Public-Mirror/Textractor
1
0
Fork 1
mirror of https://github.com/Artikash/Textractor.git synced 2024-12-23 08:54:12 +08:00
Code Issues Projects Releases Wiki Activity

huge refactor

Browse Source
This commit is contained in:
Akash Mozumdar 2019-02-10 21:46:39 -05:00
parent 6c47a27351
commit 72b9a1f8ed
8 changed files with 150 additions and 160 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

96
extensions/bingtranslate.cpp
View File

@ -1,6 +1,7 @@
#include "extension.h"
#include "extension.h"
#include "defs.h"
#include "text.h"
#include <winhttp.h>
#include "network.h"
#include <QInputDialog>
#include <QTimer>
@ -53,7 +54,7 @@ QStringList languages
"Welsh: cy"
};
std::wstring translateTo;
std::wstring translateTo = L"en";
BOOL WINAPI DllMain(HMODULE hModule, DWORD ul_reason_for_call, LPVOID lpReserved)
{
@ -82,49 +83,33 @@ BOOL WINAPI DllMain(HMODULE hModule, DWORD ul_reason_for_call, LPVOID lpReserved
return TRUE;
}
// This function detects language and puts it in translateFrom if it's empty
std::wstring Translate(std::wstring text, std::wstring& translateFrom, std::wstring translateTo)
// This function detects language and returns it if translateFrom is empty
std::wstring Translate(const std::wstring& text, std::wstring translateFrom, std::wstring translateTo)
{
static std::atomic<HINTERNET> internet = NULL;
if (!internet) internet = WinHttpOpen(L"Mozilla/5.0 Textractor", WINHTTP_ACCESS_TYPE_DEFAULT_PROXY, NULL, NULL, 0);
char utf8[10000] = {};
WideCharToMultiByte(CP_UTF8, 0, text.c_str(), -1, utf8, 10000, NULL, NULL);
text.clear();
for (int i = 0; utf8[i];)
std::wstring escapedText;
for (unsigned char ch : WideStringToString(text))
{
wchar_t utf8char[3] = {};
swprintf_s<3>(utf8char, L"%02X", (int)(unsigned char)utf8[i++]);
text += L"%" + std::wstring(utf8char);
wchar_t escapedChar[4] = {};
swprintf_s<4>(escapedChar, L"%%%02X", (int)ch);
escapedText = escapedChar;
}
std::wstring location = translateFrom.empty()
? L"/tdetect?text=" + text
: L"/ttranslate?from=" + translateFrom + L"&to=" + translateTo + L"&text=" + text;
std::wstring translation;
if (internet)
{
std::wstring location = translateFrom.empty()
? L"/tdetect?text=" + text
: L"/ttranslate?from=" + translateFrom + L"&to=" + translateTo + L"&text=" + text;
if (HINTERNET connection = WinHttpConnect(internet, L"www.bing.com", INTERNET_DEFAULT_HTTPS_PORT, 0))
{
if (HINTERNET request = WinHttpOpenRequest(connection, L"POST", location.c_str(), NULL, NULL, NULL, WINHTTP_FLAG_ESCAPE_DISABLE | WINHTTP_FLAG_SECURE))
{
if (InternetHandle connection = WinHttpConnect(internet, L"www.bing.com", INTERNET_DEFAULT_HTTPS_PORT, 0))
