2018-12-31 20:04:05 +01:00
|
|
|
|
//
|
2022-12-31 22:28:08 +01:00
|
|
|
|
// Copyright Aliaksei Levin (levlam@telegram.org), Arseny Smirnov (arseny30@gmail.com) 2014-2023
|
2018-12-31 20:04:05 +01:00
|
|
|
|
//
|
|
|
|
|
// 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)
|
|
|
|
|
//
|
2022-10-26 07:53:09 +02:00
|
|
|
|
#include "td/telegram/CustomEmojiId.h"
|
2018-12-31 20:04:05 +01:00
|
|
|
|
#include "td/telegram/MessageEntity.h"
|
2022-08-19 14:05:49 +02:00
|
|
|
|
#include "td/telegram/UserId.h"
|
|
|
|
|
|
2021-08-05 00:23:16 +02:00
|
|
|
|
#include "td/utils/algorithm.h"
|
2019-10-03 01:31:06 +02:00
|
|
|
|
#include "td/utils/common.h"
|
2018-12-31 20:04:05 +01:00
|
|
|
|
#include "td/utils/format.h"
|
|
|
|
|
#include "td/utils/logging.h"
|
2019-12-23 18:48:30 +01:00
|
|
|
|
#include "td/utils/misc.h"
|
2020-02-21 14:22:40 +01:00
|
|
|
|
#include "td/utils/Random.h"
|
2019-10-22 19:01:15 +02:00
|
|
|
|
#include "td/utils/Slice.h"
|
2021-05-17 14:21:11 +02:00
|
|
|
|
#include "td/utils/SliceBuilder.h"
|
2018-12-31 20:04:05 +01:00
|
|
|
|
#include "td/utils/tests.h"
|
2019-09-20 01:14:15 +02:00
|
|
|
|
#include "td/utils/utf8.h"
|
2018-12-31 20:04:05 +01:00
|
|
|
|
|
2019-02-04 04:50:47 +01:00
|
|
|
|
#include <algorithm>
|
2021-08-05 00:23:16 +02:00
|
|
|
|
#include <utility>
|
2019-02-04 04:50:47 +01:00
|
|
|
|
|
2019-09-26 17:36:45 +02:00
|
|
|
|
static void check_mention(const td::string &str, const td::vector<td::string> &expected) {
|
2019-09-20 01:22:42 +02:00
|
|
|
|
auto result_slice = td::find_mentions(str);
|
|
|
|
|
td::vector<td::string> result;
|
2018-12-31 20:04:05 +01:00
|
|
|
|
for (auto &it : result_slice) {
|
|
|
|
|
result.push_back(it.str());
|
|
|
|
|
}
|
|
|
|
|
if (result != expected) {
|
2019-09-20 01:22:42 +02:00
|
|
|
|
LOG(FATAL) << td::tag("text", str) << td::tag("got", td::format::as_array(result))
|
|
|
|
|
<< td::tag("expected", td::format::as_array(expected));
|
2018-12-31 20:04:05 +01:00
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
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("нет@mention", {});
|
2021-03-20 20:57:34 +01:00
|
|
|
|
check_mention(
|
|
|
|
|
"@ya @gif @wiki @vid @bing @pic @bold @imdb @ImDb @coub @like @vote @giff @cap ya cap @y @yar @bingg @bin",
|
2022-10-25 18:25:15 +02:00
|
|
|
|
{"@gif", "@wiki", "@vid", "@bing", "@pic", "@bold", "@imdb", "@ImDb", "@coub", "@like", "@vote", "@giff",
|
|
|
|
|
"@bingg"});
|
2021-08-30 22:27:32 +02:00
|
|
|
|
}
|
2018-12-31 20:04:05 +01:00
|
|
|
|
|
2019-09-26 17:36:45 +02:00
|
|
|
|
static void check_bot_command(const td::string &str, const td::vector<td::string> &expected) {
|
2019-09-20 01:22:42 +02:00
|
|
|
|
auto result_slice = td::find_bot_commands(str);
|
|
|
|
|
td::vector<td::string> result;
|
2018-12-31 20:04:05 +01:00
|
|
|
|
for (auto &it : result_slice) {
|
|
|
|
|
result.push_back(it.str());
|
|
|
|
|
}
|
|
|
|
|
if (result != expected) {
|
2019-09-20 01:22:42 +02:00
|
|
|
|
LOG(FATAL) << td::tag("text", str) << td::tag("got", td::format::as_array(result))
|
|
|
|
|
<< td::tag("expected", td::format::as_array(expected));
|
2018-12-31 20:04:05 +01:00
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
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/", {});
|
|
|
|
|
}
|
|
|
|
|
|
2019-09-26 17:36:45 +02:00
|
|
|
|
static void check_hashtag(const td::string &str, const td::vector<td::string> &expected) {
|
2019-09-20 01:22:42 +02:00
|
|
|
|
auto result_slice = td::find_hashtags(str);
|
|
|
|
|
td::vector<td::string> result;
|
2018-12-31 20:04:05 +01:00
|
|
|
|
for (auto &it : result_slice) {
|
|
|
|
|
result.push_back(it.str());
|
|
|
|
|
}
|
|
|
|
|
if (result != expected) {
|
2019-09-20 01:22:42 +02:00
|
|
|
|
LOG(FATAL) << td::tag("text", str) << td::tag("got", td::format::as_array(result))
|
|
|
|
|
<< td::tag("expected", td::format::as_array(expected));
|
2018-12-31 20:04:05 +01:00
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
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(" #я ", {"#я"});
|
|
|
|
|
check_hashtag(" я#a ", {});
|
|
|
|
|
check_hashtag(" #a# ", {});
|
|
|
|
|
check_hashtag(" #123 ", {});
|
|
|
|
|
check_hashtag(" #123a ", {"#123a"});
|
|
|
|
|
check_hashtag(" #a123 ", {"#a123"});
|
|
|
|
|
check_hashtag(" #123a# ", {});
|
2019-09-20 01:22:42 +02:00
|
|
|
|
check_hashtag(" #" + td::string(300, '1'), {});
|
|
|
|
|
check_hashtag(" #" + td::string(256, '1'), {});
|
|
|
|
|
check_hashtag(" #" + td::string(256, '1') + "a ", {});
|
|
|
|
|
check_hashtag(" #" + td::string(255, '1') + "a", {"#" + td::string(255, '1') + "a"});
|
|
|
|
|
check_hashtag(" #" + td::string(255, '1') + "Я", {"#" + td::string(255, '1') + "Я"});
|
|
|
|
|
check_hashtag(" #" + td::string(255, '1') + "a" + td::string(255, 'b') + "# ", {});
|
2018-12-31 20:04:05 +01:00
|
|
|
|
check_hashtag("#a#b #c #d", {"#c", "#d"});
|
|
|
|
|
check_hashtag("#test", {"#test"});
|
2022-01-31 00:19:17 +01:00
|
|
|
|
check_hashtag("#te·st", {"#te·st"});
|
2019-09-20 01:22:42 +02:00
|
|
|
|
check_hashtag(u8"\U0001F604\U0001F604\U0001F604\U0001F604 \U0001F604\U0001F604\U0001F604#" + td::string(200, '1') +
|
|
|
|
|
"ООО" + td::string(200, '2'),
|
|
|
|
|
{"#" + td::string(200, '1') + "ООО" + td::string(53, '2')});
|
2018-12-31 20:04:05 +01:00
|
|
|
|
check_hashtag(u8"#a\u2122", {"#a"});
|
2022-07-29 15:23:05 +02:00
|
|
|
|
check_hashtag("#a൹", {"#a"});
|
|
|
|
|
check_hashtag("#aඁං෴ก฿", {"#aඁං෴ก"});
|
2018-12-31 20:04:05 +01:00
|
|
|
|
}
|
|
|
|
|
|
2019-09-26 17:36:45 +02:00
|
|
|
|
static void check_cashtag(const td::string &str, const td::vector<td::string> &expected) {
|
2019-09-20 01:22:42 +02:00
|
|
|
|
auto result_slice = td::find_cashtags(str);
|
|
|
|
|
td::vector<td::string> result;
|
2018-03-07 18:29:33 +01:00
|
|
|
|
for (auto &it : result_slice) {
|
|
|
|
|
result.push_back(it.str());
|
|
|
|
|
}
|
|
|
|
|
if (result != expected) {
|
2019-09-20 01:22:42 +02:00
|
|
|
|
LOG(FATAL) << td::tag("text", str) << td::tag("got", td::format::as_array(result))
|
|
|
|
|
<< td::tag("expected", td::format::as_array(expected));
|
2018-03-07 18:29:33 +01:00
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
TEST(MessageEntities, cashtag) {
|
|
|
|
|
check_cashtag("", {});
|
|
|
|
|
check_cashtag("$", {});
|
|
|
|
|
check_cashtag("$$", {});
|
|
|
|
|
check_cashtag("$$$", {});
|
|
|
|
|
check_cashtag("$a", {});
|
|
|
|
|
check_cashtag(" $a", {});
|
|
|
|
|
check_cashtag("$a ", {});
|
|
|
|
|
check_cashtag(" $я ", {});
|
|
|
|
|
check_cashtag("$ab", {});
|
|
|
|
|
check_cashtag("$abc", {});
|
|
|
|
|
check_cashtag("$", {});
|
2021-08-03 01:58:12 +02:00
|
|
|
|
check_cashtag("$A", {"$A"});
|
|
|
|
|
check_cashtag("$AB", {"$AB"});
|
|
|
|
|
check_cashtag("$ABС", {});
|
2018-03-07 18:29:33 +01:00
|
|
|
|
check_cashtag("$АBC", {});
|
|
|
|
|
check_cashtag("$АВС", {});
|
|
|
|
|
check_cashtag("$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(" А$ABC ", {});
|
|
|
|
|
check_cashtag("$ABC$DEF $GHI $KLM", {"$GHI", "$KLM"});
|
|
|
|
|
check_cashtag("$TEST", {"$TEST"});
|
2021-08-03 01:58:12 +02:00
|
|
|
|
check_cashtag("$1INC", {});
|
|
|
|
|
check_cashtag("$1INCH", {"$1INCH"});
|
|
|
|
|
check_cashtag("...$1INCH...", {"$1INCH"});
|
|
|
|
|
check_cashtag("$1inch", {});
|
|
|
|
|
check_cashtag("$1INCHA", {});
|
|
|
|
|
check_cashtag("$1INCHА", {});
|
2018-03-07 18:29:33 +01:00
|
|
|
|
check_cashtag(u8"$ABC\u2122", {"$ABC"});
|
|
|
|
|
check_cashtag(u8"\u2122$ABC", {"$ABC"});
|
|
|
|
|
check_cashtag(u8"\u2122$ABC\u2122", {"$ABC"});
|
2022-07-29 15:23:05 +02:00
|
|
|
|
check_cashtag("$ABC൹", {"$ABC"});
|
|
|
|
|
check_cashtag("$ABCඁ", {});
|
|
|
|
|
check_cashtag("$ABCං", {});
|
|
|
|
|
check_cashtag("$ABC෴", {});
|
|
|
|
|
check_cashtag("$ABCก", {});
|
|
|
|
|
check_cashtag("$ABC฿", {"$ABC"});
|
2018-03-07 18:29:33 +01:00
|
|
|
|
}
|
|
|
|
|
|
2021-08-05 00:23:16 +02:00
|
|
|
|
static void check_media_timestamp(const td::string &str, const td::vector<std::pair<td::string, td::int32>> &expected) {
|
|
|
|
|
auto result = td::transform(td::find_media_timestamps(str),
|
|
|
|
|
[](auto &&entity) { return std::make_pair(entity.first.str(), entity.second); });
|
2021-07-28 07:30:22 +02:00
|
|
|
|
if (result != expected) {
|
|
|
|
|
LOG(FATAL) << td::tag("text", str) << td::tag("got", td::format::as_array(result))
|
|
|
|
|
<< td::tag("expected", td::format::as_array(expected));
|
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
TEST(MessageEntities, media_timestamp) {
|
|
|
|
|
check_media_timestamp("", {});
|
|
|
|
|
check_media_timestamp(":", {});
|
|
|
|
|
check_media_timestamp(":1", {});
|
|
|
|
|
check_media_timestamp("a:1", {});
|
|
|
|
|
check_media_timestamp("01", {});
|
|
|
|
|
check_media_timestamp("01:", {});
|
|
|
|
|
check_media_timestamp("01::", {});
|
|
|
|
|
check_media_timestamp("01::", {});
|
|
|
|
|
check_media_timestamp("a1:1a", {});
|
|
|
|
|
check_media_timestamp("a1::01a", {});
|
|
|
|
|
check_media_timestamp("2001:db8::8a2e:f70:13a4", {});
|
|
|
|
|
check_media_timestamp("0:00", {{"0:00", 0}});
|
|
|
|
|
check_media_timestamp("+0:00", {{"0:00", 0}});
|
|
|
|
|
check_media_timestamp("0:00+", {{"0:00", 0}});
|
|
|
|
|
check_media_timestamp("a0:00", {});
|
|
|
|
|
check_media_timestamp("0:00a", {});
|
|
|
|
|
check_media_timestamp("б0:00", {});
|
|
|
|
|
check_media_timestamp("0:00б", {});
|
|
|
|
|
check_media_timestamp("_0:00", {});
|
|
|
|
|
check_media_timestamp("0:00_", {});
|
|
|
|
|
check_media_timestamp("00:00:00:00", {});
|
|
|
|
|
check_media_timestamp("1:1:01 1:1:1", {{"1:1:01", 3661}});
|
|
|
|
|
check_media_timestamp("0:0:00 00:00 000:00 0000:00 00000:00 00:00:00 000:00:00 00:000:00 00:00:000",
|
|
|
|
|
{{"0:0:00", 0}, {"00:00", 0}, {"000:00", 0}, {"0000:00", 0}, {"00:00:00", 0}});
|
|
|
|
|
check_media_timestamp("00:0:00 0:00:00 00::00 :00:00 00:00: 00:00:0 00:00:", {{"00:0:00", 0}, {"0:00:00", 0}});
|
|
|
|
|
check_media_timestamp("1:1:59 1:1:-1 1:1:60", {{"1:1:59", 3719}});
|
|
|
|
|
check_media_timestamp("1:59:00 1:-1:00 1:60:00", {{"1:59:00", 7140}, {"1:00", 60}});
|
|
|
|
|
check_media_timestamp("59:59 60:00", {{"59:59", 3599}, {"60:00", 3600}});
|
|
|
|
|
check_media_timestamp("9999:59 99:59:59 99:60:59", {{"9999:59", 599999}, {"99:59:59", 360000 - 1}});
|
|
|
|
|
check_media_timestamp("2001:db8::8a2e:f70:13a4", {});
|
|
|
|
|
}
|
|
|
|
|
|
2020-02-15 02:07:19 +01:00
|
|
|
|
static void check_bank_card_number(const td::string &str, const td::vector<td::string> &expected) {
|
|
|
|
|
auto result_slice = td::find_bank_card_numbers(str);
|
|
|
|
|
td::vector<td::string> result;
|
|
|
|
|
for (auto &it : result_slice) {
|
|
|
|
|
result.push_back(it.str());
|
|
|
|
|
}
|
|
|
|
|
if (result != expected) {
|
|
|
|
|
LOG(FATAL) << td::tag("text", str) << td::tag("got", td::format::as_array(result))
|
|
|
|
|
<< td::tag("expected", td::format::as_array(expected));
|
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
TEST(MessageEntities, bank_card_number) {
|
|
|
|
|
check_bank_card_number("", {});
|
|
|
|
|
check_bank_card_number("123456789015", {});
|
|
|
|
|
check_bank_card_number("1234567890120", {});
|
|
|
|
|
check_bank_card_number("1234567890121", {});
|
|
|
|
|
check_bank_card_number("1234567890122", {});
|
|
|
|
|
check_bank_card_number("1234567890123", {});
|
|
|
|
|
check_bank_card_number("1234567890124", {});
|
|
|
|
|
check_bank_card_number("1234567890125", {});
|
|
|
|
|
check_bank_card_number("1234567890126", {});
|
|
|
|
|
check_bank_card_number("1234567890127", {});
|
|
|
|
|
check_bank_card_number("1234567890128", {"1234567890128"});
|
|
|
|
|
check_bank_card_number("1234567890129", {});
|
|
|
|
|
check_bank_card_number("12345678901500", {"12345678901500"});
|
|
|
|
|
check_bank_card_number("123456789012800", {"123456789012800"});
|
|
|
|
|
check_bank_card_number("1234567890151800", {"1234567890151800"});
|
|
|
|
|
check_bank_card_number("12345678901280000", {"12345678901280000"});
|
|
|
|
|
check_bank_card_number("123456789015009100", {"123456789015009100"});
|
|
|
|
|
check_bank_card_number("1234567890128000000", {"1234567890128000000"});
|
|
|
|
|
check_bank_card_number("12345678901500910000", {});
|
2020-05-05 02:15:30 +02:00
|
|
|
|
check_bank_card_number(" - - - - 1 - -- 2 - - -- 34 - - - 56- - 7890150000 - - - -", {});
|
|
|
|
|
check_bank_card_number(" - - - - 1 - -- 234 - - 56- - 7890150000 - - - -", {"1 - -- 234 - - 56- - 7890150000"});
|
2020-02-15 02:07:19 +01:00
|
|
|
|
check_bank_card_number("4916-3385-0608-2832; 5280 9342 8317 1080 ;345936346788903",
|
|
|
|
|
{"4916-3385-0608-2832", "5280 9342 8317 1080", "345936346788903"});
|
2020-05-07 22:11:54 +02:00
|
|
|
|
check_bank_card_number("4556728228023269, 4916141675244747020, 49161416752447470, 4556728228023269",
|
2020-02-15 02:07:19 +01:00
|
|
|
|
{"4556728228023269", "4916141675244747020", "4556728228023269"});
|
2020-05-05 02:15:30 +02:00
|
|
|
|
check_bank_card_number("a1234567890128", {});
|
|
|
|
|
check_bank_card_number("1234567890128a", {});
|
|
|
|
|
check_bank_card_number("1234567890128а", {});
|
|
|
|
|
check_bank_card_number("а1234567890128", {});
|
|
|
|
|
check_bank_card_number("1234567890128_", {});
|
|
|
|
|
check_bank_card_number("_1234567890128", {});
|
|
|
|
|
check_bank_card_number("1234567890128/", {"1234567890128"});
|
2020-05-07 22:11:54 +02:00
|
|
|
|
