, result != expected) << "Expected " << expected << " as result of is_email_address(" << str << ")";
}
TEST(MessageEntities, is_email_address) {
check_is_email_address("telegram.org", false);
check_is_email_address("security@telegram.org", true);
check_is_email_address("security.telegram.org", false);
check_is_email_address("", false);
check_is_email_address("@", false);
check_is_email_address("A@a.a.a.ab", true);
check_is_email_address("A@a.ab", true);
check_is_email_address("Test@aa.aa.aa.aa", true);
check_is_email_address("Test@test.abd", true);
check_is_email_address("a@a.a.a.ab", true);
check_is_email_address("test@test.abd", true);
check_is_email_address("test@test.com", true);
check_is_email_address("test.abd", false);
check_is_email_address("a.ab", false);
check_is_email_address("a.bc@d.ef", true);
vector<string> bad_userdatas = {"",
"a.a.a.a.a.a.a.a.a.a.a.a",
"+.+.+.+.+.+",
"*.a.a",
"a.*.a",
"a.a.*",
"a.a.",
"a.a.abcdefghijklmnopqrstuvwxyz0123456789",
"a.abcdefghijklmnopqrstuvwxyz0.a",
"abcdefghijklmnopqrstuvwxyz0.a.a"};
vector<string> good_userdatas = {"a.a.a.a.a.a.a.a.a.a.a",
"a+a+a+a+a+a+a+a+a+a+a",
"+.+.+.+.+._",
"aozAQZ0-5-9_+-aozAQZ0-5-9_.aozAQZ0-5-9_.-._.+-",
"a.a.a",
"a.a.abcdefghijklmnopqrstuvwxyz012345678",
"a.abcdefghijklmnopqrstuvwxyz.a",
"a..a",
"abcdefghijklmnopqrstuvwxyz.a.a",
".a.a"};
vector<string> bad_domains = {"",
".",
"abc",
"localhost",
"a.a.a.a.a.a.a.ab",
".......",
"a.a.a.a.a.a+ab",
"a+a.a.a.a.a.ab",
"a.a.a.a.a.a.a",
"a.a.a.a.a.a.abcdefg",
"a.a.a.a.a.a.ab0yz",
"a.a.a.a.a.a.ab9yz",
"a.a.a.a.a.a.ab-yz",
"a.a.a.a.a.a.ab_yz",
"a.a.a.a.a.a.ab*yz",
".ab",
".a.ab",
"a..ab",
"a.a.a..a.ab",
".a.a.a.a.ab",
"abcdefghijklmnopqrstuvwxyz01234.ab",
"ab0cd.abd.aA*sd.0.9.0-9.ABOYZ",
"ab*cd.abd.aAasd.0.9.0-9.ABOYZ",
"ab0cd.abd.aAasd.0.9.0*9.ABOYZ",
"*b0cd.ab_d.aA-sd.0.9.0-9.ABOYZ",
"ab0c*.ab_d.aA-sd.0.9.0-9.ABOYZ",
"ab0cd.ab_d.aA-sd.0.9.0-*.ABOYZ",
"ab0cd.ab_d.aA-sd.0.9.*-9.ABOYZ",
"-b0cd.ab_d.aA-sd.0.9.0-9.ABOYZ",
"ab0c-.ab_d.aA-sd.0.9.0-9.ABOYZ",
"ab0cd.ab_d.aA-sd.-.9.0-9.ABOYZ",
"ab0cd.ab_d.aA-sd.0.9.--9.ABOYZ",
"ab0cd.ab_d.aA-sd.0.9.0--.ABOYZ",
"_b0cd.ab_d.aA-sd.0.9.0-9.ABOYZ",
"ab0c_.ab_d.aA-sd.0.9.0-9.ABOYZ",
"ab0cd.ab_d.aA-sd._.9.0-9.ABOYZ",
"ab0cd.ab_d.aA-sd.0.9._-9.ABOYZ",
"ab0cd.ab_d.aA-sd.0.9.0-_.ABOYZ",
"-.ab_d.aA-sd.0.9.0-9.ABOYZ",
"ab0cd.ab_d.-.0.9.0-9.ABOYZ",
"ab0cd.ab_d.aA-sd.0.9.-.ABOYZ",
"_.ab_d.aA-sd.0.9.0-9.ABOYZ",
"ab0cd.ab_d._.0.9.0-9.ABOYZ",
"ab0cd.ab_d.aA-sd.0.9._.ABOYZ"};
vector<string> good_domains = {"a.a.a.a.a.a.ab",
"a.a.a.a.a.a.abcdef",
"a.a.a.a.a.a.aboyz",
"a.a.a.a.a.a.ABOYZ",
"a.a.a.a.a.a.AbOyZ",
"abcdefghijklmnopqrstuvwxyz0123.ab",
"ab0cd.ab_d.aA-sd.0.9.0-9.ABOYZ",
"A.Z.aA-sd.a.z.0-9.ABOYZ"};
for (auto &userdata : bad_userdatas) {