if (InternetHandle request = WinHttpOpenRequest(connection, L"POST", location.c_str(), NULL, NULL, NULL, WINHTTP_FLAG_ESCAPE_DISABLE | WINHTTP_FLAG_SECURE))
if (WinHttpSendRequest(request, NULL, 0, NULL, 0, 0, NULL))
{
DWORD bytesRead;
char buffer[10000] = {};
WinHttpReceiveResponse(request, NULL);
WinHttpReadData(request, buffer, 10000, &bytesRead);
wchar_t wbuffer[10000] = {};
MultiByteToWideChar(CP_UTF8, 0, buffer, -1, wbuffer, 10000);
if (translateFrom.empty()) translateFrom = wbuffer;
// Response formatted as JSON: translation starts with :" and ends with "}
if (std::wcmatch results; std::regex_search(wbuffer, results, std::wregex(L":\"(.+)\"\\}"))) translation = results[1];
}
WinHttpCloseHandle(request);
}
WinHttpCloseHandle(connection);
}
}
if (auto response = ReceiveHttpRequest(request))
if (translateFrom.empty()) translation = response.value();
else if (std::wsmatch results; std::regex_search(response.value(), results, std::wregex(L":\"(.+)\"\\}"))) translation = results[1];
Escape(translation);
return translation;
}
@ -132,35 +117,20 @@ bool ProcessSentence(std::wstring& sentence, SentenceInfo sentenceInfo)
{
if (sentenceInfo["hook address"] == -1) return false;
{
static std::mutex m;
static std::vector<DWORD> requestTimes;
std::lock_guard l(m);
requestTimes.push_back(GetTickCount());
requestTimes.erase(std::remove_if(requestTimes.begin(), requestTimes.end(), [&](DWORD requestTime) { return GetTickCount() - requestTime > 60 * 1000; }), requestTimes.end());
if (!sentenceInfo["current select"] && requestTimes.size() > 30)
{
sentence += TOO_MANY_TRANS_REQUESTS;
return true;
}
}
std::wstring translation, translateFrom;
Translate(sentence, translateFrom, translateTo);
translation = Translate(sentence, translateFrom, translateTo);
for (int i = 0; i < translation.size(); ++i)
{
if (translation[i] == L'\\')
{
translation[i] = 0x200b;
if (translation[i + 1] == L'r') translation[i + 1] = 0x200b; // for some reason \r gets displayed as a newline
if (translation[i + 1] == L'n') translation[i + 1] = L'\n';
if (translation[i + 1] == L't') translation[i + 1] = L'\t';
}
}
static RateLimiter rateLimiter(30, 60 * 1000);
std::wstring translation;
if (!(rateLimiter.Request() || sentenceInfo["current select"])) translation = TOO_MANY_TRANS_REQUESTS;
else translation = Translate(sentence, Translate(sentence, L"", translateTo), translateTo);
if (translation.empty()) translation = TRANSLATION_ERROR;
sentence += L"\n" + translation;
return true;
}
TEST(
{
std::wstring test = L"こんにちは";
ProcessSentence(test, SentenceInfo{ SentenceInfo::DUMMY });
assert(test.find(L"Hello") != std::wstring::npos);
}
);

