Compare commits

..

13 Commits

7 changed files with 250 additions and 82 deletions

View File

@@ -1,6 +1,27 @@
cmake_minimum_required(VERSION 3.5)
project(mmcs-quotes-bridge LANGUAGES CXX)
option(BUILD_SHARED_LIBS "Build shared libraries by default" NO)
if(CMAKE_COMPILER_IS_GNUCXX)
set(CMAKE_EXE_LINKER_FLAGS "-static-libgcc -static-libstdc++")
endif()
set(CURL_DISABLE_AWS ON)
set(CURL_DISABLE_DICT ON)
set(CURL_DISABLE_FILE ON)
set(CURL_DISABLE_FTP ON)
set(CURL_DISABLE_GOPHER ON)
set(CURL_DISABLE_IMAP ON)
set(CURL_DISABLE_LDAP ON)
set(CURL_DISABLE_LDAPS ON)
set(CURL_DISABLE_MQTT ON)
set(CURL_DISABLE_NTLM ON)
set(CURL_DISABLE_POP3 ON)
set(CURL_DISABLE_RTSP ON)
set(CURL_DISABLE_SMB ON)
set(CURL_DISABLE_TFTP ON)
set(CURL_DISABLE_WEBSOCKETS ON)
add_subdirectory(curl)
add_subdirectory(json)
add_subdirectory(libuv)
@@ -9,6 +30,6 @@ add_subdirectory(td)
add_executable(${PROJECT_NAME} main.cpp config.cpp http.cpp manager.cpp posts.cpp state.cpp tg.cpp vk.cpp)
target_compile_options(${PROJECT_NAME} PRIVATE -std=c++2b)
target_compile_options(${PROJECT_NAME} PRIVATE -std=c++2a)
target_link_libraries(${PROJECT_NAME} PRIVATE CURL::libcurl nlohmann_json::nlohmann_json uv spdlog::spdlog Td::TdStatic $<$<BOOL:${MINGW}>:ws2_32>)
target_link_libraries(${PROJECT_NAME} PRIVATE CURL::libcurl nlohmann_json::nlohmann_json uv_a spdlog::spdlog Td::TdStatic $<$<BOOL:${MINGW}>:ws2_32>)

23
README.md Normal file
View File

@@ -0,0 +1,23 @@
# MMCS Quotes Bridge
> Текущее название является рабочим вариантом, имеет мало общего с сутью проекта, все совпадения аббревиатур с реальными случайны и не несут никакой смысловой нагрузки.
>
> *P.S. кто знает тот знает короче :DDD*
Этот проект - реализация бота, предназначенного для репоста текстовых постов из источников в VK и Telegram в один выходной Telegram-канал. Изначальная цель - репостить цитаты из разных цитатников в один канал, но в принципе вы можете придумать своё применение.
## Платформа
**Операционная система.** Предполагается работа на Linux, но теоретически можно заставить бота работать и на других системах.
**Компилятор.** Проект написан на C++20 и соответственно требует более-менее современный компилятор.
## Зависимости
Большая часть необходимых библиотек присутствует в репозитории, однако для того, чтобы скачать их вместе с проектом, при клонировании требуется указать --recursive.
CMakeLists настроен на максимально портативный билд, однако всё же не все библиотеки встраиваются в исполняемый файл.
Для работы бота достаточно установленной библиотеки curl.
Для сборки (помимо CMake) нужны следующие дополнительные пакеты:
- gperf
- заголовочные файлы для OpenSSL, zlib
По вопросам сборки можно писать в issues.
## Конфигурация
Для настройки используется файл `bridge_config.json`. В качестве примера настройки оставлен `bridge_config.example.json`, который нужно отредактировать под себя.

View File

@@ -0,0 +1,11 @@
{
"vk_service_key": "service_key_service_key_service_key_service_key_service_key_service_key",
"tg_api_id": 1234567,
"tg_api_hash": "53a513d108f2c7d9edc67dadd3ce4185",
"tg_phone_number": "+79528125252",
"vk_sources": [{"id": "source1"}, {"id":123451234, "link": "source2"}],
"tg_sources": [{"id": -1002282282280, "link": "tgsource"}],
"tg_destination_id": -1005252525252
}