for (auto &domain : bad_domains) {
check_is_email_address(userdata + '@' + domain, false);
check_is_email_address(userdata + domain, false);
}
for (auto &domain : good_domains) {
check_is_email_address(userdata + '@' + domain, false);
check_is_email_address(userdata + domain, false);
}
}
for (auto &userdata : good_userdatas) {
for (auto &domain : bad_domains) {
check_is_email_address(userdata + '@' + domain, false);
check_is_email_address(userdata + domain, false);
}
for (auto &domain : good_domains) {
check_is_email_address(userdata + '@' + domain, true);
check_is_email_address(userdata + domain, false);
}
}
}
static void check_url(string str, std::vector<string> expected_urls,
std::vector<string> expected_email_addresses = {}) {
auto result_slice = find_urls(str);
std::vector<string> result_urls;
std::vector<string> result_email_addresses;
for (auto &it : result_slice) {
if (!it.second) {
result_urls.push_back(it.first.str());
} else {
result_email_addresses.push_back(it.first.str());
}
}
if (result_urls != expected_urls) {
LOG(FATAL) << tag("text", str) << tag("got", format::as_array(result_urls))
<< tag("expected", format::as_array(expected_urls));
}
if (result_email_addresses != expected_email_addresses) {
LOG(FATAL) << tag("text", str) << tag("got", format::as_array(result_email_addresses))
<< tag("expected", format::as_array(expected_email_addresses));
}
}
TEST(MessageEntities, url) {
check_url("telegram.org", {"telegram.org"});
check_url("(telegram.org)", {"telegram.org"});
check_url("\ntelegram.org)", {"telegram.org"});
check_url(" telegram.org)", {"telegram.org"});
check_url(".telegram.org)", {});
check_url("()telegram.org/?q=()", {"telegram.org/?q=()"});
check_url("\"telegram.org\"", {"telegram.org"});
check_url(" telegram. org. www. com... telegram.org... ...google.com...", {"telegram.org"});
check_url(" telegram.org ", {"telegram.org"});
check_url("ะขะฐะบะพะน ัะฐะนั‚: http://www.google.com ะธะปะธ ั‚ะฐะบะพะน telegram.org ", {"http://www.google.com", "telegram.org"});
check_url(" telegram.org. ", {"telegram.org"});
check_url("http://google,.com", {});
check_url("http://telegram.org/?asd=123#123.", {"http://telegram.org/?asd=123#123"});
check_url("[http://google.com](test)", {"http://google.com"});
check_url("", {});
check_url(".", {});
check_url("http://@google.com", {});
check_url("http://@goog.com", {}); // TODO: server fix
check_url("http://@@google.com", {"http://@@google.com"});
check_url("http://a@google.com", {"http://a@google.com"});
check_url("http://test@google.com", {"http://test@google.com"});
check_url("google.com:แช‰แช‰แช‰แช‰แช‰", {"google.com"});
check_url("https://telegram.org", {"https://telegram.org"});
check_url("http://telegram.org", {"http://telegram.org"});