2
extensions/extension.h
View File

@ -17,6 +17,8 @@ struct SentenceInfo
for (auto info = infoArray; info->name != nullptr; ++info) if (propertyName == info->name) return info->value;
throw;
}
inline static InfoForExtension DUMMY[2] = { { "hook address", 0 } };
};
struct SKIP {};

1
extensions/extensiontester.cpp
View File

@ -8,4 +8,5 @@ int main()
*(wcsrchr(path, L'\\') + 1) = 0;
for (auto file : std::filesystem::directory_iterator(path))
if (file.path().extension() == L".dll") LoadLibraryW(file.path().c_str());
Sleep(10000);
}

128
extensions/googletranslate.cpp
View File

@ -1,6 +1,8 @@
#include "extension.h"
#include "extension.h"
#include "defs.h"
#include "text.h"
#include <winhttp.h>
#include "util.h"
#include "network.h"
#include <ctime>
#include <QInputDialog>
#include <QTimer>
@ -69,7 +71,7 @@ QStringList languages
"Zulu: zu"
};
std::wstring translateTo;
std::wstring translateTo = L"en";
BOOL WINAPI DllMain(HMODULE hModule, DWORD ul_reason_for_call, LPVOID lpReserved)
{
@ -98,19 +100,20 @@ BOOL WINAPI DllMain(HMODULE hModule, DWORD ul_reason_for_call, LPVOID lpReserved
return TRUE;
}
std::wstring GetTranslationUri(std::wstring text, unsigned TKK)
std::wstring GetTranslationUri(const std::wstring& text, unsigned TKK)
{
// If no TKK available, use this uri. Can't use too much or google will detect unauthorized access
if (!TKK) return L"/translate_a/single?client=gtx&dt=ld&dt=rm&dt=t&tl=" + translateTo + L"&q=" + text;
// Artikash 8/19/2018: reverse engineered from translate.google.com
char utf8[10000] = {};
WideCharToMultiByte(CP_UTF8, 0, text.c_str(), -1, utf8, 10000, NULL, NULL);
unsigned a = (unsigned)(_time64(NULL) / 3600), b = a; // <- the first part of TKK
for (int i = 0; utf8[i];)
std::wstring escapedText;
unsigned a = _time64(NULL) / 3600, b = a; // <- the first part of TKK
for (unsigned char ch : WideStringToString(text))
{
a += (unsigned char)utf8[i++];
wchar_t escapedChar[4] = {};
swprintf_s<4>(escapedChar, L"%%%02X", (int)ch);
escapedText += escapedChar;
a += ch;
a += a << 10;
a ^= a >> 6;
}
@ -121,104 +124,49 @@ std::wstring GetTranslationUri(std::wstring text, unsigned TKK)
a %= 1000000;
b ^= a;
text.clear();
for (int i = 0; utf8[i];)
{
wchar_t utf8char[3] = {};
swprintf_s<3>(utf8char, L"%02X", (int)(unsigned char)utf8[i++]);
text += L"%" + std::wstring(utf8char);
}
return L"/translate_a/single?client=t&dt=ld&dt=rm&dt=t&tl=" + translateTo + L"&tk=" + std::to_wstring(a) + L"." + std::to_wstring(b) + L"&q=" + text;
return L"/translate_a/single?client=t&dt=ld&dt=rm&dt=t&tl=" + translateTo + L"&tk=" + std::to_wstring(a) + L"." + std::to_wstring(b) + L"&q=" + escapedText;
}
bool ProcessSentence(std::wstring& sentence, SentenceInfo sentenceInfo)
{
if (sentenceInfo["hook address"] == -1) return false;
{
static std::mutex m;
static std::vector<DWORD> requestTimes;
std::lock_guard l(m);
requestTimes.push_back(GetTickCount());
requestTimes.erase(std::remove_if(requestTimes.begin(), requestTimes.end(), [&](DWORD requestTime) { return GetTickCount() - requestTime > 60 * 1000; }), requestTimes.end());
if (!sentenceInfo["current select"] && requestTimes.size() > 30)
{
sentence += TOO_MANY_TRANS_REQUESTS;
return true;
}
}
static std::atomic<HINTERNET> internet = NULL;
if (!internet) internet = WinHttpOpen(L"Mozilla/5.0 Textractor", WINHTTP_ACCESS_TYPE_DEFAULT_PROXY, NULL, NULL, 0);
static std::atomic<unsigned> TKK = 0;
static RateLimiter rateLimiter(30, 60 * 1000);
std::wstring translation;
if (internet)