View File

@@ -1,6 +1,7 @@
#include "http.h"
#include "curl/curl.h"
#include "curl/easy.h"
#include "curl/multi.h"
#include <spdlog/spdlog.h>
#include <spdlog/sinks/stdout_color_sinks.h>
@@ -45,13 +46,31 @@ HttpClient::~HttpClient() {
});
}
bool HttpClient::send_request(std::string method, std::string url, HttpOptions opts, ResponseCallback cb) {
void HttpClient::cancel_request(request_id id) {
CURL *requestHandle = reinterpret_cast<CURL*>(id);
auto request = m_requests.find(requestHandle);
if (request == m_requests.end()) {
m_logger->warn("cancel_request: not found");
return;
}
curl_multi_remove_handle(m_curlMulti, requestHandle);
if (request->second.socketData->pollHandle) {
m_logger->debug("closing poll handle");
uv_poll_stop(request->second.socketData->pollHandle);
uv_close((uv_handle_t*)request->second.socketData->pollHandle, [](uv_handle_t *h){
delete h;
});
}
m_requests.erase(request);
}
request_id HttpClient::send_request(std::string method, std::string url, HttpOptions opts, ResponseCallback cb) {
m_logger->debug("send request {} {}", method, url);
CURL *requestHandle = curl_easy_init();
std::pair<decltype(m_requests)::iterator, bool> insertResult = m_requests.emplace(requestHandle, this);
if (!insertResult.second) {
curl_easy_cleanup(requestHandle);
return false;
return nullptr;
}
auto requestData = insertResult.first;
requestData->second.callback = cb;
@@ -83,7 +102,8 @@ bool HttpClient::send_request(std::string method, std::string url, HttpOptions o
if (opts.body) {
curl_easy_setopt(requestHandle, CURLOPT_POSTFIELDS, opts.body->c_str());
}
return CURLM_OK == curl_multi_add_handle(m_curlMulti, requestHandle);
curl_multi_add_handle(m_curlMulti, requestHandle);
return reinterpret_cast<void*>(requestHandle);
}
int HttpClient::curl_socket_cb(CURL *curl, curl_socket_t curlSocket, int action, HttpClient *self, void *socketPtr) {

5
http.h
View File

@@ -9,6 +9,8 @@
#include <vector>
namespace http {
typedef void *request_id;
struct HttpResponse {
int status;
std::string body;
@@ -24,7 +26,8 @@ namespace http {
HttpClient(HttpClient&&) = delete;
HttpClient(HttpClient&) = delete;
~HttpClient();
bool send_request(std::string method, std::string url, HttpOptions opts, ResponseCallback cb);
request_id send_request(std::string method, std::string url, HttpOptions opts, ResponseCallback cb);
void cancel_request(request_id id);
private:
void check_curl_messages();
static int curl_socket_cb(CURL *curl, curl_socket_t curlSocket, int action, HttpClient *self, void *socketPtr);