check_url("ftp://telegram.org", {"ftp://telegram.org"});
check_url("ftps://telegram.org", {});
check_url("sftp://telegram.org", {"sftp://telegram.org"});
check_url("hTtPs://telegram.org", {"hTtPs://telegram.org"});
check_url("HTTP://telegram.org", {"HTTP://telegram.org"});
check_url("ะฐHTTP://telegram.org", {"HTTP://telegram.org"});
check_url("sHTTP://telegram.org", {});
check_url("://telegram.org", {});
check_url("google.com:แช€แช€", {"google.com"});
check_url(
"http://"
"abcdefghijkabcdefghijkabcdefghijkabcdefghijkabcdefghijkabcdefghijkabcdefghijkabcdefghijkabcdefghijkabcdefghijkab"
"cdefghijkabcdefghijkabcdefghijkabcdefghijkabcdefghijkabcdefghijkabcdefghijkabcdefghijkabcdefghijkabcdefghijkabcd"
"efghijkabcdefghijkabcdefghijkabcdefghijkabcdefghijkabcdefghijkabcdefghijkabcdefghijkabcdefghijkabcdefghijkabcdef"
"ghijkabcdefghijkabcdefghijkabcdefghijkabcdefghijkabcdefghijkabcdefghijkabcdefghijkabcdefghijkabcdefghijkabcdefgh"
"ijkabcdefghijkabcdefghijkabcdefghijkabcdefghijkabcdefghijkabcdefghijkabcdefghijkabcdefghijkabcdefghijkabcdefghij"
"kabcdefghijkabcdefghijkabcdefghijkabcdefghijkabcdefghijkabcdefghijkabcdefghijkabcdefghijkabcdefghijkabcdefghijka"
"bcdefghijkabcdefghijkabcdefghijkabcdefghijkabcdefghijkabcdefghijkabcdefghijkabcdefghijkabcdefghijkabcdefghijkabc"
"defghijkabcdefghijkabcdefghijkabcdefghijkabcdefghijkabcdefghijkabcdefghijkabcdefghijk.com",
{});
check_url("http://โ€‰โ€‰.com", {});
check_url("URL: โ€‰โ€‰โ€‰โ€‰.com", {});
check_url("URL:โ€‰.com", {});
check_url(".com", {});
check_url("http://โ€‰โ€‰.", {});
check_url("http://.", {});
check_url("http://.com", {});
check_url("http:// .", {});
check_url(",ahttp://google.com", {"http://google.com"});
check_url(".ahttp://google.com", {});
check_url("http://1.0", {});
check_url("http://a.0", {});
check_url("http://a.a", {});
check_url("google.com:1#ab c", {"google.com:1#ab"});
check_url("google.com:1#", {"google.com:1"});
check_url("google.com:1#1", {"google.com:1#1"});
check_url("google.com:00000001/abs", {"google.com:00000001/abs"});
check_url("google.com:000000065535/abs", {"google.com:000000065535/abs"});
check_url("google.com:000000065536/abs", {"google.com"});
check_url("google.com:000000080/abs", {"google.com:000000080/abs"});
check_url("google.com:0000000/abs", {"google.com"});
check_url("google.com:0/abs", {"google.com"});
check_url("google.com:/abs", {"google.com"});
check_url("google.com:65535", {"google.com:65535"});
check_url("google.com:65536", {"google.com"});
check_url("google.com:99999", {"google.com"});
check_url("google.com:100000", {"google.com"});
check_url("127.001", {});
check_url("127.0.0.1", {"127.0.0.1"});
check_url("127.0.0.01", {});
check_url("127.0.0.256", {});
check_url("127.0.0.300", {});