check_bank_card_number("\"1234567890128", {"1234567890128"});
|
2020-05-05 02:15:30 +02:00
|
|
|
|
check_bank_card_number("+1234567890128", {});
|
2020-02-15 02:07:19 +01:00
|
|
|
|
}
|
|
|
|
|
|
2021-06-03 17:27:40 +02:00
|
|
|
|
static void check_tg_url(const td::string &str, const td::vector<td::string> &expected) {
|
|
|
|
|
auto result_slice = td::find_tg_urls(str);
|
|
|
|
|
td::vector<td::string> result;
|
|
|
|
|
for (auto &it : result_slice) {
|
|
|
|
|
result.push_back(it.str());
|
|
|
|
|
}
|
|
|
|
|
if (result != expected) {
|
|
|
|
|
LOG(FATAL) << td::tag("text", str) << td::tag("got", td::format::as_array(result))
|
|
|
|
|
<< td::tag("expected", td::format::as_array(expected));
|
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
TEST(MessageEntities, tg_url) {
|
|
|
|
|
check_tg_url("", {});
|
|
|
|
|
check_tg_url("tg://", {});
|
|
|
|
|
check_tg_url("tg://a", {"tg://a"});
|
|
|
|
|
check_tg_url("a", {});
|
|
|
|
|
check_tg_url("stg://a", {"tg://a"});
|
|
|
|
|
check_tg_url("asd asdas das ton:asd tg:test ton://resolve tg://resolve TON://_-RESOLVE_- TG://-_RESOLVE-_",
|
|
|
|
|
{"ton://resolve", "tg://resolve", "TON://_-RESOLVE_-", "TG://-_RESOLVE-_"});
|
|
|
|
|
check_tg_url("tg:test/", {});
|
|
|
|
|
check_tg_url("tg:/test/", {});
|
|
|
|
|
check_tg_url("tg://test/", {"tg://test/"});
|
|
|
|
|
check_tg_url("tg://test/?", {"tg://test/"});
|
|
|
|
|
check_tg_url("tg://test/#", {"tg://test/#"});
|
|
|
|
|
check_tg_url("tg://test?", {"tg://test"});
|
|
|
|
|
check_tg_url("tg://test#", {"tg://test"});
|
|
|
|
|
check_tg_url("tg://test/―asd―?asd=asd&asdas=―#――――", {"tg://test/―asd―?asd=asd&asdas=―#――――"});
|
|
|
|
|
check_tg_url("tg://test/?asd", {"tg://test/?asd"});
|
|
|
|
|
check_tg_url("tg://test/?.:;,('?!`.:;,('?!`", {"tg://test/"});
|
|
|
|
|
check_tg_url("tg://test/#asdf", {"tg://test/#asdf"});
|
|
|
|
|
check_tg_url("tg://test?asdf", {"tg://test?asdf"});
|
|
|
|
|
check_tg_url("tg://test#asdf", {"tg://test#asdf"});
|
|
|
|
|
check_tg_url("tg://test?as‖df", {"tg://test?as"});
|
2021-06-23 02:22:23 +02:00
|
|
|
|
check_tg_url("tg://test?sa<df", {"tg://test?sa"});
|
2021-06-03 17:27:40 +02:00
|
|
|
|
check_tg_url("tg://test?as>df", {"tg://test?as"});
|
|
|
|
|
check_tg_url("tg://test?as\"df", {"tg://test?as"});
|
|
|
|
|
check_tg_url("tg://test?as«df", {"tg://test?as"});
|
|
|
|
|
check_tg_url("tg://test?as»df", {"tg://test?as"});
|
|
|
|
|
check_tg_url("tg://test?as(df", {"tg://test?as(df"});
|
|
|
|
|
check_tg_url("tg://test?as)df", {"tg://test?as)df"});
|
|
|
|
|
check_tg_url("tg://test?as[df", {"tg://test?as[df"});
|
|
|
|
|
check_tg_url("tg://test?as]df", {"tg://test?as]df"});
|
|
|
|
|
check_tg_url("tg://test?as{df", {"tg://test?as{df"});
|
|
|
|
|
check_tg_url("tg://test?as'df", {"tg://test?as'df"});
|
|
|
|
|
check_tg_url("tg://test?as}df", {"tg://test?as}df"});
|
|
|
|
|
check_tg_url("tg://test?as$df", {"tg://test?as$df"});
|
|
|
|
|
check_tg_url("tg://test?as%df", {"tg://test?as%df"});
|
|
|
|
|
check_tg_url("tg://%30/sccct", {});
|
|
|
|
|
check_tg_url("tg://test:asd@google.com:80", {"tg://test"});
|
|
|
|
|
check_tg_url("tg://google.com", {"tg://google"});
|
|
|
|
|
check_tg_url("tg://google/.com", {"tg://google/.com"});
|
|
|
|
|
check_tg_url("tg://127.0.0.1", {"tg://127"});
|
|
|
|
|
check_tg_url("tg://б.а.н.а.на", {});
|
|
|
|
|
}
|
|
|
|
|
|
2019-09-26 17:36:45 +02:00
|
|
|
|
static void check_is_email_address(const td::string &str, bool expected) {
|
2019-09-20 01:22:42 +02:00
|
|
|
|
bool result = td::is_email_address(str);
|
2018-12-31 20:04:05 +01:00
|
|
|
|
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);
|
2018-02-16 23:29:13 +01:00
|
|
|
|
check_is_email_address("Test@test.abd", true);
|
2018-12-31 20:04:05 +01:00
|
|
|
|
check_is_email_address("a@a.a.a.ab", true);
|
2018-02-16 23:29:13 +01:00
|
|
|
|
check_is_email_address("test@test.abd", true);
|
2018-12-31 20:04:05 +01:00
|
|
|
|
check_is_email_address("test@test.com", true);
|
2018-02-16 23:29:13 +01:00
|
|
|
|
check_is_email_address("test.abd", false);
|
2018-12-31 20:04:05 +01:00
|
|
|
|
check_is_email_address("a.ab", false);
|
|
|
|
|
check_is_email_address("a.bc@d.ef", true);
|
|
|
|
|
|
2019-09-20 01:22:42 +02:00
|
|
|
|
td::vector<td::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"};
|
|
|
|
|
td::vector<td::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"};
|
|
|
|
|
|
|
|
|
|
td::vector<td::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",
|
2022-01-30 11:51:28 +01:00
|
|
|
|
"a.a.a.a.a.a.abcdefghi",
|
2019-09-20 01:22:42 +02:00
|
|
|
|
"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"};
|
|
|
|
|
td::vector<td::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"};
|
2018-12-31 20:04:05 +01:00
|
|
|
|
|
|
|
|
|
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);
|
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
|
2019-09-26 17:36:45 +02:00
|
|
|
|
static void check_url(const td::string &str, const td::vector<td::string> &expected_urls,
|
2019-09-20 01:22:42 +02:00
|
|
|
|
td::vector<td::string> expected_email_addresses = {}) {
|
|
|
|
|
auto result_slice = td::find_urls(str);
|
|
|
|
|
td::vector<td::string> result_urls;
|
|
|
|
|
td::vector<td::string> result_email_addresses;
|
2018-12-31 20:04:05 +01:00
|
|
|
|
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) {
|
2019-09-20 01:22:42 +02:00
|
|
|
|
LOG(FATAL) << td::tag("text", str) << td::tag("got", td::format::as_array(result_urls))
|
|
|
|
|
<< td::tag("expected", td::format::as_array(expected_urls));
|
2018-12-31 20:04:05 +01:00
|
|
|
|
}
|
|
|
|
|
if (result_email_addresses != expected_email_addresses) {
|
2019-09-20 01:22:42 +02:00
|
|
|
|
LOG(FATAL) << td::tag("text", str) << td::tag("got", td::format::as_array(result_email_addresses))
|
|
|
|
|
<< td::tag("expected", td::format::as_array(expected_email_addresses));
|
2018-12-31 20:04:05 +01:00
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
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("Такой сайт: http://www.google.com или такой 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
|
2021-09-09 16:59:24 +02:00
|
|
|
|
check_url("http://@@google.com", {});
|
2018-12-31 20:04:05 +01:00
|
|
|
|
check_url("http://a@google.com", {"http://a@google.com"});
|
|
|
|
|
check_url("http://test@google.com", {"http://test@google.com"});
|
|
|
|
|
check_url("google.com:᪉᪉᪉᪉᪉", {"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", {});
|
2021-02-03 19:08:01 +01:00
|
|
|
|
check_url("sftp://telegram.org", {});
|
2018-12-31 20:04:05 +01:00
|
|
|
|
check_url("hTtPs://telegram.org", {"hTtPs://telegram.org"});
|
|
|
|
|
check_url("HTTP://telegram.org", {"HTTP://telegram.org"});
|
|
|
|
|
check_url("аHTTP://telegram.org", {"HTTP://telegram.org"});
|
|
|
|
|
check_url("sHTTP://telegram.org", {});
|
|
|
|
|
check_url("://telegram.org", {});
|
|
|
|
|
check_url("google.com:᪀᪀", {"google.com"});
|
|
|
|
|
check_url(
|
|
|
|
|
"http://"
|
|
|
|
|
"abcdefghijkabcdefghijkabcdefghijkabcdefghijkabcdefghijkabcdefghijkabcdefghijkabcdefghijkabcdefghijkabcdefghijkab"
|
|
|
|
|
"cdefghijkabcdefghijkabcdefghijkabcdefghijkabcdefghijkabcdefghijkabcdefghijkabcdefghijkabcdefghijkabcdefghijkabcd"
|
|
|
|
|
"efghijkabcdefghijkabcdefghijkabcdefghijkabcdefghijkabcdefghijkabcdefghijkabcdefghijkabcdefghijkabcdefghijkabcdef"
|
|
|
|
|
"ghijkabcdefghijkabcdefghijkabcdefghijkabcdefghijkabcdefghijkabcdefghijkabcdefghijkabcdefghijkabcdefghijkabcdefgh"
|
|
|
|
|
"ijkabcdefghijkabcdefghijkabcdefghijkabcdefghijkabcdefghijkabcdefghijkabcdefghijkabcdefghijkabcdefghijkabcdefghij"
|
|
|
|
|
"kabcdefghijkabcdefghijkabcdefghijkabcdefghijkabcdefghijkabcdefghijkabcdefghijkabcdefghijkabcdefghijkabcdefghijka"
|
|
|
|
|
"bcdefghijkabcdefghijkabcdefghijkabcdefghijkabcdefghijkabcdefghijkabcdefghijkabcdefghijkabcdefghijkabcdefghijkabc"
|
|
|
|
|
"defghijkabcdefghijkabcdefghijkabcdefghijkabcdefghijkabcdefghijkabcdefghijkabcdefghijk.com",
|
|
|
|
|
{});
|
|
|
|
|
check_url("http:// .com", {});
|
|
|
|
|
check_url("URL: .com", {});
|
|
|
|
|
check_url("URL: .com", {});
|
|
|
|
|
check_url(".com", {});
|
|
|
|
|
check_url("http:// .", {});
|
|
|
|
|
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"});
|
2018-06-25 02:22:56 +02:00
|
|
|
|
check_url("google.com:1#", {"google.com:1"});
|
2018-12-31 20:04:05 +01:00
|
|
|
|
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.🤙.tk", {"www.🤙.tk"});
|
|
|
|
|
check_url("a.ab", {});
|
2018-02-16 23:29:13 +01:00
|
|
|
|
check_url("test.abd", {});
|
2018-12-31 20:04:05 +01:00
|
|
|
|
check_url("ТеСт.Москва", {});
|
|
|
|
|
check_url("ТеСт.МоСкВΑ", {});
|
|
|
|
|
check_url("ТеСт.МоСкВа", {"ТеСт.МоСкВа"});
|
|
|
|
|
check_url("ТеСт.МоСкВач", {});
|
|
|
|
|
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"});
|
2022-08-20 12:44:55 +02:00
|
|
|
|
check_url("teiegram.org/test", {});
|
|
|
|
|
check_url("TeiegraM.org/test", {});
|
2018-12-31 20:04:05 +01:00
|
|
|
|
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
|
2021-06-03 17:27:40 +02:00
|
|
|
|
check_url("tg://resolve", {});
|
2018-02-16 23:29:13 +01:00
|
|
|
|
check_url("test.abd", {});
|
2018-12-31 20:04:05 +01:00
|
|
|
|
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"});
|
2018-06-25 02:22:56 +02:00
|
|
|
|
check_url("http://google.com#", {"http://google.com"});
|
|
|
|
|
check_url("http://google.com##", {"http://google.com##"});
|
2018-12-31 20:04:05 +01:00
|
|
|
|
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://códuia.de/ dffdg,\" 12)(cpia.de/())(\" http://гришка.рф/ sdufhdf "
|
2020-06-14 04:27:49 +02:00
|
|
|
|
"http://xn--80afpi2a3c.xn--p1ai/ I have a good time.Thanks, guys!\n\n(hdfughidufhgdis) go#ogle.com гришка.рф "
|
2018-12-31 20:04:05 +01:00
|
|
|
|
"hsighsdf gi почта.рф\n\n✪df.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_Вася)\n(Something like "
|
|
|
|
|
"http://foo.com/blah_blah_(Стакан_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://✪df.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://➡️.ws/䨹\nwww.➡️.ws/"
|
|
|
|
|
"䨹\n<tag>http://example.com</tag>\nJust a www.example.com "
|
|
|
|
|
"link.\n\n➡️.ws/"
|
|
|
|
|
"䨹\n\nabcdefghijklmnopqrstuvwxyz0123456789qwe_sdfsdf.aweawe-sdfs.com\nwww.🤙.tk:1\ngoogle.com:"
|
|
|
|
|
"᪉᪉᪉᪉\ngoogle."
|
|
|
|
|
"com:᪀᪀\nhttp:// .com\nURL: .com\nURL: "
|
|
|
|
|
".com\n\ngoogle.com?qwe\ngoogle.com#qwe\ngoogle.com/?\ngoogle.com/#\ngoogle.com?\ngoogle.com#\n",
|
|
|
|
|
{"a.b.google.com",
|
|
|
|
|
"http://códuia.de/",
|
|
|
|
|
"cpia.de/()",
|
|
|
|
|
"http://гришка.рф/",
|
|
|
|
|
"http://xn--80afpi2a3c.xn--p1ai/",
|
|
|
|
|
"гришка.рф",
|
|
|
|
|
"почта.рф",
|
|
|
|
|
"✪df.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_Вася)",
|
|
|
|
|
"http://foo.com/blah_blah_(Стакан_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://✪df.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://➡️.ws/䨹",
|
|
|
|
|
"www.➡️.ws/䨹",
|
|
|
|
|
"http://example.com",
|
|
|
|
|
"www.example.com",
|
|
|
|
|
"➡️.ws/䨹",
|
|
|
|
|
"abcdefghijklmnopqrstuvwxyz0123456789qwe_sdfsdf.aweawe-sdfs.com",
|
|
|
|
|
"www.🤙.tk:1",
|
|
|
|
|
"google.com",
|
|
|
|
|
"google.com",
|
|
|
|
|
"google.com?qwe",
|
|
|
|
|
"google.com#qwe",
|
|
|
|
|
"google.com/",
|
|
|
|
|
"google.com/#",
|
|
|
|
|
"google.com",
|
2018-06-25 02:22:56 +02:00
|
|
|
|
"google.com"});
|
2018-12-31 20:04:05 +01:00
|
|
|
|
check_url("https://t.…", {});
|
|
|
|
|
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"});
|
2022-08-19 20:01:39 +02:00
|
|
|
|
check_url("https://a.debc@c.com", {"https://a.debc@c.com"});
|
2018-12-31 20:04:05 +01:00
|
|
|
|
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"});
|
2022-08-19 20:01:39 +02:00
|
|
|
|
check_url("https://a.bcde.fg@c.com", {"https://a.bcde.fg@c.com"});
|
|
|
|
|
check_url("https://a:h.bcde.fg@c.com", {"https://a:h.bcde.fg@c.com"});
|
2018-12-31 20:04:05 +01:00
|
|
|
|
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"}
|
2018-06-25 02:15:38 +02:00
|
|
|
|
check_url("Look mailto:test@example.com", {}, {"test@example.com"});
|
2018-12-31 20:04:05 +01:00
|
|
|
|
check_url("http://test.com#a", {"http://test.com#a"});
|
2018-06-25 02:22:56 +02:00
|
|
|
|
check_url("http://test.com#", {"http://test.com"});
|
2018-12-31 20:04:05 +01:00
|
|
|
|
check_url("http://test.com?#", {"http://test.com?#"});
|
|
|
|
|
check_url("http://test.com/?#", {"http://test.com/?#"});
|
|
|
|
|
check_url("https://t.me/abcdef…", {"https://t.me/abcdef"});
|
|
|
|
|
check_url("https://t.me…", {"https://t.me"});
|
|
|
|
|
check_url("https://t.m…", {});
|
|
|
|
|
check_url("https://t.…", {});
|
|
|
|
|
check_url("https://t…", {});
|
|
|
|
|
check_url("👉http://ab.com/cdefgh-1IJ", {"http://ab.com/cdefgh-1IJ"});
|
|
|
|
|
check_url("...👉http://ab.com/cdefgh-1IJ", {}); // TODO
|
2021-05-14 12:18:52 +02:00
|
|
|
|
check_url(".?", {});
|
2021-06-03 14:21:26 +02:00
|
|
|
|
check_url("http://test―‑@―google―.―com―/―–―‐―/―/―/―?―‑―#―――", {"http://test―‑@―google―.―com―/―–―‐―/―/―/―?―‑―#―――"});
|
|
|
|
|
check_url("http://google.com/‖", {"http://google.com/"});
|
2021-09-09 16:59:24 +02:00
|
|
|
|
check_url("a@b@c.com", {}, {});
|
2022-08-20 10:34:21 +02:00
|
|
|
|
check_url("abc@c.com@d.com", {});
|
2021-09-09 16:59:24 +02:00
|
|
|
|
check_url("a@b.com:c@1", {}, {"a@b.com"});
|
2022-01-30 11:51:28 +01:00
|
|
|
|
check_url("test@test.software", {}, {"test@test.software"});