View File

@@ -17,7 +17,8 @@ const unsigned long REPOST_INTERVAL = 2000;
const unsigned long VK_CHECK_INTERVAL = 600000;
RepostManager::RepostManager(uv_loop_t *eventLoop, tg::AuthCodeProvider tgCodeProvider, tg::PasswordProvider tgPasswordProvider, state::AppState *appState, config::AppConfig *config)
: m_vk(eventLoop), m_tg(eventLoop, config->tgApiId, config->tgApiHash, config->tgPhoneNumber) {
: m_vk(eventLoop), m_tg(eventLoop, config->tgApiId, config->tgApiHash, config->tgPhoneNumber),
m_fetcher(this) {
m_appState = appState;
m_appConfig = config;
m_tg.authCodeProvider = tgCodeProvider;
@@ -40,6 +41,7 @@ RepostManager::~RepostManager() {
}
if (m_checkTimer) {
uv_timer_stop(m_checkTimer);
m_checkTimerStarted = false;
uv_close((uv_handle_t*)m_checkTimer, [](uv_handle_t *h){ delete h; });
}
}
@@ -58,6 +60,12 @@ void RepostManager::load_more_telegram_chats() {
void RepostManager::start() {
m_nRequiredChats = 100500;//m_appConfig->tgSources.size() + 1;
for (auto &appState : m_appState->vkRepostState) {
appState.lastLoadedPostDate = appState.lastForwardedPostDate;
}
for (auto &appState : m_appState->tgRepostState) {
appState.lastLoadedPostDate = appState.lastForwardedPostDate;
}
m_tg.add_update_handler([this](void*, td_api::Object &obj){
if (obj.get_id() == td_api::updateAuthorizationState::ID) {
auto &authState = (td_api::updateAuthorizationState&)obj;
@@ -67,7 +75,6 @@ void RepostManager::start() {
}
} else if (obj.get_id() == td_api::updateNewChat::ID) {
auto &update = (td_api::updateNewChat&)obj;
spdlog::debug("chat {}", update.chat_->id_);
if (update.chat_->id_ == m_appConfig->tgDestinationId) {
++m_nLoadedRequiredChats;
spdlog::info("destination chat {} loaded (loaded {}/{} chats)", m_appConfig->tgDestinationId, m_nLoadedRequiredChats, m_nRequiredChats);
@@ -95,24 +102,39 @@ void RepostManager::start() {
m_tg.start();
}
NewPostFetcher::NewPostFetcher(RepostManager *m, bool fetchVk, bool fetchTg) : mgr(m) {
void NewPostFetcher::fetch(bool fetchVk, bool fetchTg, decltype(onDone) onDone, decltype(onError) onError) {
if (working) return;
working = true;
if (fetchVk) {
for (int i = 0; i < m->m_appConfig->vkSources.size(); ++i) {
for (int i = 0; i < mgr->m_appConfig->vkSources.size(); ++i) {
fetcher_state &&state {};
state.sourceIndex = i;
vkState.emplace_back(state);
}
}
if (fetchTg) {
for (int i = 0; i < m->m_appConfig->tgSources.size(); ++i) {
for (int i = 0; i < mgr->m_appConfig->tgSources.size(); ++i) {
fetcher_state &&state {};
state.sourceIndex = i;
tgState.emplace_back(state);
}
}
this->onDone = onDone;
this->onError = onError;
continue_fetch();
}
void NewPostFetcher::fetch() {
void NewPostFetcher::reset_state() {
if (!working) return;
for (int i = 0; i < vkState.size(); ++i) {
vkState[i] = fetcher_state();
}
for (int i = 0; i < tgState.size(); ++i) {
tgState[i] = fetcher_state();
}
}
void NewPostFetcher::continue_fetch() {
bool vkReady = true;
spdlog::info("fetch called");
for (int i = 0; i < vkState.size(); ++i) {
@@ -147,7 +169,7 @@ void NewPostFetcher::fetch() {
for (auto i = aposts.begin(), e = aposts.end(); i != e; ++i) {
state.posts.emplace_back(std::move(*i));
}
fetch();
continue_fetch();
});
}
}
@@ -167,8 +189,10 @@ void NewPostFetcher::fetch() {
for (auto &p : posts) {
spdlog::debug("[tg:{}] got post dated {}", i, p->date_);
}
if (posts.size() > 0) {
state.offset = posts[posts.size() - 1]->id_;
spdlog::info("[tg:{}] setting from to id {}", i, posts[posts.size() - 1]->id_);
}
check_tg_posts(i, std::move(posts));
});
state.count = state.count * 3 / 2;
@@ -186,7 +210,7 @@ void NewPostFetcher::fetch() {
for (auto i = aposts.begin(), e = aposts.end(); i != e; ++i) {
state.posts.emplace_back(std::move(*i));
}
fetch();
continue_fetch();
});
}
}
@@ -203,7 +227,7 @@ void NewPostFetcher::fetch() {
int total =
std::accumulate(vkState.begin(), vkState.end(), 0, addPostCount)
+ std::accumulate(tgState.begin(), tgState.end(), 0, addPostCount);
merged.reserve(total);
merged.reserve(total + mgr->m_unprocessedTgPosts.size());
std::vector<int> indexes;
indexes.reserve(nLists);
@@ -241,6 +265,29 @@ void NewPostFetcher::fetch() {
--indexes[minPostListIdx];
++k;
}
while (mgr->m_unprocessedTgPosts.size() > 0) {
AbstractPost tgPost = mgr->m_unprocessedTgPosts.front();
bool duplicate = false, inserted = false;
for (int i = 0; i < merged.size(); ++i) {
if (merged[i].date == tgPost.date) {
spdlog::debug("not inserting duplicate unproc tg post");
duplicate = true;
break;
} else if (merged[i].date > tgPost.date) {
spdlog::debug("inserting unproc tg post at pos {}", i);
merged.insert(merged.begin() + i, std::move(tgPost));
inserted = true;
break;
}
}
if (!duplicate && !inserted) {
spdlog::debug("appending unproc tg post");
merged.push_back(std::move(tgPost));
}
mgr->m_unprocessedTgPosts.pop();
}
working = false;
onDone(std::move(merged));
}
}
@@ -252,7 +299,6 @@ void NewPostFetcher::check_vk_posts(int index, std::vector<vk::Post> posts) {
long oldLastPostDate = appState.lastLoadedPostDate;
if (posts.size() > 0) {
spdlog::info("[vk:{}] last post date is now {}", index, posts[0].date);
}
std::vector<AbstractPost> aposts = mgr->to_abstract_posts(posts, state.sourceIndex);
spdlog::info("[vk:{}] looking for date {}, have {} - {}", index, oldLastPostDate, aposts[0].date, aposts[aposts.size() - 1].date);
if (mgr->drop_posts_older_than(aposts, oldLastPostDate)) {