check_url("127.0.0.1000", {});
check_url("127.0.0.260", {});
check_url("1.0", {});
check_url("www.๐Ÿค™.tk", {"www.๐Ÿค™.tk"});
check_url("a.ab", {});
check_url("test.abd", {});
check_url("ะขะตะกั‚.ะœะพัะบะฒะฐ", {});
check_url("ะขะตะกั‚.ะœะพะกะบะ’ฮ‘", {});
check_url("ะขะตะกั‚.ะœะพะกะบะ’ะฐ", {"ะขะตะกั‚.ะœะพะกะบะ’ะฐ"});
check_url("ะขะตะกั‚.ะœะพะกะบะ’ะฐั‡", {});
check_url("http://ร€ะขะตะกั‚.ะœะพะกะบะ’ะฐั‡", {"http://ร€ะขะตะกั‚.ะœะพะกะบะ’ะฐั‡"});
check_url("ร€ร.com. ร€ร.com.", {"ร€ร.com", "ร€ร.com"});
check_url("ร€ร.com,ร€ร.com.", {"ร€ร.com", "ร€ร.com"});
check_url("teiegram.org", {});
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(){}"});
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", {});
check_url("http://test_.google.com", {"http://test_.google.com"});
check_url("http://google_.com", {});
check_url("http://google._com_", {});
check_url("http://[2001:4860:0:2001::68]/", {}); // TODO
check_url("test.abd", {});
check_url("/.b/..a @.....@/. a.ba", {"a.ba"});
check_url("bbbbbbbbbbbbbb.@.@", {});
check_url("http://google.com/", {"http://google.com/"});
check_url("http://google.com?", {"http://google.com"});
check_url("http://google.com#", {"http://google.com"});
check_url("http://google.com##", {"http://google.com##"});
check_url("http://google.com/?", {"http://google.com/"});
check_url("https://www.google.com/ab,", {"https://www.google.com/ab"});
check_url("@.", {});
check_url(
"a.b.google.com dfsknnfs gsdfgsg http://cรณduia.de/ dffdg,\" 12)(cpia.de/())(\" http://ะณั€ะธัˆะบะฐ.ั€ั„/ sdufhdf "
"http://xn--80afpi2a3c.xn--p1ai/ I have a good time.Thanks, guys!\n\n(hdfughidufhgdis)go#ogle.com ะณั€ะธัˆะบะฐ.ั€ั„ "
"hsighsdf gi ะฟะพั‡ั‚ะฐ.ั€ั„\n\nโœชdf.ws/123 "
"xn--80afpi2a3c.xn--p1ai\n\nhttp://foo.com/blah_blah\nhttp://foo.com/blah_blah/\n(Something like "
"http://foo.com/blah_blah)\nhttp://foo.com/blah_blah_(wikipedi8989a_ะ’ะฐัั)\n(Something like "
"http://foo.com/blah_blah_(ะกั‚ะฐะบะฐะฝ_007))\nhttp://foo.com/blah_blah.\nhttp://foo.com/blah_blah/.\n<http://foo.com/"
"blah_blah>\n<http://fo@@@@@@@@@^%#*@^&@$#*@#%^*&!^#o.com/blah_blah/>\nhttp://foo.com/blah_blah,\nhttp://"
"www.example.com/wpstyle/?p=364.\nhttp://โœชdf.ws/123\nrdar://1234\nrdar:/1234\nhttp://"
"userid:password@example.com:8080\nhttp://userid@example.com\nhttp://userid@example.com:8080\nhttp://"
"userid:password@example.com\nhttp://example.com:8080 "
"x-yojimbo-item://6303E4C1-xxxx-45A6-AB9D-3A908F59AE0E\nmessage://"
"%3c330e7f8409726r6a4ba78dkf1fd71420c1bf6ff@mail.gmail.com%3e\nhttp://โžก๏ธ.ws/ไจน\nwww.โžก๏ธ.ws/"
"ไจน\n<tag>http://example.com</tag>\nJust a www.example.com "
"link.\n\nโžก๏ธ.ws/"
"ไจน\n\nabcdefghijklmnopqrstuvwxyz0123456789qwe_sdfsdf.aweawe-sdfs.com\nwww.๐Ÿค™.tk:1\ngoogle.com:"
"แช‰แช‰แช‰แช‰\ngoogle."