|
2022-02-03 13:53:36 +01:00
|
|
|
|
check_url("a:b?@gmail.com", {});
|
|
|
|
|
check_url("a?:b@gmail.com", {});
|
|
|
|
|
check_url("a#:b@gmail.com", {});
|
|
|
|
|
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"});
|
2022-08-20 12:44:55 +02:00
|
|
|
|
check_url("http://test_.com", {});
|
|
|
|
|
check_url("test_.com", {});
|
|
|
|
|
check_url("_test.com", {});
|
|
|
|
|
check_url("_.test.com", {"_.test.com"});
|
2018-12-31 20:04:05 +01:00
|
|
|
|
}
|
2019-01-16 19:18:26 +01:00
|
|
|
|
|
2019-09-26 17:36:45 +02:00
|
|
|
|
static void check_fix_formatted_text(td::string str, td::vector<td::MessageEntity> entities,
|
|
|
|
|
const td::string &expected_str,
|
2020-02-21 12:56:26 +01:00
|
|
|
|
const td::vector<td::MessageEntity> &expected_entities, bool allow_empty = true,
|
|
|
|
|
bool skip_new_entities = false, bool skip_bot_commands = false,
|
2023-02-11 20:37:43 +01:00
|
|
|
|
bool skip_trim = true) {
|
|
|
|
|
ASSERT_TRUE(td::fix_formatted_text(str, entities, allow_empty, skip_new_entities, skip_bot_commands, true, skip_trim)
|
2021-07-26 21:48:33 +02:00
|
|
|
|
.is_ok());
|
2019-01-16 19:18:26 +01:00
|
|
|
|
ASSERT_STREQ(expected_str, str);
|
|
|
|
|
ASSERT_EQ(expected_entities, entities);
|
|
|
|
|
}
|
|
|
|
|
|
2019-09-20 01:22:42 +02:00
|
|
|
|
static void check_fix_formatted_text(td::string str, td::vector<td::MessageEntity> entities, bool allow_empty,
|
2023-02-11 20:37:43 +01:00
|
|
|
|
bool skip_new_entities, bool skip_bot_commands, bool skip_trim) {
|
|
|
|
|
ASSERT_TRUE(td::fix_formatted_text(str, entities, allow_empty, skip_new_entities, skip_bot_commands, true, skip_trim)
|
2021-07-26 21:48:33 +02:00
|
|
|
|
.is_error());
|
2019-01-16 19:18:26 +01:00
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
TEST(MessageEntities, fix_formatted_text) {
|
2019-09-20 01:22:42 +02:00
|
|
|
|
td::string str;
|
|
|
|
|
td::string fixed_str;
|
2019-01-16 19:18:26 +01:00
|
|
|
|
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);
|
|
|
|
|
|
2021-08-18 20:37:31 +02:00
|
|
|
|
check_fix_formatted_text(" aba\n ", {}, " aba\n ", {}, true, true, true, true);
|
|
|
|
|
check_fix_formatted_text(" aba\n ", {}, "aba", {}, true, true, true, false);
|
|
|
|
|
check_fix_formatted_text(" \n ", {}, "", {}, true, true, true, true);
|
|
|
|
|
check_fix_formatted_text(" \n ", {}, "", {}, true, true, true, false);
|
|
|
|
|
check_fix_formatted_text(" \n ", {}, false, true, true, false);
|
|
|
|
|
|
2019-01-16 19:18:26 +01:00
|
|
|
|
str += "a \r\n ";
|
|
|
|
|
fixed_str += "a \n ";
|
|
|
|
|
|
2020-05-11 00:31:29 +02:00
|
|
|
|
for (td::int32 i = 33; i <= 35; i++) {
|
|
|
|
|
td::vector<td::MessageEntity> entities;
|
|
|
|
|
entities.emplace_back(td::MessageEntity::Type::Pre, 0, i);
|
|
|
|
|
|
|
|
|
|
td::vector<td::MessageEntity> fixed_entities = entities;
|
|
|
|
|
fixed_entities.back().length = i - 1;
|
|
|
|
|
check_fix_formatted_text(str, entities, fixed_str, fixed_entities, true, false, false, true);
|
|
|
|
|
|
|
|
|
|
td::string expected_str = fixed_str.substr(0, 33);
|
|
|
|
|
fixed_entities.back().length = i == 33 ? 32 : 33;
|
|
|
|
|
check_fix_formatted_text(str, entities, expected_str, fixed_entities, false, false, false, false);
|
|
|
|
|
}
|
|
|
|
|
|
2019-09-20 01:22:42 +02:00
|
|
|
|
for (td::int32 i = 33; i <= 35; i++) {
|
|
|
|
|
td::vector<td::MessageEntity> entities;
|
|
|
|
|
entities.emplace_back(td::MessageEntity::Type::Bold, 0, i);
|
2019-01-16 19:18:26 +01:00
|
|
|
|
|
2019-09-20 01:22:42 +02:00
|
|
|
|
td::vector<td::MessageEntity> fixed_entities;
|
2019-01-16 19:18:26 +01:00
|
|
|
|
if (i != 33) {
|
2020-05-11 00:31:29 +02:00
|
|
|
|
fixed_entities.emplace_back(td::MessageEntity::Type::Bold, 32, i - 33);
|
2019-01-16 19:18:26 +01:00
|
|
|
|
}
|
|
|
|
|
check_fix_formatted_text(str, entities, fixed_str, fixed_entities, true, false, false, true);
|
|
|
|
|
|
2019-09-20 01:22:42 +02:00
|
|
|
|
td::string expected_str;
|
2019-01-16 19:18:26 +01:00
|
|
|
|
if (i != 33) {
|
2020-05-11 00:31:29 +02:00
|
|
|
|
fixed_entities.back().offset = 0;
|
|
|
|
|
fixed_entities.back().length = 1;
|
2019-01-16 19:18:26 +01:00
|
|
|
|
}
|
2020-05-11 00:31:29 +02:00
|
|
|
|
expected_str = "a";
|
2019-01-16 19:18:26 +01:00
|
|
|
|
check_fix_formatted_text(str, entities, expected_str, fixed_entities, false, false, false, false);
|
|
|
|
|
}
|
|
|
|
|
|
2019-12-02 02:03:43 +01:00
|
|
|
|
str = "👉 👉 ";
|
2019-01-16 19:18:26 +01:00
|
|
|
|
for (int i = 0; i < 10; i++) {
|
2019-09-20 01:22:42 +02:00
|
|
|
|
td::vector<td::MessageEntity> entities;
|
|
|
|
|
entities.emplace_back(td::MessageEntity::Type::Bold, i, 1);
|
2019-12-02 02:03:43 +01:00
|
|
|
|
if (i != 2 && i != 5 && i != 6) {
|
2019-01-16 19:18:26 +01:00
|
|
|
|
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);
|
2019-12-02 02:03:43 +01:00
|
|
|
|
check_fix_formatted_text(str, entities, str.substr(0, str.size() - 2), {}, false, false, false, false);
|
2019-01-16 19:18:26 +01:00
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
str = " /test @abaca #ORD $ABC telegram.org ";
|
2023-02-11 20:37:43 +01:00
|
|
|
|
for (auto skip_trim : {false, true}) {
|
|
|
|
|
td::int32 shift = skip_trim ? 2 : 0;
|
|
|
|
|
td::string expected_str = skip_trim ? str : str.substr(2, str.size() - 3);
|
2019-01-16 19:18:26 +01:00
|
|
|
|
|
|
|
|
|
for (auto skip_new_entities : {false, true}) {
|
|
|
|
|
for (auto skip_bot_commands : {false, true}) {
|
2019-09-20 01:22:42 +02:00
|
|
|
|
td::vector<td::MessageEntity> entities;
|
2019-01-16 19:18:26 +01:00
|
|
|
|
if (!skip_new_entities) {
|
|
|
|
|
if (!skip_bot_commands) {
|
2019-09-20 01:22:42 +02:00
|
|
|
|
entities.emplace_back(td::MessageEntity::Type::BotCommand, shift, 5);
|
2019-01-16 19:18:26 +01:00
|
|
|
|
}
|
2019-09-20 01:22:42 +02:00
|
|
|
|
entities.emplace_back(td::MessageEntity::Type::Mention, shift + 6, 6);
|
|
|
|
|
entities.emplace_back(td::MessageEntity::Type::Hashtag, shift + 13, 4);
|
|
|
|
|
entities.emplace_back(td::MessageEntity::Type::Cashtag, shift + 18, 4);
|
|
|
|
|
entities.emplace_back(td::MessageEntity::Type::Url, shift + 24, 12);
|
2019-01-16 19:18:26 +01:00
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
check_fix_formatted_text(str, {}, expected_str, entities, true, skip_new_entities, skip_bot_commands,
|
2023-02-11 20:37:43 +01:00
|
|
|
|
skip_trim);
|
2019-01-16 19:18:26 +01:00
|
|
|
|
check_fix_formatted_text(str, {}, expected_str, entities, false, skip_new_entities, skip_bot_commands,
|
2023-02-11 20:37:43 +01:00
|
|
|
|
skip_trim);
|
2019-01-16 19:18:26 +01:00
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
str = "aba \r\n caba ";
|
2022-08-19 14:05:49 +02:00
|
|
|
|
td::UserId user_id(static_cast<td::int64>(1));
|
2019-09-20 01:22:42 +02:00
|
|
|
|
for (td::int32 length = 1; length <= 3; length++) {
|
|
|
|
|
for (td::int32 offset = 0; static_cast<size_t>(offset + length) <= str.size(); offset++) {
|
|
|
|
|
for (auto type : {td::MessageEntity::Type::Bold, td::MessageEntity::Type::Url, td::MessageEntity::Type::TextUrl,
|
|
|
|
|
td::MessageEntity::Type::MentionName}) {
|
2023-02-11 20:37:43 +01:00
|
|
|
|
for (auto skip_trim : {false, true}) {
|
|
|
|
|
fixed_str = skip_trim ? "aba \n caba " : "aba \n caba";
|
2019-01-16 19:18:26 +01:00
|
|
|
|
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--;
|
|
|
|
|
}
|
2020-05-11 00:31:29 +02:00
|
|
|
|
if (type == td::MessageEntity::Type::Bold || type == td::MessageEntity::Type::Url) {
|
|
|
|
|
while (fixed_length > 0 && (fixed_str[fixed_offset] == ' ' || fixed_str[fixed_offset] == '\n')) {
|
|
|
|
|
fixed_offset++;
|
|
|
|
|
fixed_length--;
|
|
|
|
|
}
|
|
|
|
|
}
|
2019-01-16 19:18:26 +01:00
|
|
|
|
|
2019-09-20 01:22:42 +02:00
|
|
|
|
td::vector<td::MessageEntity> entities;
|
2019-01-16 19:18:26 +01:00
|
|
|
|
entities.emplace_back(type, offset, length);
|
2022-08-19 14:05:49 +02:00
|
|
|
|
if (type == td::MessageEntity::Type::TextUrl) {
|
|
|
|
|
entities.back().argument = "t.me";
|
|
|
|
|
} else if (type == td::MessageEntity::Type::MentionName) {
|
|
|
|
|
entities.back().user_id = user_id;
|
|
|
|
|
}
|
2019-09-20 01:22:42 +02:00
|
|
|
|
td::vector<td::MessageEntity> fixed_entities;
|
2019-01-16 19:18:26 +01:00
|
|
|
|
if (fixed_length > 0) {
|
|
|
|
|
for (auto i = 0; i < length; i++) {
|
2021-06-09 19:35:02 +02:00
|
|
|
|
if (!td::is_space(str[offset + i]) || type == td::MessageEntity::Type::TextUrl ||
|
|
|
|
|
type == td::MessageEntity::Type::MentionName) {
|
2019-01-16 19:18:26 +01:00
|
|
|
|
fixed_entities.emplace_back(type, fixed_offset, fixed_length);
|
2022-08-19 14:05:49 +02:00
|
|
|
|
if (type == td::MessageEntity::Type::TextUrl) {
|
|
|
|
|
fixed_entities.back().argument = "t.me";
|
|
|
|
|
} else if (type == td::MessageEntity::Type::MentionName) {
|
|
|
|
|
fixed_entities.back().user_id = user_id;
|
|
|
|
|
}
|
2019-01-16 19:18:26 +01:00
|
|
|
|
break;
|
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
}
|
2023-02-11 20:37:43 +01:00
|
|
|
|
check_fix_formatted_text(str, entities, fixed_str, fixed_entities, true, false, false, skip_trim);
|
2019-01-16 19:18:26 +01:00
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
str = "aba caba";
|
2019-09-20 01:22:42 +02:00
|
|
|
|
for (td::int32 length = -10; length <= 10; length++) {
|
|
|
|
|
for (td::int32 offset = -10; offset <= 10; offset++) {
|
|
|
|
|
td::vector<td::MessageEntity> entities;
|
|
|
|
|
entities.emplace_back(td::MessageEntity::Type::Bold, offset, length);
|
2020-02-20 00:26:36 +01:00
|
|
|
|
if (length < 0 || offset < 0 || (length > 0 && static_cast<size_t>(length + offset) > str.size())) {
|
2019-12-02 02:03:43 +01:00
|
|
|
|
check_fix_formatted_text(str, entities, true, false, false, false);
|
|
|
|
|
check_fix_formatted_text(str, entities, false, false, false, true);
|
|
|
|
|
continue;
|
|
|
|
|
}
|
|
|
|
|
|
2020-05-11 00:31:29 +02:00
|
|
|
|
td::vector<td::MessageEntity> fixed_entities;
|
|
|
|
|
if (length > 0) {
|
|
|
|
|
if (offset == 3) {
|
|
|
|
|
if (length >= 2) {
|
|
|
|
|
fixed_entities.emplace_back(td::MessageEntity::Type::Bold, offset + 1, length - 1);
|
|
|
|
|
}
|
|
|
|
|
} else {
|
|
|
|
|
fixed_entities.emplace_back(td::MessageEntity::Type::Bold, offset, length);
|
|
|
|
|
}
|
2019-01-16 19:18:26 +01:00
|
|
|
|
}
|
2020-05-11 00:31:29 +02:00
|
|
|
|
|
2019-01-16 19:18:26 +01:00
|
|
|
|
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);
|
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
|
2020-05-11 00:31:29 +02:00
|
|
|
|
str = "abadcaba";
|
2019-09-20 01:22:42 +02:00
|
|
|
|
for (td::int32 length = 1; length <= 7; length++) {
|
|
|
|
|
for (td::int32 offset = 0; offset <= 8 - length; offset++) {
|
|
|
|
|
for (td::int32 length2 = 1; length2 <= 7; length2++) {
|
|
|
|
|
for (td::int32 offset2 = 0; offset2 <= 8 - length2; offset2++) {
|
2019-01-16 19:18:26 +01:00
|
|
|
|
if (offset != offset2) {
|
2019-09-20 01:22:42 +02:00
|
|
|
|
td::vector<td::MessageEntity> entities;
|
2022-08-19 14:05:49 +02:00
|
|
|
|
entities.emplace_back(td::MessageEntity::Type::TextUrl, offset, length, "t.me");
|
|
|
|
|
entities.emplace_back(td::MessageEntity::Type::TextUrl, offset2, length2, "t.me");
|
|
|
|
|
entities.emplace_back(td::MessageEntity::Type::TextUrl, offset2 + length2, 1);
|
2019-09-20 01:22:42 +02:00
|
|
|
|
td::vector<td::MessageEntity> fixed_entities = entities;
|
2022-08-19 14:05:49 +02:00
|
|
|
|
fixed_entities.pop_back();
|
2019-01-16 19:18:26 +01:00
|
|
|
|
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);
|
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
}
|
2019-09-20 01:14:15 +02:00
|
|
|
|
|
|
|
|
|
for (auto text : {" \n ➡️ ➡️ ➡️ ➡️ \n ", "\n\n\nab cd ef gh "}) {
|
|
|
|
|
str = text;
|
2019-09-20 01:22:42 +02:00
|
|
|
|
td::vector<td::MessageEntity> entities;
|
|
|
|
|
td::vector<td::MessageEntity> fixed_entities;
|
2019-09-20 01:14:15 +02:00
|
|
|
|
|
2019-12-02 02:03:43 +01:00
|
|
|
|
auto length = td::narrow_cast<int>(td::utf8_utf16_length(str));
|
2019-09-20 01:14:15 +02:00
|
|
|
|
for (int i = 0; i < 10; i++) {
|
2019-12-02 02:03:43 +01:00
|
|
|
|
if ((i + 1) * 3 + 2 <= length) {
|
|
|
|
|
entities.emplace_back(td::MessageEntity::Type::Bold, (i + 1) * 3, 2);
|
|
|
|
|
}
|
|
|
|
|
if ((i + 2) * 3 <= length) {
|
|
|
|
|
entities.emplace_back(td::MessageEntity::Type::Italic, (i + 1) * 3 + 2, 1);
|
|
|
|
|
}
|
2019-09-20 01:14:15 +02:00
|
|
|
|
|
|
|
|
|
if (i < 4) {
|
2019-09-20 01:22:42 +02:00
|
|
|
|
fixed_entities.emplace_back(td::MessageEntity::Type::Bold, i * 3, 2);
|
2019-09-20 01:14:15 +02:00
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
|
2022-09-22 11:08:34 +02:00
|
|
|
|
check_fix_formatted_text(str, entities, td::utf8_utf16_substr(str, 3, 11).str(), fixed_entities, false, false,
|
|
|
|
|
false, false);
|
2019-09-20 01:14:15 +02:00
|
|
|
|
}
|
2020-01-02 15:46:19 +01:00
|
|
|
|
|
|
|
|
|
for (td::string text : {"\t", "\r", "\n", "\t ", "\r ", "\n "}) {
|
|
|
|
|
for (auto type : {td::MessageEntity::Type::Bold, td::MessageEntity::Type::TextUrl}) {
|
|
|
|
|
check_fix_formatted_text(text, {{type, 0, 1, "http://telegram.org/"}}, "", {}, true, false, false, true);
|
|
|
|
|
}
|
|
|
|
|
}
|
2020-07-13 14:34:56 +02:00
|
|
|
|
check_fix_formatted_text("\r ", {{td::MessageEntity::Type::Bold, 0, 2}, {td::MessageEntity::Type::Underline, 0, 1}},
|
|
|
|
|
"", {}, true, false, false, true);