@@ -268,7 +314,11 @@ void NewPostFetcher::check_vk_posts(int index, std::vector<vk::Post> posts) {
spdlog::debug("[vk:{}] last loaded post date is now {}", index, state.posts[0].date);
appState.lastLoadedPostDate = state.posts[0].date;
}
fetch();
} else {
state.ready = true;
}
continue_fetch();
}
void NewPostFetcher::check_tg_posts(int index, std::vector<td::tl::unique_ptr<td_api::message>> posts) {
@@ -278,7 +328,6 @@ void NewPostFetcher::check_tg_posts(int index, std::vector<td::tl::unique_ptr<td
long oldLastPostDate = appState.lastLoadedPostDate;
if (posts.size() > 0) {
spdlog::info("[tg:{}] last post date is now {}", index, posts[0]->date_);
}
std::vector<AbstractPost> aposts = mgr->to_abstract_posts(posts, state.sourceIndex);
if (mgr->drop_posts_older_than(aposts, oldLastPostDate)) {
spdlog::info("[tg:{}] found last remembered post", index);
@@ -293,37 +342,36 @@ void NewPostFetcher::check_tg_posts(int index, std::vector<td::tl::unique_ptr<td
spdlog::debug("[tg:{}] last loaded post date is now {}", index, state.posts[0].date);
appState.lastLoadedPostDate = state.posts[0].id;
}
fetch();
} else {
state.ready = true;
}
continue_fetch();
}
void RepostManager::on_clients_ready() {
for (auto &appState : m_appState->vkRepostState) {
appState.lastLoadedPostDate = appState.lastForwardedPostDate;
}
for (auto &appState : m_appState->tgRepostState) {
appState.lastLoadedPostDate = appState.lastForwardedPostDate;
}
NewPostFetcher *f = new NewPostFetcher(this, true, true);
f->onDone = [this, f](auto posts){
delete f;
spdlog::info("checking all sources");
m_fetcher.fetch(
true, true,
[this](auto posts){
on_new_posts(posts);
};
f->onError = [f](){
delete f;
};
f->fetch();
},
[](){
// TODO error handling
spdlog::error("first post check failed");
});
}
void RepostManager::on_new_posts(std::vector<AbstractPost> posts) {
spdlog::info("collected {} new posts", posts.size());
enqueue_for_repost(posts);
if (!m_checkTimerStarted) {
spdlog::info("scheduling next check");
uv_timer_start(m_checkTimer, &RepostManager::check_timer_callback, VK_CHECK_INTERVAL, 0);
m_checkTimerStarted = true;
}
}
void RepostManager::collect_all_vk_posts(const std::variant<long, std::string> wall, std::function<void(std::vector<vk::Post>)> callback) {
@@ -421,10 +469,25 @@ bool RepostManager::drop_posts_older_than(std::vector<AbstractPost> &posts, long
}
}
std::optional<AbstractPost> RepostManager::to_abstract_post(const vk::Post &post, int sourceIndex) {
if (post.text.empty()) return {};
else return { AbstractPost(posts::SRC_VK, sourceIndex, post.id, post.date, post.text) };
}
std::optional<AbstractPost> RepostManager::to_abstract_post(const td_api::message &post, int sourceIndex) {
if (post.content_->get_id() == td_api::messageText::ID) {
auto &content = (td_api::messageText&) *post.content_;
return { AbstractPost(posts::SRC_TELEGRAM, sourceIndex, post.id_, post.date_, content.text_->text_) };
} else {
return {};
}
}
std::vector<AbstractPost> RepostManager::to_abstract_posts(std::vector<vk::Post> &posts, int sourceIndex) {
std::vector<AbstractPost> result;
result.reserve(posts.size());
for (auto &post : posts) {
if (post.text.empty()) continue;
result.emplace_back(posts::SRC_VK, sourceIndex, post.id, post.date, post.text);
}
return result;
@@ -474,24 +537,30 @@ void RepostManager::check_timer_callback(uv_timer_t *h) {
self->recheck_vk_posts({});
}
void RepostManager::recheck_vk_posts(std::function<void()> onDone) {
bool RepostManager::recheck_vk_posts(std::function<void()> onDone) {
if (m_fetcher.working) {
spdlog::error("can't recheck VK posts: another check is in progress");
return false;
}
spdlog::info("checking VK posts");
NewPostFetcher *f = new NewPostFetcher(this, true, false);
f->onDone = [this, f, onDone](std::vector<AbstractPost> &&posts){
auto onFetchDone = [this, onDone](std::vector<AbstractPost> &&posts){
spdlog::info("checked VK posts");
this->on_new_posts(posts);
if (onDone)
onDone();
delete f;
};
f->onError = [f](){
delete f;
auto onFetchError = [](){
spdlog::error("failed to check VK posts");
};
f->fetch();
m_fetcher.fetch(true, false, onFetchDone, onFetchError);
return true;
}
void RepostManager::repost(AbstractPost &post) {
if (post.text.length() == 0) {
spdlog::error("repost(): zero post length (post id {} date {})", post.id, post.date);
return;
}
spdlog::debug("reposting (post length {})", post.text.length());
std::string_view signature = posts::add_signature(post, m_appConfig);
int signatureStart = post.text.length() - signature.length();
@@ -530,11 +599,22 @@ void RepostManager::on_tg_message(td_api::updateNewMessage &update) {
if (sourceIndex == m_appConfig->tgSources.size()) {
return;
}
uv_timer_stop(m_checkTimer);
std::vector<td::tl::unique_ptr<td_api::message>> v;
v.push_back(std::move(update.message_));
recheck_vk_posts([this, post = to_abstract_posts(v, sourceIndex)](){
on_new_posts(post);
});
std::optional<AbstractPost> post = to_abstract_post(*update.message_, sourceIndex);
if (!post) {
spdlog::debug("tg message is not a post");
return;
}
spdlog::debug("adding tg post to the unprocessed tg post queue");
m_unprocessedTgPosts.push(*post);
if (!m_fetcher.working) {
if (m_checkTimerStarted) {
uv_timer_stop(m_checkTimer);
m_checkTimerStarted = false;
}
std::vector<AbstractPost> posts = { *post };
spdlog::debug("rechecking vk posts before processing the new tg post");
recheck_vk_posts({});
}
}