"com:แช€แช€\nhttp://โ€‰โ€‰.com\nURL: โ€‰โ€‰โ€‰โ€‰.com\nURL: "
".com\n\ngoogle.com?qwe\ngoogle.com#qwe\ngoogle.com/?\ngoogle.com/#\ngoogle.com?\ngoogle.com#\n",
{"a.b.google.com",
"http://cรณduia.de/",
"cpia.de/()",
"http://ะณั€ะธัˆะบะฐ.ั€ั„/",
"http://xn--80afpi2a3c.xn--p1ai/",
"ะณั€ะธัˆะบะฐ.ั€ั„",
"ะฟะพั‡ั‚ะฐ.ั€ั„",
"โœชdf.ws/123",
"xn--80afpi2a3c.xn--p1ai",
"http://foo.com/blah_blah",
"http://foo.com/blah_blah/",
"http://foo.com/blah_blah",
"http://foo.com/blah_blah_(wikipedi8989a_ะ’ะฐัั)",
"http://foo.com/blah_blah_(ะกั‚ะฐะบะฐะฝ_007)",
"http://foo.com/blah_blah",
"http://foo.com/blah_blah/",
"http://foo.com/blah_blah",
"http://foo.com/blah_blah",
"http://www.example.com/wpstyle/?p=364",
"http://โœชdf.ws/123",
"http://userid:password@example.com:8080",
"http://userid@example.com",
"http://userid@example.com:8080",
"http://userid:password@example.com",
"http://example.com:8080",
"http://โžก๏ธ.ws/ไจน",
"www.โžก๏ธ.ws/ไจน",
"http://example.com",
"www.example.com",
"โžก๏ธ.ws/ไจน",
"abcdefghijklmnopqrstuvwxyz0123456789qwe_sdfsdf.aweawe-sdfs.com",
"www.๐Ÿค™.tk:1",
"google.com",
"google.com",
"google.com?qwe",
"google.com#qwe",
"google.com/",
"google.com/#",
"google.com",
"google.com"});
check_url("https://t.โ€ฆ", {});
check_url("('http://telegram.org/a-b/?br=ie&lang=en',)", {"http://telegram.org/a-b/?br=ie&lang=en"});
check_url("https://ai.telegram.org/bot%20bot/test-...", {"https://ai.telegram.org/bot%20bot/test-"});
check_url("a.bc@c.com", {}, {"a.bc@c.com"});
check_url("https://a.bc@c.com", {"https://a.bc@c.com"});
check_url("https://a.de[bc@c.com", {"https://a.de"}, {"bc@c.com"});
check_url("https://a.de/bc@c.com", {"https://a.de/bc@c.com"});
check_url("https://a.de]bc@c.com", {"https://a.de"}, {"bc@c.com"});
check_url("https://a.de{bc@c.com", {"https://a.de"}, {"bc@c.com"});
check_url("https://a.de}bc@c.com", {"https://a.de"}, {"bc@c.com"});
check_url("https://a.de(bc@c.com", {"https://a.de"}, {"bc@c.com"});
check_url("https://a.de)bc@c.com", {"https://a.de"}, {"bc@c.com"});
check_url("https://a.de\\bc@c.com", {"https://a.de\\bc@c.com"});
check_url("https://a.de'bc@c.com", {"https://a.de"}, {"bc@c.com"});
check_url("https://a.de`bc@c.com", {"https://a.de"}, {"bc@c.com"});
check_url("https://a.bc:de.fg@c.com", {"https://a.bc:de.fg@c.com"});
check_url("https://a:h.bc:de.fg@c.com", {"https://a:h.bc:de.fg@c.com"});
check_url("https://abc@c.com", {"https://abc@c.com"});
check_url("https://de[bc@c.com", {}, {"bc@c.com"});
check_url("https://de/bc@c.com", {});
check_url("https://de]bc@c.com", {}, {"bc@c.com"});
check_url("https://de{bc@c.com", {}, {"bc@c.com"});
check_url("https://de}bc@c.com", {}, {"bc@c.com"});
check_url("https://de(bc@c.com", {}, {"bc@c.com"});
check_url("https://de)bc@c.com", {}, {"bc@c.com"});
check_url("https://de\\bc@c.com", {"https://de\\bc@c.com"});
check_url("https://de'bc@c.com", {}, {"bc@c.com"});
check_url("https://de`bc@c.com", {}, {"bc@c.com"});
check_url("https://bc:defg@c.com", {"https://bc:defg@c.com"});
check_url("https://a:hbc:defg@c.com", {"https://a:hbc:defg@c.com"});
check_url("https://a.bc@test.com:cd.com", {"https://a.bc@test.com", "cd.com"});
check_url("telegram.Org", {});
check_url("telegram.ORG", {"telegram.ORG"});
check_url("a.b.c.com.a.b.c", {});
check_url("File '/usr/views.py'", {}); // TODO server fix
check_url("@views.py'", {}); // TODO server fix
check_url("#views.py'", {}); // TODO server fix
check_url("/views.py'", {}); // TODO server fix
check_url(".views.py", {});
check_url("'views.py'", {"views.py"});