if (!(rateLimiter.Request() || sentenceInfo["current select"])) translation = TOO_MANY_TRANS_REQUESTS;
else if (internet)
{
if (!TKK)
{
if (HINTERNET connection = WinHttpConnect(internet, L"translate.google.com", INTERNET_DEFAULT_HTTPS_PORT, 0))
{
if (HINTERNET request = WinHttpOpenRequest(connection, L"GET", L"/", NULL, NULL, NULL, WINHTTP_FLAG_SECURE))
{
if (InternetHandle connection = WinHttpConnect(internet, L"translate.google.com", INTERNET_DEFAULT_HTTPS_PORT, 0))
if (InternetHandle request = WinHttpOpenRequest(connection, L"GET", L"/", NULL, NULL, NULL, WINHTTP_FLAG_SECURE))
if (WinHttpSendRequest(request, NULL, 0, NULL, 0, 0, NULL))
{
DWORD bytesRead;
char buffer[200000] = {}; // Google Translate page is ~180kb (good god the bloat >_>)
WinHttpReceiveResponse(request, NULL);
WinHttpReadData(request, buffer, 200000, &bytesRead);
if (std::cmatch results; std::regex_search(buffer, results, std::regex("(\\d{7,})'"))) TKK = stoll(results[1]);
}
WinHttpCloseHandle(request);
}
WinHttpCloseHandle(connection);
}
}
if (auto response = ReceiveHttpRequest(request))
if (std::wsmatch results; std::regex_search(response.value(), results, std::wregex(L"(\\d{7,})'"))) TKK = stoll(results[1]);
if (HINTERNET connection = WinHttpConnect(internet, L"translate.google.com", INTERNET_DEFAULT_HTTPS_PORT, 0))
{
if (HINTERNET request = WinHttpOpenRequest(connection, L"GET", GetTranslationUri(sentence, TKK).c_str(), NULL, NULL, NULL, WINHTTP_FLAG_ESCAPE_DISABLE | WINHTTP_FLAG_SECURE))
{
if (InternetHandle connection = WinHttpConnect(internet, L"translate.google.com", INTERNET_DEFAULT_HTTPS_PORT, 0))
if (InternetHandle request = WinHttpOpenRequest(connection, L"GET", GetTranslationUri(sentence, TKK).c_str(), NULL, NULL, NULL, WINHTTP_FLAG_ESCAPE_DISABLE | WINHTTP_FLAG_SECURE))
if (WinHttpSendRequest(request, NULL, 0, NULL, 0, 0, NULL))
{
DWORD bytesRead;
char buffer[10000] = {};
WinHttpReceiveResponse(request, NULL);
WinHttpReadData(request, buffer, 10000, &bytesRead);
// Response formatted as JSON: starts with '[[["'
if (buffer[0] == '[')
{
wchar_t wbuffer[10000] = {};
MultiByteToWideChar(CP_UTF8, 0, buffer, -1, wbuffer, 10000);
std::wstring response(wbuffer);
for (std::wsmatch results; std::regex_search(response, results, std::wregex(L"\\[\"(.*?)\",[n\"]")); response = results.suffix())
translation += std::wstring(results[1]) + L" ";
for (int i = 0; i < translation.size(); ++i)
{
if (translation[i] == L'\\')
{
translation[i] = 0x200b;
if (translation[i + 1] == L'r') translation[i + 1] = 0x200b; // for some reason \r gets displayed as a newline
if (translation[i + 1] == L'n') translation[i + 1] = L'\n';
if (translation[i + 1] == L't') translation[i + 1] = L'\t';
}
}
}
else
{
translation = TRANSLATION_ERROR + (L" (TKK=" + std::to_wstring(TKK) + L")");
TKK = 0;
}
}
WinHttpCloseHandle(request);
}
WinHttpCloseHandle(connection);
}
if (auto response = ReceiveHttpRequest(request))
if (response.value()[0] == L'[')
for (std::wsmatch results; std::regex_search(response.value(), results, std::wregex(L"\\[\"(.*?)\",[n\"]")); response = results.suffix())
translation += std::wstring(results[1]) + L" ";
else std::tie(translation, TKK) = std::tuple(TRANSLATION_ERROR + (L" (TKK=" + std::to_wstring(TKK) + L")"), 0);
}
if (translation.empty()) translation = TRANSLATION_ERROR;
Escape(translation);
sentence += L"\n" + translation;
return true;
}
TEST(
{
std::wstring test = L"こんにちは";
ProcessSentence(test, SentenceInfo{ SentenceInfo::DUMMY });
assert(test.find(L"Hello") != std::wstring::npos);
}
);