View File

@@ -3,6 +3,7 @@
#include "config.h"
#include "posts.h"
#include "state.h"
#include "td/tl/TlObject.h"
#include "tg.h"
#include "vk.h"
#include <functional>
@@ -25,13 +26,17 @@ namespace manager {
std::vector<AbstractPost> posts;
};
bool working = false;
RepostManager *mgr;
std::vector<fetcher_state> vkState, tgState;
std::function<void(std::vector<AbstractPost>&&)> onDone;
std::function<void()> onError;
NewPostFetcher(RepostManager *m, bool fetchVk, bool fetchTg);
void fetch();
inline NewPostFetcher(RepostManager *m) : mgr(m) {};
void fetch(bool fetchVk, bool fetchTg, decltype(onDone) onDone, decltype(onError) onError);
private:
void reset_state();
void continue_fetch();
void check_vk_posts(int index, std::vector<vk::Post> posts);
void check_tg_posts(int index, std::vector<td::tl::unique_ptr<td_api::message>> posts);
};
@@ -48,7 +53,7 @@ namespace manager {
void load_more_telegram_chats();
void on_new_posts(std::vector<AbstractPost> posts);
void on_tg_message(td_api::updateNewMessage &update);
void recheck_vk_posts(std::function<void()> onDone);
bool recheck_vk_posts(std::function<void()> onDone);
void collect_all_vk_posts(const std::variant<long, std::string> wall, std::function<void(std::vector<vk::Post>)> callback);
void collect_all_tg_posts(long channel, std::function<void(std::vector<td::tl::unique_ptr<td_api::message>>)> callback);
@@ -62,6 +67,8 @@ namespace manager {
bool drop_posts_older_than(std::vector<AbstractPost> &posts, long lastPostId);
std::optional<AbstractPost> to_abstract_post(const vk::Post &post, int sourceIndex);
std::optional<AbstractPost> to_abstract_post(const td_api::message &post, int sourceIndex);
std::vector<AbstractPost> to_abstract_posts(std::vector<vk::Post> &posts, int sourceIndex);
std::vector<AbstractPost> to_abstract_posts(std::vector<td::tl::unique_ptr<td_api::message>> &posts, int sourceIndex);
@@ -74,8 +81,11 @@ namespace manager {
config::AppConfig *m_appConfig;
vk::VKClient m_vk;
tg::TelegramClient m_tg;
NewPostFetcher m_fetcher;
std::queue<AbstractPost> m_repostQueue;
std::queue<AbstractPost> m_unprocessedTgPosts;
uv_timer_t *m_repostTimer = nullptr;
bool m_checkTimerStarted = false;
uv_timer_t *m_checkTimer = nullptr;
int m_nRequiredChats;
int m_nLoadedRequiredChats = 0;