|
|
|
|
|
check_fix_formatted_text("a \r", {{td::MessageEntity::Type::Bold, 0, 3}, {td::MessageEntity::Type::Underline, 2, 1}},
|
|
|
|
|
"a ", {{td::MessageEntity::Type::Bold, 0, 2}}, true, false, false, true);
|
|
|
|
|
check_fix_formatted_text("a \r ", {{td::MessageEntity::Type::Bold, 0, 4}, {td::MessageEntity::Type::Underline, 2, 1}},
|
|
|
|
|
"a ", {{td::MessageEntity::Type::Bold, 0, 2}}, true, false, false, true);
|
|
|
|
|
check_fix_formatted_text(
|
|
|
|
|
"a \r b", {{td::MessageEntity::Type::Bold, 0, 5}, {td::MessageEntity::Type::Underline, 2, 1}}, "a b",
|
|
|
|
|
{{td::MessageEntity::Type::Bold, 0, 2}, {td::MessageEntity::Type::Bold, 3, 1}}, true, false, false, true);
|
2020-02-21 12:56:26 +01:00
|
|
|
|
|
2020-03-10 15:06:22 +01:00
|
|
|
|
check_fix_formatted_text("a\rbc\r",
|
|
|
|
|
{{td::MessageEntity::Type::Italic, 0, 1},
|
|
|
|
|
{td::MessageEntity::Type::Bold, 0, 2},
|
|
|
|
|
{td::MessageEntity::Type::Italic, 3, 2},
|
|
|
|
|
{td::MessageEntity::Type::Bold, 3, 1}},
|
|
|
|
|
"abc",
|
|
|
|
|
{{td::MessageEntity::Type::Bold, 0, 1},
|
|
|
|
|
{td::MessageEntity::Type::Italic, 0, 1},
|
|
|
|
|
{td::MessageEntity::Type::Bold, 2, 1},
|
|
|
|
|
{td::MessageEntity::Type::Italic, 2, 1}});
|
2020-03-10 15:22:57 +01:00
|
|
|
|
check_fix_formatted_text("a ", {{td::MessageEntity::Type::Italic, 0, 2}, {td::MessageEntity::Type::Bold, 0, 1}}, "a",
|
|
|
|
|
{{td::MessageEntity::Type::Bold, 0, 1}, {td::MessageEntity::Type::Italic, 0, 1}}, false,
|
|
|
|
|
false, false, false);
|
2020-02-21 12:56:26 +01:00
|
|
|
|
check_fix_formatted_text("abc", {{td::MessageEntity::Type::Italic, 1, 1}, {td::MessageEntity::Type::Italic, 0, 1}},
|
|
|
|
|
"abc", {{td::MessageEntity::Type::Italic, 0, 2}});
|
|
|
|
|
check_fix_formatted_text("abc", {{td::MessageEntity::Type::Italic, 1, 1}, {td::MessageEntity::Type::Italic, 1, 1}},
|
|
|
|
|
"abc", {{td::MessageEntity::Type::Italic, 1, 1}});
|
|
|
|
|
check_fix_formatted_text("abc", {{td::MessageEntity::Type::Italic, 0, 2}, {td::MessageEntity::Type::Italic, 1, 2}},
|
|
|
|
|
"abc", {{td::MessageEntity::Type::Italic, 0, 3}});
|
|
|
|
|
check_fix_formatted_text("abc", {{td::MessageEntity::Type::Italic, 0, 2}, {td::MessageEntity::Type::Italic, 2, 1}},
|
|
|
|
|
"abc", {{td::MessageEntity::Type::Italic, 0, 3}});
|
|
|
|
|
check_fix_formatted_text("abc", {{td::MessageEntity::Type::Italic, 0, 1}, {td::MessageEntity::Type::Italic, 2, 1}},
|
|
|
|
|
"abc", {{td::MessageEntity::Type::Italic, 0, 1}, {td::MessageEntity::Type::Italic, 2, 1}});
|
|
|
|
|
check_fix_formatted_text("abc", {{td::MessageEntity::Type::Italic, 0, 2}, {td::MessageEntity::Type::Bold, 1, 2}},
|
|
|
|
|
"abc",
|
|
|
|
|
{{td::MessageEntity::Type::Italic, 0, 1},
|
|
|
|
|
{td::MessageEntity::Type::Bold, 1, 2},
|
|
|
|
|
{td::MessageEntity::Type::Italic, 1, 1}});
|
|
|
|
|
check_fix_formatted_text("abc", {{td::MessageEntity::Type::Italic, 0, 2}, {td::MessageEntity::Type::Bold, 2, 1}},
|
|
|
|
|
"abc", {{td::MessageEntity::Type::Italic, 0, 2}, {td::MessageEntity::Type::Bold, 2, 1}});
|
|
|
|
|
check_fix_formatted_text("abc", {{td::MessageEntity::Type::Italic, 0, 1}, {td::MessageEntity::Type::Bold, 2, 1}},
|
|
|
|
|
"abc", {{td::MessageEntity::Type::Italic, 0, 1}, {td::MessageEntity::Type::Bold, 2, 1}});
|
2020-04-21 23:57:00 +02:00
|
|
|
|
check_fix_formatted_text("@tests @tests", {{td::MessageEntity::Type::Italic, 0, 13}}, "@tests @tests",
|
|
|
|
|
{{td::MessageEntity::Type::Mention, 0, 6},
|
|
|
|
|
{td::MessageEntity::Type::Italic, 0, 6},
|
|
|
|
|
{td::MessageEntity::Type::Mention, 7, 6},
|
|
|
|
|
{td::MessageEntity::Type::Italic, 7, 6}});
|
2020-02-21 12:56:26 +01:00
|
|
|
|
|
2021-12-28 18:41:37 +01:00
|
|
|
|
// __a~b~__
|
2020-02-21 12:56:26 +01:00
|
|
|
|
check_fix_formatted_text(
|
|
|
|
|
"ab", {{td::MessageEntity::Type::Underline, 0, 2}, {td::MessageEntity::Type::Strikethrough, 1, 1}}, "ab",
|
2020-03-12 14:42:39 +01:00
|
|
|
|
{{td::MessageEntity::Type::Underline, 0, 1},
|
|
|
|
|
{td::MessageEntity::Type::Underline, 1, 1},
|
|
|
|
|
{td::MessageEntity::Type::Strikethrough, 1, 1}});
|
2020-02-21 12:56:26 +01:00
|
|
|
|
check_fix_formatted_text("ab",
|
|
|
|
|
{{td::MessageEntity::Type::Underline, 0, 1},
|
|
|
|
|
{td::MessageEntity::Type::Underline, 1, 1},
|
|
|
|
|
{td::MessageEntity::Type::Strikethrough, 1, 1}},
|
|
|
|
|
"ab",
|
|
|
|
|
{{td::MessageEntity::Type::Underline, 0, 1},
|
|
|
|
|
{td::MessageEntity::Type::Underline, 1, 1},
|
|
|
|
|
{td::MessageEntity::Type::Strikethrough, 1, 1}});
|
|
|
|
|
check_fix_formatted_text(
|
|
|
|
|
"ab", {{td::MessageEntity::Type::Strikethrough, 0, 2}, {td::MessageEntity::Type::Underline, 1, 1}}, "ab",
|
2020-03-12 14:42:39 +01:00
|
|
|
|
{{td::MessageEntity::Type::Strikethrough, 0, 1},
|
|
|
|
|
{td::MessageEntity::Type::Underline, 1, 1},
|
|
|
|
|
{td::MessageEntity::Type::Strikethrough, 1, 1}});
|
2020-02-21 12:56:26 +01:00
|
|
|
|
check_fix_formatted_text("ab",
|
|
|
|
|
{{td::MessageEntity::Type::Strikethrough, 0, 1},
|
|
|
|
|
{td::MessageEntity::Type::Strikethrough, 1, 1},
|
|
|
|
|
{td::MessageEntity::Type::Underline, 1, 1}},
|
|
|
|
|
"ab",
|
|
|
|
|
{{td::MessageEntity::Type::Strikethrough, 0, 1},
|
|
|
|
|
{td::MessageEntity::Type::Underline, 1, 1},
|
|
|
|
|
{td::MessageEntity::Type::Strikethrough, 1, 1}});
|
|
|
|
|
|
2021-12-28 18:41:37 +01:00
|
|
|
|
// __||a||b__
|
|
|
|
|
check_fix_formatted_text("ab", {{td::MessageEntity::Type::Underline, 0, 2}, {td::MessageEntity::Type::Spoiler, 0, 1}},
|
|
|
|
|
"ab",
|
|
|
|
|
{{td::MessageEntity::Type::Underline, 0, 2}, {td::MessageEntity::Type::Spoiler, 0, 1}});
|
|
|
|
|
check_fix_formatted_text("ab",
|
|
|
|
|
{{td::MessageEntity::Type::Underline, 0, 1},
|
|
|
|
|
{td::MessageEntity::Type::Underline, 1, 1},
|
|
|
|
|
{td::MessageEntity::Type::Spoiler, 0, 1}},
|
|
|
|
|
"ab",
|
|
|
|
|
{{td::MessageEntity::Type::Underline, 0, 2}, {td::MessageEntity::Type::Spoiler, 0, 1}});
|
2020-02-21 12:56:26 +01:00
|
|
|
|
|
|
|
|
|
// _*a*_\r_*b*_
|
|
|
|
|
check_fix_formatted_text("a\rb",
|
|
|
|
|
{{td::MessageEntity::Type::Bold, 0, 1},
|
2021-12-28 18:41:37 +01:00
|
|
|
|
{td::MessageEntity::Type::Italic, 0, 1},
|
2020-02-21 12:56:26 +01:00
|
|
|
|
{td::MessageEntity::Type::Bold, 2, 1},
|
2021-12-28 18:41:37 +01:00
|
|
|
|
{td::MessageEntity::Type::Italic, 2, 1}},
|
|
|
|
|
"ab", {{td::MessageEntity::Type::Bold, 0, 2}, {td::MessageEntity::Type::Italic, 0, 2}});
|
2020-02-21 12:56:26 +01:00
|
|
|
|
check_fix_formatted_text("a\nb",
|
|
|
|
|
{{td::MessageEntity::Type::Bold, 0, 1},
|
2021-12-28 18:41:37 +01:00
|
|
|
|
{td::MessageEntity::Type::Italic, 0, 1},
|
2020-02-21 12:56:26 +01:00
|
|
|
|
{td::MessageEntity::Type::Bold, 2, 1},
|
2021-12-28 18:41:37 +01:00
|
|
|
|
{td::MessageEntity::Type::Italic, 2, 1}},
|
2020-02-21 12:56:26 +01:00
|
|
|
|
"a\nb",
|
|
|
|
|
{{td::MessageEntity::Type::Bold, 0, 1},
|
2021-12-28 18:41:37 +01:00
|
|
|
|
{td::MessageEntity::Type::Italic, 0, 1},
|
2020-02-21 12:56:26 +01:00
|
|
|
|
{td::MessageEntity::Type::Bold, 2, 1},
|
2021-12-28 18:41:37 +01:00
|
|
|
|
{td::MessageEntity::Type::Italic, 2, 1}});
|
|
|
|
|
|
|
|
|
|
// ||`a`||
|
|
|
|
|
check_fix_formatted_text("a", {{td::MessageEntity::Type::Pre, 0, 1}, {td::MessageEntity::Type::Spoiler, 0, 1}}, "a",
|
|
|
|
|
{{td::MessageEntity::Type::Pre, 0, 1}});
|
|
|
|
|
check_fix_formatted_text("a", {{td::MessageEntity::Type::Spoiler, 0, 1}, {td::MessageEntity::Type::Pre, 0, 1}}, "a",
|
|
|
|
|
{{td::MessageEntity::Type::Pre, 0, 1}});
|
2020-02-21 12:56:26 +01:00
|
|
|
|
|
|
|
|
|
check_fix_formatted_text("abc",
|
|
|
|
|
{{td::MessageEntity::Type::Pre, 0, 3}, {td::MessageEntity::Type::Strikethrough, 1, 1}},
|
|
|
|
|
"abc", {{td::MessageEntity::Type::Pre, 0, 3}});
|
|
|
|
|
check_fix_formatted_text(
|
|
|
|
|
"abc", {{td::MessageEntity::Type::Pre, 1, 1}, {td::MessageEntity::Type::Strikethrough, 0, 3}}, "abc",
|
|
|
|
|
{{td::MessageEntity::Type::Strikethrough, 0, 1},
|
|
|
|
|
{td::MessageEntity::Type::Pre, 1, 1},
|
|
|
|
|
{td::MessageEntity::Type::Strikethrough, 2, 1}});
|
|
|
|
|
check_fix_formatted_text(
|
|
|
|
|
"abc", {{td::MessageEntity::Type::Pre, 1, 1}, {td::MessageEntity::Type::Strikethrough, 1, 2}}, "abc",
|
|
|
|
|
{{td::MessageEntity::Type::Pre, 1, 1}, {td::MessageEntity::Type::Strikethrough, 2, 1}});
|
|
|
|
|
check_fix_formatted_text(
|
|
|
|
|
"abc", {{td::MessageEntity::Type::Pre, 1, 1}, {td::MessageEntity::Type::Strikethrough, 0, 2}}, "abc",
|
|
|
|
|
{{td::MessageEntity::Type::Strikethrough, 0, 1}, {td::MessageEntity::Type::Pre, 1, 1}});
|
|
|
|
|
check_fix_formatted_text("abc", {{td::MessageEntity::Type::Pre, 0, 3}, {td::MessageEntity::Type::BlockQuote, 1, 1}},
|
|
|
|
|
"abc", {{td::MessageEntity::Type::BlockQuote, 1, 1}});
|
|
|
|
|
check_fix_formatted_text("abc", {{td::MessageEntity::Type::BlockQuote, 0, 3}, {td::MessageEntity::Type::Pre, 1, 1}},
|
|
|
|
|
"abc", {{td::MessageEntity::Type::BlockQuote, 0, 3}, {td::MessageEntity::Type::Pre, 1, 1}});
|
2020-02-21 14:22:40 +01:00
|
|
|
|
|
|
|
|
|
check_fix_formatted_text("example.com", {}, "example.com", {{td::MessageEntity::Type::Url, 0, 11}});
|
2020-02-24 18:26:08 +01:00
|
|
|
|
check_fix_formatted_text("example.com", {{td::MessageEntity::Type::Pre, 0, 3}}, "example.com",
|
|
|
|
|
{{td::MessageEntity::Type::Pre, 0, 3}});
|
|
|
|
|
check_fix_formatted_text("example.com", {{td::MessageEntity::Type::BlockQuote, 0, 3}}, "example.com",
|
|
|
|
|
{{td::MessageEntity::Type::BlockQuote, 0, 3}});
|
|
|
|
|
check_fix_formatted_text("example.com", {{td::MessageEntity::Type::BlockQuote, 0, 11}}, "example.com",
|
|
|
|
|
{{td::MessageEntity::Type::BlockQuote, 0, 11}, {td::MessageEntity::Type::Url, 0, 11}});
|
|
|
|
|
check_fix_formatted_text("example.com", {{td::MessageEntity::Type::Italic, 0, 11}}, "example.com",
|
|
|
|
|
{{td::MessageEntity::Type::Url, 0, 11}, {td::MessageEntity::Type::Italic, 0, 11}});
|
|
|
|
|
check_fix_formatted_text("example.com", {{td::MessageEntity::Type::Italic, 0, 3}}, "example.com",
|
|
|
|
|
{{td::MessageEntity::Type::Url, 0, 11}, {td::MessageEntity::Type::Italic, 0, 3}});
|
|
|
|
|
check_fix_formatted_text("example.com a", {{td::MessageEntity::Type::Italic, 0, 13}}, "example.com a",
|
|
|
|
|
{{td::MessageEntity::Type::Url, 0, 11},
|
|
|
|
|
{td::MessageEntity::Type::Italic, 0, 11},
|
2020-05-11 00:31:29 +02:00
|
|
|
|
{td::MessageEntity::Type::Italic, 12, 1}});
|
2020-02-24 18:26:08 +01:00
|
|
|
|
check_fix_formatted_text("a example.com", {{td::MessageEntity::Type::Italic, 0, 13}}, "a example.com",
|
|
|
|
|
{{td::MessageEntity::Type::Italic, 0, 2},
|
|
|
|
|
{td::MessageEntity::Type::Url, 2, 11},
|
|
|
|
|
{td::MessageEntity::Type::Italic, 2, 11}});
|
2020-02-21 14:22:40 +01:00
|
|
|
|
|
2020-02-24 23:28:33 +01:00
|
|
|
|
for (size_t test_n = 0; test_n < 100000; test_n++) {
|
2020-10-08 11:56:08 +02:00
|
|
|
|
bool is_url = td::Random::fast_bool();
|
2020-02-24 18:26:08 +01:00
|
|
|
|
td::int32 url_offset = 0;
|
|
|
|
|
td::int32 url_end = 0;
|
|
|
|
|
if (is_url) {
|
|
|
|
|
str = td::string(td::Random::fast(1, 5), 'a') + ":example.com:" + td::string(td::Random::fast(1, 5), 'a');
|
|
|
|
|
url_offset = static_cast<td::int32>(str.find('e'));
|
|
|
|
|
url_end = url_offset + 11;
|
|
|
|
|
} else {
|
|
|
|
|
str = td::string(td::Random::fast(1, 20), 'a');
|
|
|
|
|
}
|
2020-02-21 14:22:40 +01:00
|
|
|
|
|
2020-02-21 15:10:10 +01:00
|
|
|
|
auto n = td::Random::fast(1, 20);
|
2020-02-21 14:22:40 +01:00
|
|
|
|
td::vector<td::MessageEntity> entities;
|
|
|
|
|
for (int j = 0; j < n; j++) {
|
2023-10-18 12:14:07 +02:00
|
|
|
|
td::int32 type = td::Random::fast(4, static_cast<int>(td::MessageEntity::Type::Size) - 1);
|
2020-02-21 14:22:40 +01:00
|
|
|
|
td::int32 offset = td::Random::fast(0, static_cast<int>(str.size()) - 1);
|
|
|
|
|
auto max_length = static_cast<int>(str.size() - offset);
|
2020-02-24 23:28:33 +01:00
|
|
|
|
if ((test_n & 1) != 0 && max_length > 4) {
|
2020-02-21 14:22:40 +01:00
|
|
|
|
max_length = 4;
|
|
|
|
|
}
|
|
|
|
|
td::int32 length = td::Random::fast(0, max_length);
|
|
|
|
|
entities.emplace_back(static_cast<td::MessageEntity::Type>(type), offset, length);
|
|
|
|
|
}
|
2020-02-21 15:10:10 +01:00
|
|
|
|
|
|
|
|
|
auto get_type_mask = [](std::size_t length, const td::vector<td::MessageEntity> &entities) {
|
|
|
|
|
td::vector<td::int32> result(length);
|
|
|
|
|
for (auto &entity : entities) {
|
|
|
|
|
for (auto pos = 0; pos < entity.length; pos++) {
|
2020-02-24 18:26:08 +01:00
|
|
|
|
result[entity.offset + pos] |= (1 << static_cast<td::int32>(entity.type));
|
2020-02-21 15:10:10 +01:00
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
return result;
|
|
|
|
|
};
|
|
|
|
|
auto old_type_mask = get_type_mask(str.size(), entities);
|
2021-07-26 21:48:33 +02:00
|
|
|
|
ASSERT_TRUE(td::fix_formatted_text(str, entities, false, false, true, true, false).is_ok());
|
2020-02-21 15:10:10 +01:00
|
|
|
|
auto new_type_mask = get_type_mask(str.size(), entities);
|
2023-10-18 12:14:07 +02:00
|
|
|
|
auto splittable_mask = (1 << 5) | (1 << 6) | (1 << 14) | (1 << 15) | (1 << 19);
|
2020-02-24 23:28:33 +01:00
|
|
|
|
auto pre_mask = (1 << 7) | (1 << 8) | (1 << 9);
|
2020-02-21 15:10:10 +01:00
|
|
|
|
for (std::size_t pos = 0; pos < str.size(); pos++) {
|
2020-02-24 23:28:33 +01:00
|
|
|
|
if ((new_type_mask[pos] & pre_mask) != 0) {