check_url("bug.http://test.com/test/+#", {}); // TODO {"http://test.com/test/+#"}
check_url("//AB.C.D.E.F.GH//", {});
check_url("<http://www.ics.uci.edu/pub/ietf/uri/historical.html#WARNING>",
{"http://www.ics.uci.edu/pub/ietf/uri/historical.html#WARNING"});
check_url("Look :test@example.com", {":test@example.com"}, {}); // TODO {}, {"test@example.com"}
check_url("Look mailto:test@example.com", {}, {"test@example.com"});
check_url("http://test.com#a", {"http://test.com#a"});
check_url("http://test.com#", {"http://test.com"});
check_url("http://test.com?#", {"http://test.com?#"});
check_url("http://test.com/?#", {"http://test.com/?#"});
check_url("https://t.me/abcdefโ€ฆ", {"https://t.me/abcdef"});
check_url("https://t.meโ€ฆ", {"https://t.me"});
check_url("https://t.mโ€ฆ", {});
check_url("https://t.โ€ฆ", {});
check_url("https://tโ€ฆ", {});
check_url("๐Ÿ‘‰http://ab.com/cdefgh-1IJ", {"http://ab.com/cdefgh-1IJ"});
check_url("...๐Ÿ‘‰http://ab.com/cdefgh-1IJ", {}); // TODO
}
static void check_fix_formatted_text(string str, vector<MessageEntity> entities, string expected_str,
vector<MessageEntity> expected_entities, bool allow_empty, bool skip_new_entities,
bool skip_bot_commands, bool for_draft) {
ASSERT_TRUE(fix_formatted_text(str, entities, allow_empty, skip_new_entities, skip_bot_commands, for_draft).is_ok());
ASSERT_STREQ(expected_str, str);
ASSERT_EQ(expected_entities, entities);
}
static void check_fix_formatted_text(string str, vector<MessageEntity> entities, bool allow_empty,
bool skip_new_entities, bool skip_bot_commands, bool for_draft) {
ASSERT_TRUE(
fix_formatted_text(str, entities, allow_empty, skip_new_entities, skip_bot_commands, for_draft).is_error());
}
TEST(MessageEntities, fix_formatted_text) {
string str;
string fixed_str;
for (auto i = 0; i <= 32; i++) {
str += static_cast<char>(i);
if (i != 13) {
if (i != 10) {
fixed_str += ' ';
} else {
fixed_str += str.back();
}
}
}
check_fix_formatted_text(str, {}, "", {}, true, true, true, true);
check_fix_formatted_text(str, {}, "", {}, true, true, false, true);
check_fix_formatted_text(str, {}, "", {}, true, false, true, true);
check_fix_formatted_text(str, {}, "", {}, true, false, false, true);
check_fix_formatted_text(str, {}, "", {}, true, false, false, false);
check_fix_formatted_text(str, {}, false, false, false, false);
check_fix_formatted_text(str, {}, false, false, false, true);
str += "a \r\n ";
fixed_str += "a \n ";
for (int32 i = 33; i <= 35; i++) {
vector<MessageEntity> entities;
entities.emplace_back(MessageEntity::Type::Bold, 0, i);
vector<MessageEntity> fixed_entities;
if (i != 33) {
fixed_entities = entities;
fixed_entities.back().length = i - 1;
}
check_fix_formatted_text(str, entities, fixed_str, fixed_entities, true, false, false, true);
string expected_str;
if (i != 33) {
fixed_entities = entities;
fixed_entities.back().length = 33;
expected_str = fixed_str.substr(0, 33);
} else {
fixed_entities.clear();
expected_str = "a";
}
check_fix_formatted_text(str, entities, expected_str, fixed_entities, false, false, false, false);
}
str = "๐Ÿ‘‰ ๐Ÿ‘‰";
for (int i = 0; i < 10; i++) {
vector<MessageEntity> entities;
entities.emplace_back(MessageEntity::Type::Bold, i, 1);
if (i == 0 || i == 1 || i == 3 || i == 4) {
check_fix_formatted_text(str, entities, true, true, true, true);
check_fix_formatted_text(str, entities, false, false, false, false);
} else {
check_fix_formatted_text(str, entities, str, {}, true, true, true, true);
check_fix_formatted_text(str, entities, str, {}, false, false, false, false);
}
}