39
extensions/network.h Normal file
View File

@ -0,0 +1,39 @@
#pragma once
#include "util.h"
#include <winhttp.h>
using InternetHandle = AutoHandle<Functor<WinHttpCloseHandle>>;
std::optional<std::wstring> ReceiveHttpRequest(HINTERNET request)
{
WinHttpReceiveResponse(request, NULL);
std::string data;
DWORD dwSize, dwDownloaded;
do
{
dwSize = 0;
WinHttpQueryDataAvailable(request, &dwSize);
if (!dwSize) break;
std::vector<char> buffer(dwSize);
WinHttpReadData(request, buffer.data(), dwSize, &dwDownloaded);
data += std::string(buffer.data(), dwDownloaded);
} while (dwSize > 0);
if (data.empty()) return {};
return StringToWideString(data);
}
void Escape(std::wstring& text)
{
for (int i = 0; i < text.size(); ++i)
{
if (text[i] == L'\\')
{
text[i] = 0x200b;
if (text[i + 1] == L'r') text[i + 1] = 0x200b; // for some reason \r gets displayed as a newline
if (text[i + 1] == L'n') text[i + 1] = L'\n';
if (text[i + 1] == L't') text[i + 1] = L'\t';
}
}
}

33
extensions/util.h Normal file
View File

@ -0,0 +1,33 @@
#pragma once
#include "common.h"
#include "types.h"
class RateLimiter
{
public:
RateLimiter(int requests, int delay) : requestsLeft(requests), delay(delay) {}
bool Request() { CreateTimerQueueTimer(&DUMMY, timerQueue, [](void* This, BOOLEAN) { ++((RateLimiter*)This)->requestsLeft; }, this, delay, 0, 0); return --requestsLeft > 0; }
int delay;
private:
std::atomic<int> requestsLeft;
AutoHandle<Functor<DeleteTimerQueue>> timerQueue = CreateTimerQueue();
HANDLE DUMMY;
};
inline std::wstring StringToWideString(const std::string& text)
{
std::vector<wchar_t> buffer(text.size() + 1);
MultiByteToWideChar(CP_UTF8, 0, text.c_str(), -1, buffer.data(), buffer.size());
return buffer.data();
}
inline std::string WideStringToString(const std::wstring& text)
{
std::vector<char> buffer((text.size() + 1) * 4);
WideCharToMultiByte(CP_UTF8, 0, text.c_str(), -1, buffer.data(), buffer.size(), nullptr, nullptr);
return buffer.data();
}
inline const std::wstring NEWLINE = L"\n";

2
include/defs.h
View File

@ -42,7 +42,7 @@ inline void FORMAT_MESSAGE(const char* format, Args... args)
}
#ifdef _DEBUG
#define TEST(...) inline auto TEST__RUNNER__DUMMY = std::invoke([]{ __VA_ARGS__; return 0; })
#define TEST(...) inline auto TEST__RUNNER__DUMMY = (CloseHandle(CreateThread(nullptr, 0, [](auto) { __VA_ARGS__; return 0UL; }, NULL, 0, nullptr)), 0);
#else
#define TEST(...)
#endif

9
text.cpp
View File

@ -76,8 +76,7 @@ const char* COULD_NOT_FIND = u8"Textractor: could not find text";
const char* SELECT_LANGUAGE = u8"Select Language";
const char* BING_PROMPT = u8"What language should Bing translate to?";
const char* GOOGLE_PROMPT = u8"What language should Google translate to?";
const wchar_t* TOO_MANY_TRANS_REQUESTS = LR"(
Too many translation requests: refuse to make more)";
const wchar_t* TOO_MANY_TRANS_REQUESTS = L"Too many translation requests: refuse to make more";
const wchar_t* TRANSLATION_ERROR = L"Error while translating";
const char* EXTRA_WINDOW_INFO = u8R"(Right click to change settings
Click and drag on window edges to move, or the bottom right corner to resize)";
@ -208,8 +207,7 @@ Estoy buscando un nuevo trabajo: por favor envíame un correo si estás contrata
SELECT_LANGUAGE = u8"Seleccionar lenguaje";
BING_PROMPT = u8"¿A qué idioma debe traducir Bing?";
GOOGLE_PROMPT = u8"¿A qué idioma debe traducir Google?";
TOO_MANY_TRANS_REQUESTS = LR"(
Demasiadas peticiones de traducción: no se puede hacer más)";
TOO_MANY_TRANS_REQUESTS = L"Demasiadas peticiones de traducción: no se puede hacer más";
TRANSLATION_ERROR = L"Error al traducir";
EXTRA_WINDOW_INFO = u8R"(Clic derecho para configurar
Clic y arrastra los bordes de la ventana para moverla, o en la esquina inferior derecha para cambiar el tamaño)";
@ -294,8 +292,7 @@ S/Q/V: 代码页/UTF-16/UTF-8 字符串
SELECT_LANGUAGE = u8"选择语言";
BING_PROMPT = u8"想要使用 Bing 翻译到哪种语言?";
GOOGLE_PROMPT = u8"想要使用 Google 翻译到哪种语言?";
TOO_MANY_TRANS_REQUESTS = LR"(
太多翻译请求: 拒绝生成更多)";
TOO_MANY_TRANS_REQUESTS = L"太多翻译请求: 拒绝生成更多";
TRANSLATION_ERROR = L"翻译时出错";
EXTRA_WINDOW_INFO = u8R"(右键修改设置
在窗口边缘点击并拖拽来移动，或在右下角点击并拖拽来调整大小)";