|
2020-02-24 18:26:08 +01:00
|
|
|
|
ASSERT_EQ(0, new_type_mask[pos] & splittable_mask);
|
2020-02-21 15:10:10 +01:00
|
|
|
|
} else {
|
2020-02-24 18:26:08 +01:00
|
|
|
|
ASSERT_EQ(old_type_mask[pos] & splittable_mask, new_type_mask[pos] & splittable_mask);
|
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
bool keep_url = is_url;
|
|
|
|
|
td::MessageEntity url_entity(td::MessageEntity::Type::Url, url_offset, url_end - url_offset);
|
|
|
|
|
for (auto &entity : entities) {
|
|
|
|
|
if (entity == url_entity) {
|
|
|
|
|
continue;
|
|
|
|
|
}
|
|
|
|
|
td::int32 offset = entity.offset;
|
|
|
|
|
td::int32 end = offset + entity.length;
|
|
|
|
|
|
|
|
|
|
if (keep_url && ((1 << static_cast<td::int32>(entity.type)) & splittable_mask) == 0 &&
|
|
|
|
|
!(end <= url_offset || url_end <= offset)) {
|
|
|
|
|
keep_url = (entity.type == td::MessageEntity::Type::BlockQuote && offset <= url_offset && url_end <= end);
|
2020-02-21 15:10:10 +01:00
|
|
|
|
}
|
|
|
|
|
}
|
2020-02-24 18:26:08 +01:00
|
|
|
|
ASSERT_EQ(keep_url, std::count(entities.begin(), entities.end(), url_entity) == 1);
|
2020-02-24 23:28:33 +01:00
|
|
|
|
|
|
|
|
|
for (size_t i = 0; i < entities.size(); i++) {
|
|
|
|
|
auto type_mask = 1 << static_cast<td::int32>(entities[i].type);
|
|
|
|
|
for (size_t j = i + 1; j < entities.size(); j++) {
|
|
|
|
|
// sorted
|
|
|
|
|
ASSERT_TRUE(entities[j].offset > entities[i].offset ||
|
|
|
|
|
(entities[j].offset == entities[i].offset && entities[j].length <= entities[i].length));
|
|
|
|
|
|
|
|
|
|
// not intersecting
|
|
|
|
|
ASSERT_TRUE(entities[j].offset >= entities[i].offset + entities[i].length ||
|
|
|
|
|
entities[j].offset + entities[j].length <= entities[i].offset + entities[i].length);
|
|
|
|
|
|
|
|
|
|
if (entities[j].offset < entities[i].offset + entities[i].length) { // if nested
|
|
|
|
|
// types are different
|
|
|
|
|
ASSERT_TRUE(entities[j].type != entities[i].type);
|
|
|
|
|
|
|
|
|
|
// pre can't contain other entities
|
|
|
|
|
ASSERT_TRUE((type_mask & pre_mask) == 0);
|
|
|
|
|
|
|
|
|
|
if ((type_mask & splittable_mask) == 0 && entities[i].type != td::MessageEntity::Type::BlockQuote) {
|
|
|
|
|
// continuous entities can contain only splittable entities
|
|
|
|
|
ASSERT_TRUE(((1 << static_cast<td::int32>(entities[j].type)) & splittable_mask) != 0);
|
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
}
|
2020-02-21 14:22:40 +01:00
|
|
|
|
}
|
2020-04-08 00:23:05 +02:00
|
|
|
|
|
|
|
|
|
check_fix_formatted_text(
|
|
|
|
|
"\xe2\x80\x8f\xe2\x80\x8f \xe2\x80\x8e\xe2\x80\x8e\xe2\x80\x8e\xe2\x80\x8c \xe2\x80\x8f\xe2\x80\x8e "
|
|
|
|
|
"\xe2\x80\x8f",
|
|
|
|
|
{},
|
|
|
|
|
"\xe2\x80\x8c\xe2\x80\x8f \xe2\x80\x8c\xe2\x80\x8c\xe2\x80\x8e\xe2\x80\x8c \xe2\x80\x8c\xe2\x80\x8e "
|
|
|
|
|
"\xe2\x80\x8f",
|
|
|
|
|
{});
|
2019-01-16 19:18:26 +01:00
|
|
|
|
}
|
2019-09-26 17:36:45 +02:00
|
|
|
|
|
2023-10-19 12:45:39 +02:00
|
|
|
|
TEST(MessageEntities, is_visible_url) {
|
2023-10-19 12:47:41 +02:00
|
|
|
|
td::string str = "a telegram.org telegran.org telegrao.org telegram.orc telegrap.org c";
|
2023-10-19 12:45:39 +02:00
|
|
|
|
td::vector<td::MessageEntity> entities;
|
|
|
|
|
entities.emplace_back(td::MessageEntity::Type::TextUrl, 0, 1, "telegrab.org");
|
2023-10-19 12:47:41 +02:00
|
|
|
|
entities.emplace_back(td::MessageEntity::Type::TextUrl, str.size() - 1, 1, "telegrax.org");
|
2023-10-19 12:45:39 +02:00
|
|
|
|
td::fix_formatted_text(str, entities, false, false, false, false, true).ensure();
|
|
|
|
|
td::FormattedText text{std::move(str), std::move(entities)};
|
2023-10-19 12:46:36 +02:00
|
|
|
|
ASSERT_EQ(td::get_first_url(text), "telegrab.org");
|
2023-10-19 12:45:39 +02:00
|
|
|
|
ASSERT_TRUE(!td::is_visible_url(text, "telegrab.org"));
|
|
|
|
|
ASSERT_TRUE(td::is_visible_url(text, "telegram.org"));
|
|
|
|
|
ASSERT_TRUE(td::is_visible_url(text, "telegran.org"));
|
|
|
|
|
ASSERT_TRUE(td::is_visible_url(text, "telegrao.org"));
|
|
|
|
|
ASSERT_TRUE(!td::is_visible_url(text, "telegram.orc"));
|
|
|
|
|
ASSERT_TRUE(td::is_visible_url(text, "telegrap.org"));
|
|
|
|
|
ASSERT_TRUE(!td::is_visible_url(text, "telegraf.org"));
|
2023-10-19 12:47:41 +02:00
|
|
|
|
ASSERT_TRUE(!td::is_visible_url(text, "telegrax.org"));
|
2023-10-19 12:45:39 +02:00
|
|
|
|
}
|
|
|
|
|
|
2019-09-26 17:36:45 +02:00
|
|
|
|
static void check_parse_html(td::string text, const td::string &result, const td::vector<td::MessageEntity> &entities) {
|
|
|
|
|
auto r_entities = td::parse_html(text);
|
|
|
|
|
ASSERT_TRUE(r_entities.is_ok());
|
|
|
|
|
ASSERT_EQ(entities, r_entities.ok());
|
|
|
|
|
ASSERT_STREQ(result, text);
|
|
|
|
|
}
|
|
|
|
|
|
2019-10-03 01:31:06 +02:00
|
|
|
|
static void check_parse_html(td::string text, td::Slice error_message) {
|
2019-09-26 17:36:45 +02:00
|
|
|
|
auto r_entities = td::parse_html(text);
|
|
|
|
|
ASSERT_TRUE(r_entities.is_error());
|
|
|
|
|
ASSERT_EQ(400, r_entities.error().code());
|
|
|
|
|
ASSERT_STREQ(error_message, r_entities.error().message());
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
TEST(MessageEntities, parse_html) {
|
|
|
|
|
td::string invalid_surrogate_pair_error_message =
|
|
|
|
|
"Text contains invalid Unicode characters after decoding HTML entities, check for unmatched surrogate code units";
|
|
|
|
|
check_parse_html("�", invalid_surrogate_pair_error_message);
|
|
|
|
|
check_parse_html("�", invalid_surrogate_pair_error_message);
|
|
|
|
|
check_parse_html("�", invalid_surrogate_pair_error_message);
|
|
|
|
|
check_parse_html("🏟 🏟<<abacaba", "Unclosed start tag at byte offset 13");
|
|
|
|
|
check_parse_html("🏟 🏟<<abac aba>", "Unsupported start tag \"abac\" at byte offset 13");
|
|
|
|
|
check_parse_html("🏟 🏟<<abac>", "Unsupported start tag \"abac\" at byte offset 13");
|
|
|
|
|
check_parse_html("🏟 🏟<<i =aba>", "Empty attribute name in the tag \"i\" at byte offset 13");
|
|
|
|
|
check_parse_html("🏟 🏟<<i aba>",
|
|
|
|
|
"Expected equal sign in declaration of an attribute of the tag \"i\" at byte offset 13");
|
|
|
|
|
check_parse_html("🏟 🏟<<i aba = ", "Unclosed start tag \"i\" at byte offset 13");
|
|
|
|
|
check_parse_html("🏟 🏟<<i aba = 190azAz-.,", "Unexpected end of name token at byte offset 27");
|
|
|
|
|
check_parse_html("🏟 🏟<<i aba = \"<>">", "Unclosed start tag at byte offset 13");
|
|
|
|
|
check_parse_html("🏟 🏟<<i aba = \'<>">", "Unclosed start tag at byte offset 13");
|
|
|
|
|
check_parse_html("🏟 🏟<</", "Unexpected end tag at byte offset 13");
|
|
|
|
|
check_parse_html("🏟 🏟<<b></b></", "Unexpected end tag at byte offset 20");
|
|
|
|
|
check_parse_html("🏟 🏟<<i>a</i ", "Unclosed end tag at byte offset 17");
|
2023-02-08 13:28:36 +01:00
|
|
|
|
check_parse_html("🏟 🏟<<i>a</em >", "Unmatched end tag at byte offset 17, expected \"</i>\", found \"</em>\"");
|
2019-09-26 17:36:45 +02:00
|
|
|
|
|
|
|
|
|
check_parse_html("", "", {});
|
|
|
|
|
check_parse_html("➡️ ➡️", "➡️ ➡️", {});
|
2023-01-09 15:07:08 +01:00
|
|
|
|
check_parse_html("≥<>&"«»�", "≥<>&\"«»�", {});
|
|
|
|
|
check_parse_html("⩔", "⩔", {});
|
2023-02-08 13:28:36 +01:00
|
|
|
|
check_parse_html("➡️ ➡️<i>➡️ ➡️</i>", "➡️ ➡️➡️ ➡️", {{td::MessageEntity::Type::Italic, 5, 5}});
|
|
|
|
|
check_parse_html("➡️ ➡️<em>➡️ ➡️</em>", "➡️ ➡️➡️ ➡️", {{td::MessageEntity::Type::Italic, 5, 5}});
|
|
|
|
|
check_parse_html("➡️ ➡️<b>➡️ ➡️</b>", "➡️ ➡️➡️ ➡️", {{td::MessageEntity::Type::Bold, 5, 5}});
|
|
|
|
|
check_parse_html("➡️ ➡️<strong>➡️ ➡️</strong>", "➡️ ➡️➡️ ➡️", {{td::MessageEntity::Type::Bold, 5, 5}});
|
|
|
|
|
check_parse_html("➡️ ➡️<u>➡️ ➡️</u>", "➡️ ➡️➡️ ➡️", {{td::MessageEntity::Type::Underline, 5, 5}});
|
|
|
|
|
check_parse_html("➡️ ➡️<ins>➡️ ➡️</ins>", "➡️ ➡️➡️ ➡️", {{td::MessageEntity::Type::Underline, 5, 5}});
|
|
|
|
|
check_parse_html("➡️ ➡️<s>➡️ ➡️</s>", "➡️ ➡️➡️ ➡️", {{td::MessageEntity::Type::Strikethrough, 5, 5}});
|
|
|
|
|
check_parse_html("➡️ ➡️<strike>➡️ ➡️</strike>", "➡️ ➡️➡️ ➡️", {{td::MessageEntity::Type::Strikethrough, 5, 5}});
|
|
|
|
|
check_parse_html("➡️ ➡️<del>➡️ ➡️</del>", "➡️ ➡️➡️ ➡️", {{td::MessageEntity::Type::Strikethrough, 5, 5}});
|
2019-10-03 01:31:06 +02:00
|
|
|
|
check_parse_html("➡️ ➡️<i>➡️ ➡️</i><b>➡️ ➡️</b>", "➡️ ➡️➡️ ➡️➡️ ➡️",
|
|
|
|
|
{{td::MessageEntity::Type::Italic, 5, 5}, {td::MessageEntity::Type::Bold, 10, 5}});
|
2019-09-26 17:36:45 +02:00
|
|
|
|
check_parse_html("🏟 🏟<i>🏟 <🏟</i>", "🏟 🏟🏟 <🏟", {{td::MessageEntity::Type::Italic, 5, 6}});
|
|
|
|
|
check_parse_html("🏟 🏟<i>🏟 ><b aba = caba><🏟</b></i>", "🏟 🏟🏟 ><🏟",
|
|
|
|
|
{{td::MessageEntity::Type::Italic, 5, 7}, {td::MessageEntity::Type::Bold, 9, 3}});
|
2023-02-08 13:28:36 +01:00
|
|
|
|
check_parse_html("🏟 🏟<<i aba = 190azAz-. >a</i>", "🏟 🏟<a", {{td::MessageEntity::Type::Italic, 6, 1}});
|
|
|
|
|
check_parse_html("🏟 🏟<<i aba = 190azAz-.>a</i>", "🏟 🏟<a", {{td::MessageEntity::Type::Italic, 6, 1}});
|
2019-09-26 17:36:45 +02:00
|
|
|
|
check_parse_html("🏟 🏟<<i aba = \"<>"\">a</i>", "🏟 🏟<a",
|
|
|
|
|
{{td::MessageEntity::Type::Italic, 6, 1}});
|
|
|
|
|
check_parse_html("🏟 🏟<<i aba = '<>"'>a</i>", "🏟 🏟<a",
|
|
|
|
|
{{td::MessageEntity::Type::Italic, 6, 1}});
|
|
|
|
|
check_parse_html("🏟 🏟<<i aba = '<>"'>a</>", "🏟 🏟<a",
|
|
|
|
|
{{td::MessageEntity::Type::Italic, 6, 1}});
|
2023-02-08 13:28:36 +01:00
|
|
|
|
check_parse_html("🏟 🏟<<i>🏟 🏟<</>", "🏟 🏟<🏟 🏟<", {{td::MessageEntity::Type::Italic, 6, 6}});
|
2019-09-26 17:36:45 +02:00
|
|
|
|
check_parse_html("🏟 🏟<<i>a</ >", "🏟 🏟<a", {{td::MessageEntity::Type::Italic, 6, 1}});
|
|
|
|
|
check_parse_html("🏟 🏟<<i>a</i >", "🏟 🏟<a", {{td::MessageEntity::Type::Italic, 6, 1}});
|
|
|
|
|
check_parse_html("🏟 🏟<<b></b>", "🏟 🏟<", {});
|
2020-01-02 15:46:19 +01:00
|
|
|
|
check_parse_html("<i>\t</i>", "\t", {{td::MessageEntity::Type::Italic, 0, 1}});
|
|
|
|
|
check_parse_html("<i>\r</i>", "\r", {{td::MessageEntity::Type::Italic, 0, 1}});
|
|
|
|
|
check_parse_html("<i>\n</i>", "\n", {{td::MessageEntity::Type::Italic, 0, 1}});
|
2023-02-08 13:28:36 +01:00
|
|
|
|
check_parse_html("➡️ ➡️<span class = \"tg-spoiler\">➡️ ➡️</span><b>➡️ ➡️</b>", "➡️ ➡️➡️ ➡️➡️ ➡️",
|
2021-12-28 18:41:37 +01:00
|
|
|
|
{{td::MessageEntity::Type::Spoiler, 5, 5}, {td::MessageEntity::Type::Bold, 10, 5}});
|
|
|
|
|
check_parse_html("🏟 🏟<span class=\"tg-spoiler\">🏟 <🏟</span>", "🏟 🏟🏟 <🏟",
|
|
|
|
|
{{td::MessageEntity::Type::Spoiler, 5, 6}});
|
2023-02-08 13:28:36 +01:00
|
|
|
|
check_parse_html("🏟 🏟<span class=\"tg-spoiler\">🏟 ><b aba = caba><🏟</b></span>", "🏟 🏟🏟 ><🏟",
|
2021-12-28 18:41:37 +01:00
|
|
|
|
{{td::MessageEntity::Type::Spoiler, 5, 7}, {td::MessageEntity::Type::Bold, 9, 3}});
|
2023-02-08 13:28:36 +01:00
|
|
|
|
check_parse_html("➡️ ➡️<tg-spoiler>➡️ ➡️</tg-spoiler><b>➡️ ➡️</b>", "➡️ ➡️➡️ ➡️➡️ ➡️",
|
2021-12-31 15:15:11 +01:00
|
|
|
|
{{td::MessageEntity::Type::Spoiler, 5, 5}, {td::MessageEntity::Type::Bold, 10, 5}});
|
2023-02-08 13:28:36 +01:00
|
|
|
|
check_parse_html("🏟 🏟<tg-spoiler>🏟 <🏟</tg-spoiler>", "🏟 🏟🏟 <🏟", {{td::MessageEntity::Type::Spoiler, 5, 6}});
|
2021-12-31 15:15:11 +01:00
|
|
|
|
check_parse_html("🏟 🏟<tg-spoiler>🏟 ><b aba = caba><🏟</b></tg-spoiler>", "🏟 🏟🏟 ><🏟",
|
|
|
|
|
{{td::MessageEntity::Type::Spoiler, 5, 7}, {td::MessageEntity::Type::Bold, 9, 3}});
|
2020-01-02 15:46:19 +01:00
|
|
|
|
check_parse_html("<a href=telegram.org>\t</a>", "\t",
|
|
|
|
|
{{td::MessageEntity::Type::TextUrl, 0, 1, "http://telegram.org/"}});
|
|
|
|
|
check_parse_html("<a href=telegram.org>\r</a>", "\r",
|
|
|
|
|
{{td::MessageEntity::Type::TextUrl, 0, 1, "http://telegram.org/"}});
|
|
|
|
|
check_parse_html("<a href=telegram.org>\n</a>", "\n",
|
|
|
|
|
{{td::MessageEntity::Type::TextUrl, 0, 1, "http://telegram.org/"}});
|
2019-09-26 17:36:45 +02:00
|
|
|
|
check_parse_html("<code><i><b> </b></i></code><i><b><code> </code></b></i>", " ",
|
|
|
|
|
{{td::MessageEntity::Type::Code, 0, 1},
|
|
|
|
|
{td::MessageEntity::Type::Bold, 0, 1},
|
|
|
|
|
{td::MessageEntity::Type::Italic, 0, 1},
|
|
|
|
|
{td::MessageEntity::Type::Code, 1, 1},
|
|
|
|
|
{td::MessageEntity::Type::Bold, 1, 1},
|
|
|
|
|
{td::MessageEntity::Type::Italic, 1, 1}});
|
|
|
|
|
check_parse_html("<i><b> </b> <code> </code></i>", " ",
|
|
|
|
|
{{td::MessageEntity::Type::Italic, 0, 3},
|
|
|
|
|
{td::MessageEntity::Type::Bold, 0, 1},
|
|
|
|
|
{td::MessageEntity::Type::Code, 2, 1}});
|
|
|
|
|
check_parse_html("<a href=telegram.org> </a>", " ",
|
|
|
|
|
{{td::MessageEntity::Type::TextUrl, 0, 1, "http://telegram.org/"}});
|
|
|
|
|
check_parse_html("<a href =\"telegram.org\" > </a>", " ",
|
|
|
|
|
{{td::MessageEntity::Type::TextUrl, 0, 1, "http://telegram.org/"}});
|
|
|
|
|
check_parse_html("<a href= 'telegram.org' > </a>", " ",
|
|
|
|
|
{{td::MessageEntity::Type::TextUrl, 0, 1, "http://telegram.org/"}});
|
|
|
|
|
check_parse_html("<a href= 'telegram.org?<' > </a>", " ",
|
|
|
|
|
{{td::MessageEntity::Type::TextUrl, 0, 1, "http://telegram.org/?<"}});
|
|
|
|
|
check_parse_html("<a> </a>", " ", {});
|
|
|
|
|
check_parse_html("<a>telegram.org </a>", "telegram.org ", {});
|
|
|
|
|
check_parse_html("<a>telegram.org</a>", "telegram.org",
|
|
|
|
|
{{td::MessageEntity::Type::TextUrl, 0, 12, "http://telegram.org/"}});
|
|
|
|
|
check_parse_html("<a>https://telegram.org/asdsa?asdasdwe#12e3we</a>", "https://telegram.org/asdsa?asdasdwe#12e3we",
|
|
|
|
|