str = " /test @abaca #ORD $ABC telegram.org ";
for (auto for_draft : {false, true}) {
int32 shift = for_draft ? 2 : 0;
string expected_str = for_draft ? str : str.substr(2, str.size() - 3);
for (auto skip_new_entities : {false, true}) {
for (auto skip_bot_commands : {false, true}) {
vector<MessageEntity> entities;
if (!skip_new_entities) {
if (!skip_bot_commands) {
entities.emplace_back(MessageEntity::Type::BotCommand, shift, 5);
}
entities.emplace_back(MessageEntity::Type::Mention, shift + 6, 6);
entities.emplace_back(MessageEntity::Type::Hashtag, shift + 13, 4);
entities.emplace_back(MessageEntity::Type::Cashtag, shift + 18, 4);
entities.emplace_back(MessageEntity::Type::Url, shift + 24, 12);
}
check_fix_formatted_text(str, {}, expected_str, entities, true, skip_new_entities, skip_bot_commands,
for_draft);
check_fix_formatted_text(str, {}, expected_str, entities, false, skip_new_entities, skip_bot_commands,
for_draft);
}
}
}
str = "aba \r\n caba ";
for (int32 length = 1; length <= 3; length++) {
for (int32 offset = 0; static_cast<size_t>(offset + length) <= str.size(); offset++) {
for (auto type : {MessageEntity::Type::Bold, MessageEntity::Type::Url, MessageEntity::Type::TextUrl,
MessageEntity::Type::MentionName}) {
for (auto for_draft : {false, true}) {
fixed_str = for_draft ? "aba \n caba " : "aba \n caba";
auto fixed_length = offset <= 4 && offset + length >= 5 ? length - 1 : length;
auto fixed_offset = offset >= 5 ? offset - 1 : offset;
if (static_cast<size_t>(fixed_offset) >= fixed_str.size()) {
fixed_length = 0;
}
while (static_cast<size_t>(fixed_offset + fixed_length) > fixed_str.size()) {
fixed_length--;
}
vector<MessageEntity> entities;
entities.emplace_back(type, offset, length);
vector<MessageEntity> fixed_entities;
if (fixed_length > 0) {
for (auto i = 0; i < length; i++) {
if (str[offset + i] != '\r' && str[offset + i] != '\n' &&
(str[offset + i] != ' ' || type == MessageEntity::Type::TextUrl ||
type == MessageEntity::Type::MentionName)) {
fixed_entities.emplace_back(type, fixed_offset, fixed_length);
break;
}
}
}
check_fix_formatted_text(str, entities, fixed_str, fixed_entities, true, false, false, for_draft);
}
}
}
}
str = "aba caba";
for (int32 length = -10; length <= 10; length++) {
for (int32 offset = -10; offset <= 10; offset++) {
vector<MessageEntity> entities;
entities.emplace_back(MessageEntity::Type::Bold, offset, length);
vector<MessageEntity> fixed_entities;
if (length > 0 && offset >= 0 && static_cast<size_t>(length + offset) <= str.size() &&
(length >= 2 || offset != 3)) {
fixed_entities.emplace_back(MessageEntity::Type::Bold, offset, length);
}
check_fix_formatted_text(str, entities, str, fixed_entities, true, false, false, false);
check_fix_formatted_text(str, entities, str, fixed_entities, false, false, false, true);
}
}
str = "aba caba";
for (int32 length = 1; length <= 7; length++) {
for (int32 offset = 0; offset <= 8 - length; offset++) {
for (int32 length2 = 1; length2 <= 7; length2++) {
for (int32 offset2 = 0; offset2 <= 8 - length2; offset2++) {
if (offset != offset2) {
vector<MessageEntity> entities;
entities.emplace_back(MessageEntity::Type::TextUrl, offset, length);
entities.emplace_back(MessageEntity::Type::TextUrl, offset2, length2);
vector<MessageEntity> fixed_entities = entities;
std::sort(fixed_entities.begin(), fixed_entities.end());
if (fixed_entities[0].offset + fixed_entities[0].length > fixed_entities[1].offset) {
fixed_entities.pop_back();
}
check_fix_formatted_text(str, entities, str, fixed_entities, false, false, false, false);
}
}
}
}
}
}