{{td::MessageEntity::Type::TextUrl, 0, 42, "https://telegram.org/asdsa?asdasdwe#12e3we"}});
|
2023-02-08 13:28:36 +01:00
|
|
|
|
check_parse_html("🏟 🏟<<pre >🏟 🏟<</>", "🏟 🏟<🏟 🏟<", {{td::MessageEntity::Type::Pre, 6, 6}});
|
|
|
|
|
check_parse_html("🏟 🏟<<code >🏟 🏟<</>", "🏟 🏟<🏟 🏟<", {{td::MessageEntity::Type::Code, 6, 6}});
|
2019-10-03 15:39:50 +02:00
|
|
|
|
check_parse_html("🏟 🏟<<pre><code>🏟 🏟<</code></>", "🏟 🏟<🏟 🏟<",
|
|
|
|
|
{{td::MessageEntity::Type::Pre, 6, 6}, {td::MessageEntity::Type::Code, 6, 6}});
|
|
|
|
|
check_parse_html("🏟 🏟<<pre><code class=\"language-\">🏟 🏟<</code></>", "🏟 🏟<🏟 🏟<",
|
|
|
|
|
{{td::MessageEntity::Type::Pre, 6, 6}, {td::MessageEntity::Type::Code, 6, 6}});
|
|
|
|
|
check_parse_html("🏟 🏟<<pre><code class=\"language-fift\">🏟 🏟<</></>", "🏟 🏟<🏟 🏟<",
|
|
|
|
|
{{td::MessageEntity::Type::PreCode, 6, 6, "fift"}});
|
|
|
|
|
check_parse_html("🏟 🏟<<code class=\"language-fift\"><pre>🏟 🏟<</></>", "🏟 🏟<🏟 🏟<",
|
|
|
|
|
{{td::MessageEntity::Type::PreCode, 6, 6, "fift"}});
|
|
|
|
|
check_parse_html("🏟 🏟<<pre><code class=\"language-fift\">🏟 🏟<</> </>", "🏟 🏟<🏟 🏟< ",
|
|
|
|
|
{{td::MessageEntity::Type::Pre, 6, 7}, {td::MessageEntity::Type::Code, 6, 6}});
|
|
|
|
|
check_parse_html("🏟 🏟<<pre> <code class=\"language-fift\">🏟 🏟<</></>", "🏟 🏟< 🏟 🏟<",
|
|
|
|
|
{{td::MessageEntity::Type::Pre, 6, 7}, {td::MessageEntity::Type::Code, 7, 6}});
|
2023-02-08 13:28:36 +01:00
|
|
|
|
check_parse_html("➡️ ➡️<tg-emoji emoji-id = \"12345\">➡️ ➡️</tg-emoji><b>➡️ ➡️</b>", "➡️ ➡️➡️ ➡️➡️ ➡️",
|
2022-10-26 07:53:09 +02:00
|
|
|
|
{{td::MessageEntity::Type::CustomEmoji, 5, 5, td::CustomEmojiId(static_cast<td::int64>(12345))},
|
2022-07-19 13:13:37 +02:00
|
|
|
|
{td::MessageEntity::Type::Bold, 10, 5}});
|
|
|
|
|
check_parse_html("🏟 🏟<tg-emoji emoji-id=\"54321\">🏟 <🏟</tg-emoji>", "🏟 🏟🏟 <🏟",
|
2022-10-26 07:53:09 +02:00
|
|
|
|
{{td::MessageEntity::Type::CustomEmoji, 5, 6, td::CustomEmojiId(static_cast<td::int64>(54321))}});
|
|
|
|
|
check_parse_html("🏟 🏟<b aba = caba><tg-emoji emoji-id=\"1\">🏟</tg-emoji>1</b>", "🏟 🏟🏟1",
|
|
|
|
|
{{td::MessageEntity::Type::Bold, 5, 3},
|
|
|
|
|
{td::MessageEntity::Type::CustomEmoji, 5, 2, td::CustomEmojiId(static_cast<td::int64>(1))}});
|
2019-09-26 17:36:45 +02:00
|
|
|
|
}
|
2019-10-03 01:31:06 +02:00
|
|
|
|
|
|
|
|
|
static void check_parse_markdown(td::string text, const td::string &result,
|
|
|
|
|
const td::vector<td::MessageEntity> &entities) {
|
|
|
|
|
auto r_entities = td::parse_markdown_v2(text);
|
|
|
|
|
ASSERT_TRUE(r_entities.is_ok());
|
|
|
|
|
ASSERT_EQ(entities, r_entities.ok());
|
|
|
|
|
ASSERT_STREQ(result, text);
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
static void check_parse_markdown(td::string text, td::Slice error_message) {
|
|
|
|
|
auto r_entities = td::parse_markdown_v2(text);
|
|
|
|
|
ASSERT_TRUE(r_entities.is_error());
|
|
|
|
|
ASSERT_EQ(400, r_entities.error().code());
|
|
|
|
|
ASSERT_STREQ(error_message, r_entities.error().message());
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
TEST(MessageEntities, parse_markdown) {
|
2019-12-24 23:57:33 +01:00
|
|
|
|
td::Slice reserved_characters("]()>#+-=|{}.!");
|
2019-10-03 01:31:06 +02:00
|
|
|
|
td::Slice begin_characters("_*[~`");
|
|
|
|
|
for (char c = 1; c < 126; c++) {
|
|
|
|
|
if (begin_characters.find(c) != td::Slice::npos) {
|
|
|
|
|
continue;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
td::string text(1, c);
|
|
|
|
|
if (reserved_characters.find(c) == td::Slice::npos) {
|
|
|
|
|
check_parse_markdown(text, text, {});
|
|
|
|
|
} else {
|
|
|
|
|
check_parse_markdown(
|
|
|
|
|
text, PSLICE() << "Character '" << c << "' is reserved and must be escaped with the preceding '\\'");
|
|
|
|
|
|
|
|
|
|
td::string escaped_text = "\\" + text;
|
|
|
|
|
check_parse_markdown(escaped_text, text, {});
|
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
check_parse_markdown("🏟 🏟_abacaba", "Can't find end of Italic entity at byte offset 9");
|
|
|
|
|
check_parse_markdown("🏟 🏟_abac * asd ", "Can't find end of Bold entity at byte offset 15");
|
|
|
|
|
check_parse_markdown("🏟 🏟_abac * asd _", "Can't find end of Italic entity at byte offset 21");
|
|
|
|
|
check_parse_markdown("🏟 🏟`", "Can't find end of Code entity at byte offset 9");
|
|
|
|
|
check_parse_markdown("🏟 🏟```", "Can't find end of Pre entity at byte offset 9");
|
|
|
|
|
check_parse_markdown("🏟 🏟```a", "Can't find end of Pre entity at byte offset 9");
|
|
|
|
|
check_parse_markdown("🏟 🏟```a ", "Can't find end of PreCode entity at byte offset 9");
|
|
|
|
|
check_parse_markdown("🏟 🏟__🏟 🏟_", "Can't find end of Italic entity at byte offset 20");
|
|
|
|
|
check_parse_markdown("🏟 🏟_🏟 🏟__", "Can't find end of Underline entity at byte offset 19");
|
|
|
|
|
check_parse_markdown("🏟 🏟```🏟 🏟`", "Can't find end of Code entity at byte offset 21");
|
|
|
|
|
check_parse_markdown("🏟 🏟```🏟 🏟_", "Can't find end of PreCode entity at byte offset 9");
|
|
|
|
|
check_parse_markdown("🏟 🏟```🏟 🏟\\`", "Can't find end of PreCode entity at byte offset 9");
|
|
|
|
|
check_parse_markdown("[telegram\\.org](asd\\)", "Can't find end of a URL at byte offset 16");
|
|
|
|
|
check_parse_markdown("[telegram\\.org](", "Can't find end of a URL at byte offset 16");
|
|
|
|
|
check_parse_markdown("[telegram\\.org](asd", "Can't find end of a URL at byte offset 16");
|
|
|
|
|
check_parse_markdown("🏟 🏟__🏟 _🏟___", "Can't find end of Italic entity at byte offset 23");
|
|
|
|
|
check_parse_markdown("🏟 🏟__", "Can't find end of Underline entity at byte offset 9");
|
2021-12-28 18:41:37 +01:00
|
|
|
|
check_parse_markdown("🏟 🏟||test\\|", "Can't find end of Spoiler entity at byte offset 9");
|
2022-07-18 22:40:57 +02:00
|
|
|
|
check_parse_markdown("🏟 🏟!", "Character '!' is reserved and must be escaped with the preceding '\\'");
|
|
|
|
|
check_parse_markdown("🏟 🏟![", "Can't find end of CustomEmoji entity at byte offset 9");
|
|
|
|
|
check_parse_markdown("🏟 🏟![👍", "Can't find end of CustomEmoji entity at byte offset 9");
|
|
|
|
|
check_parse_markdown("🏟 🏟![👍]", "Custom emoji entity must contain a tg://emoji URL");
|
|
|
|
|
check_parse_markdown("🏟 🏟![👍](tg://emoji?id=1234", "Can't find end of a custom emoji URL at byte offset 17");
|
|
|
|
|
check_parse_markdown("🏟 🏟![👍](t://emoji?id=1234)", "Custom emoji URL must have scheme tg");
|
|
|
|
|
check_parse_markdown("🏟 🏟![👍](tg:emojis?id=1234)", "Custom emoji URL must have host \"emoji\"");
|
|
|
|
|
check_parse_markdown("🏟 🏟![👍](tg://emoji#test)", "Custom emoji URL must have an emoji identifier");
|
|
|
|
|
check_parse_markdown("🏟 🏟![👍](tg://emoji?test=1#&id=25)", "Custom emoji URL must have an emoji identifier");
|
|
|
|
|
check_parse_markdown("🏟 🏟![👍](tg://emoji?test=1231&id=025)", "Invalid custom emoji identifier specified");
|
2019-10-03 01:31:06 +02:00
|
|
|
|
|
|
|
|
|
check_parse_markdown("", "", {});
|
|
|
|
|
check_parse_markdown("\\\\", "\\", {});
|
|
|
|
|
check_parse_markdown("\\\\\\", "\\\\", {});
|
|
|
|
|
check_parse_markdown("\\\\\\\\\\_\\*\\`", "\\\\_*`", {});
|
|
|
|
|
check_parse_markdown("➡️ ➡️", "➡️ ➡️", {});
|
|
|
|
|
check_parse_markdown("🏟 🏟``", "🏟 🏟", {});
|
|
|
|
|
check_parse_markdown("🏟 🏟_abac \\* asd _", "🏟 🏟abac * asd ", {{td::MessageEntity::Type::Italic, 5, 11}});
|
|
|
|
|
check_parse_markdown("🏟 \\.🏟_🏟\\. 🏟_", "🏟 .🏟🏟. 🏟", {{td::MessageEntity::Type::Italic, 6, 6}});
|
|
|
|
|
check_parse_markdown("\\\\\\a\\b\\c\\d\\e\\f\\1\\2\\3\\4\\➡️\\", "\\abcdef1234\\➡️\\", {});
|
2023-02-08 13:28:36 +01:00
|
|
|
|
check_parse_markdown("➡️ ➡️_➡️ ➡️_", "➡️ ➡️➡️ ➡️", {{td::MessageEntity::Type::Italic, 5, 5}});
|
2019-10-03 01:31:06 +02:00
|
|
|
|
check_parse_markdown("➡️ ➡️_➡️ ➡️_*➡️ ➡️*", "➡️ ➡️➡️ ➡️➡️ ➡️",
|
|
|
|
|
{{td::MessageEntity::Type::Italic, 5, 5}, {td::MessageEntity::Type::Bold, 10, 5}});
|
|
|
|
|
check_parse_markdown("🏟 🏟_🏟 \\.🏟_", "🏟 🏟🏟 .🏟", {{td::MessageEntity::Type::Italic, 5, 6}});
|
|
|
|
|
check_parse_markdown("🏟 🏟_🏟 *🏟*_", "🏟 🏟🏟 🏟",
|
|
|
|
|
{{td::MessageEntity::Type::Italic, 5, 5}, {td::MessageEntity::Type::Bold, 8, 2}});
|
|
|
|
|
check_parse_markdown("🏟 🏟_🏟 __🏟___", "🏟 🏟🏟 🏟",
|
|
|
|
|
{{td::MessageEntity::Type::Italic, 5, 5}, {td::MessageEntity::Type::Underline, 8, 2}});
|
|
|
|
|
check_parse_markdown("🏟 🏟__🏟 _🏟_ __", "🏟 🏟🏟 🏟 ",
|
|
|
|
|
{{td::MessageEntity::Type::Underline, 5, 6}, {td::MessageEntity::Type::Italic, 8, 2}});
|
|
|
|
|
check_parse_markdown("🏟 🏟__🏟 _🏟_\\___", "🏟 🏟🏟 🏟_",
|
|
|
|
|
{{td::MessageEntity::Type::Underline, 5, 6}, {td::MessageEntity::Type::Italic, 8, 2}});
|
|
|
|
|
check_parse_markdown("🏟 🏟`🏟 🏟```", "🏟 🏟🏟 🏟", {{td::MessageEntity::Type::Code, 5, 5}});
|
2023-02-08 13:28:36 +01:00
|
|
|
|
check_parse_markdown("🏟 🏟```🏟 🏟```", "🏟 🏟 🏟", {{td::MessageEntity::Type::PreCode, 5, 3, "🏟"}});
|
|
|
|
|
check_parse_markdown("🏟 🏟```🏟\n🏟```", "🏟 🏟🏟", {{td::MessageEntity::Type::PreCode, 5, 2, "🏟"}});
|
|
|
|
|
check_parse_markdown("🏟 🏟```🏟\r🏟```", "🏟 🏟🏟", {{td::MessageEntity::Type::PreCode, 5, 2, "🏟"}});
|
|
|
|
|
check_parse_markdown("🏟 🏟```🏟\n\r🏟```", "🏟 🏟🏟", {{td::MessageEntity::Type::PreCode, 5, 2, "🏟"}});
|
|
|
|
|
check_parse_markdown("🏟 🏟```🏟\r\n🏟```", "🏟 🏟🏟", {{td::MessageEntity::Type::PreCode, 5, 2, "🏟"}});
|
|
|
|
|
check_parse_markdown("🏟 🏟```🏟\n\n🏟```", "🏟 🏟\n🏟", {{td::MessageEntity::Type::PreCode, 5, 3, "🏟"}});
|
|
|
|
|
check_parse_markdown("🏟 🏟```🏟\r\r🏟```", "🏟 🏟\r🏟", {{td::MessageEntity::Type::PreCode, 5, 3, "🏟"}});
|
|
|
|
|
check_parse_markdown("🏟 🏟```🏟 \\\\\\`🏟```", "🏟 🏟 \\`🏟", {{td::MessageEntity::Type::PreCode, 5, 5, "🏟"}});
|
2019-10-03 01:31:06 +02:00
|
|
|
|
check_parse_markdown("🏟 🏟**", "🏟 🏟", {});
|
2021-12-28 18:41:37 +01:00
|
|
|
|
check_parse_markdown("||test||", "test", {{td::MessageEntity::Type::Spoiler, 0, 4}});
|
2019-10-03 01:31:06 +02:00
|
|
|
|
check_parse_markdown("🏟 🏟``", "🏟 🏟", {});
|
|
|
|
|
check_parse_markdown("🏟 🏟``````", "🏟 🏟", {});
|
|
|
|
|
check_parse_markdown("🏟 🏟____", "🏟 🏟", {});
|
|
|
|
|
check_parse_markdown("`_* *_`__*` `*__", "_* *_ ",
|
|
|
|
|
{{td::MessageEntity::Type::Code, 0, 5},
|
|
|
|
|
{td::MessageEntity::Type::Code, 5, 1},
|
|
|
|
|
{td::MessageEntity::Type::Bold, 5, 1},
|
|
|
|
|
{td::MessageEntity::Type::Underline, 5, 1}});
|
|
|
|
|
check_parse_markdown("_* * ` `_", " ",
|
|
|
|
|
{{td::MessageEntity::Type::Italic, 0, 3},
|
|
|
|
|
{td::MessageEntity::Type::Bold, 0, 1},
|
|
|
|
|
{td::MessageEntity::Type::Code, 2, 1}});
|
|
|
|
|
check_parse_markdown("[](telegram.org)", "", {});
|
|
|
|
|
check_parse_markdown("[ ](telegram.org)", " ", {{td::MessageEntity::Type::TextUrl, 0, 1, "http://telegram.org/"}});
|
|
|
|
|
check_parse_markdown("[ ](as)", " ", {});
|
|
|
|
|
check_parse_markdown("[telegram\\.org]", "telegram.org",
|
|
|
|
|
{{td::MessageEntity::Type::TextUrl, 0, 12, "http://telegram.org/"}});
|
|
|
|
|
check_parse_markdown("[telegram\\.org]a", "telegram.orga",
|
|
|
|
|
{{td::MessageEntity::Type::TextUrl, 0, 12, "http://telegram.org/"}});
|
|
|
|
|
check_parse_markdown("[telegram\\.org](telegram.dog)", "telegram.org",
|
|
|
|
|
{{td::MessageEntity::Type::TextUrl, 0, 12, "http://telegram.dog/"}});
|
|
|
|
|
check_parse_markdown("[telegram\\.org](https://telegram.dog?)", "telegram.org",
|
|
|
|
|
{{td::MessageEntity::Type::TextUrl, 0, 12, "https://telegram.dog/?"}});
|
|
|
|
|
check_parse_markdown("[telegram\\.org](https://telegram.dog?\\\\\\()", "telegram.org",
|
|
|
|
|
{{td::MessageEntity::Type::TextUrl, 0, 12, "https://telegram.dog/?\\("}});
|
|
|
|
|
check_parse_markdown("[telegram\\.org]()", "telegram.org", {});
|
|
|
|
|
check_parse_markdown("[telegram\\.org](asdasd)", "telegram.org", {});
|
2021-09-03 11:27:59 +02:00
|
|
|
|
check_parse_markdown("[telegram\\.org](tg:user?id=123456)", "telegram.org",
|
|
|
|
|
{{0, 12, td::UserId(static_cast<td::int64>(123456))}});
|
2022-07-18 22:40:57 +02:00
|
|
|
|
check_parse_markdown("🏟 🏟![👍](TG://EMoJI/?test=1231&id=25#id=32)a", "🏟 🏟👍a",
|
2022-10-26 07:53:09 +02:00
|
|
|
|
{{td::MessageEntity::Type::CustomEmoji, 5, 2, td::CustomEmojiId(static_cast<td::int64>(25))}});
|
2019-10-03 01:31:06 +02:00
|
|
|
|
}
|
2020-03-09 21:55:32 +01:00
|
|
|
|
|
2020-03-12 04:22:14 +01:00
|
|
|
|
static void check_parse_markdown_v3(td::string text, td::vector<td::MessageEntity> entities,
|
|
|
|
|
const td::string &result_text, const td::vector<td::MessageEntity> &result_entities,
|
|
|
|
|
bool fix = false) {
|
2020-03-09 21:55:32 +01:00
|
|
|
|
auto parsed_text = td::parse_markdown_v3({std::move(text), std::move(entities)});
|
|
|
|
|
if (fix) {
|
2021-07-26 21:48:33 +02:00
|
|
|
|
ASSERT_TRUE(td::fix_formatted_text(parsed_text.text, parsed_text.entities, true, true, true, true, true).is_ok());
|
2020-03-09 21:55:32 +01:00
|
|
|
|
}
|
2020-03-12 04:22:14 +01:00
|
|
|
|
ASSERT_STREQ(result_text, parsed_text.text);
|
2020-03-09 21:55:32 +01:00
|
|
|
|
ASSERT_EQ(result_entities, parsed_text.entities);
|
2020-03-12 04:22:14 +01:00
|
|
|
|
if (fix) {
|
|
|
|
|
auto markdown_text = td::get_markdown_v3(parsed_text);
|
|
|
|
|
ASSERT_TRUE(parsed_text == markdown_text || parsed_text == td::parse_markdown_v3(markdown_text));
|
|
|
|
|
}
|
2020-03-09 21:55:32 +01:00
|
|
|
|
}
|
|
|
|
|
|
2020-03-12 04:22:14 +01:00
|
|
|
|
static void check_parse_markdown_v3(td::string text, const td::string &result_text,
|
2020-03-09 21:55:32 +01:00
|
|
|
|
const td::vector<td::MessageEntity> &result_entities, bool fix = false) {
|
2020-03-12 04:22:14 +01:00
|
|
|
|
check_parse_markdown_v3(std::move(text), td::vector<td::MessageEntity>(), result_text, result_entities, fix);
|
2020-03-09 21:55:32 +01:00
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
TEST(MessageEntities, parse_markdown_v3) {
|
|
|
|
|
check_parse_markdown_v3("🏟````🏟``🏟`aba🏟```c🏟`aba🏟 daba🏟```c🏟`aba🏟```🏟 `🏟``🏟```",
|
|
|
|
|
"🏟````🏟``🏟aba🏟```c🏟aba🏟 daba🏟c🏟`aba🏟🏟 `🏟``🏟```",
|
|
|
|
|
{{td::MessageEntity::Type::Code, 12, 11}, {td::MessageEntity::Type::Pre, 35, 9}});
|
2023-02-08 13:28:36 +01:00
|
|
|
|
check_parse_markdown_v3("🏟````🏟``🏟`aba🏟```c🏟`aba🏟 daba🏟```c🏟`aba🏟🏟```🏟 `🏟``🏟```",
|
|
|
|
|
{{td::MessageEntity::Type::Italic, 12, 1},
|
|
|
|
|
{td::MessageEntity::Type::Italic, 44, 1},
|
|
|
|
|
{td::MessageEntity::Type::Bold, 45, 1},
|
|
|
|
|
{td::MessageEntity::Type::Bold, 49, 2}},
|
|
|
|
|
"🏟````🏟``🏟`aba🏟c🏟`aba🏟 daba🏟c🏟`aba🏟🏟🏟 `🏟``🏟",
|
|
|
|
|
{{td::MessageEntity::Type::Italic, 12, 1},
|
|
|
|
|
{td::MessageEntity::Type::Pre, 18, 16},
|
|
|
|
|
{td::MessageEntity::Type::Italic, 38, 1},
|
|
|
|
|
{td::MessageEntity::Type::Bold, 39, 1},
|
|
|
|
|
{td::MessageEntity::Type::Bold, 43, 2},
|
|
|
|
|
{td::MessageEntity::Type::Pre, 45, 10}});
|
2020-03-09 21:55:32 +01:00
|
|
|
|
check_parse_markdown_v3("` `", " ", {{td::MessageEntity::Type::Code, 0, 1}});
|
|
|
|
|
check_parse_markdown_v3("`\n`", "\n", {{td::MessageEntity::Type::Code, 0, 1}});
|
|
|
|
|
check_parse_markdown_v3("` `a", " a", {{td::MessageEntity::Type::Code, 0, 1}}, true);
|
2021-06-09 19:35:02 +02:00
|
|
|
|
check_parse_markdown_v3("`\n`a", "\na", {{td::MessageEntity::Type::Code, 0, 1}}, true);
|
2020-03-09 21:55:32 +01:00
|
|
|
|
check_parse_markdown_v3("``", "``", {});
|
2020-03-12 04:22:14 +01:00
|
|
|
|
check_parse_markdown_v3("`a````b```", "`a````b```", {});
|
|
|
|
|
check_parse_markdown_v3("ab", {{td::MessageEntity::Type::Code, 0, 1}, {td::MessageEntity::Type::Pre, 1, 1}}, "ab",
|
|
|
|
|
{{td::MessageEntity::Type::Code, 0, 1}, {td::MessageEntity::Type::Pre, 1, 1}});
|
2020-03-09 21:55:32 +01:00
|
|
|
|
|
|
|
|
|
check_parse_markdown_v3("[a](b[c](t.me)", "[a](b[c](t.me)", {});
|
|
|
|
|
check_parse_markdown_v3("[](t.me)", "[](t.me)", {});
|
|
|
|
|
check_parse_markdown_v3("[ ](t.me)", " ", {{td::MessageEntity::Type::TextUrl, 0, 1, "http://t.me/"}});
|
|
|
|
|
check_parse_markdown_v3("[ ](t.me)", "", {}, true);
|
|
|
|
|
check_parse_markdown_v3("[ ](t.me)a", " a", {{td::MessageEntity::Type::TextUrl, 0, 1, "http://t.me/"}}, true);
|
|
|
|
|
check_parse_markdown_v3(
|
2021-09-03 11:27:59 +02:00
|
|
|
|
"[ ](t.me) [ ](t.me)",
|
|
|
|
|
{{td::MessageEntity::Type::TextUrl, 8, 1, "http://t.me/"}, {10, 1, td::UserId(static_cast<td::int64>(1))}},
|
|
|
|
|
"[ ](t.me) [ ](t.me)",
|
|
|
|
|
{{td::MessageEntity::Type::TextUrl, 8, 1, "http://t.me/"}, {10, 1, td::UserId(static_cast<td::int64>(1))}});
|
2020-03-09 21:55:32 +01:00
|
|
|
|
check_parse_markdown_v3("[\n](t.me)", "\n", {{td::MessageEntity::Type::TextUrl, 0, 1, "http://t.me/"}});
|
2021-06-09 19:35:02 +02:00
|
|
|
|
check_parse_markdown_v3("[\n](t.me)a", "\na", {{td::MessageEntity::Type::TextUrl, 0, 1, "http://t.me/"}}, true);
|
2020-03-09 21:55:32 +01:00
|
|
|
|
check_parse_markdown_v3("asd[abcd](google.com)", {{td::MessageEntity::Type::Italic, 0, 5}}, "asdabcd",
|
|
|
|
|
{{td::MessageEntity::Type::Italic, 0, 3},
|
|
|
|
|
{td::MessageEntity::Type::TextUrl, 3, 4, "http://google.com/"},
|
|
|
|
|
{td::MessageEntity::Type::Italic, 3, 1}});
|
|
|
|
|
check_parse_markdown_v3("asd[abcd](google.com)efg[hi](https://t.me?t=1#h)e",
|
|
|
|
|
{{td::MessageEntity::Type::Italic, 0, 5}, {td::MessageEntity::Type::Italic, 18, 31}},
|
|
|
|
|
"asdabcdefghie",
|
|
|
|
|
{{td::MessageEntity::Type::Italic, 0, 3},
|
|
|
|
|
{td::MessageEntity::Type::TextUrl, 3, 4, "http://google.com/"},
|
|
|
|
|
{td::MessageEntity::Type::Italic, 3, 1},
|
|
|
|
|
{td::MessageEntity::Type::Italic, 7, 3},
|
|
|
|
|
{td::MessageEntity::Type::TextUrl, 10, 2, "https://t.me/?t=1#h"},
|
|
|
|
|
{td::MessageEntity::Type::Italic, 10, 2},
|
|
|
|
|
{td::MessageEntity::Type::Italic, 12, 1}});
|
|
|
|
|
check_parse_markdown_v3(
|
|
|
|
|
"🏟🏟🏟[🏟🏟🏟🏟🏟](www.🤙.tk#1)🤙🤙🤙[🏟🏟🏟🏟](www.🤙.tk#2)🤙🤙🤙["
|
|
|
|
|
"🏟🏟🏟🏟](www.🤙.tk#3)🏟🏟🏟[🏟🏟🏟🏟](www.🤙.tk#4)🤙🤙",
|
|
|
|
|
"🏟🏟🏟🏟🏟🏟🏟🏟🤙🤙🤙🏟🏟🏟🏟🤙🤙🤙🏟🏟🏟🏟🏟🏟🏟🏟🏟🏟"
|
|
|
|
|
"🏟🤙🤙",
|
|
|
|
|
{{td::MessageEntity::Type::TextUrl, 6, 10, "http://www.🤙.tk/#1"},
|
|
|
|
|
{td::MessageEntity::Type::TextUrl, 22, 8, "http://www.🤙.tk/#2"},
|
|
|
|
|
{td::MessageEntity::Type::TextUrl, 36, 8, "http://www.🤙.tk/#3"},
|
|
|
|
|
{td::MessageEntity::Type::TextUrl, 50, 8, "http://www.🤙.tk/#4"}});
|
|
|
|
|
check_parse_markdown_v3(
|
|
|
|
|
"[🏟🏟🏟🏟🏟](www.🤙.tk#1)[🏟🏟🏟🏟](www.🤙.tk#2)[🏟🏟🏟🏟](www.🤙.tk#3)["
|
|
|
|
|
"🏟🏟🏟🏟](www.🤙.tk#4)",
|
|
|
|
|
"🏟🏟🏟🏟🏟🏟🏟🏟🏟🏟🏟🏟🏟🏟🏟🏟🏟",
|
|
|
|
|
{{td::MessageEntity::Type::TextUrl, 0, 10, "http://www.🤙.tk/#1"},
|
|
|
|
|
{td::MessageEntity::Type::TextUrl, 10, 8, "http://www.🤙.tk/#2"},
|
|
|
|
|
{td::MessageEntity::Type::TextUrl, 18, 8, "http://www.🤙.tk/#3"},
|
|
|
|
|
{td::MessageEntity::Type::TextUrl, 26, 8, "http://www.🤙.tk/#4"}});
|
|
|
|
|
check_parse_markdown_v3(
|
|
|
|
|
"🏟🏟🏟[🏟🏟🏟🏟🏟](www.🤙.tk)🤙🤙🤙[🏟🏟🏟🏟](www.🤙.tk)🤙🤙🤙["
|
|
|
|
|
"🏟🏟🏟🏟](www.🤙.tk)🏟🏟🏟[🏟🏟🏟🏟](www.🤙.tk)🤙🤙",
|
|
|
|
|
{{td::MessageEntity::Type::Bold, 0, 2},
|
|
|
|
|
{td::MessageEntity::Type::Bold, 4, 2},
|
|
|
|
|
{td::MessageEntity::Type::Bold, 7, 2},
|
|
|
|
|
{td::MessageEntity::Type::Bold, 11, 2},
|
|
|
|
|
{td::MessageEntity::Type::Bold, 15, 2},
|
|
|
|
|
{td::MessageEntity::Type::Bold, 18, 2},
|
|
|
|
|
{td::MessageEntity::Type::Bold, 26, 2},
|
2021-09-03 11:27:59 +02:00
|
|
|
|
{31, 2, td::UserId(static_cast<td::int64>(1))},
|
2020-03-09 21:55:32 +01:00
|
|
|
|
{td::MessageEntity::Type::Bold, 35, 1},
|
|
|
|
|
{td::MessageEntity::Type::Bold, 44, 2},
|
|
|
|
|
{td::MessageEntity::Type::Bold, 50, 2},
|
|
|
|
|
{td::MessageEntity::Type::Bold, 54, 2},
|
2021-09-03 11:27:59 +02:00
|
|
|
|
{56, 2, td::UserId(static_cast<td::int64>(2))},
|
2020-03-09 21:55:32 +01:00
|
|
|
|
{td::MessageEntity::Type::Bold, 58, 7},
|
2021-09-03 11:27:59 +02:00
|
|
|
|
{60, 2, td::UserId(static_cast<td::int64>(3))},
|
2020-03-09 21:55:32 +01:00
|
|
|
|
{td::MessageEntity::Type::Bold, 67, 7},
|
|
|
|
|
{td::MessageEntity::Type::Bold, 80, 7},
|
|
|
|
|
{td::MessageEntity::Type::Bold, 89, 25}},
|
|
|
|
|
"🏟🏟🏟🏟🏟🏟🏟🏟🤙🤙🤙🏟🏟🏟🏟🤙🤙🤙🏟🏟🏟🏟🏟🏟🏟🏟🏟🏟"
|
|
|
|
|
"🏟🤙🤙",
|
|
|
|
|
{{td::MessageEntity::Type::Bold, 0, 2},
|
|
|
|
|
{td::MessageEntity::Type::Bold, 4, 2},
|
|
|
|
|
{td::MessageEntity::Type::TextUrl, 6, 10, "http://www.🤙.tk/"},
|
|
|
|
|
{td::MessageEntity::Type::Bold, 6, 2},
|
|
|
|
|
{td::MessageEntity::Type::Bold, 10, 2},
|
|
|
|
|
{td::MessageEntity::Type::Bold, 14, 2},
|
2021-09-03 11:27:59 +02:00
|
|
|
|
{18, 2, td::UserId(static_cast<td::int64>(1))},
|
2020-03-09 21:55:32 +01:00
|
|
|
|
{td::MessageEntity::Type::TextUrl, 22, 8, "http://www.🤙.tk/"},
|
2021-09-03 11:27:59 +02:00
|
|
|
|
{30, 2, td::UserId(static_cast<td::int64>(2))},
|
2020-03-09 21:55:32 +01:00
|
|
|
|
{td::MessageEntity::Type::Bold, 32, 2},
|
2021-09-03 11:27:59 +02:00
|
|
|
|
{34, 2, td::UserId(static_cast<td::int64>(3))},
|
2020-03-09 21:55:32 +01:00
|
|
|
|
{td::MessageEntity::Type::Bold, 34, 2},
|
|
|
|
|
{td::MessageEntity::Type::TextUrl, 36, 8, "http://www.🤙.tk/"},
|
|
|
|
|
{td::MessageEntity::Type::Bold, 36, 2},
|
|
|
|
|
{td::MessageEntity::Type::Bold, 40, 4},
|
|
|
|
|
{td::MessageEntity::Type::Bold, 44, 4},
|
|
|
|
|
{td::MessageEntity::Type::TextUrl, 50, 8, "http://www.🤙.tk/"},
|
|
|
|
|
{td::MessageEntity::Type::Bold, 50, 8},
|
|
|
|
|
{td::MessageEntity::Type::Bold, 58, 4}});
|
|
|
|
|
check_parse_markdown_v3("[`a`](t.me) [b](t.me)", {{td::MessageEntity::Type::Code, 13, 1}}, "[a](t.me) [b](t.me)",
|
|
|
|
|
{{td::MessageEntity::Type::Code, 1, 1}, {td::MessageEntity::Type::Code, 11, 1}});
|
|
|
|
|
check_parse_markdown_v3(
|
|
|
|
|
"[text](example.com)",
|
|
|
|
|
{{td::MessageEntity::Type::Strikethrough, 0, 1}, {td::MessageEntity::Type::Strikethrough, 5, 14}}, "text",
|
|
|
|
|
{{td::MessageEntity::Type::TextUrl, 0, 4, "http://example.com/"}});
|
2021-12-28 18:41:37 +01:00
|
|
|
|
check_parse_markdown_v3("[text](example.com)",
|
|
|
|
|
{{td::MessageEntity::Type::Spoiler, 0, 1}, {td::MessageEntity::Type::Spoiler, 5, 14}}, "text",
|
|
|
|
|
{{td::MessageEntity::Type::TextUrl, 0, 4, "http://example.com/"}});
|
2020-03-09 21:55:32 +01:00
|
|
|
|
|
|
|
|
|
check_parse_markdown_v3("🏟[🏟](t.me) `🏟` [🏟](t.me) `a`", "🏟🏟 🏟 🏟 a",
|
|
|
|
|
{{td::MessageEntity::Type::TextUrl, 2, 2, "http://t.me/"},
|
|
|
|
|
{td::MessageEntity::Type::Code, 5, 2},
|
|
|
|
|
{td::MessageEntity::Type::TextUrl, 8, 2, "http://t.me/"},
|
|
|
|
|
{td::MessageEntity::Type::Code, 11, 1}});
|
|
|
|
|
|
|
|
|
|
check_parse_markdown_v3("__ __", " ", {{td::MessageEntity::Type::Italic, 0, 1}});
|
|
|
|
|
check_parse_markdown_v3("__\n__", "\n", {{td::MessageEntity::Type::Italic, 0, 1}});
|
|
|
|
|
check_parse_markdown_v3("__ __a", " a", {}, true);
|
|
|
|
|
check_parse_markdown_v3("__\n__a", "\na", {}, true);
|
2021-12-28 18:41:37 +01:00
|
|
|
|
check_parse_markdown_v3("**** __a__ **b** ~~c~~ ||d||", "**** a b c d",
|
2020-03-09 21:55:32 +01:00
|
|
|
|
{{td::MessageEntity::Type::Italic, 5, 1},
|
|
|
|
|
{td::MessageEntity::Type::Bold, 7, 1},
|
2021-12-28 18:41:37 +01:00
|
|
|
|
{td::MessageEntity::Type::Strikethrough, 9, 1},
|
|
|
|
|
{td::MessageEntity::Type::Spoiler, 11, 1}});
|
|
|
|
|
check_parse_markdown_v3("тест __аааа__ **бббб** ~~вввв~~ ||гггг||", "тест аааа бббб вввв гггг",
|
2020-03-09 21:55:32 +01:00
|
|
|
|
{{td::MessageEntity::Type::Italic, 5, 4},
|
|
|
|
|
{td::MessageEntity::Type::Bold, 10, 4},
|
2021-12-28 18:41:37 +01:00
|
|
|
|
{td::MessageEntity::Type::Strikethrough, 15, 4},
|
|
|
|
|
{td::MessageEntity::Type::Spoiler, 20, 4}});
|
2020-03-09 21:55:32 +01:00
|
|
|
|
check_parse_markdown_v3("___a___ ***b** ~c~~", "___a___ ***b** ~c~~", {});
|
|
|
|
|
check_parse_markdown_v3(
|
|
|
|
|
"__asd[ab__cd](t.me)", "asdabcd",
|
|
|
|
|
{{td::MessageEntity::Type::Italic, 0, 5}, {td::MessageEntity::Type::TextUrl, 3, 4, "http://t.me/"}});
|
|
|
|
|
check_parse_markdown_v3("__asd[ab__cd](t.me)", "asdabcd",
|
|
|
|
|
{{td::MessageEntity::Type::Italic, 0, 3},
|
|
|
|
|
{td::MessageEntity::Type::TextUrl, 3, 4, "http://t.me/"},
|
|
|
|
|
{td::MessageEntity::Type::Italic, 3, 2}},
|
|
|
|
|
true);
|
2020-03-12 04:22:14 +01:00
|
|
|
|
check_parse_markdown_v3("__a #test__test", "__a #test__test", {});
|
|
|
|
|
check_parse_markdown_v3("a #testtest", {{td::MessageEntity::Type::Italic, 0, 7}}, "a #testtest",
|
|
|
|
|
{{td::MessageEntity::Type::Italic, 0, 7}});
|
2020-03-09 21:55:32 +01:00
|
|
|
|
|
2020-03-12 00:49:16 +01:00
|
|
|
|
// TODO parse_markdown_v3 is not idempotent now, which is bad
|
|
|
|
|
check_parse_markdown_v3(
|
|
|
|
|
"~~**~~__**a__", {{td::MessageEntity::Type::Strikethrough, 2, 1}, {td::MessageEntity::Type::Bold, 6, 1}},
|
|
|
|
|
"**__**a__", {{td::MessageEntity::Type::Strikethrough, 0, 2}, {td::MessageEntity::Type::Bold, 2, 1}}, true);
|
|
|
|
|
check_parse_markdown_v3("**__**a__",
|
|
|
|
|
{{td::MessageEntity::Type::Strikethrough, 0, 2}, {td::MessageEntity::Type::Bold, 2, 1}},
|
|
|
|
|
"__a__", {{td::MessageEntity::Type::Bold, 0, 2}}, true);
|
|
|
|
|
check_parse_markdown_v3("__a__", {{td::MessageEntity::Type::Bold, 0, 2}}, "a",
|
|
|
|
|
{{td::MessageEntity::Type::Italic, 0, 1}}, true);
|
|
|
|
|
check_parse_markdown_v3("~~__~~#test__test", "__#test__test", {{td::MessageEntity::Type::Strikethrough, 0, 2}});
|
|
|
|
|
check_parse_markdown_v3("__#test__test", {{td::MessageEntity::Type::Strikethrough, 0, 2}}, "#testtest",
|
|
|
|
|
{{td::MessageEntity::Type::Italic, 0, 5}});
|
|
|
|
|
|
2021-12-28 18:41:37 +01:00
|
|
|
|
check_parse_markdown_v3(
|
|
|
|
|
"~~**~~||**a||", {{td::MessageEntity::Type::Strikethrough, 2, 1}, {td::MessageEntity::Type::Bold, 6, 1}},
|
|
|
|
|
"**||**a||", {{td::MessageEntity::Type::Strikethrough, 0, 2}, {td::MessageEntity::Type::Bold, 2, 1}}, true);
|
|
|
|
|
check_parse_markdown_v3("**||**a||",
|
|
|
|
|
{{td::MessageEntity::Type::Strikethrough, 0, 2}, {td::MessageEntity::Type::Bold, 2, 1}},
|
|
|
|
|
"||a||", {{td::MessageEntity::Type::Bold, 0, 2}}, true);
|
|
|
|
|
check_parse_markdown_v3("||a||", {{td::MessageEntity::Type::Bold, 0, 2}}, "a",
|
|
|
|
|
{{td::MessageEntity::Type::Spoiler, 0, 1}}, true);
|
|
|
|
|
check_parse_markdown_v3("~~||~~#test||test", "#testtest", {{td::MessageEntity::Type::Spoiler, 0, 5}});
|
|
|
|
|
check_parse_markdown_v3("||#test||test", {{td::MessageEntity::Type::Strikethrough, 0, 2}}, "#testtest",
|
|
|
|
|
{{td::MessageEntity::Type::Spoiler, 0, 5}});
|
|
|
|
|
|
2020-03-09 21:55:32 +01:00
|
|
|
|
check_parse_markdown_v3("__[ab_](t.me)_", "__ab__", {{td::MessageEntity::Type::TextUrl, 2, 3, "http://t.me/"}});
|
|
|
|
|
check_parse_markdown_v3(
|
|
|
|
|
"__[ab__](t.me)_", "ab_",
|
|
|
|
|
{{td::MessageEntity::Type::TextUrl, 0, 2, "http://t.me/"}, {td::MessageEntity::Type::Italic, 0, 2}});
|
|
|
|
|
check_parse_markdown_v3("__[__ab__](t.me)__", "____ab____",
|
|
|
|
|
{{td::MessageEntity::Type::TextUrl, 2, 6, "http://t.me/"}});
|
|
|
|
|
check_parse_markdown_v3(
|
|
|
|
|
"__[__ab__](t.me)a__", "____aba",
|
|
|
|
|
{{td::MessageEntity::Type::TextUrl, 2, 4, "http://t.me/"}, {td::MessageEntity::Type::Italic, 6, 1}});
|
|
|
|
|
check_parse_markdown_v3("`a` __ab__", {{td::MessageEntity::Type::Bold, 6, 3}}, "a __ab__",
|
|
|
|
|
{{td::MessageEntity::Type::Code, 0, 1}, {td::MessageEntity::Type::Bold, 4, 3}});
|
|
|
|
|
check_parse_markdown_v3("`a` __ab__", {{td::MessageEntity::Type::Underline, 5, 1}}, "a __ab__",
|
|
|
|
|
{{td::MessageEntity::Type::Code, 0, 1}, {td::MessageEntity::Type::Underline, 3, 1}});
|
|
|
|
|
|
2021-12-28 18:41:37 +01:00
|
|
|
|
check_parse_markdown_v3("||[ab|](t.me)|", "||ab||", {{td::MessageEntity::Type::TextUrl, 2, 3, "http://t.me/"}});
|
|
|
|
|
check_parse_markdown_v3(
|
|
|
|
|
"||[ab||](t.me)|", "ab|",
|
|
|
|
|
{{td::MessageEntity::Type::TextUrl, 0, 2, "http://t.me/"}, {td::MessageEntity::Type::Spoiler, 0, 2}});
|
|
|
|
|
check_parse_markdown_v3("||[||ab||](t.me)||", "||||ab||||",
|
|
|
|
|
{{td::MessageEntity::Type::TextUrl, 2, 6, "http://t.me/"}});
|
|
|
|
|
check_parse_markdown_v3(
|
|
|
|
|
"||[||ab||](t.me)a||", "||||aba",
|
|
|
|
|
{{td::MessageEntity::Type::TextUrl, 2, 4, "http://t.me/"}, {td::MessageEntity::Type::Spoiler, 6, 1}});
|
|
|
|
|
check_parse_markdown_v3("`a` ||ab||", {{td::MessageEntity::Type::Bold, 6, 3}}, "a ||ab||",
|
|
|
|
|
{{td::MessageEntity::Type::Code, 0, 1}, {td::MessageEntity::Type::Bold, 4, 3}});
|
|
|
|
|
check_parse_markdown_v3("`a` ||ab||", {{td::MessageEntity::Type::Underline, 5, 1}}, "a ||ab||",
|
|
|
|
|
{{td::MessageEntity::Type::Code, 0, 1}, {td::MessageEntity::Type::Underline, 3, 1}});
|
|
|
|
|
|
2020-03-09 21:55:32 +01:00
|
|
|
|
check_parse_markdown_v3("`a` @test__test__test", "a @test__test__test", {{td::MessageEntity::Type::Code, 0, 1}});
|
|
|
|
|
check_parse_markdown_v3("`a` #test__test__test", "a #test__test__test", {{td::MessageEntity::Type::Code, 0, 1}});
|
|
|
|
|
check_parse_markdown_v3("`a` __@test_test_test__", "a @test_test_test",
|
|
|
|
|
{{td::MessageEntity::Type::Code, 0, 1}, {td::MessageEntity::Type::Italic, 2, 15}});
|
|
|
|
|
check_parse_markdown_v3("`a` __#test_test_test__", "a #test_test_test",
|
|
|
|
|
{{td::MessageEntity::Type::Code, 0, 1}, {td::MessageEntity::Type::Italic, 2, 15}});
|
|
|
|
|
check_parse_markdown_v3("[a](t.me) __@test**test**test__", "a @testtesttest",
|
|
|
|
|
{{td::MessageEntity::Type::TextUrl, 0, 1, "http://t.me/"},
|
|
|
|
|
{td::MessageEntity::Type::Italic, 2, 13},
|
|
|
|
|
{td::MessageEntity::Type::Bold, 7, 4}});
|
|
|
|
|
check_parse_markdown_v3("[a](t.me) __#test~~test~~test__", "a #testtesttest",
|
|
|
|
|
{{td::MessageEntity::Type::TextUrl, 0, 1, "http://t.me/"},
|
|
|
|
|
{td::MessageEntity::Type::Italic, 2, 13},
|
|
|
|
|
{td::MessageEntity::Type::Strikethrough, 7, 4}});
|
|
|
|
|
check_parse_markdown_v3("[a](t.me) __@test__test__test__", "a @testtesttest",
|
|
|
|
|
{{td::MessageEntity::Type::TextUrl, 0, 1, "http://t.me/"},
|
|
|
|
|
{td::MessageEntity::Type::Italic, 2, 5},
|
|
|
|
|
{td::MessageEntity::Type::Italic, 11, 4}});
|
|
|
|
|
check_parse_markdown_v3("__**~~__gh**~~", "gh",
|
|
|
|
|
{{td::MessageEntity::Type::Bold, 0, 2}, {td::MessageEntity::Type::Strikethrough, 0, 2}});
|
|
|
|
|
check_parse_markdown_v3("__ab**cd~~ef__gh**ij~~", "abcdefghij",
|
|
|
|
|
{{td::MessageEntity::Type::Italic, 0, 6},
|
|
|
|
|
{td::MessageEntity::Type::Bold, 2, 6},
|
|
|
|
|
{td::MessageEntity::Type::Strikethrough, 4, 6}});
|
2021-12-28 18:41:37 +01:00
|
|
|
|
check_parse_markdown_v3("__ab**cd~~ef||gh__ij**kl~~mn||", "abcdefghijklmn",
|
2020-03-09 21:55:32 +01:00
|
|
|
|
{{td::MessageEntity::Type::Italic, 0, 2},
|
|
|
|
|
{td::MessageEntity::Type::Bold, 2, 2},
|
|
|
|
|
{td::MessageEntity::Type::Italic, 2, 2},
|
2021-12-28 18:41:37 +01:00
|
|
|
|
{td::MessageEntity::Type::Bold, 4, 2},
|
|
|
|
|
{td::MessageEntity::Type::Italic, 4, 2},
|
|
|
|
|
{td::MessageEntity::Type::Strikethrough, 4, 2},
|
|
|
|
|
{td::MessageEntity::Type::Spoiler, 6, 8},
|
|
|
|
|
{td::MessageEntity::Type::Strikethrough, 6, 6},
|
|
|
|
|
{td::MessageEntity::Type::Bold, 6, 4},
|
|
|
|
|
{td::MessageEntity::Type::Italic, 6, 2}},
|
2020-03-09 21:55:32 +01:00
|
|
|
|
true);
|
|
|
|
|
check_parse_markdown_v3("__ab**[cd~~ef__](t.me)gh**ij~~", "abcdefghij",
|
|
|
|
|
{{td::MessageEntity::Type::Italic, 0, 6},
|
|
|
|
|
{td::MessageEntity::Type::Bold, 2, 6},
|
|
|
|
|
{td::MessageEntity::Type::TextUrl, 2, 4, "http://t.me/"},
|
|
|
|
|
{td::MessageEntity::Type::Strikethrough, 4, 6}});
|
|
|
|
|
check_parse_markdown_v3("__ab**[cd~~e](t.me)f__gh**ij~~", "abcdefghij",
|
|
|
|
|
{{td::MessageEntity::Type::Italic, 0, 6},
|
|
|
|
|
{td::MessageEntity::Type::Bold, 2, 6},
|
|
|
|
|
{td::MessageEntity::Type::TextUrl, 2, 3, "http://t.me/"},
|
|
|
|
|
{td::MessageEntity::Type::Strikethrough, 4, 6}});
|
|
|
|
|
check_parse_markdown_v3("__ab**[cd~~](t.me)ef__gh**ij~~", "abcdefghij",
|
|
|
|
|
{{td::MessageEntity::Type::Italic, 0, 6},
|
|
|
|
|
{td::MessageEntity::Type::Bold, 2, 6},
|
|
|
|
|
{td::MessageEntity::Type::TextUrl, 2, 2, "http://t.me/"},
|
|
|
|
|
{td::MessageEntity::Type::Strikethrough, 4, 6}});
|
|
|
|
|
check_parse_markdown_v3("[__**bold italic link**__](example.com)", "bold italic link",
|
|
|
|
|
{{td::MessageEntity::Type::TextUrl, 0, 16, "http://example.com/"},
|
|
|
|
|
{td::MessageEntity::Type::Bold, 0, 16},
|
|
|
|
|
{td::MessageEntity::Type::Italic, 0, 16}});
|
|
|
|
|
check_parse_markdown_v3(
|
|
|
|
|
"__italic__ ~~strikethrough~~ **bold** `code` ```pre``` __[italic__ text_url](telegram.org) __italic**bold "
|
|
|
|
|
"italic__bold**__italic__ ~~strikethrough~~ **bold** `code` ```pre``` __[italic__ text_url](telegram.org) "
|
2021-12-28 18:41:37 +01:00
|
|
|
|
"__italic**bold italic__bold** ||spoiler||",
|
2020-03-09 21:55:32 +01:00
|
|
|
|
"italic strikethrough bold code pre italic text_url italicbold italicbolditalic strikethrough bold code pre "
|
2021-12-28 18:41:37 +01:00
|
|
|
|
"italic text_url italicbold italicbold spoiler",
|
2020-03-09 21:55:32 +01:00
|
|
|
|
{{td::MessageEntity::Type::Italic, 0, 6},
|
|
|
|
|
{td::MessageEntity::Type::Strikethrough, 7, 13},
|
|
|
|
|
{td::MessageEntity::Type::Bold, 21, 4},
|
|
|
|
|
{td::MessageEntity::Type::Code, 26, 4},
|
|
|
|
|
{td::MessageEntity::Type::Pre, 31, 3},
|
|
|
|
|
{td::MessageEntity::Type::TextUrl, 35, 15, "http://telegram.org/"},
|
|
|
|
|
{td::MessageEntity::Type::Italic, 35, 6},
|
|
|
|
|
{td::MessageEntity::Type::Italic, 51, 17},
|
|
|
|
|
{td::MessageEntity::Type::Bold, 57, 15},
|
|
|
|
|
{td::MessageEntity::Type::Italic, 72, 6},
|
|
|
|
|
{td::MessageEntity::Type::Strikethrough, 79, 13},
|
|
|
|
|
{td::MessageEntity::Type::Bold, 93, 4},
|
|
|
|
|
{td::MessageEntity::Type::Code, 98, 4},
|
|
|
|
|
{td::MessageEntity::Type::Pre, 103, 3},
|
|
|
|
|
{td::MessageEntity::Type::TextUrl, 107, 15, "http://telegram.org/"},
|
|
|
|
|
{td::MessageEntity::Type::Italic, 107, 6},
|
|
|
|
|
{td::MessageEntity::Type::Italic, 123, 17},
|
2021-12-28 18:41:37 +01:00
|
|
|
|
{td::MessageEntity::Type::Bold, 129, 15},
|
|
|
|
|
{td::MessageEntity::Type::Spoiler, 145, 7}});
|
2020-03-09 21:55:32 +01:00
|
|
|
|
|
2021-12-28 18:41:37 +01:00
|
|
|
|
td::vector<td::string> parts{"a", " #test__a", "__", "**", "~~", "||", "[", "](t.me)", "`"};
|
2020-03-09 21:55:32 +01:00
|
|
|
|
td::vector<td::MessageEntity::Type> types{
|
|
|
|
|
td::MessageEntity::Type::Bold, td::MessageEntity::Type::Italic, td::MessageEntity::Type::Underline,
|
2021-12-28 18:41:37 +01:00
|
|
|
|
td::MessageEntity::Type::Strikethrough, td::MessageEntity::Type::Spoiler, td::MessageEntity::Type::Code,
|
|
|
|
|
td::MessageEntity::Type::Pre, td::MessageEntity::Type::PreCode, td::MessageEntity::Type::TextUrl,
|
|
|
|
|
td::MessageEntity::Type::MentionName, td::MessageEntity::Type::Cashtag};
|
2020-03-09 21:55:32 +01:00
|
|
|
|
for (size_t test_n = 0; test_n < 1000; test_n++) {
|
|
|
|
|
td::string str;
|
|
|
|
|
int part_n = td::Random::fast(1, 200);
|
|
|
|
|
for (int i = 0; i < part_n; i++) {
|
|
|
|
|
str += parts[td::Random::fast(0, static_cast<int>(parts.size()) - 1)];
|
|
|
|
|
}
|
|
|
|
|
td::vector<td::MessageEntity> entities;
|
|
|
|
|
int entity_n = td::Random::fast(1, 20);
|
|
|
|
|
for (int i = 0; i < entity_n; i++) {
|
|
|
|
|
auto type = types[td::Random::fast(0, static_cast<int>(types.size()) - 1)];
|
|
|
|
|
td::int32 offset = td::Random::fast(0, static_cast<int>(str.size()) - 1);
|
|
|
|
|
auto max_length = static_cast<int>(str.size() - offset);
|
|
|
|
|
if ((test_n & 1) != 0 && max_length > 4) {
|
|
|
|
|
max_length = 4;
|
|
|
|
|
}
|
|
|
|
|
td::int32 length = td::Random::fast(0, max_length);
|
|
|
|
|
entities.emplace_back(type, offset, length);
|
|
|
|
|
}
|
|
|
|
|
|
2020-03-11 23:44:42 +01:00
|
|
|
|
td::FormattedText text{std::move(str), std::move(entities)};
|
|
|
|
|
while (true) {
|
2021-07-26 21:48:33 +02:00
|
|
|
|
ASSERT_TRUE(td::fix_formatted_text(text.text, text.entities, true, true, true, true, true).is_ok());
|
2020-03-11 23:44:42 +01:00
|
|
|
|
auto parsed_text = td::parse_markdown_v3(text);
|
2021-07-26 21:48:33 +02:00
|
|
|
|
ASSERT_TRUE(td::fix_formatted_text(parsed_text.text, parsed_text.entities, true, true, true, true, true).is_ok());
|
2020-03-11 23:44:42 +01:00
|
|
|
|
if (parsed_text == text) {
|
|
|
|
|
break;
|
|
|
|
|
}
|
|
|
|
|
text = std::move(parsed_text);
|
|
|
|
|
}
|
|
|
|
|
ASSERT_EQ(text, td::parse_markdown_v3(text));
|
2020-03-12 04:22:14 +01:00
|
|
|
|
auto markdown_text = td::get_markdown_v3(text);
|
|
|
|
|
ASSERT_TRUE(text == markdown_text || text == td::parse_markdown_v3(markdown_text));
|
2020-03-09 21:55:32 +01:00
|
|
|
|
}
|
|
|
|
|
}
|
2020-03-12 04:22:14 +01:00
|
|
|
|
|
2021-10-21 11:51:16 +02:00
|
|
|
|
static void check_get_markdown_v3(const td::string &result_text, const td::vector<td::MessageEntity> &result_entities,
|
|
|
|
|
td::string text, td::vector<td::MessageEntity> entities) {
|
2020-03-12 04:22:14 +01:00
|
|
|
|
auto markdown_text = td::get_markdown_v3({std::move(text), std::move(entities)});
|
|
|
|
|
ASSERT_STREQ(result_text, markdown_text.text);
|
|
|
|
|
ASSERT_EQ(result_entities, markdown_text.entities);
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
TEST(MessageEntities, get_markdown_v3) {
|
|
|
|
|
check_get_markdown_v3("``` ```", {}, " ", {{td::MessageEntity::Type::Pre, 0, 1}});
|
|
|
|
|
check_get_markdown_v3("` `", {}, " ", {{td::MessageEntity::Type::Code, 0, 1}});
|
|
|
|
|
check_get_markdown_v3("`\n`", {}, "\n", {{td::MessageEntity::Type::Code, 0, 1}});
|
|
|
|
|
check_get_markdown_v3("ab", {{td::MessageEntity::Type::Code, 0, 1}, {td::MessageEntity::Type::Pre, 1, 1}}, "ab",
|
|
|
|
|
{{td::MessageEntity::Type::Code, 0, 1}, {td::MessageEntity::Type::Pre, 1, 1}});
|
|
|
|
|
|
|
|
|
|
check_get_markdown_v3("[ ](http://t.me/)", {}, " ", {{td::MessageEntity::Type::TextUrl, 0, 1, "http://t.me/"}});
|
2021-09-03 11:27:59 +02:00
|
|
|
|
check_get_markdown_v3(
|
|
|
|
|
"[ ]t.me[)](http://t.me/) [ ](t.me)", {{25, 1, td::UserId(static_cast<td::int64>(1))}}, "[ ]t.me) [ ](t.me)",
|
|
|
|
|
{{td::MessageEntity::Type::TextUrl, 7, 1, "http://t.me/"}, {9, 1, td::UserId(static_cast<td::int64>(1))}});
|
2020-03-12 04:22:14 +01:00
|
|
|
|
|
|
|
|
|
check_get_markdown_v3("__ __", {}, " ", {{td::MessageEntity::Type::Italic, 0, 1}});
|
|
|
|
|
check_get_markdown_v3("** **", {}, " ", {{td::MessageEntity::Type::Bold, 0, 1}});
|
|
|
|
|
check_get_markdown_v3("~~ ~~", {}, " ", {{td::MessageEntity::Type::Strikethrough, 0, 1}});
|
2021-12-28 18:41:37 +01:00
|
|
|
|
check_get_markdown_v3("|| ||", {}, " ", {{td::MessageEntity::Type::Spoiler, 0, 1}});
|
|
|
|
|
check_get_markdown_v3("__a__ **b** ~~c~~ ||d|| e", {{td::MessageEntity::Type::PreCode, 24, 1, "C++"}}, "a b c d e",
|
2020-03-12 04:22:14 +01:00
|
|
|
|
{{td::MessageEntity::Type::Italic, 0, 1},
|
|
|
|
|
{td::MessageEntity::Type::Bold, 2, 1},
|
|
|
|
|
{td::MessageEntity::Type::Strikethrough, 4, 1},
|
2021-12-28 18:41:37 +01:00
|
|
|
|
{td::MessageEntity::Type::Spoiler, 6, 1},
|
|
|
|
|
{td::MessageEntity::Type::PreCode, 8, 1, "C++"}});
|
2020-03-12 04:22:14 +01:00
|
|
|
|
check_get_markdown_v3("`ab` ```cd``` ef", {{td::MessageEntity::Type::PreCode, 14, 2, "C++"}}, "ab cd ef",
|
|
|
|
|
{{td::MessageEntity::Type::Code, 0, 2},
|
|
|
|
|
{td::MessageEntity::Type::Pre, 3, 2},
|
|
|
|
|
{td::MessageEntity::Type::PreCode, 6, 2, "C++"}});
|
|
|
|
|
check_get_markdown_v3("__asd__[__ab__cd](http://t.me/)", {}, "asdabcd",
|
|
|
|
|
{{td::MessageEntity::Type::Italic, 0, 3},
|
|
|
|
|
{td::MessageEntity::Type::TextUrl, 3, 4, "http://t.me/"},
|
|
|
|
|
{td::MessageEntity::Type::Italic, 3, 2}});
|
|
|
|
|
|
|
|
|
|
check_get_markdown_v3("__ab", {{td::MessageEntity::Type::Italic, 3, 1}}, "__ab",
|
|
|
|
|
{{td::MessageEntity::Type::Italic, 3, 1}});
|
|
|
|
|
check_get_markdown_v3("__ab__**__cd__**~~**__ef__gh**ij~~", {}, "abcdefghij",
|
|
|
|
|
{{td::MessageEntity::Type::Italic, 0, 2},
|
|
|
|
|
{td::MessageEntity::Type::Bold, 2, 2},
|
|
|
|
|
{td::MessageEntity::Type::Italic, 2, 2},
|
|
|
|
|
{td::MessageEntity::Type::Strikethrough, 4, 6},
|
|
|
|
|
{td::MessageEntity::Type::Bold, 4, 4},
|
|
|
|
|
{td::MessageEntity::Type::Italic, 4, 2}});
|
|
|
|
|
check_get_markdown_v3("[**__bold italic link__**](http://example.com/)", {}, "bold italic link",
|
|
|
|
|
{{td::MessageEntity::Type::TextUrl, 0, 16, "http://example.com/"},
|
|
|
|
|
{td::MessageEntity::Type::Bold, 0, 16},
|
|
|
|
|
{td::MessageEntity::Type::Italic, 0, 16}});
|
|
|
|
|
}
|